(deploy) Redeploy all services.

This reverts commit 39ce40bfeb.

ROADMAP.md

@@ -1,4 +1,4 @@
-# Roadmap 2024-2025
+# Roadmap 2025
 
 This is a roadmap with major features planned for Marginalia Search.
 
@@ -30,12 +30,6 @@ Retaining the ability to independently crawl the web is still strongly desirable
 The search engine has a bit of a problem showing spicy content mixed in with the results.  It would be desirable to have a way to filter this out.  It's likely something like a URL blacklist (e.g. [UT1](https://dsi.ut-capitole.fr/blacklists/index_en.php) )
 combined with naive bayesian filter would go a long way, or something more sophisticated...?
 
-## Web Design Overhaul
-
-The design is kinda clunky and hard to maintain, and needlessly outdated-looking.  
-
-In progress: PR [#127](https://github.com/MarginaliaSearch/MarginaliaSearch/pull/127)  -- demo available at https://test.marginalia.nu/
-
 ## Additional Language Support
 
 It would be desirable if the search engine supported more languages than English.  This is partially about
@@ -62,8 +56,31 @@ filter for any API consumer.
 
 I've talked to the stract dev and he does not think it's a good idea to mimic their optics language, which is quite ad-hoc, but instead to work together to find some new common description language for this. 
 
+## Show favicons next to search results
+
+This is expected from search engines.  Basic proof of concept sketch of fetching this data has been done, but the feature is some way from being reality. 
+
+## Specialized crawler for github
+
+One of the search engine's biggest limitations right now is that it does not index github at all.   A specialized crawler that fetches at least the readme.md would go a long way toward providing search capabilities in this domain.
+
 # Completed
 
+## Web Design Overhaul (COMPLETED 2025-01)
+
+The design is kinda clunky and hard to maintain, and needlessly outdated-looking.  
+
+PR [#127](https://github.com/MarginaliaSearch/MarginaliaSearch/pull/127)
+
+## Finalize RSS support (COMPLETED 2024-11)
+
+Marginalia has experimental RSS preview support for a few domains.  This works well and
+it should be extended to all domains.  It would also be interesting to offer search of the
+RSS data itself, or use the RSS set to feed a special live index that updates faster than the
+main dataset. 
+
+Completed with PR [#122](https://github.com/MarginaliaSearch/MarginaliaSearch/pull/122) and PR [#125](https://github.com/MarginaliaSearch/MarginaliaSearch/pull/125)
+
 ## Proper Position Index (COMPLETED 2024-09)
 
 The search engine uses a fixed width bit mask to indicate word positions.  It has the benefit
@@ -76,11 +93,3 @@ list, as is the civilized way of doing this.
 
 Completed with PR [#99](https://github.com/MarginaliaSearch/MarginaliaSearch/pull/99)
 
-## Finalize RSS support (COMPLETED 2024-11)
-
-Marginalia has experimental RSS preview support for a few domains.  This works well and
-it should be extended to all domains.  It would also be interesting to offer search of the
-RSS data itself, or use the RSS set to feed a special live index that updates faster than the
-main dataset. 
-
-Completed with PR [#122](https://github.com/MarginaliaSearch/MarginaliaSearch/pull/122) and PR [#125](https://github.com/MarginaliaSearch/MarginaliaSearch/pull/125)

build.gradle

@@ -5,7 +5,7 @@ plugins {
 
     // This is a workaround for a bug in the Jib plugin that causes it to stall randomly
     // https://github.com/GoogleContainerTools/jib/issues/3347
-    id 'com.google.cloud.tools.jib' version '3.4.3' apply(false)
+    id 'com.google.cloud.tools.jib' version '3.4.5' apply(false)
 }
 
 group 'marginalia'
@@ -43,12 +43,11 @@ subprojects.forEach {it ->
 }
 
 ext {
-    jvmVersion=23
+    jvmVersion = 24
-    dockerImageBase='container-registry.oracle.com/graalvm/jdk:23'
+    dockerImageBase='container-registry.oracle.com/graalvm/jdk:24'
     dockerImageTag='latest'
     dockerImageRegistry='marginalia'
-    jibVersion = '3.4.3'
+    jibVersion = '3.4.5'
-
 }
 
 idea {

code/common/config/java/nu/marginalia/LanguageModels.java

@@ -24,58 +24,4 @@ public class LanguageModels {
         this.fasttextLanguageModel = fasttextLanguageModel;
         this.segments = segments;
     }
-
-    public static LanguageModelsBuilder builder() {
-        return new LanguageModelsBuilder();
-    }
-
-    public static class LanguageModelsBuilder {
-        private Path termFrequencies;
-        private Path openNLPSentenceDetectionData;
-        private Path posRules;
-        private Path posDict;
-        private Path fasttextLanguageModel;
-        private Path segments;
-
-        LanguageModelsBuilder() {
-        }
-
-        public LanguageModelsBuilder termFrequencies(Path termFrequencies) {
-            this.termFrequencies = termFrequencies;
-            return this;
-        }
-
-        public LanguageModelsBuilder openNLPSentenceDetectionData(Path openNLPSentenceDetectionData) {
-            this.openNLPSentenceDetectionData = openNLPSentenceDetectionData;
-            return this;
-        }
-
-        public LanguageModelsBuilder posRules(Path posRules) {
-            this.posRules = posRules;
-            return this;
-        }
-
-        public LanguageModelsBuilder posDict(Path posDict) {
-            this.posDict = posDict;
-            return this;
-        }
-
-        public LanguageModelsBuilder fasttextLanguageModel(Path fasttextLanguageModel) {
-            this.fasttextLanguageModel = fasttextLanguageModel;
-            return this;
-        }
-
-        public LanguageModelsBuilder segments(Path segments) {
-            this.segments = segments;
-            return this;
-        }
-
-        public LanguageModels build() {
-            return new LanguageModels(this.termFrequencies, this.openNLPSentenceDetectionData, this.posRules, this.posDict, this.fasttextLanguageModel, this.segments);
-        }
-
-        public String toString() {
-            return "LanguageModels.LanguageModelsBuilder(termFrequencies=" + this.termFrequencies + ", openNLPSentenceDetectionData=" + this.openNLPSentenceDetectionData + ", posRules=" + this.posRules + ", posDict=" + this.posDict + ", fasttextLanguageModel=" + this.fasttextLanguageModel + ", segments=" + this.segments + ")";
-        }
-    }
 }

code/common/db/java/nu/marginalia/db/DbDomainQueries.java

@@ -8,18 +8,23 @@ import com.google.inject.Inject;
 import com.google.inject.Singleton;
 import com.zaxxer.hikari.HikariDataSource;
 import nu.marginalia.model.EdgeDomain;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
 
 import java.sql.SQLException;
-import java.util.NoSuchElementException;
+import java.util.*;
-import java.util.Optional;
-import java.util.OptionalInt;
 import java.util.concurrent.ExecutionException;
 
 @Singleton
 public class DbDomainQueries {
     private final HikariDataSource dataSource;
 
+    private static final Logger logger = LoggerFactory.getLogger(DbDomainQueries.class);
+
     private final Cache<EdgeDomain, Integer> domainIdCache = CacheBuilder.newBuilder().maximumSize(10_000).build();
+    private final Cache<EdgeDomain, DomainIdWithNode> domainWithNodeCache = CacheBuilder.newBuilder().maximumSize(10_000).build();
+    private final Cache<Integer, EdgeDomain> domainNameCache = CacheBuilder.newBuilder().maximumSize(10_000).build();
+    private final Cache<String, List<DomainWithNode>> siblingsCache = CacheBuilder.newBuilder().maximumSize(10_000).build();
 
     @Inject
     public DbDomainQueries(HikariDataSource dataSource)
@@ -29,16 +34,21 @@ public class DbDomainQueries {
 
 
     public Integer getDomainId(EdgeDomain domain) throws NoSuchElementException {
-        try (var connection = dataSource.getConnection()) {
+        try {
-
             return domainIdCache.get(domain, () -> {
-                try (var stmt = connection.prepareStatement("SELECT ID FROM EC_DOMAIN WHERE DOMAIN_NAME=?")) {
+                try (var connection = dataSource.getConnection();
+                     var stmt = connection.prepareStatement("SELECT ID FROM EC_DOMAIN WHERE DOMAIN_NAME=?")) {
+
                     stmt.setString(1, domain.toString());
                     var rsp = stmt.executeQuery();
                     if (rsp.next()) {
                         return rsp.getInt(1);
                     }
                 }
+                catch (SQLException ex) {
+                    throw new RuntimeException(ex);
+                }
+
                 throw new NoSuchElementException();
             });
         }
@@ -48,8 +58,33 @@ public class DbDomainQueries {
         catch (ExecutionException ex) {
             throw new RuntimeException(ex.getCause());
         }
-        catch (SQLException ex) {
+    }
-            throw new RuntimeException(ex);
+
+
+    public DomainIdWithNode getDomainIdWithNode(EdgeDomain domain) throws NoSuchElementException {
+        try {
+            return domainWithNodeCache.get(domain, () -> {
+                try (var connection = dataSource.getConnection();
+                     var stmt = connection.prepareStatement("SELECT ID, NODE_AFFINITY FROM EC_DOMAIN WHERE DOMAIN_NAME=?")) {
+
+                    stmt.setString(1, domain.toString());
+                    var rsp = stmt.executeQuery();
+                    if (rsp.next()) {
+                        return new DomainIdWithNode(rsp.getInt(1), rsp.getInt(2));
+                    }
+                }
+                catch (SQLException ex) {
+                    throw new RuntimeException(ex);
+                }
+
+                throw new NoSuchElementException();
+            });
+        }
+        catch (UncheckedExecutionException ex) {
+            throw new NoSuchElementException();
+        }
+        catch (ExecutionException ex) {
+            throw new RuntimeException(ex.getCause());
         }
     }
 
@@ -83,22 +118,62 @@ public class DbDomainQueries {
     }
 
     public Optional<EdgeDomain> getDomain(int id) {
-        try (var connection = dataSource.getConnection()) {
 
+        EdgeDomain existing = domainNameCache.getIfPresent(id);
+        if (existing != null) {
+            return Optional.of(existing);
+        }
+
+        try (var connection = dataSource.getConnection()) {
             try (var stmt = connection.prepareStatement("SELECT DOMAIN_NAME FROM EC_DOMAIN WHERE ID=?")) {
                 stmt.setInt(1, id);
                 var rsp = stmt.executeQuery();
                 if (rsp.next()) {
-                    return Optional.of(new EdgeDomain(rsp.getString(1)));
+                    var val = new EdgeDomain(rsp.getString(1));
+                    domainNameCache.put(id, val);
+                    return Optional.of(val);
                 }
                 return Optional.empty();
             }
         }
-        catch (UncheckedExecutionException ex) {
-            throw new RuntimeException(ex.getCause());
-        }
         catch (SQLException ex) {
             throw new RuntimeException(ex);
         }
     }
+
+    public List<DomainWithNode> otherSubdomains(EdgeDomain domain, int cnt) throws ExecutionException {
+        String topDomain = domain.topDomain;
+
+        return siblingsCache.get(topDomain, () -> {
+            List<DomainWithNode> ret = new ArrayList<>();
+
+            try (var conn = dataSource.getConnection();
+                 var stmt = conn.prepareStatement("SELECT DOMAIN_NAME, NODE_AFFINITY FROM EC_DOMAIN WHERE DOMAIN_TOP = ? LIMIT ?")) {
+                stmt.setString(1, topDomain);
+                stmt.setInt(2, cnt);
+
+                var rs = stmt.executeQuery();
+                while (rs.next()) {
+                    var sibling = new EdgeDomain(rs.getString(1));
+
+                    if (sibling.equals(domain))
+                        continue;
+
+                    ret.add(new DomainWithNode(sibling, rs.getInt(2)));
+                }
+            } catch (SQLException e) {
+                logger.error("Failed to get domain neighbors");
+            }
+            return ret;
+        });
+
+    }
+
+    public record DomainWithNode (EdgeDomain domain, int nodeAffinity) {
+        public boolean isIndexed() {
+            return nodeAffinity > 0;
+        }
+    }
+
+    public record DomainIdWithNode (int domainId, int nodeAffinity) { }
 }

code/common/db/java/nu/marginalia/db/DbDomainStatsExportMultitool.java

@@ -1,118 +0,0 @@
-package nu.marginalia.db;
-
-import com.zaxxer.hikari.HikariDataSource;
-
-import java.sql.Connection;
-import java.sql.PreparedStatement;
-import java.sql.SQLException;
-import java.util.ArrayList;
-import java.util.List;
-import java.util.OptionalInt;
-
-/** Class used in exporting data.  This is intended to be used for a brief time
- * and then discarded, not kept around as a service.
- */
-public class DbDomainStatsExportMultitool implements AutoCloseable {
-    private final Connection connection;
-    private final int nodeId;
-    private final PreparedStatement knownUrlsQuery;
-    private final PreparedStatement visitedUrlsQuery;
-    private final PreparedStatement goodUrlsQuery;
-    private final PreparedStatement domainNameToId;
-
-    private final PreparedStatement allDomainsQuery;
-    private final PreparedStatement crawlQueueDomains;
-    private final PreparedStatement indexedDomainsQuery;
-
-    public DbDomainStatsExportMultitool(HikariDataSource dataSource, int nodeId) throws SQLException {
-        this.connection = dataSource.getConnection();
-        this.nodeId = nodeId;
-
-        knownUrlsQuery = connection.prepareStatement("""
-                SELECT KNOWN_URLS
-                FROM EC_DOMAIN INNER JOIN DOMAIN_METADATA
-                    ON EC_DOMAIN.ID=DOMAIN_METADATA.ID
-                WHERE DOMAIN_NAME=?
-                """);
-        visitedUrlsQuery = connection.prepareStatement("""
-                SELECT VISITED_URLS
-                FROM EC_DOMAIN INNER JOIN DOMAIN_METADATA
-                    ON EC_DOMAIN.ID=DOMAIN_METADATA.ID
-                WHERE DOMAIN_NAME=?
-                """);
-        goodUrlsQuery = connection.prepareStatement("""
-                SELECT GOOD_URLS
-                FROM EC_DOMAIN INNER JOIN DOMAIN_METADATA
-                    ON EC_DOMAIN.ID=DOMAIN_METADATA.ID
-                WHERE DOMAIN_NAME=?
-                """);
-        domainNameToId = connection.prepareStatement("""
-                SELECT ID
-                FROM EC_DOMAIN
-                WHERE DOMAIN_NAME=?
-                """);
-        allDomainsQuery = connection.prepareStatement("""
-                SELECT DOMAIN_NAME
-                FROM EC_DOMAIN
-                """);
-        crawlQueueDomains = connection.prepareStatement("""
-                SELECT DOMAIN_NAME
-                FROM CRAWL_QUEUE
-                """);
-        indexedDomainsQuery = connection.prepareStatement("""
-                SELECT DOMAIN_NAME
-                FROM EC_DOMAIN
-                WHERE INDEXED > 0
-                """);
-    }
-
-    public OptionalInt getVisitedUrls(String domainName) throws SQLException {
-        return executeNameToIntQuery(domainName, visitedUrlsQuery);
-    }
-
-    public OptionalInt getDomainId(String domainName) throws SQLException {
-        return executeNameToIntQuery(domainName, domainNameToId);
-    }
-
-    public List<String> getCrawlQueueDomains() throws SQLException {
-        return executeListQuery(crawlQueueDomains, 100);
-    }
-    public List<String> getAllIndexedDomains() throws SQLException {
-        return executeListQuery(indexedDomainsQuery, 100_000);
-    }
-
-    private OptionalInt executeNameToIntQuery(String domainName, PreparedStatement statement)
-            throws SQLException {
-        statement.setString(1, domainName);
-        var rs = statement.executeQuery();
-
-        if (rs.next()) {
-            return OptionalInt.of(rs.getInt(1));
-        }
-
-        return OptionalInt.empty();
-    }
-
-    private List<String> executeListQuery(PreparedStatement statement, int sizeHint) throws SQLException {
-        List<String> ret = new ArrayList<>(sizeHint);
-
-        var rs = statement.executeQuery();
-
-        while (rs.next()) {
-            ret.add(rs.getString(1));
-        }
-
-        return ret;
-    }
-
-    @Override
-    public void close() throws SQLException {
-        knownUrlsQuery.close();
-        goodUrlsQuery.close();
-        visitedUrlsQuery.close();
-        allDomainsQuery.close();
-        crawlQueueDomains.close();
-        domainNameToId.close();
-        connection.close();
-    }
-}

code/common/model/java/nu/marginalia/model/DocumentFormat.java

@@ -0,0 +1,24 @@
+package nu.marginalia.model;
+
+public enum DocumentFormat {
+    PLAIN(0, 1, "text"),
+    PDF(0, 1, "pdf"),
+    UNKNOWN(0, 1, "???"),
+    HTML123(0, 1, "html"),
+    HTML4(-0.1, 1.05, "html"),
+    XHTML(-0.1, 1.05, "html"),
+    HTML5(0.5, 1.1, "html");
+
+    /** Used to tune quality score */
+    public final double offset;
+    /** Used to tune quality score */
+    public final double scale;
+    public final String shortFormat;
+
+    DocumentFormat(double offset, double scale, String shortFormat) {
+        this.offset = offset;
+        this.scale = scale;
+        this.shortFormat = shortFormat;
+    }
+
+}

code/common/model/java/nu/marginalia/model/EdgeDomain.java

@@ -14,7 +14,7 @@ public class EdgeDomain implements Serializable {
     @Nonnull
     public final String topDomain;
 
-    public EdgeDomain(String host) {
+    public EdgeDomain(@Nonnull String host) {
         Objects.requireNonNull(host, "domain name must not be null");
 
         host = host.toLowerCase();
@@ -61,6 +61,10 @@ public class EdgeDomain implements Serializable {
         this.topDomain = topDomain;
     }
 
+    public static String getTopDomain(String host) {
+        return new EdgeDomain(host).topDomain;
+    }
+
     private boolean looksLikeGovTld(String host) {
         if (host.length() < 8)
             return false;
@@ -108,32 +112,6 @@ public class EdgeDomain implements Serializable {
         return topDomain;
     }
 
-    public String getDomainKey() {
-        int cutPoint = topDomain.indexOf('.');
-        if (cutPoint < 0) {
-            return topDomain;
-        }
-        return topDomain.substring(0, cutPoint).toLowerCase();
-    }
-
-    public String getLongDomainKey() {
-        StringBuilder ret = new StringBuilder();
-
-        int cutPoint = topDomain.indexOf('.');
-        if (cutPoint < 0) {
-            ret.append(topDomain);
-        } else {
-            ret.append(topDomain, 0, cutPoint);
-        }
-
-        if (!subDomain.isEmpty() && !"www".equals(subDomain)) {
-            ret.append(":");
-            ret.append(subDomain);
-        }
-
-        return ret.toString().toLowerCase();
-    }
-
     /** If possible, try to provide an alias domain,
      * i.e. a domain name that is very likely to link to this one
      * */

code/common/model/java/nu/marginalia/model/EdgeUrl.java

@@ -1,16 +1,14 @@
 package nu.marginalia.model;
 
 import nu.marginalia.util.QueryParams;
+import org.apache.commons.lang3.StringUtils;
 
 import javax.annotation.Nullable;
 import java.io.Serializable;
-import java.net.MalformedURLException;
+import java.net.*;
-import java.net.URI;
+import java.nio.charset.StandardCharsets;
-import java.net.URISyntaxException;
-import java.net.URL;
 import java.util.Objects;
 import java.util.Optional;
-import java.util.regex.Pattern;
 
 public class EdgeUrl implements Serializable {
     public final String proto;
@@ -33,7 +31,7 @@ public class EdgeUrl implements Serializable {
 
     private static URI parseURI(String url) throws URISyntaxException {
         try {
-            return new URI(urlencodeFixer(url));
+            return EdgeUriFactory.parseURILenient(url);
         } catch (URISyntaxException ex) {
             throw new URISyntaxException("Failed to parse URI '" + url + "'", ex.getMessage());
         }
@@ -51,58 +49,6 @@ public class EdgeUrl implements Serializable {
         }
     }
 
-    private static Pattern badCharPattern = Pattern.compile("[ \t\n\"<>\\[\\]()',|]");
-
-    /* Java's URI parser is a bit too strict in throwing exceptions when there's an error.
-
-       Here on the Internet, standards are like the picture on the box of the frozen pizza,
-       and what you get is more like what's on the inside, we try to patch things instead,
-       just give it a best-effort attempt att cleaning out broken or unnecessary constructions
-       like bad or missing URLEncoding
-     */
-    public static String urlencodeFixer(String url) throws URISyntaxException {
-        var s = new StringBuilder();
-        String goodChars = "&.?:/-;+$#";
-        String hexChars = "0123456789abcdefABCDEF";
-
-        int pathIdx = findPathIdx(url);
-        if (pathIdx < 0) { // url looks like http://marginalia.nu
-            return url + "/";
-        }
-        s.append(url, 0, pathIdx);
-
-        // We don't want the fragment, and multiple fragments breaks the Java URIParser for some reason
-        int end = url.indexOf("#");
-        if (end < 0) end = url.length();
-
-        for (int i = pathIdx; i < end; i++) {
-            int c = url.charAt(i);
-
-            if (goodChars.indexOf(c) >= 0 || (c >= 'A' && c <= 'Z') || (c >= 'a' && c <= 'z') || (c >= '0' && c <= '9')) {
-                s.appendCodePoint(c);
-            } else if (c == '%' && i + 2 < end) {
-                int cn = url.charAt(i + 1);
-                int cnn = url.charAt(i + 2);
-                if (hexChars.indexOf(cn) >= 0 && hexChars.indexOf(cnn) >= 0) {
-                    s.appendCodePoint(c);
-                } else {
-                    s.append("%25");
-                }
-            } else {
-                s.append(String.format("%%%02X", c));
-            }
-        }
-
-        return s.toString();
-    }
-
-    private static int findPathIdx(String url) throws URISyntaxException {
-        int colonIdx = url.indexOf(':');
-        if (colonIdx < 0 || colonIdx + 2 >= url.length()) {
-            throw new URISyntaxException(url, "Lacking protocol");
-        }
-        return url.indexOf('/', colonIdx + 2);
-    }
 
     public EdgeUrl(URI URI) {
         try {
@@ -166,11 +112,32 @@ public class EdgeUrl implements Serializable {
             sb.append(port);
         }
 
+        EdgeUriFactory.urlencodePath(sb, path);
+
+        if (param != null) {
+            EdgeUriFactory.urlencodeQuery(sb, param);
+        }
+
+        return sb.toString();
+    }
+
+
+    public String toDisplayString() {
+        StringBuilder sb = new StringBuilder(256);
+
+        sb.append(proto);
+        sb.append("://");
+        sb.append(domain);
+
+        if (port != null) {
+            sb.append(':');
+            sb.append(port);
+        }
+
         sb.append(path);
 
         if (param != null) {
-            sb.append('?');
+            sb.append('?').append(param);
-            sb.append(param);
         }
 
         return sb.toString();
@@ -247,3 +214,244 @@ public class EdgeUrl implements Serializable {
     }
 
 }
+
+class EdgeUriFactory {
+    public static URI parseURILenient(String url) throws URISyntaxException {
+
+        if (shouldOmitUrlencodeRepair(url)) {
+            try {
+                return new URI(url);
+            }
+            catch (URISyntaxException ex) {
+                // ignore and run the lenient parser
+            }
+        }
+
+        var s = new StringBuilder(url.length()+8);
+
+        int pathIdx = findPathIdx(url);
+        if (pathIdx < 0) { // url looks like http://marginalia.nu
+            return new URI(url + "/");
+        }
+        s.append(url, 0, pathIdx);
+
+        // We don't want the fragment, and multiple fragments breaks the Java URIParser for some reason
+        int end = url.indexOf("#");
+        if (end < 0) end = url.length();
+
+        int queryIdx = url.indexOf('?');
+        if (queryIdx < 0) queryIdx = end;
+
+        urlencodePath(s, url.substring(pathIdx, queryIdx));
+        if (queryIdx < end) {
+            urlencodeQuery(s, url.substring(queryIdx + 1, end));
+        }
+        return new URI(s.toString());
+    }
+
+    /** Break apart the path element of an URI into its components, and then
+     * urlencode any component that needs it, and recombine it into a single
+     * path element again.
+     */
+    public static void urlencodePath(StringBuilder sb, String path) {
+        if (path == null || path.isEmpty()) {
+            return;
+        }
+
+        String[] pathParts = StringUtils.split(path, '/');
+        if (pathParts.length == 0) {
+            sb.append('/');
+            return;
+        }
+
+        boolean shouldUrlEncode = false;
+        for (String pathPart : pathParts) {
+            if (pathPart.isEmpty()) continue;
+
+            if (needsUrlEncode(pathPart)) {
+                shouldUrlEncode = true;
+                break;
+            }
+        }
+
+        for (String pathPart : pathParts) {
+            if (pathPart.isEmpty()) continue;
+
+            if (shouldUrlEncode) {
+                sb.append('/');
+                sb.append(URLEncoder.encode(pathPart, StandardCharsets.UTF_8).replace("+", "%20"));
+            } else {
+                sb.append('/');
+                sb.append(pathPart);
+            }
+        }
+
+        if (path.endsWith("/")) {
+            sb.append('/');
+        }
+
+    }
+
+    /** Break apart the query element of a URI into its components, and then
+     * urlencode any component that needs it, and recombine it into a single
+     * query element again.
+     */
+    public static void urlencodeQuery(StringBuilder sb, String param) {
+        if (param == null || param.isEmpty()) {
+            return;
+        }
+
+        String[] queryParts = StringUtils.split(param, '&');
+
+        boolean shouldUrlEncode = false;
+        for (String queryPart : queryParts) {
+            if (queryPart.isEmpty()) continue;
+
+            if (needsUrlEncode(queryPart)) {
+                shouldUrlEncode = true;
+                break;
+            }
+        }
+
+        boolean first = true;
+        for (String queryPart : queryParts) {
+            if (queryPart.isEmpty()) continue;
+
+            if (first) {
+                sb.append('?');
+                first = false;
+            } else {
+                sb.append('&');
+            }
+
+            if (shouldUrlEncode) {
+                int idx = queryPart.indexOf('=');
+                if (idx < 0) {
+                    sb.append(URLEncoder.encode(queryPart, StandardCharsets.UTF_8));
+                } else {
+                    sb.append(URLEncoder.encode(queryPart.substring(0, idx), StandardCharsets.UTF_8));
+                    sb.append('=');
+                    sb.append(URLEncoder.encode(queryPart.substring(idx + 1), StandardCharsets.UTF_8));
+                }
+            } else {
+                sb.append(queryPart);
+            }
+        }
+    }
+
+    /** Test if the url element needs URL encoding.
+     * <p></p>
+     * Note we may have been given an already encoded path element,
+     * so we include % and + in the list of good characters
+     */
+    static boolean needsUrlEncode(String urlElement) {
+        for (int i = 0; i < urlElement.length(); i++) {
+            char c = urlElement.charAt(i);
+
+            if (isUrlSafe(c)) continue;
+            if ("+".indexOf(c) >= 0) continue;
+            if (c == '%' && i + 2 < urlElement.length()) {
+                char c1 = urlElement.charAt(i + 1);
+                char c2 = urlElement.charAt(i + 2);
+                if (isHexDigit(c1) && isHexDigit(c2)) {
+                    i += 2;
+                    continue;
+                }
+            }
+
+            return true;
+        }
+
+        return false;
+    }
+
+
+    static boolean isUrlSafe(int c) {
+        if (c >= 'a' && c <= 'z') return true;
+        if (c >= 'A' && c <= 'Z') return true;
+        if (c >= '0' && c <= '9') return true;
+        if (c == '-' || c == '_' || c == '.' || c == '~') return true;
+
+        return false;
+    }
+
+    /** Test if the URL is a valid URL that does not need to be
+     * urlencoded.
+     * <p></p>
+     * This is a very simple heuristic test that does not guarantee
+     * that the URL is valid, but it will identify cases where we
+     * are fairly certain that the URL does not need encoding,
+     * so we can skip a bunch of allocations and string operations
+     * that would otherwise be needed to fix the URL.
+     */
+    static boolean shouldOmitUrlencodeRepair(String url) {
+        int idx = 0;
+        final int len = url.length();
+
+        // Validate the scheme
+        while (idx < len - 2) {
+            char c = url.charAt(idx++);
+            if (c == ':') break;
+            if (!isAsciiAlphabetic(c)) return false;
+        }
+        if (url.charAt(idx++) != '/') return false;
+        if (url.charAt(idx++) != '/') return false;
+
+        // Validate the authority
+        while (idx < len) {
+            char c = url.charAt(idx++);
+            if (c == '/') break;
+            if (c == ':') continue;
+            if (c == '@') continue;
+            if (!isUrlSafe(c)) return false;
+        }
+
+        // Validate the path
+        if (idx >= len) return true;
+
+        while (idx < len) {
+            char c = url.charAt(idx++);
+            if (c == '?') break;
+            if (c == '/') continue;
+            if (c == '#') return true;
+            if (!isUrlSafe(c)) return false;
+        }
+
+        if (idx >= len) return true;
+
+        // Validate the query
+        while (idx < len) {
+            char c = url.charAt(idx++);
+            if (c == '&') continue;
+            if (c == '=') continue;
+            if (c == '#') return true;
+            if (!isUrlSafe(c)) return false;
+        }
+
+        return true;
+    }
+
+
+    private static boolean isAsciiAlphabetic(int c) {
+        return (c >= 'a' && c <= 'f') || (c >= 'A' && c <= 'F');
+    }
+
+    private static boolean isHexDigit(int c) {
+        return (c >= '0' && c <= '9') || (c >= 'a' && c <= 'f') || (c >= 'A' && c <= 'F');
+    }
+
+    /** Find the index of the path element in a URL.
+     * <p></p>
+     * The path element starts after the scheme and authority part of the URL,
+     * which is everything up to and including the first slash after the colon.
+     */
+    private static int findPathIdx(String url) throws URISyntaxException {
+        int colonIdx = url.indexOf(':');
+        if (colonIdx < 0 || colonIdx + 3 >= url.length()) {
+            throw new URISyntaxException(url, "Lacking scheme");
+        }
+        return url.indexOf('/', colonIdx + 3);
+    }
+
+
+}

code/common/model/java/nu/marginalia/model/crawl/HtmlFeature.java

@@ -28,6 +28,8 @@ public enum HtmlFeature {
 
     GA_SPAM("special:gaspam"),
 
+    PDF("format:pdf"),
+
     /** For fingerprinting and ranking */
     OPENGRAPH("special:opengraph"),
     OPENGRAPH_IMAGE("special:opengraph:image"),

code/common/model/java/nu/marginalia/model/html/HtmlStandard.java

@@ -1,22 +0,0 @@
-package nu.marginalia.model.html;
-
-// This class really doesn't belong anywhere, but will squat here for now
-public enum HtmlStandard {
-    PLAIN(0, 1),
-    UNKNOWN(0, 1),
-    HTML123(0, 1),
-    HTML4(-0.1, 1.05),
-    XHTML(-0.1, 1.05),
-    HTML5(0.5, 1.1);
-
-    /** Used to tune quality score */
-    public final double offset;
-    /** Used to tune quality score */
-    public final double scale;
-
-    HtmlStandard(double offset, double scale) {
-        this.offset = offset;
-        this.scale = scale;
-    }
-
-}

code/common/model/java/nu/marginalia/model/idx/DocumentFlags.java

@@ -9,7 +9,7 @@ public enum DocumentFlags {
     GeneratorForum,
     GeneratorWiki,
     Sideloaded,
-    Unused7,
+    PdfFile,
     Unused8,
     ;
 

code/common/model/java/nu/marginalia/util/QueryParams.java

@@ -83,6 +83,11 @@ public class QueryParams {
         if (path.endsWith("StoryView.py")) { // folklore.org is neat
             return param.startsWith("project=") || param.startsWith("story=");
         }
+
+        // www.perseus.tufts.edu:
+        if (param.startsWith("collection=")) return true;
+        if (param.startsWith("doc=")) return true;
+
         return false;
     }
 }

code/common/model/test/nu/marginalia/model/EdgeDomainTest.java

@@ -8,14 +8,6 @@ import static org.junit.jupiter.api.Assertions.assertEquals;
 
 class EdgeDomainTest {
 
-    @Test
-    public void testSkepdic() throws URISyntaxException {
-        var domain = new EdgeUrl("http://www.skepdic.com/astrology.html");
-        assertEquals("skepdic", domain.getDomain().getDomainKey());
-        var domain2 = new EdgeUrl("http://skepdic.com/astrology.html");
-        assertEquals("skepdic", domain2.getDomain().getDomainKey());
-    }
-
     @Test
     public void testHkDomain() throws URISyntaxException {
         var domain = new EdgeUrl("http://l7072i3.l7c.net");

code/common/model/test/nu/marginalia/model/EdgeUrlTest.java

@@ -1,6 +1,6 @@
 package nu.marginalia.model;
 
-import nu.marginalia.model.EdgeUrl;
+import org.junit.jupiter.api.Assertions;
 import org.junit.jupiter.api.Test;
 
 import java.net.URISyntaxException;
@@ -21,25 +21,70 @@ class EdgeUrlTest {
                 new EdgeUrl("https://memex.marginalia.nu/#here")
         );
     }
+
     @Test
-    public void testParam() throws URISyntaxException {
+    void testUriFromString() throws URISyntaxException {
-        System.out.println(new EdgeUrl("https://memex.marginalia.nu/index.php?id=1").toString());
+        // We test these URLs several times as we perform URLEncode-fixing both when parsing the URL and when
-        System.out.println(new EdgeUrl("https://memex.marginalia.nu/showthread.php?id=1&count=5&tracking=123").toString());
+        // converting it back to a string, we want to ensure there is no changes along the way.
-    }
+
-    @Test
+        Assertions.assertEquals("/", EdgeUriFactory.parseURILenient("https://www.example.com/").getPath());
-    void urlencodeFixer() throws URISyntaxException {
+        Assertions.assertEquals("https://www.example.com/", EdgeUriFactory.parseURILenient("https://www.example.com/").toString());
-        System.out.println(EdgeUrl.urlencodeFixer("https://www.example.com/#heredoc"));
+        Assertions.assertEquals("https://www.example.com/", new EdgeUrl("https://www.example.com/").toString());
-        System.out.println(EdgeUrl.urlencodeFixer("https://www.example.com/%-sign"));
+
-        System.out.println(EdgeUrl.urlencodeFixer("https://www.example.com/%22-sign"));
+        Assertions.assertEquals("/", EdgeUriFactory.parseURILenient("https://www.example.com/#heredoc").getPath());
-        System.out.println(EdgeUrl.urlencodeFixer("https://www.example.com/\n \"huh\""));
+        Assertions.assertEquals("https://www.example.com/", EdgeUriFactory.parseURILenient("https://www.example.com/#heredoc").toString());
+        Assertions.assertEquals("https://www.example.com/", new EdgeUrl("https://www.example.com/#heredoc").toString());
+
+        Assertions.assertEquals("/trailingslash/", EdgeUriFactory.parseURILenient("https://www.example.com/trailingslash/").getPath());
+        Assertions.assertEquals("https://www.example.com/trailingslash/", EdgeUriFactory.parseURILenient("https://www.example.com/trailingslash/").toString());
+        Assertions.assertEquals("https://www.example.com/trailingslash/", new EdgeUrl("https://www.example.com/trailingslash/").toString());
+
+        Assertions.assertEquals("/%-sign", EdgeUriFactory.parseURILenient("https://www.example.com/%-sign").getPath());
+        Assertions.assertEquals("https://www.example.com/%25-sign", EdgeUriFactory.parseURILenient("https://www.example.com/%-sign").toString());
+        Assertions.assertEquals("https://www.example.com/%25-sign", new EdgeUrl("https://www.example.com/%-sign").toString());
+
+        Assertions.assertEquals("/%-sign/\"-sign", EdgeUriFactory.parseURILenient("https://www.example.com//%-sign/\"-sign").getPath());
+        Assertions.assertEquals("https://www.example.com/%25-sign/%22-sign", EdgeUriFactory.parseURILenient("https://www.example.com//%-sign/\"-sign").toString());
+        Assertions.assertEquals("https://www.example.com/%25-sign/%22-sign", new EdgeUrl("https://www.example.com//%-sign/\"-sign").toString());
+
+        Assertions.assertEquals("/\"-sign", EdgeUriFactory.parseURILenient("https://www.example.com/%22-sign").getPath());
+        Assertions.assertEquals("https://www.example.com/%22-sign", EdgeUriFactory.parseURILenient("https://www.example.com/%22-sign").toString());
+        Assertions.assertEquals("https://www.example.com/%22-sign", new EdgeUrl("https://www.example.com/%22-sign").toString());
+
+        Assertions.assertEquals("/\n \"huh\"", EdgeUriFactory.parseURILenient("https://www.example.com/\n \"huh\"").getPath());
+        Assertions.assertEquals("https://www.example.com/%0A%20%22huh%22", EdgeUriFactory.parseURILenient("https://www.example.com/\n \"huh\"").toString());
+        Assertions.assertEquals("https://www.example.com/%0A%20%22huh%22", new EdgeUrl("https://www.example.com/\n \"huh\"").toString());
+
+        Assertions.assertEquals("/wiki/Sámi", EdgeUriFactory.parseURILenient("https://en.wikipedia.org/wiki/Sámi").getPath());
+        Assertions.assertEquals("https://en.wikipedia.org/wiki/S%C3%A1mi", EdgeUriFactory.parseURILenient("https://en.wikipedia.org/wiki/Sámi").toString());
+        Assertions.assertEquals("https://en.wikipedia.org/wiki/S%C3%A1mi", new EdgeUrl("https://en.wikipedia.org/wiki/Sámi").toString());
+
+        Assertions.assertEquals("https://www.prijatelji-zivotinja.hr/index.en.php?id=2301k", new EdgeUrl("https://www.prijatelji-zivotinja.hr/index.en.php?id=2301k").toString());
     }
 
     @Test
     void testParms() throws URISyntaxException {
-        System.out.println(new EdgeUrl("https://search.marginalia.nu/?id=123"));
+        Assertions.assertEquals("id=123", new EdgeUrl("https://search.marginalia.nu/?id=123").param);
-        System.out.println(new EdgeUrl("https://search.marginalia.nu/?t=123"));
+        Assertions.assertEquals("https://search.marginalia.nu/?id=123", new EdgeUrl("https://search.marginalia.nu/?id=123").toString());
-        System.out.println(new EdgeUrl("https://search.marginalia.nu/?v=123"));
+
-        System.out.println(new EdgeUrl("https://search.marginalia.nu/?m=123"));
+        Assertions.assertEquals("t=123", new EdgeUrl("https://search.marginalia.nu/?t=123").param);
-        System.out.println(new EdgeUrl("https://search.marginalia.nu/?follow=123"));
+        Assertions.assertEquals("https://search.marginalia.nu/?t=123", new EdgeUrl("https://search.marginalia.nu/?t=123").toString());
+
+        Assertions.assertEquals("v=123", new EdgeUrl("https://search.marginalia.nu/?v=123").param);
+        Assertions.assertEquals("https://search.marginalia.nu/?v=123", new EdgeUrl("https://search.marginalia.nu/?v=123").toString());
+
+        Assertions.assertEquals("id=1", new EdgeUrl("https://memex.marginalia.nu/showthread.php?id=1&count=5&tracking=123").param);
+        Assertions.assertEquals("https://memex.marginalia.nu/showthread.php?id=1",
+                new EdgeUrl("https://memex.marginalia.nu/showthread.php?id=1&count=5&tracking=123").toString());
+
+
+        Assertions.assertEquals("id=1&t=5", new EdgeUrl("https://memex.marginalia.nu/shöwthrëad.php?id=1&t=5&tracking=123").param);
+        Assertions.assertEquals("https://memex.marginalia.nu/sh%C3%B6wthr%C3%ABad.php?id=1&t=5", new EdgeUrl("https://memex.marginalia.nu/shöwthrëad.php?id=1&t=5&tracking=123").toString());
+
+        Assertions.assertEquals("id=1&t=5", new EdgeUrl("https://memex.marginalia.nu/shöwthrëad.php?trëaking=123&id=1&t=5&").param);
+        Assertions.assertEquals("https://memex.marginalia.nu/sh%C3%B6wthr%C3%ABad.php?id=1&t=5", new EdgeUrl("https://memex.marginalia.nu/shöwthrëad.php?trëaking=123&id=1&t=5&").toString());
+
+        Assertions.assertNull(new EdgeUrl("https://search.marginalia.nu/?m=123").param);
+        Assertions.assertNull(new EdgeUrl("https://search.marginalia.nu/?follow=123").param);
     }
 }

code/common/service/java/nu/marginalia/process/control/ProcessAdHocTaskHeartbeatImpl.java

@@ -59,16 +59,13 @@ public class ProcessAdHocTaskHeartbeatImpl implements AutoCloseable, ProcessAdHo
      */
     @Override
     public void progress(String step, int stepProgress, int stepCount) {
+        int lastProgress = this.progress;
         this.step = step;
-
-
-        // off by one since we calculate the progress based on the number of steps,
-        // and Enum.ordinal() is zero-based (so the 5th step in a 5 step task is 4, not 5; resulting in the
-        // final progress being 80% and not 100%)
-
         this.progress = (int) Math.round(100. * stepProgress / (double) stepCount);
 
-        logger.info("ProcessTask {} progress: {}%", taskBase, progress);
+        if (this.progress / 10 != lastProgress / 10) {
+            logger.info("ProcessTask {} progress: {}%", taskBase, progress);
+        }
     }
 
     /** Wrap a collection to provide heartbeat progress updates as it's iterated through */

code/common/service/java/nu/marginalia/process/log/WorkLog.java

@@ -10,7 +10,9 @@ import java.nio.charset.StandardCharsets;
 import java.nio.file.Files;
 import java.nio.file.Path;
 import java.time.LocalDateTime;
-import java.util.*;
+import java.util.HashSet;
+import java.util.Optional;
+import java.util.Set;
 import java.util.function.Function;
 
 /** WorkLog is a journal of work done by a process,
@@ -61,6 +63,12 @@ public class WorkLog implements AutoCloseable, Closeable {
         return new WorkLoadIterable<>(logFile, mapper);
     }
 
+    public static int countEntries(Path crawlerLog) throws IOException{
+        try (var linesStream = Files.lines(crawlerLog)) {
+            return (int) linesStream.filter(WorkLogEntry::isJobId).count();
+        }
+    }
+
     // Use synchro over concurrent set to avoid competing writes
     // - correct is better than fast here, it's sketchy enough to use
     // a PrintWriter

code/common/service/java/nu/marginalia/service/control/ServiceAdHocTaskHeartbeatImpl.java

@@ -57,16 +57,13 @@ public class ServiceAdHocTaskHeartbeatImpl implements AutoCloseable, ServiceAdHo
      */
     @Override
     public void progress(String step, int stepProgress, int stepCount) {
+        int lastProgress = this.progress;
         this.step = step;
-
-
-        // off by one since we calculate the progress based on the number of steps,
-        // and Enum.ordinal() is zero-based (so the 5th step in a 5 step task is 4, not 5; resulting in the
-        // final progress being 80% and not 100%)
-
         this.progress = (int) Math.round(100. * stepProgress / (double) stepCount);
 
-        logger.info("ServiceTask {} progress: {}%", taskBase, progress);
+        if (this.progress / 10 != lastProgress / 10) {
+            logger.info("ProcessTask {} progress: {}%", taskBase, progress);
+        }
     }
 
     public void shutDown() {

code/common/service/java/nu/marginalia/service/module/DatabaseModule.java

@@ -89,7 +89,7 @@ public class DatabaseModule extends AbstractModule {
             config.addDataSourceProperty("prepStmtCacheSize", "250");
             config.addDataSourceProperty("prepStmtCacheSqlLimit", "2048");
 
-            config.setMaximumPoolSize(5);
+            config.setMaximumPoolSize(Integer.getInteger("db.poolSize", 5));
             config.setMinimumIdle(2);
 
             config.setMaxLifetime(Duration.ofMinutes(9).toMillis());

code/common/service/java/nu/marginalia/service/module/ServiceConfigurationModule.java

@@ -6,6 +6,7 @@ import nu.marginalia.service.ServiceId;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 
+import java.io.IOException;
 import java.net.InetAddress;
 import java.net.NetworkInterface;
 import java.util.Enumeration;
@@ -115,11 +116,12 @@ public class ServiceConfigurationModule extends AbstractModule {
         }
     }
 
-    public static String getLocalNetworkIP() throws Exception {
+    public static String getLocalNetworkIP() throws IOException {
         Enumeration<NetworkInterface> nets = NetworkInterface.getNetworkInterfaces();
 
         while (nets.hasMoreElements()) {
             NetworkInterface netif = nets.nextElement();
+            logger.info("Considering network interface {}:  Up? {},  Loopback? {}", netif.getDisplayName(), netif.isUp(), netif.isLoopback());
             if (!netif.isUp() || netif.isLoopback()) {
                 continue;
             }
@@ -127,6 +129,7 @@ public class ServiceConfigurationModule extends AbstractModule {
             Enumeration<InetAddress> inetAddresses = netif.getInetAddresses();
             while (inetAddresses.hasMoreElements()) {
                 InetAddress addr = inetAddresses.nextElement();
+                logger.info("Considering address {}: SiteLocal? {}, Loopback? {}", addr.getHostAddress(), addr.isSiteLocalAddress(), addr.isLoopbackAddress());
                 if (addr.isSiteLocalAddress() && !addr.isLoopbackAddress()) {
                     return addr.getHostAddress();
                 }

code/common/service/java/nu/marginalia/service/server/JoobyService.java

@@ -15,6 +15,7 @@ import org.slf4j.LoggerFactory;
 import org.slf4j.Marker;
 import org.slf4j.MarkerFactory;
 
+import java.nio.file.Files;
 import java.nio.file.Path;
 import java.nio.file.Paths;
 import java.util.List;
@@ -106,9 +107,12 @@ public class JoobyService {
                 config.externalAddress());
 
         // FIXME:  This won't work outside of docker, may need to submit a PR to jooby to allow classpaths here
-        jooby.install(new JteModule(Path.of("/app/resources/jte"), Path.of("/app/classes/jte-precompiled")));
+        if (Files.exists(Path.of("/app/resources/jte")) || Files.exists(Path.of("/app/classes/jte-precompiled"))) {
-        jooby.assets("/*", Paths.get("/app/resources/static"));
+            jooby.install(new JteModule(Path.of("/app/resources/jte"), Path.of("/app/classes/jte-precompiled")));
-
+        }
+        if (Files.exists(Path.of("/app/resources/static"))) {
+            jooby.assets("/*", Paths.get("/app/resources/static"));
+        }
         var options = new ServerOptions();
         options.setHost(config.bindAddress());
         options.setPort(restEndpoint.port());
@@ -118,6 +122,11 @@ public class JoobyService {
         // single digit percentage difference since HTML already compresses very well with level = 1.
         options.setCompressionLevel(1);
 
+        // Set a cap on the number of worker threads, as Jooby's default value does not seem to consider
+        // multi-tenant servers with high thread counts, and spins up an exorbitant number of threads in that
+        // scenario
+        options.setWorkerThreads(Math.min(128, options.getWorkerThreads()));
+
 
         jooby.setServerOptions(options);
 

code/common/service/java/nu/marginalia/service/server/MetricsServer.java

@@ -6,25 +6,36 @@ import nu.marginalia.service.module.ServiceConfiguration;
 import org.eclipse.jetty.server.Server;
 import org.eclipse.jetty.servlet.ServletContextHandler;
 import org.eclipse.jetty.servlet.ServletHolder;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
 
 import java.net.InetSocketAddress;
 
 public class MetricsServer {
 
+    private static final Logger logger = LoggerFactory.getLogger(MetricsServer.class);
+
     @Inject
-    public MetricsServer(ServiceConfiguration configuration) throws Exception {
+    public MetricsServer(ServiceConfiguration configuration) {
         // If less than zero, we forego setting up a metrics server
         if (configuration.metricsPort() < 0)
             return;
 
-        Server server = new Server(new InetSocketAddress(configuration.bindAddress(), configuration.metricsPort()));
+        try {
+            Server server = new Server(new InetSocketAddress(configuration.bindAddress(), configuration.metricsPort()));
 
-        ServletContextHandler context = new ServletContextHandler();
+            ServletContextHandler context = new ServletContextHandler();
-        context.setContextPath("/");
+            context.setContextPath("/");
-        server.setHandler(context);
+            server.setHandler(context);
 
-        context.addServlet(new ServletHolder(new MetricsServlet()), "/metrics");
+            context.addServlet(new ServletHolder(new MetricsServlet()), "/metrics");
 
-        server.start();
+            logger.info("MetricsServer listening on {}:{}", configuration.bindAddress(), configuration.metricsPort());
+
+            server.start();
+        }
+        catch (Exception|NoSuchMethodError ex) {
+            logger.error("Failed to set up metrics server", ex);
+        }
     }
 }

code/common/service/java/nu/marginalia/service/server/RateLimiter.java

@@ -35,21 +35,8 @@ public class RateLimiter {
     }
 
 
-    public static RateLimiter forExpensiveRequest() {
-        return new RateLimiter(5, 10);
-    }
-
     public static RateLimiter custom(int perMinute) {
-        return new RateLimiter(perMinute, 60);
+        return new RateLimiter(4 * perMinute, perMinute);
-    }
-
-    public static RateLimiter forSpamBots() {
-        return new RateLimiter(120, 3600);
-    }
-
-
-    public static RateLimiter forLogin() {
-        return new RateLimiter(3, 15);
     }
 
     private void cleanIdleBuckets() {
@@ -62,7 +49,7 @@ public class RateLimiter {
     }
 
     private Bucket createBucket() {
-        var refill = Refill.greedy(1, Duration.ofSeconds(refillRate));
+        var refill = Refill.greedy(refillRate, Duration.ofSeconds(60));
         var bw = Bandwidth.classic(capacity, refill);
         return Bucket.builder().addLimit(bw).build();
     }

code/common/service/resources/log4j2-json.xml

@@ -3,8 +3,16 @@
         <Console name="Console" target="SYSTEM_OUT">
             <PatternLayout pattern="%d{HH:mm:ss,SSS} %style{%-8markerSimpleName}{FG_Cyan} %highlight{%-5level}{FATAL=red, ERROR=red, WARN=yellow} %-24t %-20c{1}  --  %msg%n"/>
             <Filters>
+                <MarkerFilter marker="PROCESS" onMatch="DENY" onMismatch="NEUTRAL" />
                 <MarkerFilter marker="QUERY" onMatch="DENY" onMismatch="NEUTRAL" />
                 <MarkerFilter marker="HTTP" onMatch="DENY" onMismatch="NEUTRAL" />
+                <MarkerFilter marker="CRAWLER" onMatch="DENY" onMismatch="NEUTRAL" />
+            </Filters>
+        </Console>
+        <Console name="ProcessConsole" target="SYSTEM_OUT">
+            <PatternLayout pattern="%style{P}{FG_Cyan} %msg%n"/>
+            <Filters>
+                <MarkerFilter marker="PROCESS" onMatch="ALLOW" onMismatch="DENY" />
             </Filters>
         </Console>
         <RollingFile name="LogToFile" fileName="${env:WMSA_LOG_DIR:-/var/log/wmsa}/wmsa-${sys:service-name}-${env:WMSA_SERVICE_NODE:-0}.log" filePattern="/var/log/wmsa/wmsa-${sys:service-name}-${env:WMSA_SERVICE_NODE:-0}-log-%d{MM-dd-yy-HH-mm-ss}-%i.log.gz"
@@ -13,15 +21,29 @@
             <Filters>
                 <MarkerFilter marker="QUERY" onMatch="DENY" onMismatch="NEUTRAL" />
                 <MarkerFilter marker="HTTP" onMatch="DENY" onMismatch="NEUTRAL" />
+                <MarkerFilter marker="CRAWLER" onMatch="DENY" onMismatch="NEUTRAL" />
+                <MarkerFilter marker="PROCESS" onMatch="DENY" onMismatch="NEUTRAL" />
             </Filters>
             <SizeBasedTriggeringPolicy size="10MB" />
         </RollingFile>
+        <RollingFile name="LogToFile" fileName="${env:WMSA_LOG_DIR:-/var/log/wmsa}/crawler-audit-${env:WMSA_SERVICE_NODE:-0}.log" filePattern="/var/log/wmsa/crawler-audit-${env:WMSA_SERVICE_NODE:-0}-log-%d{MM-dd-yy-HH-mm-ss}-%i.log.gz"
+                     ignoreExceptions="false">
+            <PatternLayout>
+                <Pattern>%d{yyyy-MM-dd HH:mm:ss,SSS}: %msg{nolookups}%n</Pattern>
+            </PatternLayout>
+            <SizeBasedTriggeringPolicy size="100MB" />
+            <Filters>
+                <MarkerFilter marker="CRAWLER" onMatch="ALLOW" onMismatch="DENY" />
+            </Filters>
+        </RollingFile>
     </Appenders>
     <Loggers>
         <Logger name="org.apache.zookeeper" level="WARN" />
-
+        <Logger name="org.apache.pdfbox" level="ERROR" />
+        <Logger name="org.apache.fontbox.ttf" level="ERROR" />
         <Root level="info">
             <AppenderRef ref="Console"/>
+            <AppenderRef ref="ProcessConsole"/>
             <AppenderRef ref="LogToFile"/>
         </Root>
     </Loggers>

code/common/service/resources/log4j2-prod.xml

@@ -1,10 +1,49 @@
 <Configuration xmlns="http://logging.apache.org/log4j/2.0/config" >
     <Appenders>
-        <Console name="Console" target="SYSTEM_OUT">
+        <Console name="ConsoleInfo" target="SYSTEM_OUT">
-            <PatternLayout pattern="%d{HH:mm:ss,SSS} %style{%-8markerSimpleName}{FG_Cyan} %highlight{%-5level}{FATAL=red, ERROR=red, WARN=yellow} %-24t %-20c{1}  --  %msg%n"/>
+            <PatternLayout pattern="- %d{HH:mm:ss,SSS} %-20c{1} -- %msg%n"/>
             <Filters>
+                <LevelMatchFilter level="INFO" onMatch="ALLOW" onMismatch="DENY"/>
+                <MarkerFilter marker="PROCESS" onMatch="DENY" onMismatch="NEUTRAL" />
                 <MarkerFilter marker="QUERY" onMatch="DENY" onMismatch="NEUTRAL" />
                 <MarkerFilter marker="HTTP" onMatch="DENY" onMismatch="NEUTRAL" />
+                <MarkerFilter marker="CRAWLER" onMatch="DENY" onMismatch="NEUTRAL" />
+            </Filters>
+        </Console>
+        <Console name="ConsoleWarn" target="SYSTEM_OUT">
+            <PatternLayout pattern="⚠ %d{HH:mm:ss,SSS} %-20c{1} -- %msg%n"/>
+            <Filters>
+                <LevelMatchFilter level="WARN" onMatch="ALLOW" onMismatch="DENY"/>
+                <MarkerFilter marker="PROCESS" onMatch="DENY" onMismatch="NEUTRAL" />
+                <MarkerFilter marker="QUERY" onMatch="DENY" onMismatch="NEUTRAL" />
+                <MarkerFilter marker="HTTP" onMatch="DENY" onMismatch="NEUTRAL" />
+                <MarkerFilter marker="CRAWLER" onMatch="DENY" onMismatch="NEUTRAL" />
+            </Filters>
+        </Console>
+        <Console name="ConsoleError" target="SYSTEM_OUT">
+            <PatternLayout pattern="🔥 %d{HH:mm:ss,SSS} %-20c{1} -- %msg%n"/>
+            <Filters>
+                <LevelMatchFilter level="ERROR" onMatch="ALLOW" onMismatch="DENY"/>
+                <MarkerFilter marker="PROCESS" onMatch="DENY" onMismatch="NEUTRAL" />
+                <MarkerFilter marker="QUERY" onMatch="DENY" onMismatch="NEUTRAL" />
+                <MarkerFilter marker="HTTP" onMatch="DENY" onMismatch="NEUTRAL" />
+                <MarkerFilter marker="CRAWLER" onMatch="DENY" onMismatch="NEUTRAL" />
+            </Filters>
+        </Console>
+        <Console name="ConsoleFatal" target="SYSTEM_OUT">
+            <PatternLayout pattern="💀 %d{HH:mm:ss,SSS} %-20c{1} -- %msg%n"/>
+            <Filters>
+                <LevelMatchFilter level="FATAL" onMatch="ALLOW" onMismatch="DENY"/>
+                <MarkerFilter marker="PROCESS" onMatch="DENY" onMismatch="NEUTRAL" />
+                <MarkerFilter marker="QUERY" onMatch="DENY" onMismatch="NEUTRAL" />
+                <MarkerFilter marker="HTTP" onMatch="DENY" onMismatch="NEUTRAL" />
+                <MarkerFilter marker="CRAWLER" onMatch="DENY" onMismatch="NEUTRAL" />
+            </Filters>
+        </Console>
+        <Console name="ProcessConsole" target="SYSTEM_OUT">
+            <PatternLayout pattern="%style{%msg%n}{FG_Cyan}"/>
+            <Filters>
+                <MarkerFilter marker="PROCESS" onMatch="ALLOW" onMismatch="DENY" />
             </Filters>
         </Console>
         <RollingFile name="LogToFile" fileName="${env:WMSA_LOG_DIR:-/var/log/wmsa}/wmsa-${sys:service-name}-${env:WMSA_SERVICE_NODE:-0}.log" filePattern="/var/log/wmsa/wmsa-${sys:service-name}-${env:WMSA_SERVICE_NODE:-0}-log-%d{MM-dd-yy-HH-mm-ss}-%i.log.gz"
@@ -17,14 +56,30 @@
                 <MarkerFilter marker="PROCESS" onMatch="DENY" onMismatch="NEUTRAL" />
                 <MarkerFilter marker="QUERY" onMatch="DENY" onMismatch="NEUTRAL" />
                 <MarkerFilter marker="HTTP" onMatch="DENY" onMismatch="NEUTRAL" />
+                <MarkerFilter marker="CRAWLER" onMatch="DENY" onMismatch="NEUTRAL" />
+            </Filters>
+        </RollingFile>
+        <RollingFile name="LogToFile" fileName="${env:WMSA_LOG_DIR:-/var/log/wmsa}/crawler-audit-${env:WMSA_SERVICE_NODE:-0}.log" filePattern="/var/log/wmsa/crawler-audit-${env:WMSA_SERVICE_NODE:-0}-log-%d{MM-dd-yy-HH-mm-ss}-%i.log.gz"
+                     ignoreExceptions="false">
+            <PatternLayout>
+                <Pattern>%d{yyyy-MM-dd HH:mm:ss,SSS}: %msg{nolookups}%n</Pattern>
+            </PatternLayout>
+            <SizeBasedTriggeringPolicy size="100MB" />
+            <Filters>
+                <MarkerFilter marker="CRAWLER" onMatch="ALLOW" onMismatch="DENY" />
             </Filters>
         </RollingFile>
     </Appenders>
     <Loggers>
         <Logger name="org.apache.zookeeper" level="WARN" />
-
+        <Logger name="org.apache.pdfbox" level="ERROR" />
+        <Logger name="org.apache.fontbox.ttf" level="ERROR" />
         <Root level="info">
-            <AppenderRef ref="Console"/>
+            <AppenderRef ref="ConsoleInfo"/>
+            <AppenderRef ref="ConsoleWarn"/>
+            <AppenderRef ref="ConsoleError"/>
+            <AppenderRef ref="ConsoleFatal"/>
+            <AppenderRef ref="ProcessConsole"/>
             <AppenderRef ref="LogToFile"/>
         </Root>
     </Loggers>

code/common/service/resources/log4j2-test.xml

@@ -1,15 +1,50 @@
 <Configuration xmlns="http://logging.apache.org/log4j/2.0/config" >
     <Appenders>
-        <Console name="Console" target="SYSTEM_OUT">
+        <Console name="ConsoleInfo" target="SYSTEM_OUT">
-            <PatternLayout pattern="%d{HH:mm:ss,SSS} %style{%-8markerSimpleName}{FG_Cyan} %highlight{%-5level}{FATAL=red, ERROR=red, WARN=yellow} %-24t %-20c{1}  --  %msg%n"/>
+            <PatternLayout pattern="- %d{HH:mm:ss,SSS} %-20c{1} -- %msg%n"/>
+            <Filters>
+                <LevelMatchFilter level="INFO" onMatch="ALLOW" onMismatch="DENY"/>
+                <MarkerFilter marker="PROCESS" onMatch="DENY" onMismatch="NEUTRAL" />
+            </Filters>
+        </Console>
+        <Console name="ConsoleWarn" target="SYSTEM_OUT">
+            <PatternLayout pattern="⚠ %d{HH:mm:ss,SSS} %-20c{1} -- %msg%n"/>
+            <Filters>
+                <LevelMatchFilter level="WARN" onMatch="ALLOW" onMismatch="DENY"/>
+                <MarkerFilter marker="PROCESS" onMatch="DENY" onMismatch="NEUTRAL" />
+            </Filters>
+        </Console>
+        <Console name="ConsoleError" target="SYSTEM_OUT">
+            <PatternLayout pattern="🔥 %d{HH:mm:ss,SSS} %-20c{1} -- %msg%n"/>
+            <Filters>
+                <LevelMatchFilter level="ERROR" onMatch="ALLOW" onMismatch="DENY"/>
+                <MarkerFilter marker="PROCESS" onMatch="DENY" onMismatch="NEUTRAL" />
+            </Filters>
+        </Console>
+        <Console name="ConsoleFatal" target="SYSTEM_OUT">
+            <PatternLayout pattern="💀 %d{HH:mm:ss,SSS} %-20c{1} -- %msg%n"/>
+            <Filters>
+                <LevelMatchFilter level="FATAL" onMatch="ALLOW" onMismatch="DENY"/>
+                <MarkerFilter marker="PROCESS" onMatch="DENY" onMismatch="NEUTRAL" />
+            </Filters>
+        </Console>
+        <Console name="ProcessConsole" target="SYSTEM_OUT">
+            <PatternLayout pattern="%style{%msg%n}{FG_Cyan}"/>
+            <Filters>
+                <MarkerFilter marker="PROCESS" onMatch="ALLOW" onMismatch="DENY" />
+            </Filters>
         </Console>
     </Appenders>
     <Loggers>
         <Logger name="org.apache.zookeeper" level="WARN" />
-
+        <Logger name="org.apache.pdfbox" level="ERROR" />
+        <Logger name="org.apache.fontbox.ttf" level="ERROR" />
         <Root level="info">
-            <AppenderRef ref="Console"/>
+            <AppenderRef ref="ConsoleInfo"/>
-            <AppenderRef ref="LogToFile"/>
+            <AppenderRef ref="ConsoleWarn"/>
+            <AppenderRef ref="ConsoleError"/>
+            <AppenderRef ref="ConsoleFatal"/>
+            <AppenderRef ref="ProcessConsole"/>
         </Root>
     </Loggers>
 </Configuration>

code/common/service/test/nu/marginalia/service/discovery/ZkServiceRegistryTest.java

@@ -25,7 +25,7 @@ import static org.mockito.Mockito.when;
 class ZkServiceRegistryTest {
     private static final int ZOOKEEPER_PORT = 2181;
     private static final GenericContainer<?> zookeeper =
-            new GenericContainer<>("zookeeper:3.8.0")
+            new GenericContainer<>("zookeeper:3.8")
                     .withExposedPorts(ZOOKEEPER_PORT);
 
     List<ZkServiceRegistry> registries = new ArrayList<>();

code/execution/api/java/nu/marginalia/executor/client/ExecutorExportClient.java

@@ -48,12 +48,13 @@ public class ExecutorExportClient {
         return msgId;
     }
 
-    public void exportSampleData(int node, FileStorageId fid, int size, String name) {
+    public void exportSampleData(int node, FileStorageId fid, int size, String ctFilter, String name) {
         channelPool.call(ExecutorExportApiBlockingStub::exportSampleData)
                 .forNode(node)
                 .run(RpcExportSampleData.newBuilder()
                         .setFileStorageId(fid.id())
                         .setSize(size)
+                        .setCtFilter(ctFilter)
                         .setName(name)
                         .build());
     }

code/execution/api/src/main/protobuf/executor-api.proto

@@ -100,6 +100,7 @@ message RpcExportSampleData {
   int64 fileStorageId = 1;
   int32 size = 2;
   string name = 3;
+  string ctFilter = 4;
 }
 message RpcDownloadSampleData {
   string sampleSet = 1;

code/execution/java/nu/marginalia/actor/ExecutorActor.java

@@ -20,6 +20,7 @@ public enum ExecutorActor {
     EXPORT_FEEDS(NodeProfile.BATCH_CRAWL, NodeProfile.MIXED),
     EXPORT_SAMPLE_DATA(NodeProfile.BATCH_CRAWL, NodeProfile.MIXED),
     DOWNLOAD_SAMPLE(NodeProfile.BATCH_CRAWL, NodeProfile.MIXED),
+    MIGRATE_CRAWL_DATA(NodeProfile.BATCH_CRAWL, NodeProfile.MIXED),
 
     PROC_CONVERTER_SPAWNER(NodeProfile.BATCH_CRAWL, NodeProfile.MIXED, NodeProfile.SIDELOAD),
     PROC_LOADER_SPAWNER(NodeProfile.BATCH_CRAWL, NodeProfile.MIXED, NodeProfile.SIDELOAD),

code/execution/java/nu/marginalia/actor/ExecutorActorControlService.java

@@ -66,6 +66,7 @@ public class ExecutorActorControlService {
                                        DownloadSampleActor downloadSampleActor,
                                        ScrapeFeedsActor scrapeFeedsActor,
                                        ExecutorActorStateMachines stateMachines,
+                                       MigrateCrawlDataActor migrateCrawlDataActor,
                                        ExportAllPrecessionActor exportAllPrecessionActor,
                                        UpdateRssActor updateRssActor) throws SQLException {
         this.messageQueueFactory = messageQueueFactory;
@@ -107,6 +108,8 @@ public class ExecutorActorControlService {
         register(ExecutorActor.SCRAPE_FEEDS, scrapeFeedsActor);
         register(ExecutorActor.UPDATE_RSS, updateRssActor);
 
+        register(ExecutorActor.MIGRATE_CRAWL_DATA, migrateCrawlDataActor);
+
         if (serviceConfiguration.node() == 1) {
             register(ExecutorActor.PREC_EXPORT_ALL, exportAllPrecessionActor);
         }

code/execution/java/nu/marginalia/actor/proc/UpdateRssActor.java

@@ -14,6 +14,8 @@ import nu.marginalia.mq.persistence.MqPersistence;
 import nu.marginalia.nodecfg.NodeConfigurationService;
 import nu.marginalia.nodecfg.model.NodeProfile;
 import nu.marginalia.service.module.ServiceConfiguration;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
 
 import java.time.Duration;
 import java.time.LocalDateTime;
@@ -29,6 +31,7 @@ public class UpdateRssActor extends RecordActorPrototype {
 
     private final NodeConfigurationService nodeConfigurationService;
     private final MqPersistence persistence;
+    private static final Logger logger = LoggerFactory.getLogger(UpdateRssActor.class);
 
     @Inject
     public UpdateRssActor(Gson gson,
@@ -101,8 +104,8 @@ public class UpdateRssActor extends RecordActorPrototype {
             case UpdateRefresh(int count, long msgId) -> {
                 MqMessage msg = persistence.waitForMessageTerminalState(msgId, Duration.ofSeconds(10), Duration.ofHours(12));
                 if (msg == null) {
-                    // Retry the update
+                    logger.warn("UpdateRefresh is taking a very long time");
-                    yield new Error("Failed to update feeds: message not found");
+                    yield new UpdateRefresh(count, msgId);
                 } else if (msg.state() != MqMessageState.OK) {
                     // Retry the update
                     yield new Error("Failed to update feeds: " + msg.state());
@@ -119,8 +122,8 @@ public class UpdateRssActor extends RecordActorPrototype {
             case UpdateClean(long msgId) -> {
                 MqMessage msg = persistence.waitForMessageTerminalState(msgId, Duration.ofSeconds(10), Duration.ofHours(12));
                 if (msg == null) {
-                    // Retry the update
+                    logger.warn("UpdateClean is taking a very long time");
-                    yield new Error("Failed to update feeds: message not found");
+                    yield new UpdateClean(msgId);
                 } else if (msg.state() != MqMessageState.OK) {
                     // Retry the update
                     yield new Error("Failed to update feeds: " + msg.state());

code/execution/java/nu/marginalia/actor/task/DownloadSampleActor.java

@@ -8,6 +8,7 @@ import nu.marginalia.actor.state.ActorResumeBehavior;
 import nu.marginalia.actor.state.ActorStep;
 import nu.marginalia.actor.state.Resume;
 import nu.marginalia.service.control.ServiceEventLog;
+import nu.marginalia.service.control.ServiceHeartbeat;
 import nu.marginalia.storage.FileStorageService;
 import nu.marginalia.storage.model.FileStorage;
 import nu.marginalia.storage.model.FileStorageId;
@@ -19,6 +20,7 @@ import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 
 import java.io.*;
+import java.net.HttpURLConnection;
 import java.net.MalformedURLException;
 import java.net.URI;
 import java.net.URL;
@@ -32,6 +34,7 @@ public class DownloadSampleActor extends RecordActorPrototype {
 
     private final FileStorageService storageService;
     private final ServiceEventLog eventLog;
+    private final ServiceHeartbeat heartbeat;
     private final Logger logger = LoggerFactory.getLogger(getClass());
 
     @Resume(behavior = ActorResumeBehavior.ERROR)
@@ -66,15 +69,39 @@ public class DownloadSampleActor extends RecordActorPrototype {
 
                 Files.deleteIfExists(Path.of(tarFileName));
 
-                try (var is = new BufferedInputStream(new URI(downloadURI).toURL().openStream());
+                HttpURLConnection urlConnection = (HttpURLConnection) new URI(downloadURI).toURL().openConnection();
-                     var os = new BufferedOutputStream(Files.newOutputStream(Path.of(tarFileName), StandardOpenOption.CREATE))) {
+
-                    is.transferTo(os);
+                try (var hb = heartbeat.createServiceAdHocTaskHeartbeat("Downloading sample")) {
+                    long size = urlConnection.getContentLengthLong();
+                    byte[] buffer = new byte[8192];
+
+                    try (var is = new BufferedInputStream(urlConnection.getInputStream());
+                         var os = new BufferedOutputStream(Files.newOutputStream(Path.of(tarFileName), StandardOpenOption.CREATE))) {
+                        long copiedSize = 0;
+
+                        while (copiedSize < size) {
+                            int read = is.read(buffer);
+
+                            if (read < 0) // We've been promised a file of length 'size'
+                                throw new IOException("Unexpected end of stream");
+
+                            os.write(buffer, 0, read);
+                            copiedSize += read;
+
+                            // Update progress bar
+                            hb.progress(String.format("%d MB", copiedSize / 1024 / 1024), (int) (copiedSize / 1024), (int) (size / 1024));
+                        }
+                    }
+
                 }
                 catch (Exception ex) {
                     eventLog.logEvent(DownloadSampleActor.class, "Error downloading sample");
                     logger.error("Error downloading sample", ex);
                     yield new Error();
                 }
+                finally {
+                    urlConnection.disconnect();
+                }
 
                 eventLog.logEvent(DownloadSampleActor.class, "Download complete");
                 yield new Extract(fileStorageId, tarFileName);
@@ -170,11 +197,12 @@ public class DownloadSampleActor extends RecordActorPrototype {
     @Inject
     public DownloadSampleActor(Gson gson,
                                FileStorageService storageService,
-                               ServiceEventLog eventLog)
+                               ServiceEventLog eventLog, ServiceHeartbeat heartbeat)
     {
         super(gson);
         this.storageService = storageService;
         this.eventLog = eventLog;
+        this.heartbeat = heartbeat;
     }
 
 }

code/execution/java/nu/marginalia/actor/task/ExportSampleDataActor.java

@@ -26,32 +26,32 @@ public class ExportSampleDataActor extends RecordActorPrototype {
     private final MqOutbox exportTasksOutbox;
     private final Logger logger = LoggerFactory.getLogger(getClass());
 
-    public record Export(FileStorageId crawlId, int size, String name) implements ActorStep {}
+    public record Export(FileStorageId crawlId, int size, String ctFilter, String name) implements ActorStep {}
-    public record Run(FileStorageId crawlId, FileStorageId destId, int size, String name, long msgId) implements ActorStep {
+    public record Run(FileStorageId crawlId, FileStorageId destId, int size, String ctFilter, String name, long msgId) implements ActorStep {
-        public Run(FileStorageId crawlId, FileStorageId destId, int size, String name) {
+        public Run(FileStorageId crawlId, FileStorageId destId, int size, String name, String ctFilter) {
-            this(crawlId, destId, size, name, -1);
+            this(crawlId, destId, size, name, ctFilter,-1);
         }
     }
 
     @Override
     public ActorStep transition(ActorStep self) throws Exception {
         return switch(self) {
-            case Export(FileStorageId crawlId, int size, String name) -> {
+            case Export(FileStorageId crawlId, int size, String ctFilter, String name) -> {
                 var storage = storageService.allocateStorage(FileStorageType.EXPORT,
                         "crawl-sample-export",
                         "Crawl Data Sample " + name + "/" + size + " " + LocalDateTime.now()
                 );
 
                 if (storage == null) yield new Error("Bad storage id");
-                yield new Run(crawlId, storage.id(), size, name);
+                yield new Run(crawlId, storage.id(), size, ctFilter, name);
             }
-            case Run(FileStorageId crawlId, FileStorageId destId, int size, String name, long msgId) when msgId < 0 -> {
+            case Run(FileStorageId crawlId, FileStorageId destId, int size, String ctFilter, String name, long msgId) when msgId < 0 -> {
                 storageService.setFileStorageState(destId, FileStorageState.NEW);
 
-                long newMsgId = exportTasksOutbox.sendAsync(ExportTaskRequest.sampleData(crawlId, destId, size, name));
+                long newMsgId = exportTasksOutbox.sendAsync(ExportTaskRequest.sampleData(crawlId, destId, ctFilter, size, name));
-                yield new Run(crawlId, destId, size, name, newMsgId);
+                yield new Run(crawlId, destId, size, ctFilter, name, newMsgId);
             }
-            case Run(_, FileStorageId destId, _, _, long msgId) -> {
+            case Run(_, FileStorageId destId, _, _, _, long msgId) -> {
                 var rsp = processWatcher.waitResponse(exportTasksOutbox, ProcessService.ProcessId.EXPORT_TASKS, msgId);
 
                 if (rsp.state() != MqMessageState.OK) {
@@ -70,7 +70,7 @@ public class ExportSampleDataActor extends RecordActorPrototype {
 
     @Override
     public String describe() {
-        return "Export RSS/Atom feeds from crawl data";
+        return "Export sample crawl data";
     }
 
     @Inject

code/execution/java/nu/marginalia/actor/task/MigrateCrawlDataActor.java

@@ -0,0 +1,150 @@
+package nu.marginalia.actor.task;
+
+import com.google.gson.Gson;
+import jakarta.inject.Inject;
+import jakarta.inject.Singleton;
+import nu.marginalia.actor.prototype.RecordActorPrototype;
+import nu.marginalia.actor.state.ActorStep;
+import nu.marginalia.io.CrawlerOutputFile;
+import nu.marginalia.process.log.WorkLog;
+import nu.marginalia.process.log.WorkLogEntry;
+import nu.marginalia.service.control.ServiceHeartbeat;
+import nu.marginalia.slop.SlopCrawlDataRecord;
+import nu.marginalia.storage.FileStorageService;
+import nu.marginalia.storage.model.FileStorage;
+import nu.marginalia.storage.model.FileStorageId;
+import org.apache.logging.log4j.util.Strings;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+import java.nio.file.Files;
+import java.nio.file.Path;
+import java.nio.file.StandardCopyOption;
+import java.util.Map;
+import java.util.Optional;
+import java.util.function.Function;
+
+@Singleton
+public class MigrateCrawlDataActor extends RecordActorPrototype {
+
+    private final FileStorageService fileStorageService;
+    private final ServiceHeartbeat serviceHeartbeat;
+    private static final Logger logger = LoggerFactory.getLogger(MigrateCrawlDataActor.class);
+
+    @Inject
+    public MigrateCrawlDataActor(Gson gson, FileStorageService fileStorageService, ServiceHeartbeat serviceHeartbeat) {
+        super(gson);
+
+        this.fileStorageService = fileStorageService;
+        this.serviceHeartbeat = serviceHeartbeat;
+    }
+
+    public record Run(long fileStorageId) implements ActorStep {}
+
+    @Override
+    public ActorStep transition(ActorStep self) throws Exception {
+        return switch (self) {
+            case Run(long fileStorageId) -> {
+
+                FileStorage storage = fileStorageService.getStorage(FileStorageId.of(fileStorageId));
+                Path root = storage.asPath();
+
+                Path crawlerLog = root.resolve("crawler.log");
+                Path newCrawlerLog = Files.createTempFile(root, "crawler", ".migrate.log");
+
+                int totalEntries = WorkLog.countEntries(crawlerLog);
+
+                try (WorkLog workLog = new WorkLog(newCrawlerLog);
+                     var heartbeat = serviceHeartbeat.createServiceAdHocTaskHeartbeat("Migrating")
+                ) {
+                    int entryIdx = 0;
+
+                    for (Map.Entry<WorkLogEntry, Path> item : WorkLog.iterableMap(crawlerLog, new CrawlDataLocator(root))) {
+
+                        final WorkLogEntry entry = item.getKey();
+                        final Path inputPath = item.getValue();
+
+                        Path outputPath = inputPath;
+                        heartbeat.progress("Migrating" + inputPath.getFileName(), entryIdx++, totalEntries);
+
+                        if (inputPath.toString().endsWith(".parquet")) {
+                            String domain = entry.id();
+                            String id = Integer.toHexString(domain.hashCode());
+
+                            outputPath = CrawlerOutputFile.createSlopPath(root, id, domain);
+
+                            if (Files.exists(inputPath)) {
+                                try {
+                                    SlopCrawlDataRecord.convertFromParquet(inputPath, outputPath);
+                                    Files.deleteIfExists(inputPath);
+                                } catch (Exception ex) {
+                                    outputPath = inputPath; // don't update the work log on error
+                                    logger.error("Failed to convert " + inputPath, ex);
+                                }
+                            }
+                            else if (!Files.exists(inputPath) && !Files.exists(outputPath)) {
+                                // if the input file is missing, and the output file is missing, we just write the log
+                                // record identical to the old one
+                                outputPath = inputPath;
+                            }
+                        }
+
+                        // Write a log entry for the (possibly) converted file
+                        workLog.setJobToFinished(entry.id(), outputPath.toString(), entry.cnt());
+                    }
+                }
+
+                Path oldCrawlerLog = Files.createTempFile(root, "crawler-", ".migrate.old.log");
+                Files.move(crawlerLog, oldCrawlerLog, StandardCopyOption.REPLACE_EXISTING);
+                Files.move(newCrawlerLog, crawlerLog);
+
+                yield new End();
+            }
+            default -> new Error();
+        };
+    }
+
+    private static class CrawlDataLocator implements Function<WorkLogEntry, Optional<Map.Entry<WorkLogEntry, Path>>> {
+
+        private final Path crawlRootDir;
+
+        CrawlDataLocator(Path crawlRootDir) {
+            this.crawlRootDir = crawlRootDir;
+        }
+
+        @Override
+        public Optional<Map.Entry<WorkLogEntry, Path>> apply(WorkLogEntry entry) {
+            var path = getCrawledFilePath(crawlRootDir, entry.path());
+
+            if (!Files.exists(path)) {
+                return Optional.empty();
+            }
+
+            try {
+                return Optional.of(Map.entry(entry, path));
+            }
+            catch (Exception ex) {
+                return Optional.empty();
+            }
+        }
+
+        private Path getCrawledFilePath(Path crawlDir, String fileName) {
+            int sp = fileName.lastIndexOf('/');
+
+            // Normalize the filename
+            if (sp >= 0 && sp + 1< fileName.length())
+                fileName = fileName.substring(sp + 1);
+            if (fileName.length() < 4)
+                fileName = Strings.repeat("0", 4 - fileName.length()) + fileName;
+
+            String sp1 = fileName.substring(0, 2);
+            String sp2 = fileName.substring(2, 4);
+            return crawlDir.resolve(sp1).resolve(sp2).resolve(fileName);
+        }
+    }
+
+    @Override
+    public String describe() {
+        return "Migrates crawl data to the latest format";
+    }
+}

code/execution/java/nu/marginalia/execution/ExecutorExportGrpcService.java

@@ -49,6 +49,7 @@ public class ExecutorExportGrpcService
                     new ExportSampleDataActor.Export(
                             FileStorageId.of(request.getFileStorageId()),
                             request.getSize(),
+                            request.getCtFilter(),
                             request.getName()
                     )
             );

code/functions/favicon/api/build.gradle

@@ -0,0 +1,47 @@
+plugins {
+    id 'java'
+
+    id "com.google.protobuf" version "0.9.4"
+    id 'jvm-test-suite'
+}
+
+java {
+    toolchain {
+        languageVersion.set(JavaLanguageVersion.of(rootProject.ext.jvmVersion))
+    }
+}
+
+jar.archiveBaseName = 'favicon-api'
+
+apply from: "$rootProject.projectDir/protobuf.gradle"
+apply from: "$rootProject.projectDir/srcsets.gradle"
+
+dependencies {
+    implementation project(':code:common:model')
+    implementation project(':code:common:config')
+    implementation project(':code:common:service')
+
+    implementation libs.bundles.slf4j
+
+    implementation libs.prometheus
+    implementation libs.notnull
+    implementation libs.guava
+    implementation dependencies.create(libs.guice.get()) {
+        exclude group: 'com.google.guava'
+    }
+    implementation libs.gson
+    implementation libs.bundles.protobuf
+    implementation libs.guava
+    libs.bundles.grpc.get().each {
+        implementation dependencies.create(it) {
+            exclude group: 'com.google.guava'
+        }
+    }
+
+
+
+    testImplementation libs.bundles.slf4j.test
+    testImplementation libs.bundles.junit
+    testImplementation libs.mockito
+
+}

code/functions/favicon/api/java/nu/marginalia/api/favicon/FaviconClient.java

@@ -0,0 +1,39 @@
+package nu.marginalia.api.favicon;
+
+import com.google.inject.Inject;
+import nu.marginalia.service.client.GrpcChannelPoolFactory;
+import nu.marginalia.service.client.GrpcMultiNodeChannelPool;
+import nu.marginalia.service.discovery.property.ServiceKey;
+import nu.marginalia.service.discovery.property.ServicePartition;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+import java.util.Optional;
+
+public class FaviconClient {
+    private static final Logger logger = LoggerFactory.getLogger(FaviconClient.class);
+
+    private final GrpcMultiNodeChannelPool<FaviconAPIGrpc.FaviconAPIBlockingStub> channelPool;
+
+    @Inject
+    public FaviconClient(GrpcChannelPoolFactory factory) {
+        this.channelPool = factory.createMulti(
+                ServiceKey.forGrpcApi(FaviconAPIGrpc.class, ServicePartition.multi()),
+                FaviconAPIGrpc::newBlockingStub);
+    }
+
+    public record FaviconData(byte[] bytes, String contentType) {}
+
+
+    public Optional<FaviconData> getFavicon(String domain, int node) {
+        RpcFaviconResponse rsp = channelPool.call(FaviconAPIGrpc.FaviconAPIBlockingStub::getFavicon)
+                .forNode(node)
+                .run(RpcFaviconRequest.newBuilder().setDomain(domain).build());
+
+        if (rsp.getData().isEmpty())
+            return Optional.empty();
+
+        return Optional.of(new FaviconData(rsp.getData().toByteArray(), rsp.getContentType()));
+    }
+
+}

code/functions/favicon/api/src/main/protobuf/favicon.proto

@@ -0,0 +1,20 @@
+syntax="proto3";
+package marginalia.api.favicon;
+
+option java_package="nu.marginalia.api.favicon";
+option java_multiple_files=true;
+
+service FaviconAPI {
+  /** Fetches information about a domain. */
+  rpc getFavicon(RpcFaviconRequest) returns (RpcFaviconResponse) {}
+}
+
+message RpcFaviconRequest {
+  string domain = 1;
+}
+
+message RpcFaviconResponse {
+  string domain = 1;
+  bytes data = 2;
+  string contentType = 3;
+}

code/functions/favicon/build.gradle

@@ -0,0 +1,49 @@
+plugins {
+    id 'java'
+
+    id 'application'
+    id 'jvm-test-suite'
+}
+
+java {
+    toolchain {
+        languageVersion.set(JavaLanguageVersion.of(rootProject.ext.jvmVersion))
+    }
+}
+
+apply from: "$rootProject.projectDir/srcsets.gradle"
+
+dependencies {
+    implementation project(':code:common:config')
+    implementation project(':code:common:service')
+    implementation project(':code:common:model')
+    implementation project(':code:common:db')
+    implementation project(':code:functions:favicon:api')
+    implementation project(':code:processes:crawling-process')
+
+    implementation libs.bundles.slf4j
+
+    implementation libs.prometheus
+    implementation libs.guava
+    libs.bundles.grpc.get().each {
+        implementation dependencies.create(it) {
+            exclude group: 'com.google.guava'
+        }
+    }
+
+
+    implementation libs.notnull
+    implementation libs.guava
+    implementation dependencies.create(libs.guice.get()) {
+        exclude group: 'com.google.guava'
+    }
+    implementation dependencies.create(libs.spark.get()) {
+        exclude group: 'org.eclipse.jetty'
+    }
+
+    testImplementation libs.bundles.slf4j.test
+    testImplementation libs.bundles.junit
+    testImplementation libs.mockito
+
+
+}

code/functions/favicon/java/nu/marginalia/functions/favicon/FaviconGrpcService.java

@@ -0,0 +1,48 @@
+package nu.marginalia.functions.favicon;
+
+import com.google.inject.Inject;
+import com.google.inject.Singleton;
+import com.google.protobuf.ByteString;
+import io.grpc.stub.StreamObserver;
+import nu.marginalia.api.favicon.FaviconAPIGrpc;
+import nu.marginalia.api.favicon.RpcFaviconRequest;
+import nu.marginalia.api.favicon.RpcFaviconResponse;
+import nu.marginalia.crawl.DomainStateDb;
+import nu.marginalia.service.server.DiscoverableService;
+
+import java.util.Optional;
+
+@Singleton
+public class FaviconGrpcService extends FaviconAPIGrpc.FaviconAPIImplBase implements DiscoverableService {
+    private final DomainStateDb domainStateDb;
+
+    @Inject
+    public FaviconGrpcService(DomainStateDb domainStateDb) {
+        this.domainStateDb = domainStateDb;
+    }
+
+    public boolean shouldRegisterService() {
+        return domainStateDb.isAvailable();
+    }
+
+    @Override
+    public void getFavicon(RpcFaviconRequest request, StreamObserver<RpcFaviconResponse> responseObserver) {
+        Optional<DomainStateDb.FaviconRecord> icon = domainStateDb.getIcon(request.getDomain());
+
+        RpcFaviconResponse response;
+        if (icon.isEmpty()) {
+            response = RpcFaviconResponse.newBuilder().build();
+        }
+        else {
+            var iconRecord = icon.get();
+            response = RpcFaviconResponse.newBuilder()
+                            .setContentType(iconRecord.contentType())
+                            .setDomain(request.getDomain())
+                            .setData(ByteString.copyFrom(iconRecord.imageData()))
+                            .build();
+        }
+
+        responseObserver.onNext(response);
+        responseObserver.onCompleted();
+    }
+}

code/functions/live-capture/build.gradle

@@ -27,12 +27,14 @@ dependencies {
     implementation project(':code:processes:crawling-process:ft-content-type')
 
     implementation libs.jsoup
-    implementation libs.rssreader
+    implementation project(':third-party:rssreader')
     implementation libs.opencsv
+    implementation libs.slop
     implementation libs.sqlite
     implementation libs.bundles.slf4j
     implementation libs.commons.lang3
     implementation libs.commons.io
+    implementation libs.wiremock
 
     implementation libs.prometheus
     implementation libs.guava

code/functions/live-capture/java/nu/marginalia/livecapture/BrowserlessClient.java

@@ -1,6 +1,7 @@
 package nu.marginalia.livecapture;
 
 import com.google.gson.Gson;
+import nu.marginalia.WmsaHome;
 import nu.marginalia.model.gson.GsonFactory;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
@@ -12,10 +13,13 @@ import java.net.http.HttpRequest;
 import java.net.http.HttpResponse;
 import java.time.Duration;
 import java.util.Map;
+import java.util.Optional;
 
 /** Client for local browserless.io API */
 public class BrowserlessClient implements AutoCloseable {
+
     private static final Logger logger = LoggerFactory.getLogger(BrowserlessClient.class);
+    private static final String BROWSERLESS_TOKEN = System.getProperty("live-capture.browserless-token", "BROWSERLESS_TOKEN");
 
     private final HttpClient httpClient = HttpClient.newBuilder()
             .version(HttpClient.Version.HTTP_1_1)
@@ -25,18 +29,21 @@ public class BrowserlessClient implements AutoCloseable {
     private final URI browserlessURI;
     private final Gson gson = GsonFactory.get();
 
+    private final String userAgent = WmsaHome.getUserAgent().uaString();
+
     public BrowserlessClient(URI browserlessURI) {
         this.browserlessURI = browserlessURI;
     }
 
-    public String content(String url, GotoOptions gotoOptions) throws IOException, InterruptedException {
+    public Optional<String> content(String url, GotoOptions gotoOptions) throws IOException, InterruptedException {
         Map<String, Object> requestData = Map.of(
                 "url", url,
+                "userAgent", userAgent,
                 "gotoOptions", gotoOptions
         );
 
         var request = HttpRequest.newBuilder()
-                .uri(browserlessURI.resolve("/content"))
+                .uri(browserlessURI.resolve("/content?token="+BROWSERLESS_TOKEN))
                 .method("POST", HttpRequest.BodyPublishers.ofString(
                         gson.toJson(requestData)
                 ))
@@ -47,10 +54,10 @@ public class BrowserlessClient implements AutoCloseable {
 
         if (rsp.statusCode() >= 300) {
             logger.info("Failed to fetch content for {}, status {}", url, rsp.statusCode());
-            return null;
+            return Optional.empty();
         }
 
-        return rsp.body();
+        return Optional.of(rsp.body());
     }
 
     public byte[] screenshot(String url, GotoOptions gotoOptions, ScreenshotOptions screenshotOptions)
@@ -58,12 +65,13 @@ public class BrowserlessClient implements AutoCloseable {
 
         Map<String, Object> requestData = Map.of(
                 "url", url,
+                "userAgent", userAgent,
                 "options", screenshotOptions,
                 "gotoOptions", gotoOptions
         );
 
         var request = HttpRequest.newBuilder()
-                .uri(browserlessURI.resolve("/screenshot"))
+                .uri(browserlessURI.resolve("/screenshot?token="+BROWSERLESS_TOKEN))
                 .method("POST", HttpRequest.BodyPublishers.ofString(
                         gson.toJson(requestData)
                 ))
@@ -82,7 +90,7 @@ public class BrowserlessClient implements AutoCloseable {
     }
 
     @Override
-    public void close() throws Exception {
+    public void close() {
         httpClient.shutdownNow();
     }
 

code/functions/live-capture/java/nu/marginalia/rss/model/FeedItem.java

@@ -1,6 +1,6 @@
 package nu.marginalia.rss.model;
 
-import com.apptasticsoftware.rssreader.Item;
+import nu.marginalia.rss.svc.SimpleFeedParser;
 import org.apache.commons.lang3.StringUtils;
 import org.jetbrains.annotations.NotNull;
 import org.jsoup.Jsoup;
@@ -18,37 +18,33 @@ public record FeedItem(String title,
     public static final int MAX_DESC_LENGTH = 255;
     public static final DateTimeFormatter DATE_FORMAT = DateTimeFormatter.ofPattern("yyyy-MM-dd'T'HH:mm:ss.SSSZ");
 
-    public static FeedItem fromItem(Item item, boolean keepFragment) {
+    public static FeedItem fromItem(SimpleFeedParser.ItemData item, boolean keepFragment) {
-        String title = item.getTitle().orElse("");
+        String title = item.title();
         String date = getItemDate(item);
         String description = getItemDescription(item);
         String url;
 
-        if (keepFragment || item.getLink().isEmpty()) {
+        if (keepFragment) {
-            url = item.getLink().orElse("");
+            url = item.url();
         }
         else {
             try {
-                String link = item.getLink().get();
+                String link = item.url();
                 var linkUri = new URI(link);
                 var cleanUri = new URI(linkUri.getScheme(), linkUri.getAuthority(), linkUri.getPath(), linkUri.getQuery(), null);
                 url = cleanUri.toString();
             }
             catch (Exception e) {
                 // fallback to original link if we can't clean it, this is not a very important step
-                url = item.getLink().get();
+                url = item.url();
             }
         }
 
         return new FeedItem(title, date, description, url);
     }
 
-    private static String getItemDescription(Item item) {
+    private static String getItemDescription(SimpleFeedParser.ItemData item) {
-        Optional<String> description = item.getDescription();
+        String rawDescription = item.description();
-        if (description.isEmpty())
-            return "";
-
-        String rawDescription = description.get();
         if (rawDescription.indexOf('<') >= 0) {
             rawDescription = Jsoup.parseBodyFragment(rawDescription).text();
         }
@@ -58,15 +54,18 @@ public record FeedItem(String title,
 
     // e.g. http://fabiensanglard.net/rss.xml does dates like this:  1 Apr 2021 00:00:00 +0000
     private static final DateTimeFormatter extraFormatter = DateTimeFormatter.ofPattern("d MMM yyyy HH:mm:ss Z");
-    private static String getItemDate(Item item) {
+    private static String getItemDate(SimpleFeedParser.ItemData item) {
         Optional<ZonedDateTime> zonedDateTime = Optional.empty();
         try {
             zonedDateTime = item.getPubDateZonedDateTime();
         }
         catch (Exception e) {
-            zonedDateTime = item.getPubDate()
+            try {
-                    .map(extraFormatter::parse)
+                zonedDateTime = Optional.of(ZonedDateTime.from(extraFormatter.parse(item.pubDate())));
-                    .map(ZonedDateTime::from);
+            }
+            catch (Exception e2) {
+                // ignore
+            }
         }
 
         return zonedDateTime.map(date -> date.format(DATE_FORMAT)).orElse("");

code/functions/live-capture/java/nu/marginalia/rss/svc/FeedFetcherService.java

@@ -1,7 +1,5 @@
 package nu.marginalia.rss.svc;
 
-import com.apptasticsoftware.rssreader.Item;
-import com.apptasticsoftware.rssreader.RssReader;
 import com.google.inject.Inject;
 import com.opencsv.CSVReader;
 import nu.marginalia.WmsaHome;
@@ -20,7 +18,6 @@ import nu.marginalia.storage.FileStorageService;
 import nu.marginalia.storage.model.FileStorage;
 import nu.marginalia.storage.model.FileStorageType;
 import nu.marginalia.util.SimpleBlockingThreadPool;
-import org.apache.commons.io.input.BOMInputStream;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 
@@ -32,11 +29,11 @@ import java.net.URISyntaxException;
 import java.net.http.HttpClient;
 import java.net.http.HttpRequest;
 import java.net.http.HttpResponse;
-import java.nio.charset.StandardCharsets;
 import java.sql.SQLException;
 import java.time.*;
 import java.time.format.DateTimeFormatter;
 import java.util.*;
+import java.util.concurrent.ExecutorService;
 import java.util.concurrent.Executors;
 import java.util.concurrent.TimeUnit;
 import java.util.concurrent.atomic.AtomicInteger;
@@ -48,8 +45,6 @@ public class FeedFetcherService {
     private static final int MAX_FEED_ITEMS = 10;
     private static final Logger logger = LoggerFactory.getLogger(FeedFetcherService.class);
 
-    private final RssReader rssReader = new RssReader();
-
     private final FeedDb feedDb;
     private final FileStorageService fileStorageService;
     private final NodeConfigurationService nodeConfigurationService;
@@ -72,23 +67,12 @@ public class FeedFetcherService {
         this.nodeConfigurationService = nodeConfigurationService;
         this.serviceHeartbeat = serviceHeartbeat;
         this.executorClient = executorClient;
-
-
-        // Add support for some alternate date tags for atom
-        rssReader.addItemExtension("issued", this::setDateFallback);
-        rssReader.addItemExtension("created", this::setDateFallback);
-    }
-
-    private void setDateFallback(Item item, String value) {
-        if (item.getPubDate().isEmpty()) {
-            item.setPubDate(value);
-        }
     }
 
     public enum UpdateMode {
         CLEAN,
         REFRESH
-    };
+    }
 
     public void updateFeeds(UpdateMode updateMode) throws IOException {
         if (updating) // Prevent concurrent updates
@@ -96,6 +80,7 @@ public class FeedFetcherService {
             throw new IllegalStateException("Already updating feeds, refusing to start another update");
         }
 
+
         try (FeedDbWriter writer = feedDb.createWriter();
              HttpClient client = HttpClient.newBuilder()
                 .connectTimeout(Duration.ofSeconds(15))
@@ -103,6 +88,8 @@ public class FeedFetcherService {
                 .followRedirects(HttpClient.Redirect.NORMAL)
                 .version(HttpClient.Version.HTTP_2)
                 .build();
+             ExecutorService fetchExecutor = Executors.newCachedThreadPool();
+             FeedJournal feedJournal = FeedJournal.create();
              var heartbeat = serviceHeartbeat.createServiceAdHocTaskHeartbeat("Update Rss Feeds")
         ) {
             updating = true;
@@ -146,7 +133,7 @@ public class FeedFetcherService {
 
                         FetchResult feedData;
                         try (DomainLocks.DomainLock domainLock = domainLocks.lockDomain(new EdgeDomain(feed.domain()))) {
-                            feedData = fetchFeedData(feed, client, ifModifiedSinceDate, ifNoneMatchTag);
+                            feedData = fetchFeedData(feed, client, fetchExecutor, ifModifiedSinceDate, ifNoneMatchTag);
                         } catch (Exception ex) {
                             feedData = new FetchResult.TransientError();
                         }
@@ -155,6 +142,8 @@ public class FeedFetcherService {
                             case FetchResult.Success(String value, String etag) -> {
                                 writer.saveEtag(feed.domain(), etag);
                                 writer.saveFeed(parseFeed(value, feed));
+
+                                feedJournal.record(feed.feedUrl(), value);
                             }
                             case FetchResult.NotModified() -> {
                                 writer.saveEtag(feed.domain(), ifNoneMatchTag);
@@ -224,6 +213,7 @@ public class FeedFetcherService {
 
     private FetchResult fetchFeedData(FeedDefinition feed,
                                       HttpClient client,
+                                      ExecutorService executorService,
                                       @Nullable String ifModifiedSinceDate,
                                       @Nullable String ifNoneMatchTag)
     {
@@ -239,18 +229,27 @@ public class FeedFetcherService {
                     .timeout(Duration.ofSeconds(15))
                     ;
 
-            if (ifModifiedSinceDate != null) {
+            // Set the If-Modified-Since or If-None-Match headers if we have them
+            // though since there are certain idiosyncrasies in server implementations,
+            // we avoid setting both at the same time as that may turn a 304 into a 200.
+            if (ifNoneMatchTag != null) {
+                requestBuilder.header("If-None-Match", ifNoneMatchTag);
+            } else if (ifModifiedSinceDate != null) {
                 requestBuilder.header("If-Modified-Since", ifModifiedSinceDate);
             }
 
-            if (ifNoneMatchTag != null) {
-                requestBuilder.header("If-None-Match", ifNoneMatchTag);
-            }
 
             HttpRequest getRequest = requestBuilder.build();
 
             for (int i = 0; i < 3; i++) {
-                HttpResponse<byte[]> rs = client.send(getRequest, HttpResponse.BodyHandlers.ofByteArray());
+
+                /* Note we need to use an executor to time-limit the send() method in HttpClient, as
+                 * its support for timeouts only applies to the time until response starts to be received,
+                 * and does not catch the case when the server starts to send data but then hangs.
+                 */
+                HttpResponse<byte[]> rs = executorService.submit(
+                        () -> client.send(getRequest, HttpResponse.BodyHandlers.ofByteArray()))
+                                .get(15, TimeUnit.SECONDS);
 
                 if (rs.statusCode() == 429) { // Too Many Requests
                     int retryAfter = Integer.parseInt(rs.headers().firstValue("Retry-After").orElse("2"));
@@ -367,12 +366,7 @@ public class FeedFetcherService {
 
     public FeedItems parseFeed(String feedData, FeedDefinition definition) {
         try {
-            feedData = sanitizeEntities(feedData);
+            List<SimpleFeedParser.ItemData> rawItems = SimpleFeedParser.parse(feedData);
-
-            List<Item> rawItems = rssReader.read(
-                    // Massage the data to maximize the possibility of the flaky XML parser consuming it
-                    new BOMInputStream(new ByteArrayInputStream(feedData.trim().getBytes(StandardCharsets.UTF_8)), false)
-            ).toList();
 
             boolean keepUriFragment = rawItems.size() < 2 || areFragmentsDisparate(rawItems);
 
@@ -395,33 +389,6 @@ public class FeedFetcherService {
         }
     }
 
-    private static final Map<String, String> HTML_ENTITIES = Map.of(
-            "&raquo;", "»",
-            "&laquo;", "«",
-            "&mdash;", "--",
-            "&ndash;", "-",
-            "&rsquo;", "'",
-            "&lsquo;", "'",
-            "&quot;", "\"",
-            "&nbsp;", ""
-    );
-
-    /** The XML parser will blow up if you insert HTML entities in the feed XML,
-     * which is unfortunately relatively common.  Replace them as far as is possible
-     * with their corresponding characters
-     */
-    static String sanitizeEntities(String feedData) {
-        String result = feedData;
-        for (Map.Entry<String, String> entry : HTML_ENTITIES.entrySet()) {
-            result = result.replace(entry.getKey(), entry.getValue());
-        }
-
-        // Handle lone ampersands not part of a recognized XML entity
-        result = result.replaceAll("&(?!(amp|lt|gt|apos|quot);)", "&amp;");
-
-        return result;
-    }
-
     /** Decide whether to keep URI fragments in the feed items.
      * <p></p>
      * We keep fragments if there are multiple different fragments in the items.
@@ -429,16 +396,16 @@ public class FeedFetcherService {
      * @param items The items to check
      * @return True if we should keep the fragments, false otherwise
      */
-    private boolean areFragmentsDisparate(List<Item> items) {
+    private boolean areFragmentsDisparate(List<SimpleFeedParser.ItemData> items) {
         Set<String> seenFragments = new HashSet<>();
 
         try {
             for (var item : items) {
-                if (item.getLink().isEmpty()) {
+                if (item.url().isBlank()) {
                     continue;
                 }
 
-                var link = item.getLink().get();
+                var link = item.url();
                 if (!link.contains("#")) {
                     continue;
                 }

code/functions/live-capture/java/nu/marginalia/rss/svc/FeedJournal.java

@@ -0,0 +1,76 @@
+package nu.marginalia.rss.svc;
+
+import nu.marginalia.WmsaHome;
+import nu.marginalia.slop.SlopTable;
+import nu.marginalia.slop.column.string.StringColumn;
+import nu.marginalia.slop.desc.StorageType;
+import org.apache.commons.io.FileUtils;
+
+import java.io.IOException;
+import java.nio.charset.StandardCharsets;
+import java.nio.file.Files;
+import java.nio.file.Path;
+import java.util.function.BiConsumer;
+
+/** Utility for recording fetched feeds to a journal, useful in debugging feed parser issues.
+ */
+public interface FeedJournal extends AutoCloseable {
+    StringColumn urlColumn = new StringColumn("url");
+    StringColumn contentsColumn = new StringColumn("contents", StandardCharsets.UTF_8, StorageType.ZSTD);
+
+    void record(String url, String contents) throws IOException;
+    void close() throws IOException;
+
+
+    static FeedJournal create() throws IOException {
+        if (Boolean.getBoolean("feedFetcher.persistJournal")) {
+            Path journalPath = WmsaHome.getDataPath().resolve("feed-journal");
+            if (Files.isDirectory(journalPath)) {
+                FileUtils.deleteDirectory(journalPath.toFile());
+            }
+            Files.createDirectories(journalPath);
+            return new RecordingFeedJournal(journalPath);
+        }
+        else {
+            return new NoOpFeedJournal();
+        }
+    }
+
+    class NoOpFeedJournal implements FeedJournal {
+        @Override
+        public void record(String url, String contents) {}
+
+        @Override
+        public void close() {}
+    }
+
+    class RecordingFeedJournal extends SlopTable implements FeedJournal {
+
+        private final StringColumn.Writer urlWriter;
+        private final StringColumn.Writer contentsWriter;
+
+        public RecordingFeedJournal(Path path) throws IOException {
+            super(path, SlopTable.getNumPages(path, FeedJournal.urlColumn));
+
+            urlWriter = urlColumn.create(this);
+            contentsWriter = contentsColumn.create(this);
+        }
+
+        public synchronized void record(String url, String contents) throws IOException {
+            urlWriter.put(url);
+            contentsWriter.put(contents);
+        }
+    }
+
+    static void replay(Path journalPath, BiConsumer<String, String> urlAndContent) throws IOException {
+        try (SlopTable table = new SlopTable(journalPath)) {
+            final StringColumn.Reader urlReader = urlColumn.open(table);
+            final StringColumn.Reader contentsReader = contentsColumn.open(table);
+
+            while (urlReader.hasRemaining()) {
+                urlAndContent.accept(urlReader.get(), contentsReader.get());
+            }
+        }
+
+    }
+}

code/functions/live-capture/java/nu/marginalia/rss/svc/SimpleFeedParser.java

@@ -0,0 +1,102 @@
+package nu.marginalia.rss.svc;
+
+import com.apptasticsoftware.rssreader.DateTimeParser;
+import com.apptasticsoftware.rssreader.util.Default;
+import org.jsoup.Jsoup;
+import org.jsoup.parser.Parser;
+
+import java.time.ZonedDateTime;
+import java.util.ArrayList;
+import java.util.List;
+import java.util.Optional;
+
+public class SimpleFeedParser {
+
+    private static final DateTimeParser dateTimeParser = Default.getDateTimeParser();
+
+    public record ItemData (
+            String title,
+            String description,
+            String url,
+            String pubDate
+    ) {
+        public boolean isWellFormed() {
+            return title != null && !title.isBlank() &&
+                    description != null && !description.isBlank() &&
+                    url != null && !url.isBlank() &&
+                    pubDate != null && !pubDate.isBlank();
+        }
+
+        public Optional<ZonedDateTime> getPubDateZonedDateTime() {
+            try {
+                return Optional.ofNullable(dateTimeParser.parse(pubDate()));
+            }
+            catch (Exception e) {
+                return Optional.empty();
+            }
+        }
+
+    }
+
+    public static List<ItemData> parse(String content) {
+        var doc = Jsoup.parse(content, Parser.xmlParser());
+        List<ItemData> ret = new ArrayList<>();
+
+        doc.select("item, entry").forEach(element -> {
+            String link = "";
+            String title = "";
+            String description = "";
+            String pubDate = "";
+
+            for (String attr : List.of("title", "dc:title")) {
+                if (!title.isBlank())
+                    break;
+                var tag = element.getElementsByTag(attr).first();
+                if (tag != null) {
+                    title = tag.text();
+                }
+            }
+
+            for (String attr : List.of("title", "summary", "content", "description", "dc:description")) {
+                if (!description.isBlank())
+                    break;
+                var tag = element.getElementsByTag(attr).first();
+                if (tag != null) {
+                    description = tag.text();
+                }
+            }
+
+            for (String attr : List.of("pubDate", "published", "updated", "issued", "created", "dc:date")) {
+                if (!pubDate.isBlank())
+                    break;
+                var tag = element.getElementsByTag(attr).first();
+                if (tag != null) {
+                    pubDate = tag.text();
+                }
+            }
+
+            for (String attr : List.of("link", "url")) {
+                if (!link.isBlank())
+                    break;
+                var tag = element.getElementsByTag(attr).first();
+
+                if (tag != null) {
+                    String linkText = tag.text();
+
+                    if (linkText.isBlank()) {
+                        linkText = tag.attr("href");
+                    }
+
+                    link = linkText;
+                }
+
+            }
+
+            ret.add(new ItemData(title, description, link, pubDate));
+        });
+
+
+        return ret;
+    }
+
+}

code/functions/live-capture/test-resources/nlnet.atom

@@ -0,0 +1,27 @@
+<feed xmlns="http://www.w3.org/2005/Atom" xml:base="https://nlnet.nl">
+  <title type="text">NLnet news</title>
+  <updated>2025-01-01T00:00:00Z</updated>
+  <id>https://nlnet.nl/feed.atom</id>
+  <link rel="self" type="application/atom+xml" href="https://nlnet.nl/feed.atom"/>
+  <entry>
+    <id>https://nlnet.nl/news/2025/20250101-announcing-grantees-June-call.html</id>
+    <author>
+      <name>NLnet</name>
+    </author>
+    <title type="xhtml">
+      <div xmlns="http://www.w3.org/1999/xhtml">50 Free and Open Source Projects Selected for NGI Zero grants</div>
+    </title>
+    <link href="/news/2025/20250101-announcing-grantees-June-call.html"/>
+    <updated>2025-01-01T00:00:00Z</updated>
+    <content type="xhtml">
+      <div xmlns="http://www.w3.org/1999/xhtml">
+        <p class="paralead">Happy 2025 everyone! On this first day of the fresh new year we are happy to announce 50 project teams were selected to receive NGI Zero grants. We are welcoming projects from 18 countries involving people and organisations of various types: individuals, associations, small and medium enterprises, foundations, universities, and informal collectives. The new projects are all across the different layers of the NGI technology stack: from trustworthy open hardware to services &amp; applications which provide autonomy for end-users.</p>
+        <p>The 50 free and open source projects were selected across two funds. 19 teams will receive grants from the <a href="/commonsfund/">NGI Zero Commons Fund</a>, a broadly themed fund that supports people working on reclaiming the public nature of the internet. The other 31 projects will work within <a href="/core/">NGI Zero Core</a> which focuses on strengthening the open internet architecture. Both funds offer financial and practical support. The latter consisting of <a href="/NGI0/services/">support services</a> such as accessibility and security audits, advice on license compliance, help with testing, documentation or UX design.</p>
+        <h2>If you applied for a grant</h2>
+        <p>This is the selection for the <a href="https://nlnet.nl/news/2024/20240401-call.html">June call</a>. We always inform <em>all</em> applicants about the outcome of the review ahead of the public announcement, if the are selected or not. If you have not heard anything, you probably applied to a later call that is still under review. You can see which call you applied to by checking the application number assigned to the project when you applied. The second number in the sequence refers to the month of the call, so 06 in the case of the June call. (It should not happen, but if you did apply to the June call and did not hear anything, do contact us.)</p>
+        <h2>Meet the new projects!</h2>
+      </div>
+    </content>
+  </entry>
+
+</feed>

code/functions/live-capture/test/nu/marginalia/livecapture/BrowserlessClientTest.java

@@ -1,36 +1,97 @@
 package nu.marginalia.livecapture;
 
+import com.github.tomakehurst.wiremock.WireMockServer;
+import com.github.tomakehurst.wiremock.core.WireMockConfiguration;
+import nu.marginalia.WmsaHome;
+import nu.marginalia.service.module.ServiceConfigurationModule;
 import org.junit.jupiter.api.Assertions;
 import org.junit.jupiter.api.BeforeAll;
+import org.junit.jupiter.api.Tag;
 import org.junit.jupiter.api.Test;
 import org.testcontainers.containers.GenericContainer;
 import org.testcontainers.junit.jupiter.Testcontainers;
 import org.testcontainers.utility.DockerImageName;
 
+import java.io.IOException;
 import java.net.URI;
+import java.util.Map;
+
+import static com.github.tomakehurst.wiremock.client.WireMock.*;
+
 
 @Testcontainers
+@Tag("slow")
 public class BrowserlessClientTest {
-    static GenericContainer<?> container = new GenericContainer<>(DockerImageName.parse("browserless/chrome")).withExposedPorts(3000);
+    static GenericContainer<?> container = new GenericContainer<>(DockerImageName.parse("browserless/chrome"))
+            .withEnv(Map.of("TOKEN", "BROWSERLESS_TOKEN"))
+            .withNetworkMode("bridge")
+            .withExposedPorts(3000);
+
+    static WireMockServer wireMockServer =
+            new WireMockServer(WireMockConfiguration.wireMockConfig()
+                    .port(18089));
+
+    static String localIp;
+
+    static URI browserlessURI;
 
     @BeforeAll
-    public static void setup() {
+    public static void setup() throws IOException {
         container.start();
+
+        browserlessURI = URI.create(String.format("http://%s:%d/",
+                container.getHost(),
+                container.getMappedPort(3000))
+        );
+
+        wireMockServer.start();
+        wireMockServer.stubFor(get("/").willReturn(aResponse().withStatus(200).withBody("Ok")));
+
+        localIp = ServiceConfigurationModule.getLocalNetworkIP();
+
+    }
+
+    @Tag("flaky")
+    @Test
+    public void testInspectContentUA__Flaky() throws Exception {
+        try (var client = new BrowserlessClient(browserlessURI)) {
+            client.content("http://" + localIp + ":18089/",
+                    BrowserlessClient.GotoOptions.defaultValues()
+            );
+        }
+
+        wireMockServer.verify(getRequestedFor(urlEqualTo("/")).withHeader("User-Agent", equalTo(WmsaHome.getUserAgent().uaString())));
+    }
+
+    @Tag("flaky")
+    @Test
+    public void testInspectScreenshotUA__Flaky() throws Exception {
+        try (var client = new BrowserlessClient(browserlessURI)) {
+            client.screenshot("http://" + localIp + ":18089/",
+                    BrowserlessClient.GotoOptions.defaultValues(),
+                    BrowserlessClient.ScreenshotOptions.defaultValues()
+            );
+        }
+
+        wireMockServer.verify(getRequestedFor(urlEqualTo("/")).withHeader("User-Agent", equalTo(WmsaHome.getUserAgent().uaString())));
     }
 
     @Test
     public void testContent() throws Exception {
-        try (var client = new BrowserlessClient(URI.create("http://" + container.getHost() + ":" + container.getMappedPort(3000)))) {
+        try (var client = new BrowserlessClient(browserlessURI)) {
-            var content = client.content("https://www.marginalia.nu/", BrowserlessClient.GotoOptions.defaultValues());
+            var content = client.content("https://www.marginalia.nu/", BrowserlessClient.GotoOptions.defaultValues()).orElseThrow();
-            Assertions.assertNotNull(content, "Content should not be null");
+
             Assertions.assertFalse(content.isBlank(), "Content should not be empty");
         }
     }
 
     @Test
     public void testScreenshot() throws Exception {
-        try (var client = new BrowserlessClient(URI.create("http://" + container.getHost() + ":" + container.getMappedPort(3000)))) {
+        try (var client = new BrowserlessClient(browserlessURI)) {
-            var screenshot = client.screenshot("https://www.marginalia.nu/", BrowserlessClient.GotoOptions.defaultValues(), BrowserlessClient.ScreenshotOptions.defaultValues());
+            var screenshot = client.screenshot("https://www.marginalia.nu/",
+                    BrowserlessClient.GotoOptions.defaultValues(),
+                    BrowserlessClient.ScreenshotOptions.defaultValues());
+
             Assertions.assertNotNull(screenshot, "Screenshot should not be null");
         }
     }

code/functions/live-capture/test/nu/marginalia/rss/svc/TestXmlSanitization.java

@@ -1,30 +0,0 @@
-package nu.marginalia.rss.svc;
-
-import org.junit.jupiter.api.Assertions;
-import org.junit.jupiter.api.Test;
-
-public class TestXmlSanitization {
-
-    @Test
-    public void testPreservedEntities() {
-        Assertions.assertEquals("&", FeedFetcherService.sanitizeEntities("&"));
-        Assertions.assertEquals("<", FeedFetcherService.sanitizeEntities("<"));
-        Assertions.assertEquals(">", FeedFetcherService.sanitizeEntities(">"));
-        Assertions.assertEquals("'", FeedFetcherService.sanitizeEntities("'"));
-    }
-
-    @Test
-    public void testStrayAmpersand() {
-        Assertions.assertEquals("Bed & Breakfast", FeedFetcherService.sanitizeEntities("Bed & Breakfast"));
-    }
-
-    @Test
-    public void testTranslatedHtmlEntity() {
-        Assertions.assertEquals("Foo -- Bar", FeedFetcherService.sanitizeEntities("Foo — Bar"));
-    }
-
-    @Test
-    public void testTranslatedHtmlEntityQuot() {
-        Assertions.assertEquals("\"Bob\"", FeedFetcherService.sanitizeEntities(""Bob""));
-    }
-}

code/functions/search-query/api/java/nu/marginalia/api/searchquery/IndexProtobufCodec.java

@@ -2,9 +2,6 @@ package nu.marginalia.api.searchquery;
 
 import nu.marginalia.api.searchquery.model.query.SearchPhraseConstraint;
 import nu.marginalia.api.searchquery.model.query.SearchQuery;
-import nu.marginalia.api.searchquery.model.results.Bm25Parameters;
-import nu.marginalia.api.searchquery.model.results.ResultRankingParameters;
-import nu.marginalia.index.query.limit.QueryLimits;
 import nu.marginalia.index.query.limit.SpecificationLimit;
 import nu.marginalia.index.query.limit.SpecificationLimitType;
 
@@ -27,37 +24,19 @@ public class IndexProtobufCodec {
                 .build();
     }
 
-    public static  QueryLimits convertQueryLimits(RpcQueryLimits queryLimits) {
-        return new QueryLimits(
-                queryLimits.getResultsByDomain(),
-                queryLimits.getResultsTotal(),
-                queryLimits.getTimeoutMs(),
-                queryLimits.getFetchSize()
-        );
-    }
-
-    public static RpcQueryLimits convertQueryLimits(QueryLimits queryLimits) {
-        return RpcQueryLimits.newBuilder()
-                .setResultsByDomain(queryLimits.resultsByDomain())
-                .setResultsTotal(queryLimits.resultsTotal())
-                .setTimeoutMs(queryLimits.timeoutMs())
-                .setFetchSize(queryLimits.fetchSize())
-                .build();
-    }
-
     public static SearchQuery convertRpcQuery(RpcQuery query) {
-        List<SearchPhraseConstraint> phraeConstraints = new ArrayList<>();
+        List<SearchPhraseConstraint> phraseConstraints = new ArrayList<>();
 
         for (int j = 0; j < query.getPhrasesCount(); j++) {
             var coh = query.getPhrases(j);
             if (coh.getType() == RpcPhrases.TYPE.OPTIONAL) {
-                phraeConstraints.add(new SearchPhraseConstraint.Optional(List.copyOf(coh.getTermsList())));
+                phraseConstraints.add(new SearchPhraseConstraint.Optional(List.copyOf(coh.getTermsList())));
             }
             else if (coh.getType() == RpcPhrases.TYPE.MANDATORY) {
-                phraeConstraints.add(new SearchPhraseConstraint.Mandatory(List.copyOf(coh.getTermsList())));
+                phraseConstraints.add(new SearchPhraseConstraint.Mandatory(List.copyOf(coh.getTermsList())));
             }
             else if (coh.getType() == RpcPhrases.TYPE.FULL) {
-                phraeConstraints.add(new SearchPhraseConstraint.Full(List.copyOf(coh.getTermsList())));
+                phraseConstraints.add(new SearchPhraseConstraint.Full(List.copyOf(coh.getTermsList())));
             }
             else {
                 throw new IllegalArgumentException("Unknown phrase constraint type: " + coh.getType());
@@ -70,7 +49,7 @@ public class IndexProtobufCodec {
                 query.getExcludeList(),
                 query.getAdviceList(),
                 query.getPriorityList(),
-                phraeConstraints
+                phraseConstraints
         );
     }
 
@@ -103,60 +82,4 @@ public class IndexProtobufCodec {
         return subqueryBuilder.build();
     }
 
-    public static ResultRankingParameters convertRankingParameterss(RpcResultRankingParameters params) {
-        if (params == null)
-            return ResultRankingParameters.sensibleDefaults();
-
-        return new ResultRankingParameters(
-                new Bm25Parameters(params.getBm25K(), params.getBm25B()),
-                params.getShortDocumentThreshold(),
-                params.getShortDocumentPenalty(),
-                params.getDomainRankBonus(),
-                params.getQualityPenalty(),
-                params.getShortSentenceThreshold(),
-                params.getShortSentencePenalty(),
-                params.getBm25Weight(),
-                params.getTcfFirstPositionWeight(),
-                params.getTcfVerbatimWeight(),
-                params.getTcfProximityWeight(),
-                ResultRankingParameters.TemporalBias.valueOf(params.getTemporalBias().getBias().name()),
-                params.getTemporalBiasWeight(),
-                params.getExportDebugData()
-        );
-    }
-
-    public static RpcResultRankingParameters convertRankingParameterss(ResultRankingParameters rankingParams,
-                                                                       RpcTemporalBias temporalBias)
-    {
-        if (rankingParams == null) {
-            rankingParams = ResultRankingParameters.sensibleDefaults();
-        }
-
-        var builder = RpcResultRankingParameters.newBuilder()
-                        .setBm25B(rankingParams.bm25Params.b())
-                        .setBm25K(rankingParams.bm25Params.k())
-                        .setShortDocumentThreshold(rankingParams.shortDocumentThreshold)
-                        .setShortDocumentPenalty(rankingParams.shortDocumentPenalty)
-                        .setDomainRankBonus(rankingParams.domainRankBonus)
-                        .setQualityPenalty(rankingParams.qualityPenalty)
-                        .setShortSentenceThreshold(rankingParams.shortSentenceThreshold)
-                        .setShortSentencePenalty(rankingParams.shortSentencePenalty)
-                        .setBm25Weight(rankingParams.bm25Weight)
-                        .setTcfFirstPositionWeight(rankingParams.tcfFirstPosition)
-                        .setTcfProximityWeight(rankingParams.tcfProximity)
-                        .setTcfVerbatimWeight(rankingParams.tcfVerbatim)
-                        .setTemporalBiasWeight(rankingParams.temporalBiasWeight)
-                        .setExportDebugData(rankingParams.exportDebugData);
-
-        if (temporalBias != null && temporalBias.getBias() != RpcTemporalBias.Bias.NONE) {
-            builder.setTemporalBias(temporalBias);
-        }
-        else {
-            builder.setTemporalBias(RpcTemporalBias.newBuilder()
-                    .setBias(RpcTemporalBias.Bias.valueOf(rankingParams.temporalBias.name())));
-        }
-
-        return builder.build();
-    }
-
 }

code/functions/search-query/api/java/nu/marginalia/api/searchquery/QueryProtobufCodec.java

@@ -5,7 +5,7 @@ import nu.marginalia.api.searchquery.model.query.QueryParams;
 import nu.marginalia.api.searchquery.model.query.QueryResponse;
 import nu.marginalia.api.searchquery.model.query.SearchSpecification;
 import nu.marginalia.api.searchquery.model.results.DecoratedSearchResultItem;
-import nu.marginalia.api.searchquery.model.results.ResultRankingParameters;
+import nu.marginalia.api.searchquery.model.results.PrototypeRankingParameters;
 import nu.marginalia.api.searchquery.model.results.SearchResultItem;
 import nu.marginalia.api.searchquery.model.results.SearchResultKeywordScore;
 import nu.marginalia.api.searchquery.model.results.debug.DebugFactor;
@@ -37,7 +37,7 @@ public class QueryProtobufCodec {
         builder.setSize(IndexProtobufCodec.convertSpecLimit(query.specs.size));
         builder.setRank(IndexProtobufCodec.convertSpecLimit(query.specs.rank));
 
-        builder.setQueryLimits(IndexProtobufCodec.convertQueryLimits(query.specs.queryLimits));
+        builder.setQueryLimits(query.specs.queryLimits);
 
         // Query strategy may be overridden by the query, but if not, use the one from the request
         if (query.specs.queryStrategy != null && query.specs.queryStrategy != QueryStrategy.AUTO)
@@ -45,9 +45,27 @@ public class QueryProtobufCodec {
         else
             builder.setQueryStrategy(request.getQueryStrategy());
 
-        if (query.specs.rankingParams != null) {
+        if (request.getTemporalBias().getBias() != RpcTemporalBias.Bias.NONE) {
-            builder.setParameters(IndexProtobufCodec.convertRankingParameterss(query.specs.rankingParams, request.getTemporalBias()));
+            if (query.specs.rankingParams != null) {
+                builder.setParameters(
+                        RpcResultRankingParameters.newBuilder(query.specs.rankingParams)
+                                .setTemporalBias(request.getTemporalBias())
+                                .build()
+                );
+            } else {
+                builder.setParameters(
+                        RpcResultRankingParameters.newBuilder(PrototypeRankingParameters.sensibleDefaults())
+                                .setTemporalBias(request.getTemporalBias())
+                                .build()
+                );
+            }
+        } else if (query.specs.rankingParams != null) {
+            builder.setParameters(query.specs.rankingParams);
         }
+        // else {
+        // if we have no ranking params, we don't need to set them, the client check and use the default values
+        // so we don't need to send this huge object over the wire
+        // }
 
         return builder.build();
     }
@@ -65,18 +83,13 @@ public class QueryProtobufCodec {
         builder.setSize(IndexProtobufCodec.convertSpecLimit(query.specs.size));
         builder.setRank(IndexProtobufCodec.convertSpecLimit(query.specs.rank));
 
-        builder.setQueryLimits(IndexProtobufCodec.convertQueryLimits(query.specs.queryLimits));
+        builder.setQueryLimits(query.specs.queryLimits);
 
         // Query strategy may be overridden by the query, but if not, use the one from the request
         builder.setQueryStrategy(query.specs.queryStrategy.name());
 
         if (query.specs.rankingParams != null) {
-            builder.setParameters(IndexProtobufCodec.convertRankingParameterss(
+            builder.setParameters(query.specs.rankingParams);
-                    query.specs.rankingParams,
-                    RpcTemporalBias.newBuilder().setBias(
-                                    RpcTemporalBias.Bias.NONE)
-                            .build())
-            );
         }
 
         return builder.build();
@@ -95,10 +108,10 @@ public class QueryProtobufCodec {
                 IndexProtobufCodec.convertSpecLimit(request.getSize()),
                 IndexProtobufCodec.convertSpecLimit(request.getRank()),
                 request.getDomainIdsList(),
-                IndexProtobufCodec.convertQueryLimits(request.getQueryLimits()),
+                request.getQueryLimits(),
                 request.getSearchSetIdentifier(),
                 QueryStrategy.valueOf(request.getQueryStrategy()),
-                ResultRankingParameters.TemporalBias.valueOf(request.getTemporalBias().getBias().name()),
+                RpcTemporalBias.Bias.valueOf(request.getTemporalBias().getBias().name()),
                 request.getPagination().getPage()
         );
     }
@@ -294,9 +307,9 @@ public class QueryProtobufCodec {
                 IndexProtobufCodec.convertSpecLimit(specs.getYear()),
                 IndexProtobufCodec.convertSpecLimit(specs.getSize()),
                 IndexProtobufCodec.convertSpecLimit(specs.getRank()),
-                IndexProtobufCodec.convertQueryLimits(specs.getQueryLimits()),
+                specs.getQueryLimits(),
                 QueryStrategy.valueOf(specs.getQueryStrategy()),
-                IndexProtobufCodec.convertRankingParameterss(specs.getParameters())
+                specs.hasParameters() ? specs.getParameters() : null
         );
     }
 
@@ -307,7 +320,7 @@ public class QueryProtobufCodec {
                 .addAllTacitExcludes(params.tacitExcludes())
                 .addAllTacitPriority(params.tacitPriority())
                 .setHumanQuery(params.humanQuery())
-                .setQueryLimits(IndexProtobufCodec.convertQueryLimits(params.limits()))
+                .setQueryLimits(params.limits())
                 .setQuality(IndexProtobufCodec.convertSpecLimit(params.quality()))
                 .setYear(IndexProtobufCodec.convertSpecLimit(params.year()))
                 .setSize(IndexProtobufCodec.convertSpecLimit(params.size()))
@@ -319,7 +332,7 @@ public class QueryProtobufCodec {
                         .build())
                 .setPagination(RpcQsQueryPagination.newBuilder()
                         .setPage(params.page())
-                        .setPageSize(Math.min(100, params.limits().resultsTotal()))
+                        .setPageSize(Math.min(100, params.limits().getResultsTotal()))
                         .build());
 
         if (params.nearDomain() != null)

code/functions/search-query/api/java/nu/marginalia/api/searchquery/model/query/QueryParams.java

@@ -1,7 +1,7 @@
 package nu.marginalia.api.searchquery.model.query;
 
-import nu.marginalia.api.searchquery.model.results.ResultRankingParameters;
+import nu.marginalia.api.searchquery.RpcQueryLimits;
-import nu.marginalia.index.query.limit.QueryLimits;
+import nu.marginalia.api.searchquery.RpcTemporalBias;
 import nu.marginalia.index.query.limit.QueryStrategy;
 import nu.marginalia.index.query.limit.SpecificationLimit;
 
@@ -21,14 +21,14 @@ public record QueryParams(
         SpecificationLimit size,
         SpecificationLimit rank,
         List<Integer> domainIds,
-        QueryLimits limits,
+        RpcQueryLimits limits,
         String identifier,
         QueryStrategy queryStrategy,
-        ResultRankingParameters.TemporalBias temporalBias,
+        RpcTemporalBias.Bias temporalBias,
         int page
 )
 {
-    public QueryParams(String query, QueryLimits limits, String identifier) {
+    public QueryParams(String query, RpcQueryLimits limits, String identifier) {
         this(query, null,
                 List.of(),
                 List.of(),
@@ -42,7 +42,7 @@ public record QueryParams(
                 limits,
                 identifier,
                 QueryStrategy.AUTO,
-                ResultRankingParameters.TemporalBias.NONE,
+                RpcTemporalBias.Bias.NONE,
                 1 // page
                 );
     }

code/functions/search-query/api/java/nu/marginalia/api/searchquery/model/query/SearchSpecification.java

@@ -1,10 +1,11 @@
 package nu.marginalia.api.searchquery.model.query;
 
-import nu.marginalia.api.searchquery.model.results.ResultRankingParameters;
+import nu.marginalia.api.searchquery.RpcQueryLimits;
-import nu.marginalia.index.query.limit.QueryLimits;
+import nu.marginalia.api.searchquery.RpcResultRankingParameters;
 import nu.marginalia.index.query.limit.QueryStrategy;
 import nu.marginalia.index.query.limit.SpecificationLimit;
 
+import javax.annotation.Nullable;
 import java.util.List;
 
 public class SearchSpecification {
@@ -24,11 +25,12 @@ public class SearchSpecification {
     public SpecificationLimit size;
     public SpecificationLimit rank;
 
-    public final QueryLimits queryLimits;
+    public final RpcQueryLimits queryLimits;
 
     public final QueryStrategy queryStrategy;
 
-    public final ResultRankingParameters rankingParams;
+    @Nullable
+    public final RpcResultRankingParameters rankingParams;
 
     public SearchSpecification(SearchQuery query,
                                List<Integer> domains,
@@ -38,9 +40,9 @@ public class SearchSpecification {
                                SpecificationLimit year,
                                SpecificationLimit size,
                                SpecificationLimit rank,
-                               QueryLimits queryLimits,
+                               RpcQueryLimits queryLimits,
                                QueryStrategy queryStrategy,
-                               ResultRankingParameters rankingParams)
+                               @Nullable RpcResultRankingParameters rankingParams)
     {
         this.query = query;
         this.domains = domains;
@@ -91,7 +93,7 @@ public class SearchSpecification {
         return this.rank;
     }
 
-    public QueryLimits getQueryLimits() {
+    public RpcQueryLimits getQueryLimits() {
         return this.queryLimits;
     }
 
@@ -99,7 +101,7 @@ public class SearchSpecification {
         return this.queryStrategy;
     }
 
-    public ResultRankingParameters getRankingParams() {
+    public RpcResultRankingParameters getRankingParams() {
         return this.rankingParams;
     }
 
@@ -120,9 +122,9 @@ public class SearchSpecification {
         private boolean size$set;
         private SpecificationLimit rank$value;
         private boolean rank$set;
-        private QueryLimits queryLimits;
+        private RpcQueryLimits queryLimits;
         private QueryStrategy queryStrategy;
-        private ResultRankingParameters rankingParams;
+        private RpcResultRankingParameters rankingParams;
 
         SearchSpecificationBuilder() {
         }
@@ -171,7 +173,7 @@ public class SearchSpecification {
             return this;
         }
 
-        public SearchSpecificationBuilder queryLimits(QueryLimits queryLimits) {
+        public SearchSpecificationBuilder queryLimits(RpcQueryLimits queryLimits) {
             this.queryLimits = queryLimits;
             return this;
         }
@@ -181,7 +183,7 @@ public class SearchSpecification {
             return this;
         }
 
-        public SearchSpecificationBuilder rankingParams(ResultRankingParameters rankingParams) {
+        public SearchSpecificationBuilder rankingParams(RpcResultRankingParameters rankingParams) {
             this.rankingParams = rankingParams;
             return this;
         }

code/functions/search-query/api/java/nu/marginalia/api/searchquery/model/results/DecoratedSearchResultItem.java

@@ -1,6 +1,7 @@
 package nu.marginalia.api.searchquery.model.results;
 
 import nu.marginalia.api.searchquery.model.results.debug.ResultRankingDetails;
+import nu.marginalia.model.DocumentFormat;
 import nu.marginalia.model.EdgeUrl;
 import org.jetbrains.annotations.NotNull;
 
@@ -161,4 +162,14 @@ public class DecoratedSearchResultItem implements Comparable<DecoratedSearchResu
     public String toString() {
         return "DecoratedSearchResultItem(rawIndexResult=" + this.getRawIndexResult() + ", url=" + this.getUrl() + ", title=" + this.getTitle() + ", description=" + this.getDescription() + ", urlQuality=" + this.getUrlQuality() + ", format=" + this.getFormat() + ", features=" + this.getFeatures() + ", pubYear=" + this.getPubYear() + ", dataHash=" + this.getDataHash() + ", wordsTotal=" + this.getWordsTotal() + ", bestPositions=" + this.getBestPositions() + ", rankingScore=" + this.getRankingScore() + ", resultsFromDomain=" + this.getResultsFromDomain() + ", rankingDetails=" + this.getRankingDetails() + ")";
     }
+
+    public String getShortFormat() {
+        try {
+            var df = DocumentFormat.valueOf(format);
+            return df.shortFormat;
+        }
+        catch (IllegalArgumentException e) {
+            return DocumentFormat.UNKNOWN.shortFormat;
+        }
+    }
 }

code/functions/search-query/api/java/nu/marginalia/api/searchquery/model/results/PrototypeRankingParameters.java

@@ -0,0 +1,33 @@
+package nu.marginalia.api.searchquery.model.results;
+
+import nu.marginalia.api.searchquery.RpcResultRankingParameters;
+import nu.marginalia.api.searchquery.RpcTemporalBias;
+
+public class PrototypeRankingParameters {
+
+    /** These are the default ranking parameters that are used when no parameters are specified. */
+
+    private static final RpcResultRankingParameters _sensibleDefaults = RpcResultRankingParameters.newBuilder()
+                .setBm25B(0.5)
+                .setBm25K(1.2)
+                .setShortDocumentThreshold(2000)
+                .setShortDocumentPenalty(2.)
+                .setDomainRankBonus(1 / 100.)
+                .setQualityPenalty(1 / 15.)
+                .setShortSentenceThreshold(2)
+                .setShortSentencePenalty(5)
+                .setBm25Weight(1.)
+                .setTcfVerbatimWeight(1.)
+                .setTcfProximityWeight(1.)
+                .setTcfFirstPositionWeight(5)
+                .setTemporalBias(RpcTemporalBias.newBuilder().setBias(RpcTemporalBias.Bias.NONE))
+                .setTemporalBiasWeight(5.0)
+                .setExportDebugData(false)
+                .setDisablePenalties(false)
+                .build();
+
+    public static RpcResultRankingParameters sensibleDefaults() {
+        return _sensibleDefaults;
+    }
+
+}

code/functions/search-query/api/java/nu/marginalia/api/searchquery/model/results/ResultRankingContext.java

@@ -1,12 +1,13 @@
 package nu.marginalia.api.searchquery.model.results;
 
+import nu.marginalia.api.searchquery.RpcResultRankingParameters;
 import nu.marginalia.api.searchquery.model.compiled.CqDataInt;
 
 import java.util.BitSet;
 
 public class ResultRankingContext {
     private final int docCount;
-    public final ResultRankingParameters params;
+    public final RpcResultRankingParameters params;
 
 
     public final BitSet regularMask;
@@ -21,7 +22,7 @@ public class ResultRankingContext {
     public final CqDataInt priorityCounts;
 
     public ResultRankingContext(int docCount,
-                                ResultRankingParameters params,
+                                RpcResultRankingParameters params,
                                 BitSet ngramsMask,
                                 BitSet regularMask,
                                 CqDataInt fullCounts,

code/functions/search-query/api/java/nu/marginalia/api/searchquery/model/results/ResultRankingParameters.java

@@ -1,278 +0,0 @@
-package nu.marginalia.api.searchquery.model.results;
-
-import java.util.Objects;
-
-public class ResultRankingParameters {
-
-    /**
-     * Tuning for BM25 when applied to full document matches
-     */
-    public final Bm25Parameters bm25Params;
-
-    /**
-     * Documents below this length are penalized
-     */
-    public int shortDocumentThreshold;
-
-    public double shortDocumentPenalty;
-
-
-    /**
-     * Scaling factor associated with domain rank (unscaled rank value is 0-255; high is good)
-     */
-    public double domainRankBonus;
-
-    /**
-     * Scaling factor associated with document quality (unscaled rank value is 0-15; high is bad)
-     */
-    public double qualityPenalty;
-
-    /**
-     * Average sentence length values below this threshold are penalized, range [0-4), 2 or 3 is probably what you want
-     */
-    public int shortSentenceThreshold;
-
-    /**
-     * Magnitude of penalty for documents with low average sentence length
-     */
-    public double shortSentencePenalty;
-
-    public double bm25Weight;
-    public double tcfFirstPosition;
-    public double tcfVerbatim;
-    public double tcfProximity;
-
-    public TemporalBias temporalBias;
-    public double temporalBiasWeight;
-
-    public boolean exportDebugData;
-
-    public ResultRankingParameters(Bm25Parameters bm25Params, int shortDocumentThreshold, double shortDocumentPenalty, double domainRankBonus, double qualityPenalty, int shortSentenceThreshold, double shortSentencePenalty, double bm25Weight, double tcfFirstPosition, double tcfVerbatim, double tcfProximity, TemporalBias temporalBias, double temporalBiasWeight, boolean exportDebugData) {
-        this.bm25Params = bm25Params;
-        this.shortDocumentThreshold = shortDocumentThreshold;
-        this.shortDocumentPenalty = shortDocumentPenalty;
-        this.domainRankBonus = domainRankBonus;
-        this.qualityPenalty = qualityPenalty;
-        this.shortSentenceThreshold = shortSentenceThreshold;
-        this.shortSentencePenalty = shortSentencePenalty;
-        this.bm25Weight = bm25Weight;
-        this.tcfFirstPosition = tcfFirstPosition;
-        this.tcfVerbatim = tcfVerbatim;
-        this.tcfProximity = tcfProximity;
-        this.temporalBias = temporalBias;
-        this.temporalBiasWeight = temporalBiasWeight;
-        this.exportDebugData = exportDebugData;
-    }
-
-    public static ResultRankingParameters sensibleDefaults() {
-        return builder()
-                .bm25Params(new Bm25Parameters(1.2, 0.5))
-                .shortDocumentThreshold(2000)
-                .shortDocumentPenalty(2.)
-                .domainRankBonus(1 / 100.)
-                .qualityPenalty(1 / 15.)
-                .shortSentenceThreshold(2)
-                .shortSentencePenalty(5)
-                .bm25Weight(1.)
-                .tcfVerbatim(1.)
-                .tcfProximity(1.)
-                .tcfFirstPosition(5)
-                .temporalBias(TemporalBias.NONE)
-                .temporalBiasWeight(5.0)
-                .exportDebugData(false)
-                .build();
-    }
-
-    public static ResultRankingParametersBuilder builder() {
-        return new ResultRankingParametersBuilder();
-    }
-
-    public Bm25Parameters getBm25Params() {
-        return this.bm25Params;
-    }
-
-    public int getShortDocumentThreshold() {
-        return this.shortDocumentThreshold;
-    }
-
-    public double getShortDocumentPenalty() {
-        return this.shortDocumentPenalty;
-    }
-
-    public double getDomainRankBonus() {
-        return this.domainRankBonus;
-    }
-
-    public double getQualityPenalty() {
-        return this.qualityPenalty;
-    }
-
-    public int getShortSentenceThreshold() {
-        return this.shortSentenceThreshold;
-    }
-
-    public double getShortSentencePenalty() {
-        return this.shortSentencePenalty;
-    }
-
-    public double getBm25Weight() {
-        return this.bm25Weight;
-    }
-
-    public double getTcfFirstPosition() {
-        return this.tcfFirstPosition;
-    }
-
-    public double getTcfVerbatim() {
-        return this.tcfVerbatim;
-    }
-
-    public double getTcfProximity() {
-        return this.tcfProximity;
-    }
-
-    public TemporalBias getTemporalBias() {
-        return this.temporalBias;
-    }
-
-    public double getTemporalBiasWeight() {
-        return this.temporalBiasWeight;
-    }
-
-    public boolean isExportDebugData() {
-        return this.exportDebugData;
-    }
-
-    @Override
-    public final boolean equals(Object o) {
-        if (this == o) return true;
-        if (!(o instanceof ResultRankingParameters that)) return false;
-
-        return shortDocumentThreshold == that.shortDocumentThreshold && Double.compare(shortDocumentPenalty, that.shortDocumentPenalty) == 0 && Double.compare(domainRankBonus, that.domainRankBonus) == 0 && Double.compare(qualityPenalty, that.qualityPenalty) == 0 && shortSentenceThreshold == that.shortSentenceThreshold && Double.compare(shortSentencePenalty, that.shortSentencePenalty) == 0 && Double.compare(bm25Weight, that.bm25Weight) == 0 && Double.compare(tcfFirstPosition, that.tcfFirstPosition) == 0 && Double.compare(tcfVerbatim, that.tcfVerbatim) == 0 && Double.compare(tcfProximity, that.tcfProximity) == 0 && Double.compare(temporalBiasWeight, that.temporalBiasWeight) == 0 && exportDebugData == that.exportDebugData && Objects.equals(bm25Params, that.bm25Params) && temporalBias == that.temporalBias;
-    }
-
-    @Override
-    public int hashCode() {
-        int result = Objects.hashCode(bm25Params);
-        result = 31 * result + shortDocumentThreshold;
-        result = 31 * result + Double.hashCode(shortDocumentPenalty);
-        result = 31 * result + Double.hashCode(domainRankBonus);
-        result = 31 * result + Double.hashCode(qualityPenalty);
-        result = 31 * result + shortSentenceThreshold;
-        result = 31 * result + Double.hashCode(shortSentencePenalty);
-        result = 31 * result + Double.hashCode(bm25Weight);
-        result = 31 * result + Double.hashCode(tcfFirstPosition);
-        result = 31 * result + Double.hashCode(tcfVerbatim);
-        result = 31 * result + Double.hashCode(tcfProximity);
-        result = 31 * result + Objects.hashCode(temporalBias);
-        result = 31 * result + Double.hashCode(temporalBiasWeight);
-        result = 31 * result + Boolean.hashCode(exportDebugData);
-        return result;
-    }
-
-    public String toString() {
-        return "ResultRankingParameters(bm25Params=" + this.getBm25Params() + ", shortDocumentThreshold=" + this.getShortDocumentThreshold() + ", shortDocumentPenalty=" + this.getShortDocumentPenalty() + ", domainRankBonus=" + this.getDomainRankBonus() + ", qualityPenalty=" + this.getQualityPenalty() + ", shortSentenceThreshold=" + this.getShortSentenceThreshold() + ", shortSentencePenalty=" + this.getShortSentencePenalty() + ", bm25Weight=" + this.getBm25Weight() + ", tcfFirstPosition=" + this.getTcfFirstPosition() + ", tcfVerbatim=" + this.getTcfVerbatim() + ", tcfProximity=" + this.getTcfProximity() + ", temporalBias=" + this.getTemporalBias() + ", temporalBiasWeight=" + this.getTemporalBiasWeight() + ", exportDebugData=" + this.isExportDebugData() + ")";
-    }
-
-    public enum TemporalBias {
-        RECENT, OLD, NONE
-    }
-
-    public static class ResultRankingParametersBuilder {
-        private Bm25Parameters bm25Params;
-        private int shortDocumentThreshold;
-        private double shortDocumentPenalty;
-        private double domainRankBonus;
-        private double qualityPenalty;
-        private int shortSentenceThreshold;
-        private double shortSentencePenalty;
-        private double bm25Weight;
-        private double tcfFirstPosition;
-        private double tcfVerbatim;
-        private double tcfProximity;
-        private TemporalBias temporalBias;
-        private double temporalBiasWeight;
-        private boolean exportDebugData;
-
-        ResultRankingParametersBuilder() {
-        }
-
-        public ResultRankingParametersBuilder bm25Params(Bm25Parameters bm25Params) {
-            this.bm25Params = bm25Params;
-            return this;
-        }
-
-        public ResultRankingParametersBuilder shortDocumentThreshold(int shortDocumentThreshold) {
-            this.shortDocumentThreshold = shortDocumentThreshold;
-            return this;
-        }
-
-        public ResultRankingParametersBuilder shortDocumentPenalty(double shortDocumentPenalty) {
-            this.shortDocumentPenalty = shortDocumentPenalty;
-            return this;
-        }
-
-        public ResultRankingParametersBuilder domainRankBonus(double domainRankBonus) {
-            this.domainRankBonus = domainRankBonus;
-            return this;
-        }
-
-        public ResultRankingParametersBuilder qualityPenalty(double qualityPenalty) {
-            this.qualityPenalty = qualityPenalty;
-            return this;
-        }
-
-        public ResultRankingParametersBuilder shortSentenceThreshold(int shortSentenceThreshold) {
-            this.shortSentenceThreshold = shortSentenceThreshold;
-            return this;
-        }
-
-        public ResultRankingParametersBuilder shortSentencePenalty(double shortSentencePenalty) {
-            this.shortSentencePenalty = shortSentencePenalty;
-            return this;
-        }
-
-        public ResultRankingParametersBuilder bm25Weight(double bm25Weight) {
-            this.bm25Weight = bm25Weight;
-            return this;
-        }
-
-        public ResultRankingParametersBuilder tcfFirstPosition(double tcfFirstPosition) {
-            this.tcfFirstPosition = tcfFirstPosition;
-            return this;
-        }
-
-        public ResultRankingParametersBuilder tcfVerbatim(double tcfVerbatim) {
-            this.tcfVerbatim = tcfVerbatim;
-            return this;
-        }
-
-        public ResultRankingParametersBuilder tcfProximity(double tcfProximity) {
-            this.tcfProximity = tcfProximity;
-            return this;
-        }
-
-        public ResultRankingParametersBuilder temporalBias(TemporalBias temporalBias) {
-            this.temporalBias = temporalBias;
-            return this;
-        }
-
-        public ResultRankingParametersBuilder temporalBiasWeight(double temporalBiasWeight) {
-            this.temporalBiasWeight = temporalBiasWeight;
-            return this;
-        }
-
-        public ResultRankingParametersBuilder exportDebugData(boolean exportDebugData) {
-            this.exportDebugData = exportDebugData;
-            return this;
-        }
-
-        public ResultRankingParameters build() {
-            return new ResultRankingParameters(this.bm25Params, this.shortDocumentThreshold, this.shortDocumentPenalty, this.domainRankBonus, this.qualityPenalty, this.shortSentenceThreshold, this.shortSentencePenalty, this.bm25Weight, this.tcfFirstPosition, this.tcfVerbatim, this.tcfProximity, this.temporalBias, this.temporalBiasWeight, this.exportDebugData);
-        }
-
-        public String toString() {
-            return "ResultRankingParameters.ResultRankingParametersBuilder(bm25Params=" + this.bm25Params + ", shortDocumentThreshold=" + this.shortDocumentThreshold + ", shortDocumentPenalty=" + this.shortDocumentPenalty + ", domainRankBonus=" + this.domainRankBonus + ", qualityPenalty=" + this.qualityPenalty + ", shortSentenceThreshold=" + this.shortSentenceThreshold + ", shortSentencePenalty=" + this.shortSentencePenalty + ", bm25Weight=" + this.bm25Weight + ", tcfFirstPosition=" + this.tcfFirstPosition + ", tcfVerbatim=" + this.tcfVerbatim + ", tcfProximity=" + this.tcfProximity + ", temporalBias=" + this.temporalBias + ", temporalBiasWeight=" + this.temporalBiasWeight + ", exportDebugData=" + this.exportDebugData + ")";
-        }
-    }
-}

code/functions/search-query/api/src/main/protobuf/query-api.proto

@@ -162,6 +162,7 @@ message RpcResultRankingParameters {
   double temporalBiasWeight = 17;
 
   bool exportDebugData = 18;
+  bool disablePenalties = 19;
 
 }
 

code/functions/search-query/api/test/nu/marginalia/index/client/IndexProtobufCodecTest.java

@@ -3,8 +3,6 @@ package nu.marginalia.index.client;
 import nu.marginalia.api.searchquery.IndexProtobufCodec;
 import nu.marginalia.api.searchquery.model.query.SearchPhraseConstraint;
 import nu.marginalia.api.searchquery.model.query.SearchQuery;
-import nu.marginalia.api.searchquery.model.results.ResultRankingParameters;
-import nu.marginalia.index.query.limit.QueryLimits;
 import nu.marginalia.index.query.limit.SpecificationLimit;
 import org.junit.jupiter.api.Test;
 
@@ -22,18 +20,6 @@ class IndexProtobufCodecTest {
         verifyIsIdentityTransformation(SpecificationLimit.lessThan(1), l -> IndexProtobufCodec.convertSpecLimit(IndexProtobufCodec.convertSpecLimit(l)));
     }
 
-    @Test
-    public void testRankingParameters() {
-        verifyIsIdentityTransformation(ResultRankingParameters.sensibleDefaults(),
-                p -> IndexProtobufCodec.convertRankingParameterss(IndexProtobufCodec.convertRankingParameterss(p, null)));
-    }
-
-    @Test
-    public void testQueryLimits() {
-        verifyIsIdentityTransformation(new QueryLimits(1,2,3,4),
-                l -> IndexProtobufCodec.convertQueryLimits(IndexProtobufCodec.convertQueryLimits(l))
-                );
-    }
     @Test
     public void testSubqery() {
         verifyIsIdentityTransformation(new SearchQuery(

code/functions/search-query/java/nu/marginalia/functions/searchquery/QueryFactory.java

@@ -2,8 +2,9 @@ package nu.marginalia.functions.searchquery;
 
 import com.google.inject.Inject;
 import com.google.inject.Singleton;
+import nu.marginalia.api.searchquery.RpcQueryLimits;
+import nu.marginalia.api.searchquery.RpcResultRankingParameters;
 import nu.marginalia.api.searchquery.model.query.*;
-import nu.marginalia.api.searchquery.model.results.ResultRankingParameters;
 import nu.marginalia.functions.searchquery.query_parser.QueryExpansion;
 import nu.marginalia.functions.searchquery.query_parser.QueryParser;
 import nu.marginalia.functions.searchquery.query_parser.token.QueryToken;
@@ -36,7 +37,7 @@ public class QueryFactory {
 
 
     public ProcessedQuery createQuery(QueryParams params,
-                                      @Nullable ResultRankingParameters rankingParams) {
+                                      @Nullable RpcResultRankingParameters rankingParams) {
         final var query = params.humanQuery();
 
         if (query.length() > 1000) {
@@ -132,7 +133,9 @@ public class QueryFactory {
         var limits = params.limits();
         // Disable limits on number of results per domain if we're searching with a site:-type term
         if (domain != null) {
-            limits = limits.forSingleDomain();
+            limits = RpcQueryLimits.newBuilder(limits)
+                    .setResultsByDomain(limits.getResultsTotal())
+                    .build();
         }
 
         var expansion = queryExpansion.expandQuery(queryBuilder.searchTermsInclude);

code/functions/search-query/java/nu/marginalia/functions/searchquery/QueryGRPCService.java

@@ -9,7 +9,7 @@ import nu.marginalia.api.searchquery.*;
 import nu.marginalia.api.searchquery.model.query.ProcessedQuery;
 import nu.marginalia.api.searchquery.model.query.QueryParams;
 import nu.marginalia.api.searchquery.model.results.DecoratedSearchResultItem;
-import nu.marginalia.api.searchquery.model.results.ResultRankingParameters;
+import nu.marginalia.api.searchquery.model.results.PrototypeRankingParameters;
 import nu.marginalia.index.api.IndexClient;
 import nu.marginalia.service.server.DiscoverableService;
 import org.slf4j.Logger;
@@ -55,7 +55,7 @@ public class QueryGRPCService
                     .time(() -> {
 
                 var params = QueryProtobufCodec.convertRequest(request);
-                var query = queryFactory.createQuery(params, ResultRankingParameters.sensibleDefaults());
+                var query = queryFactory.createQuery(params, PrototypeRankingParameters.sensibleDefaults());
 
                 var indexRequest = QueryProtobufCodec.convertQuery(request, query);
 
@@ -102,7 +102,7 @@ public class QueryGRPCService
             String originalQuery,
             QueryParams params,
             IndexClient.Pagination pagination,
-            ResultRankingParameters rankingParameters) {
+            RpcResultRankingParameters rankingParameters) {
 
         var query = queryFactory.createQuery(params, rankingParameters);
         IndexClient.AggregateQueryResponse response = indexClient.executeQueries(QueryProtobufCodec.convertQuery(originalQuery, query), pagination);

code/functions/search-query/java/nu/marginalia/functions/searchquery/query_parser/QueryExpansion.java

@@ -134,6 +134,10 @@ public class QueryExpansion {
                 if (scoreCombo > scoreA + scoreB || scoreCombo > 1000) {
                     graph.addVariantForSpan(prev, qw, joinedWord);
                 }
+                else if (StringUtils.isAlpha(prev.word()) && StringUtils.isNumeric(qw.word())) { // join e.g. trs 80 to trs80 and trs-80
+                    graph.addVariantForSpan(prev, qw, prev.word() + qw.word());
+                    graph.addVariantForSpan(prev, qw, prev.word() + "-" + qw.word());
+                }
             }
 
             prev = qw;

code/functions/search-query/java/nu/marginalia/functions/searchquery/query_parser/QueryParser.java

@@ -233,9 +233,19 @@ public class QueryParser {
             entity.replace(new QueryToken.RankTerm(limit, str));
         } else if (str.startsWith("qs=")) {
             entity.replace(new QueryToken.QsTerm(str.substring(3)));
-        } else if (str.contains(":")) {
+        } else if (str.startsWith("site:")
+                || str.startsWith("format:")
+                || str.startsWith("file:")
+                || str.startsWith("tld:")
+                || str.startsWith("ip:")
+                || str.startsWith("as:")
+                || str.startsWith("asn:")
+                || str.startsWith("generator:")
+        )
+        {
             entity.replace(new QueryToken.AdviceTerm(str, t.displayStr()));
         }
+
     }
 
     private static SpecificationLimit parseSpecificationLimit(String str) {

code/functions/search-query/test/nu/marginalia/query/svc/QueryFactoryTest.java

@@ -1,12 +1,12 @@
 package nu.marginalia.query.svc;
 
 import nu.marginalia.WmsaHome;
+import nu.marginalia.api.searchquery.RpcQueryLimits;
+import nu.marginalia.api.searchquery.RpcTemporalBias;
 import nu.marginalia.api.searchquery.model.query.QueryParams;
 import nu.marginalia.api.searchquery.model.query.SearchSpecification;
-import nu.marginalia.api.searchquery.model.results.ResultRankingParameters;
 import nu.marginalia.functions.searchquery.QueryFactory;
 import nu.marginalia.functions.searchquery.query_parser.QueryExpansion;
-import nu.marginalia.index.query.limit.QueryLimits;
 import nu.marginalia.index.query.limit.QueryStrategy;
 import nu.marginalia.index.query.limit.SpecificationLimit;
 import nu.marginalia.index.query.limit.SpecificationLimitType;
@@ -49,10 +49,15 @@ public class QueryFactoryTest {
                         SpecificationLimit.none(),
                         SpecificationLimit.none(),
                         null,
-                        new QueryLimits(100, 100, 100, 100),
+                        RpcQueryLimits.newBuilder()
+                                .setResultsTotal(100)
+                                .setResultsByDomain(100)
+                                .setTimeoutMs(100)
+                                .setFetchSize(100)
+                                .build(),
                         "NONE",
                         QueryStrategy.AUTO,
-                        ResultRankingParameters.TemporalBias.NONE,
+                        RpcTemporalBias.Bias.NONE,
                         0), null).specs;
     }
 
@@ -208,6 +213,24 @@ public class QueryFactoryTest {
         System.out.println(subquery);
     }
 
+
+    @Test
+    public void testContractionWordNum() {
+        var subquery = parseAndGetSpecs("glove 80");
+
+        Assertions.assertTrue(subquery.query.compiledQuery.contains(" glove "));
+        Assertions.assertTrue(subquery.query.compiledQuery.contains(" 80 "));
+        Assertions.assertTrue(subquery.query.compiledQuery.contains(" glove-80 "));
+        Assertions.assertTrue(subquery.query.compiledQuery.contains(" glove80 "));
+    }
+
+
+    @Test
+    public void testCplusPlus() {
+        var subquery = parseAndGetSpecs("std::vector::push_back vector");
+        System.out.println(subquery);
+    }
+
     @Test
     public void testQuotedApostrophe() {
         var subquery = parseAndGetSpecs("\"bob's cars\"");

code/index/api/java/nu/marginalia/index/api/IndexClient.java

@@ -16,20 +16,19 @@ import org.slf4j.LoggerFactory;
 
 import java.util.ArrayList;
 import java.util.Comparator;
-import java.util.Iterator;
 import java.util.List;
 import java.util.concurrent.CompletableFuture;
 import java.util.concurrent.ExecutorService;
 import java.util.concurrent.Executors;
-
+import java.util.concurrent.atomic.AtomicInteger;
-import static java.lang.Math.clamp;
+import java.util.function.Consumer;
 
 @Singleton
 public class IndexClient {
     private static final Logger logger = LoggerFactory.getLogger(IndexClient.class);
     private final GrpcMultiNodeChannelPool<IndexApiGrpc.IndexApiBlockingStub> channelPool;
     private final DomainBlacklistImpl blacklist;
-    private static final ExecutorService executor = Executors.newVirtualThreadPerTaskExecutor();
+    private static final ExecutorService executor = Executors.newCachedThreadPool();
 
     @Inject
     public IndexClient(GrpcChannelPoolFactory channelPoolFactory, DomainBlacklistImpl blacklist) {
@@ -51,40 +50,37 @@ public class IndexClient {
 
     /** Execute a query on the index partitions and return the combined results. */
     public AggregateQueryResponse executeQueries(RpcIndexQuery indexRequest, Pagination pagination) {
-        List<CompletableFuture<Iterator<RpcDecoratedResultItem>>> futures =
-                channelPool.call(IndexApiGrpc.IndexApiBlockingStub::query)
-                        .async(executor)
-                        .runEach(indexRequest);
 
         final int requestedMaxResults = indexRequest.getQueryLimits().getResultsTotal();
-        final int resultsUpperBound = requestedMaxResults * channelPool.getNumNodes();
 
-        List<RpcDecoratedResultItem> results = new ArrayList<>(resultsUpperBound);
+        AtomicInteger totalNumResults = new AtomicInteger(0);
 
-        for (var future : futures) {
+        List<RpcDecoratedResultItem> results =
-            try {
+                channelPool.call(IndexApiGrpc.IndexApiBlockingStub::query)
-                future.get().forEachRemaining(results::add);
+                        .async(executor)
-            }
+                        .runEach(indexRequest)
-            catch (Exception e) {
+                        .stream()
-                logger.error("Downstream exception", e);
+                        .map(future -> future.thenApply(iterator -> {
-            }
+                            List<RpcDecoratedResultItem> ret = new ArrayList<>(requestedMaxResults);
-        }
+                            iterator.forEachRemaining(ret::add);
+                            totalNumResults.addAndGet(ret.size());
+                            return ret;
+                        }))
+                        .mapMulti((CompletableFuture<List<RpcDecoratedResultItem>> fut, Consumer<List<RpcDecoratedResultItem>> c) ->{
+                            try {
+                                c.accept(fut.join());
+                            } catch (Exception e) {
+                                logger.error("Error while fetching results", e);
+                            }
+                        })
+                        .flatMap(List::stream)
+                        .filter(item -> !isBlacklisted(item))
+                        .sorted(comparator)
+                        .skip(Math.max(0, (pagination.page - 1) * pagination.pageSize))
+                        .limit(pagination.pageSize)
+                        .toList();
 
-        // Sort the results by ranking score and remove blacklisted domains
+        return new AggregateQueryResponse(results, pagination.page(), totalNumResults.get());
-        results.sort(comparator);
-        results.removeIf(this::isBlacklisted);
-
-        int numReceivedResults = results.size();
-
-        // pagination is typically 1-indexed, so we need to adjust the start and end indices
-        int indexStart = (pagination.page - 1) * pagination.pageSize;
-        int indexEnd = (pagination.page) * pagination.pageSize;
-
-        results = results.subList(
-                clamp(indexStart, 0, Math.max(0, results.size() - 1)), // from is inclusive, so subtract 1 from size()
-                clamp(indexEnd, 0, results.size()));
-
-        return new AggregateQueryResponse(results, pagination.page(), numReceivedResults);
     }
 
     private boolean isBlacklisted(RpcDecoratedResultItem item) {

code/index/java/nu/marginalia/index/IndexGrpcService.java

@@ -10,12 +10,12 @@ import it.unimi.dsi.fastutil.longs.LongArrayList;
 import nu.marginalia.api.searchquery.IndexApiGrpc;
 import nu.marginalia.api.searchquery.RpcDecoratedResultItem;
 import nu.marginalia.api.searchquery.RpcIndexQuery;
+import nu.marginalia.api.searchquery.RpcResultRankingParameters;
 import nu.marginalia.api.searchquery.model.compiled.CompiledQuery;
 import nu.marginalia.api.searchquery.model.compiled.CompiledQueryLong;
 import nu.marginalia.api.searchquery.model.compiled.CqDataInt;
 import nu.marginalia.api.searchquery.model.query.SearchSpecification;
 import nu.marginalia.api.searchquery.model.results.ResultRankingContext;
-import nu.marginalia.api.searchquery.model.results.ResultRankingParameters;
 import nu.marginalia.array.page.LongQueryBuffer;
 import nu.marginalia.index.index.StatefulIndex;
 import nu.marginalia.index.model.SearchParameters;
@@ -211,7 +211,7 @@ public class IndexGrpcService
     /** This class is responsible for ranking the results and adding the best results to the
      * resultHeap, which depending on the state of the indexLookup threads may or may not block
      */
-    private ResultRankingContext createRankingContext(ResultRankingParameters rankingParams,
+    private ResultRankingContext createRankingContext(RpcResultRankingParameters rankingParams,
                                                       CompiledQuery<String> compiledQuery,
                                                       CompiledQueryLong compiledQueryIds)
     {

code/index/java/nu/marginalia/index/model/SearchParameters.java

@@ -2,12 +2,13 @@ package nu.marginalia.index.model;
 
 import nu.marginalia.api.searchquery.IndexProtobufCodec;
 import nu.marginalia.api.searchquery.RpcIndexQuery;
+import nu.marginalia.api.searchquery.RpcResultRankingParameters;
 import nu.marginalia.api.searchquery.model.compiled.CompiledQuery;
 import nu.marginalia.api.searchquery.model.compiled.CompiledQueryLong;
 import nu.marginalia.api.searchquery.model.compiled.CompiledQueryParser;
-import nu.marginalia.api.searchquery.model.query.SearchSpecification;
 import nu.marginalia.api.searchquery.model.query.SearchQuery;
-import nu.marginalia.api.searchquery.model.results.ResultRankingParameters;
+import nu.marginalia.api.searchquery.model.query.SearchSpecification;
+import nu.marginalia.api.searchquery.model.results.PrototypeRankingParameters;
 import nu.marginalia.index.query.IndexSearchBudget;
 import nu.marginalia.index.query.limit.QueryStrategy;
 import nu.marginalia.index.searchset.SearchSet;
@@ -23,7 +24,7 @@ public class SearchParameters {
     public final IndexSearchBudget budget;
     public final SearchQuery query;
     public final QueryParams queryParams;
-    public final ResultRankingParameters rankingParams;
+    public final RpcResultRankingParameters rankingParams;
 
     public final int limitByDomain;
     public final int limitTotal;
@@ -41,11 +42,11 @@ public class SearchParameters {
     public SearchParameters(SearchSpecification specsSet, SearchSet searchSet) {
         var limits = specsSet.queryLimits;
 
-        this.fetchSize = limits.fetchSize();
+        this.fetchSize = limits.getFetchSize();
-        this.budget = new IndexSearchBudget(limits.timeoutMs());
+        this.budget = new IndexSearchBudget(limits.getTimeoutMs());
         this.query = specsSet.query;
-        this.limitByDomain = limits.resultsByDomain();
+        this.limitByDomain = limits.getResultsByDomain();
-        this.limitTotal = limits.resultsTotal();
+        this.limitTotal = limits.getResultsTotal();
 
         queryParams = new QueryParams(
                 specsSet.quality,
@@ -62,17 +63,17 @@ public class SearchParameters {
     }
 
     public SearchParameters(RpcIndexQuery request, SearchSet searchSet) {
-        var limits = IndexProtobufCodec.convertQueryLimits(request.getQueryLimits());
+        var limits = request.getQueryLimits();
 
-        this.fetchSize = limits.fetchSize();
+        this.fetchSize = limits.getFetchSize();
 
         // The time budget is halved because this is the point when we start to
         // wrap up the search and return the results.
-        this.budget = new IndexSearchBudget(limits.timeoutMs() / 2);
+        this.budget = new IndexSearchBudget(limits.getTimeoutMs() / 2);
         this.query = IndexProtobufCodec.convertRpcQuery(request.getQuery());
 
-        this.limitByDomain = limits.resultsByDomain();
+        this.limitByDomain = limits.getResultsByDomain();
-        this.limitTotal = limits.resultsTotal();
+        this.limitTotal = limits.getResultsTotal();
 
         queryParams = new QueryParams(
                 convertSpecLimit(request.getQuality()),
@@ -85,7 +86,7 @@ public class SearchParameters {
         compiledQuery = CompiledQueryParser.parse(this.query.compiledQuery);
         compiledQueryIds = compiledQuery.mapToLong(SearchTermsUtil::getWordId);
 
-        rankingParams = IndexProtobufCodec.convertRankingParameterss(request.getParameters());
+        rankingParams = request.hasParameters() ? request.getParameters() : PrototypeRankingParameters.sensibleDefaults();
     }
 
 

code/index/java/nu/marginalia/index/results/Bm25GraphVisitor.java

@@ -2,7 +2,6 @@ package nu.marginalia.index.results;
 
 import nu.marginalia.api.searchquery.model.compiled.CqDataInt;
 import nu.marginalia.api.searchquery.model.compiled.CqExpression;
-import nu.marginalia.api.searchquery.model.results.Bm25Parameters;
 import nu.marginalia.api.searchquery.model.results.ResultRankingContext;
 
 import java.util.BitSet;
@@ -24,14 +23,14 @@ public class Bm25GraphVisitor implements CqExpression.DoubleVisitor {
 
     private final BitSet mask;
 
-    public Bm25GraphVisitor(Bm25Parameters bm25Parameters,
+    public Bm25GraphVisitor(double k1, double b,
                             float[] counts,
                             int length,
                             ResultRankingContext ctx) {
         this.length = length;
 
-        this.k1 = bm25Parameters.k();
+        this.k1 = k1;
-        this.b = bm25Parameters.b();
+        this.b = b;
 
         this.docCount = ctx.termFreqDocCount();
         this.counts = counts;

code/index/java/nu/marginalia/index/results/DomainRankingOverrides.java

@@ -0,0 +1,119 @@
+package nu.marginalia.index.results;
+
+import com.google.inject.Inject;
+import com.google.inject.Singleton;
+import gnu.trove.map.hash.TIntDoubleHashMap;
+import nu.marginalia.WmsaHome;
+import nu.marginalia.db.DbDomainQueries;
+import nu.marginalia.model.EdgeDomain;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+import java.io.IOException;
+import java.nio.file.Files;
+import java.nio.file.Path;
+import java.util.List;
+import java.util.OptionalInt;
+import java.util.concurrent.TimeUnit;
+
+@Singleton
+public class DomainRankingOverrides {
+    private final DbDomainQueries domainQueries;
+
+    private volatile TIntDoubleHashMap rankingFactors = new TIntDoubleHashMap(100, 0.75f, -1, 1.);
+
+    private static final Logger logger = LoggerFactory.getLogger(DomainRankingOverrides.class);
+
+    private final Path overrideFilePath;
+
+    @Inject
+    public DomainRankingOverrides(DbDomainQueries domainQueries) {
+        this.domainQueries = domainQueries;
+
+        overrideFilePath = WmsaHome.getDataPath().resolve("domain-ranking-factors.txt");
+
+        Thread.ofPlatform().start(this::updateRunner);
+    }
+
+    // for test access
+    public DomainRankingOverrides(DbDomainQueries domainQueries, Path overrideFilePath)
+    {
+        this.domainQueries = domainQueries;
+        this.overrideFilePath = overrideFilePath;
+    }
+
+
+    public double getRankingFactor(int domainId) {
+        return rankingFactors.get(domainId);
+    }
+
+    private void updateRunner() {
+        for (;;) {
+            reloadFile();
+
+            try {
+                TimeUnit.MINUTES.sleep(5);
+            } catch (InterruptedException ex) {
+                logger.warn("Thread interrupted", ex);
+                break;
+            }
+        }
+    }
+
+    void reloadFile() {
+        if (!Files.exists(overrideFilePath)) {
+            return;
+        }
+
+        try {
+            List<String> lines = Files.readAllLines(overrideFilePath);
+
+            double factor = 1.;
+
+            var newRankingFactors = new TIntDoubleHashMap(lines.size(), 0.75f, -1, 1.);
+
+            for (var line : lines) {
+                if (line.isBlank()) continue;
+                if (line.startsWith("#")) continue;
+
+                String[] parts = line.split("\\s+");
+                if (parts.length != 2) {
+                    logger.warn("Unrecognized format for domain overrides file: {}", line);
+                    continue;
+                }
+
+                try {
+                    switch (parts[0]) {
+                        case "value" -> {
+                            // error handle me
+                            factor = Double.parseDouble(parts[1]);
+                            if (factor < 0) {
+                                logger.error("Negative values are not permitted, found {}", factor);
+                                factor = 1;
+                            }
+                        }
+                        case "domain" -> {
+                            // error handle
+                            OptionalInt domainId = domainQueries.tryGetDomainId(new EdgeDomain(parts[1]));
+                            if (domainId.isPresent()) {
+                                newRankingFactors.put(domainId.getAsInt(), factor);
+                            }
+                            else {
+                                logger.warn("Unrecognized domain id {}", parts[1]);
+                            }
+                        }
+                        default -> {
+                            logger.warn("Unrecognized format {}", line);
+                        }
+                    }
+                } catch (Exception ex) {
+                    logger.warn("Error in parsing domain overrides file: {} ({})", line, ex.getClass().getSimpleName());
+                }
+            }
+
+            rankingFactors = newRankingFactors;
+        } catch (IOException ex) {
+            logger.error("Failed to read " + overrideFilePath, ex);
+        }
+    }
+}

code/index/java/nu/marginalia/index/results/IndexResultRankingService.java

@@ -40,13 +40,16 @@ public class IndexResultRankingService {
 
     private final DocumentDbReader documentDbReader;
     private final StatefulIndex statefulIndex;
+    private final DomainRankingOverrides domainRankingOverrides;
 
     @Inject
     public IndexResultRankingService(DocumentDbReader documentDbReader,
-                                     StatefulIndex statefulIndex)
+                                     StatefulIndex statefulIndex,
+                                     DomainRankingOverrides domainRankingOverrides)
     {
         this.documentDbReader = documentDbReader;
         this.statefulIndex = statefulIndex;
+        this.domainRankingOverrides = domainRankingOverrides;
     }
 
     public List<SearchResultItem> rankResults(SearchParameters params,
@@ -57,7 +60,7 @@ public class IndexResultRankingService {
         if (resultIds.isEmpty())
             return List.of();
 
-        IndexResultScoreCalculator resultRanker = new IndexResultScoreCalculator(statefulIndex, rankingContext, params);
+        IndexResultScoreCalculator resultRanker = new IndexResultScoreCalculator(statefulIndex, domainRankingOverrides, rankingContext, params);
 
         List<SearchResultItem> results = new ArrayList<>(resultIds.size());
 
@@ -156,7 +159,7 @@ public class IndexResultRankingService {
         // for the selected results, as this would be comically expensive to do for all the results we
         // discard along the way
 
-        if (params.rankingParams.exportDebugData) {
+        if (params.rankingParams.getExportDebugData()) {
             var combinedIdsList = new LongArrayList(resultsList.size());
             for (var item : resultsList) {
                 combinedIdsList.add(item.combinedId);

code/index/java/nu/marginalia/index/results/IndexResultScoreCalculator.java

@@ -2,10 +2,11 @@ package nu.marginalia.index.results;
 
 import it.unimi.dsi.fastutil.ints.IntIterator;
 import it.unimi.dsi.fastutil.ints.IntList;
+import nu.marginalia.api.searchquery.RpcResultRankingParameters;
+import nu.marginalia.api.searchquery.RpcTemporalBias;
 import nu.marginalia.api.searchquery.model.compiled.CompiledQuery;
 import nu.marginalia.api.searchquery.model.compiled.CompiledQueryLong;
 import nu.marginalia.api.searchquery.model.results.ResultRankingContext;
-import nu.marginalia.api.searchquery.model.results.ResultRankingParameters;
 import nu.marginalia.api.searchquery.model.results.SearchResultItem;
 import nu.marginalia.api.searchquery.model.results.debug.DebugRankingFactors;
 import nu.marginalia.index.forward.spans.DocumentSpans;
@@ -40,14 +41,17 @@ public class IndexResultScoreCalculator {
     private final CombinedIndexReader index;
     private final QueryParams queryParams;
 
+    private final DomainRankingOverrides domainRankingOverrides;
     private final ResultRankingContext rankingContext;
     private final CompiledQuery<String> compiledQuery;
 
     public IndexResultScoreCalculator(StatefulIndex statefulIndex,
+                                      DomainRankingOverrides domainRankingOverrides,
                                       ResultRankingContext rankingContext,
                                       SearchParameters params)
     {
         this.index = statefulIndex.get();
+        this.domainRankingOverrides = domainRankingOverrides;
         this.rankingContext = rankingContext;
 
         this.queryParams = params.queryParams;
@@ -116,20 +120,20 @@ public class IndexResultScoreCalculator {
 
         float proximitiyFac = getProximitiyFac(decodedPositions, searchTerms.phraseConstraints, verbatimMatches, unorderedMatches, spans);
 
-        double score_firstPosition = params.tcfFirstPosition * (1.0 / Math.sqrt(unorderedMatches.firstPosition));
+        double score_firstPosition = params.getTcfFirstPositionWeight() * (1.0 / Math.sqrt(unorderedMatches.firstPosition));
-        double score_verbatim = params.tcfVerbatim * verbatimMatches.getScore();
+        double score_verbatim = params.getTcfVerbatimWeight() * verbatimMatches.getScore();
-        double score_proximity = params.tcfProximity * proximitiyFac;
+        double score_proximity = params.getTcfProximityWeight() * proximitiyFac;
-        double score_bM25 = params.bm25Weight
+        double score_bM25 = params.getBm25Weight()
-                * wordFlagsQuery.root.visit(new Bm25GraphVisitor(params.bm25Params, unorderedMatches.getWeightedCounts(), docSize, rankingContext))
+                * wordFlagsQuery.root.visit(new Bm25GraphVisitor(params.getBm25K(), params.getBm25B(), unorderedMatches.getWeightedCounts(), docSize, rankingContext))
                 / (Math.sqrt(unorderedMatches.searchableKeywordCount + 1));
-        double score_bFlags = params.bm25Weight
+        double score_bFlags = params.getBm25Weight()
-                * wordFlagsQuery.root.visit(new TermFlagsGraphVisitor(params.bm25Params, wordFlagsQuery.data, unorderedMatches.getWeightedCounts(), rankingContext))
+                * wordFlagsQuery.root.visit(new TermFlagsGraphVisitor(params.getBm25K(), wordFlagsQuery.data, unorderedMatches.getWeightedCounts(), rankingContext))
                 / (Math.sqrt(unorderedMatches.searchableKeywordCount + 1));
 
+        double rankingAdjustment = domainRankingOverrides.getRankingFactor(UrlIdCodec.getDomainId(combinedId));
+
         double score = normalize(
-                score_firstPosition + score_proximity + score_verbatim
+                rankingAdjustment * (score_firstPosition + score_proximity + score_verbatim + score_bM25 + score_bFlags),
-                        + score_bM25
-                        + score_bFlags,
                 -Math.min(0, documentBonus) // The magnitude of documentBonus, if it is negative; otherwise 0
         );
 
@@ -245,9 +249,13 @@ public class IndexResultScoreCalculator {
     private double calculateDocumentBonus(long documentMetadata,
                                           int features,
                                           int length,
-                                          ResultRankingParameters rankingParams,
+                                          RpcResultRankingParameters rankingParams,
                                           @Nullable DebugRankingFactors debugRankingFactors) {
 
+        if (rankingParams.getDisablePenalties()) {
+            return 0.;
+        }
+
         int rank = DocumentMetadata.decodeRank(documentMetadata);
         int asl = DocumentMetadata.decodeAvgSentenceLength(documentMetadata);
         int quality = DocumentMetadata.decodeQuality(documentMetadata);
@@ -256,18 +264,18 @@ public class IndexResultScoreCalculator {
         int topology = DocumentMetadata.decodeTopology(documentMetadata);
         int year = DocumentMetadata.decodeYear(documentMetadata);
 
-        double averageSentenceLengthPenalty = (asl >= rankingParams.shortSentenceThreshold ? 0 : -rankingParams.shortSentencePenalty);
+        double averageSentenceLengthPenalty = (asl >= rankingParams.getShortSentenceThreshold() ? 0 : -rankingParams.getShortSentencePenalty());
 
         final double qualityPenalty = calculateQualityPenalty(size, quality, rankingParams);
-        final double rankingBonus = (255. - rank) * rankingParams.domainRankBonus;
+        final double rankingBonus = (255. - rank) * rankingParams.getDomainRankBonus();
         final double topologyBonus = Math.log(1 + topology);
-        final double documentLengthPenalty = length > rankingParams.shortDocumentThreshold ? 0 : -rankingParams.shortDocumentPenalty;
+        final double documentLengthPenalty = length > rankingParams.getShortDocumentThreshold() ? 0 : -rankingParams.getShortDocumentPenalty();
         final double temporalBias;
 
-        if (rankingParams.temporalBias == ResultRankingParameters.TemporalBias.RECENT) {
+        if (rankingParams.getTemporalBias().getBias() == RpcTemporalBias.Bias.RECENT) {
-            temporalBias = - Math.abs(year - PubDate.MAX_YEAR) * rankingParams.temporalBiasWeight;
+            temporalBias = - Math.abs(year - PubDate.MAX_YEAR) * rankingParams.getTemporalBiasWeight();
-        } else if (rankingParams.temporalBias == ResultRankingParameters.TemporalBias.OLD) {
+        } else if (rankingParams.getTemporalBias().getBias() == RpcTemporalBias.Bias.OLD) {
-            temporalBias = - Math.abs(year - PubDate.MIN_YEAR) * rankingParams.temporalBiasWeight;
+            temporalBias = - Math.abs(year - PubDate.MIN_YEAR) * rankingParams.getTemporalBiasWeight();
         } else {
             temporalBias = 0;
         }
@@ -506,14 +514,14 @@ public class IndexResultScoreCalculator {
     }
 
 
-    private double calculateQualityPenalty(int size, int quality, ResultRankingParameters rankingParams) {
+    private double calculateQualityPenalty(int size, int quality, RpcResultRankingParameters rankingParams) {
         if (size < 400) {
             if (quality < 5)
                 return 0;
-            return -quality * rankingParams.qualityPenalty;
+            return -quality * rankingParams.getQualityPenalty();
         }
         else {
-            return -quality * rankingParams.qualityPenalty * 20;
+            return -quality * rankingParams.getQualityPenalty() * 20;
         }
     }
 
@@ -575,3 +583,4 @@ public class IndexResultScoreCalculator {
     }
 
 }
+

code/index/java/nu/marginalia/index/results/TermFlagsGraphVisitor.java

@@ -3,7 +3,6 @@ package nu.marginalia.index.results;
 import nu.marginalia.api.searchquery.model.compiled.CqDataInt;
 import nu.marginalia.api.searchquery.model.compiled.CqDataLong;
 import nu.marginalia.api.searchquery.model.compiled.CqExpression;
-import nu.marginalia.api.searchquery.model.results.Bm25Parameters;
 import nu.marginalia.api.searchquery.model.results.ResultRankingContext;
 import nu.marginalia.model.idx.WordFlags;
 
@@ -15,15 +14,14 @@ public class TermFlagsGraphVisitor implements CqExpression.DoubleVisitor {
     private final CqDataLong wordMetaData;
     private final CqDataInt frequencies;
     private final float[] counts;
-    private final Bm25Parameters bm25Parameters;
+    private final double k1;
-
     private final int docCount;
 
-    public TermFlagsGraphVisitor(Bm25Parameters bm25Parameters,
+    public TermFlagsGraphVisitor(double k1,
                                  CqDataLong wordMetaData,
                                  float[] counts,
                                  ResultRankingContext ctx) {
-        this.bm25Parameters = bm25Parameters;
+        this.k1 = k1;
         this.counts = counts;
         this.docCount = ctx.termFreqDocCount();
         this.wordMetaData = wordMetaData;
@@ -55,7 +53,7 @@ public class TermFlagsGraphVisitor implements CqExpression.DoubleVisitor {
         int freq = frequencies.get(idx);
 
         // note we override b to zero for priority terms as they are independent of document length
-        return invFreq(docCount, freq) * f(bm25Parameters.k(), 0, count, 0);
+        return invFreq(docCount, freq) * f(k1, 0, count, 0);
     }
 
     private double evaluatePriorityScore(int idx) {

code/index/query/java/nu/marginalia/index/query/limit/QueryLimits.java

@@ -1,7 +0,0 @@
-package nu.marginalia.index.query.limit;
-
-public record QueryLimits(int resultsByDomain, int resultsTotal, int timeoutMs, int fetchSize) {
-    public QueryLimits forSingleDomain() {
-        return new QueryLimits(resultsTotal, resultsTotal, timeoutMs, fetchSize);
-    }
-}

code/index/test/nu/marginalia/index/IndexQueryServiceIntegrationSmokeTest.java

@@ -4,10 +4,11 @@ import com.google.inject.Guice;
 import com.google.inject.Inject;
 import nu.marginalia.IndexLocations;
 import nu.marginalia.api.searchquery.RpcDecoratedResultItem;
+import nu.marginalia.api.searchquery.RpcQueryLimits;
 import nu.marginalia.api.searchquery.model.query.SearchPhraseConstraint;
 import nu.marginalia.api.searchquery.model.query.SearchQuery;
 import nu.marginalia.api.searchquery.model.query.SearchSpecification;
-import nu.marginalia.api.searchquery.model.results.ResultRankingParameters;
+import nu.marginalia.api.searchquery.model.results.PrototypeRankingParameters;
 import nu.marginalia.index.construction.DocIdRewriter;
 import nu.marginalia.index.construction.full.FullIndexConstructor;
 import nu.marginalia.index.construction.prio.PrioIndexConstructor;
@@ -17,7 +18,6 @@ import nu.marginalia.index.forward.construction.ForwardIndexConverter;
 import nu.marginalia.index.index.StatefulIndex;
 import nu.marginalia.index.journal.IndexJournal;
 import nu.marginalia.index.journal.IndexJournalSlopWriter;
-import nu.marginalia.index.query.limit.QueryLimits;
 import nu.marginalia.index.query.limit.QueryStrategy;
 import nu.marginalia.index.query.limit.SpecificationLimit;
 import nu.marginalia.linkdb.docs.DocumentDbReader;
@@ -115,9 +115,16 @@ public class IndexQueryServiceIntegrationSmokeTest {
 
         var rsp = queryService.justQuery(
                 SearchSpecification.builder()
-                        .queryLimits(new QueryLimits(10, 10, Integer.MAX_VALUE, 4000))
+                        .queryLimits(
+                                RpcQueryLimits.newBuilder()
+                                        .setResultsByDomain(10)
+                                        .setResultsTotal(10)
+                                        .setTimeoutMs(Integer.MAX_VALUE)
+                                        .setFetchSize(4000)
+                                        .build()
+                        )
                         .queryStrategy(QueryStrategy.SENTENCE)
-                        .rankingParams(ResultRankingParameters.sensibleDefaults())
+                        .rankingParams(PrototypeRankingParameters.sensibleDefaults())
                         .domains(new ArrayList<>())
                         .searchSetIdentifier("NONE")
                         .query(
@@ -171,9 +178,16 @@ public class IndexQueryServiceIntegrationSmokeTest {
 
         var rsp = queryService.justQuery(
                 SearchSpecification.builder()
-                        .queryLimits(new QueryLimits(10, 10, Integer.MAX_VALUE, 4000))
+                        .queryLimits(
+                                RpcQueryLimits.newBuilder()
+                                        .setResultsByDomain(10)
+                                        .setResultsTotal(10)
+                                        .setTimeoutMs(Integer.MAX_VALUE)
+                                        .setFetchSize(4000)
+                                        .build()
+                        )
                         .queryStrategy(QueryStrategy.SENTENCE)
-                        .rankingParams(ResultRankingParameters.sensibleDefaults())
+                        .rankingParams(PrototypeRankingParameters.sensibleDefaults())
                         .domains(new ArrayList<>())
                         .searchSetIdentifier("NONE")
                         .query(
@@ -225,8 +239,15 @@ public class IndexQueryServiceIntegrationSmokeTest {
 
         var rsp = queryService.justQuery(
                 SearchSpecification.builder()
-                        .queryLimits(new QueryLimits(10, 10, Integer.MAX_VALUE, 4000))
+                        .queryLimits(
-                        .rankingParams(ResultRankingParameters.sensibleDefaults())
+                                RpcQueryLimits.newBuilder()
+                                        .setResultsByDomain(10)
+                                        .setResultsTotal(10)
+                                        .setTimeoutMs(Integer.MAX_VALUE)
+                                        .setFetchSize(4000)
+                                        .build()
+                        )
+                        .rankingParams(PrototypeRankingParameters.sensibleDefaults())
                         .queryStrategy(QueryStrategy.SENTENCE)
                         .domains(List.of(2))
                         .query(
@@ -282,11 +303,18 @@ public class IndexQueryServiceIntegrationSmokeTest {
 
         var rsp = queryService.justQuery(
                 SearchSpecification.builder()
-                        .queryLimits(new QueryLimits(10, 10, Integer.MAX_VALUE, 4000))
+                        .queryLimits(
+                                RpcQueryLimits.newBuilder()
+                                        .setResultsByDomain(10)
+                                        .setResultsTotal(10)
+                                        .setTimeoutMs(Integer.MAX_VALUE)
+                                        .setFetchSize(4000)
+                                        .build()
+                        )
                         .year(SpecificationLimit.equals(1998))
                         .queryStrategy(QueryStrategy.SENTENCE)
                         .searchSetIdentifier("NONE")
-                        .rankingParams(ResultRankingParameters.sensibleDefaults())
+                        .rankingParams(PrototypeRankingParameters.sensibleDefaults())
                         .query(
                             SearchQuery.builder()
                                 .compiledQuery("4")

code/index/test/nu/marginalia/index/IndexQueryServiceIntegrationTest.java

@@ -4,10 +4,11 @@ import com.google.inject.Guice;
 import com.google.inject.Inject;
 import it.unimi.dsi.fastutil.ints.IntList;
 import nu.marginalia.IndexLocations;
+import nu.marginalia.api.searchquery.RpcQueryLimits;
 import nu.marginalia.api.searchquery.model.query.SearchPhraseConstraint;
 import nu.marginalia.api.searchquery.model.query.SearchQuery;
 import nu.marginalia.api.searchquery.model.query.SearchSpecification;
-import nu.marginalia.api.searchquery.model.results.ResultRankingParameters;
+import nu.marginalia.api.searchquery.model.results.PrototypeRankingParameters;
 import nu.marginalia.hash.MurmurHash3_128;
 import nu.marginalia.index.construction.DocIdRewriter;
 import nu.marginalia.index.construction.full.FullIndexConstructor;
@@ -18,7 +19,6 @@ import nu.marginalia.index.forward.construction.ForwardIndexConverter;
 import nu.marginalia.index.index.StatefulIndex;
 import nu.marginalia.index.journal.IndexJournal;
 import nu.marginalia.index.journal.IndexJournalSlopWriter;
-import nu.marginalia.index.query.limit.QueryLimits;
 import nu.marginalia.index.query.limit.QueryStrategy;
 import nu.marginalia.index.query.limit.SpecificationLimit;
 import nu.marginalia.linkdb.docs.DocumentDbReader;
@@ -389,13 +389,20 @@ public class IndexQueryServiceIntegrationTest {
     SearchSpecification basicQuery(Function<SearchSpecification.SearchSpecificationBuilder, SearchSpecification.SearchSpecificationBuilder> mutator)
     {
         var builder = SearchSpecification.builder()
-                .queryLimits(new QueryLimits(10, 10, Integer.MAX_VALUE, 4000))
+                .queryLimits(
+                        RpcQueryLimits.newBuilder()
+                                .setResultsByDomain(10)
+                                .setResultsTotal(10)
+                                .setTimeoutMs(Integer.MAX_VALUE)
+                                .setFetchSize(4000)
+                                .build()
+                )
                 .queryStrategy(QueryStrategy.SENTENCE)
                 .year(SpecificationLimit.none())
                 .quality(SpecificationLimit.none())
                 .size(SpecificationLimit.none())
                 .rank(SpecificationLimit.none())
-                .rankingParams(ResultRankingParameters.sensibleDefaults())
+                .rankingParams(PrototypeRankingParameters.sensibleDefaults())
                 .domains(new ArrayList<>())
                 .searchSetIdentifier("NONE");
 

code/index/test/nu/marginalia/index/results/DomainRankingOverridesTest.java

@@ -0,0 +1,103 @@
+package nu.marginalia.index.results;
+
+import com.zaxxer.hikari.HikariConfig;
+import com.zaxxer.hikari.HikariDataSource;
+import nu.marginalia.db.DbDomainQueries;
+import nu.marginalia.model.EdgeDomain;
+import nu.marginalia.test.TestMigrationLoader;
+import org.junit.jupiter.api.Assertions;
+import org.junit.jupiter.api.BeforeAll;
+import org.junit.jupiter.api.Tag;
+import org.junit.jupiter.api.Test;
+import org.junit.jupiter.api.parallel.Execution;
+import org.junit.jupiter.api.parallel.ExecutionMode;
+import org.testcontainers.containers.MariaDBContainer;
+import org.testcontainers.junit.jupiter.Container;
+import org.testcontainers.junit.jupiter.Testcontainers;
+
+import java.io.IOException;
+import java.nio.file.Files;
+import java.nio.file.Path;
+import java.nio.file.StandardOpenOption;
+import java.sql.SQLException;
+
+@Testcontainers
+@Execution(ExecutionMode.SAME_THREAD)
+@Tag("slow")
+class DomainRankingOverridesTest {
+    @Container
+    static MariaDBContainer<?> mariaDBContainer = new MariaDBContainer<>("mariadb")
+            .withDatabaseName("WMSA_prod")
+            .withUsername("wmsa")
+            .withPassword("wmsa")
+            .withNetworkAliases("mariadb");
+
+    private static DbDomainQueries domainQueries;
+
+    @BeforeAll
+    public static void setup() throws SQLException {
+        HikariConfig config = new HikariConfig();
+        config.setJdbcUrl(mariaDBContainer.getJdbcUrl());
+        config.setUsername("wmsa");
+        config.setPassword("wmsa");
+
+        var dataSource = new HikariDataSource(config);
+
+        TestMigrationLoader.flywayMigration(dataSource);
+
+        try (var conn = dataSource.getConnection();
+             var stmt = conn.createStatement()) {
+            stmt.executeQuery("DELETE FROM EC_DOMAIN"); // Wipe any old state from other test runs
+
+            stmt.executeQuery("INSERT INTO EC_DOMAIN (DOMAIN_NAME, DOMAIN_TOP, NODE_AFFINITY) VALUES ('first.example.com', 'example.com', 1)");
+            stmt.executeQuery("INSERT INTO EC_DOMAIN (DOMAIN_NAME, DOMAIN_TOP, NODE_AFFINITY) VALUES ('second.example.com', 'example.com', 1)");
+            stmt.executeQuery("INSERT INTO EC_DOMAIN (DOMAIN_NAME, DOMAIN_TOP, NODE_AFFINITY) VALUES ('third.example.com', 'example.com', 1)");
+            stmt.executeQuery("INSERT INTO EC_DOMAIN (DOMAIN_NAME, DOMAIN_TOP, NODE_AFFINITY) VALUES ('not-added.example.com', 'example.com', 1)");
+        }
+
+        domainQueries = new DbDomainQueries(dataSource);
+
+    }
+
+    @Test
+    public void test() throws IOException {
+
+        Path overridesFile = Files.createTempFile(getClass().getSimpleName(), ".txt");
+        try {
+
+            Files.writeString(overridesFile, """
+                    # A comment
+                    value 0.75
+                    domain first.example.com
+                    domain second.example.com
+                    
+                    value 1.1
+                    domain third.example.com
+                    """,
+                    StandardOpenOption.APPEND);
+
+            var overrides = new DomainRankingOverrides(domainQueries, overridesFile);
+
+            overrides.reloadFile();
+
+            Assertions.assertEquals(0.75, overrides.getRankingFactor(
+                    domainQueries.getDomainId(new EdgeDomain("first.example.com"))
+            ));
+            Assertions.assertEquals(0.75, overrides.getRankingFactor(
+                    domainQueries.getDomainId(new EdgeDomain("second.example.com"))
+            ));
+            Assertions.assertEquals(1.1, overrides.getRankingFactor(
+                    domainQueries.getDomainId(new EdgeDomain("third.example.com"))
+            ));
+            Assertions.assertEquals(1.0, overrides.getRankingFactor(
+                    domainQueries.getDomainId(new EdgeDomain("not-added.example.com"))
+            ));
+            Assertions.assertEquals(1.0, overrides.getRankingFactor(1<<23));
+
+        }
+        finally {
+            Files.deleteIfExists(overridesFile);
+        }
+    }
+
+}

code/libraries/blocking-thread-pool/java/nu/marginalia/util/SimpleBlockingThreadPool.java

@@ -23,16 +23,33 @@ public class SimpleBlockingThreadPool {
     private final Logger logger = LoggerFactory.getLogger(SimpleBlockingThreadPool.class);
 
     public SimpleBlockingThreadPool(String name, int poolSize, int queueSize) {
+        this(name, poolSize, queueSize, ThreadType.PLATFORM);
+    }
+
+    public SimpleBlockingThreadPool(String name, int poolSize, int queueSize, ThreadType threadType) {
         tasks = new ArrayBlockingQueue<>(queueSize);
 
         for (int i = 0; i < poolSize; i++) {
-            Thread worker = new Thread(this::worker, name  + "[" + i + "]");
+
-            worker.setDaemon(true);
+            Thread.Builder threadBuilder = switch (threadType) {
-            worker.start();
+                case VIRTUAL -> Thread.ofVirtual();
+                case PLATFORM -> Thread.ofPlatform().daemon(true);
+            };
+
+            Thread worker = threadBuilder
+                    .name(name  + "[" + i + "]")
+                    .start(this::worker);
+
             workers.add(worker);
         }
 
     }
+
+    public enum ThreadType {
+        VIRTUAL,
+        PLATFORM
+    }
+
     public void submit(Task task) throws InterruptedException {
         tasks.put(task);
     }

code/libraries/coded-sequence/java/nu/marginalia/sequence/slop/GammaCodedSequenceArrayColumn.java

@@ -45,6 +45,11 @@ public class GammaCodedSequenceArrayColumn extends AbstractObjectColumn<List<Gam
         );
     }
 
+    @Override
+    public int alignmentSize() {
+        return 1;
+    }
+
     public Reader openUnregistered(URI uri, int page) throws IOException {
         return new Reader(
                 dataColumn.openUnregistered(uri, page),
@@ -109,6 +114,11 @@ public class GammaCodedSequenceArrayColumn extends AbstractObjectColumn<List<Gam
             dataReader.skip(toSkip);
         }
 
+        @Override
+        public boolean isDirect() {
+            return dataReader.isDirect();
+        }
+
         @Override
         public boolean hasRemaining() throws IOException {
             return groupsReader.hasRemaining();

code/libraries/coded-sequence/java/nu/marginalia/sequence/slop/GammaCodedSequenceColumn.java

@@ -44,6 +44,11 @@ public class GammaCodedSequenceColumn extends AbstractObjectColumn<GammaCodedSeq
         );
     }
 
+    @Override
+    public int alignmentSize() {
+        return 1;
+    }
+
     public Reader openUnregistered(URI uri, int page) throws IOException {
         return new Reader(
                 Storage.reader(uri, this, page, false),
@@ -96,6 +101,11 @@ public class GammaCodedSequenceColumn extends AbstractObjectColumn<GammaCodedSeq
             this.indexReader = indexReader;
         }
 
+        @Override
+        public boolean isDirect() {
+            return storage.isDirect();
+        }
+
         @Override
         public AbstractColumn<?, ?> columnDesc() {
             return GammaCodedSequenceColumn.this;

code/libraries/coded-sequence/java/nu/marginalia/sequence/slop/VarintCodedSequenceArrayColumn.java

@@ -45,6 +45,11 @@ public class VarintCodedSequenceArrayColumn extends AbstractObjectColumn<List<Va
         );
     }
 
+    @Override
+    public int alignmentSize() {
+        return 0;
+    }
+
     public Reader openUnregistered(URI uri, int page) throws IOException {
         return new Reader(
                 dataColumn.openUnregistered(uri, page),
@@ -109,6 +114,11 @@ public class VarintCodedSequenceArrayColumn extends AbstractObjectColumn<List<Va
             dataReader.skip(toSkip);
         }
 
+        @Override
+        public boolean isDirect() {
+            return dataReader.isDirect();
+        }
+
         @Override
         public boolean hasRemaining() throws IOException {
             return groupsReader.hasRemaining();

code/libraries/coded-sequence/java/nu/marginalia/sequence/slop/VarintCodedSequenceColumn.java

@@ -44,6 +44,11 @@ public class VarintCodedSequenceColumn extends AbstractObjectColumn<VarintCodedS
         );
     }
 
+    @Override
+    public int alignmentSize() {
+        return 1;
+    }
+
     public Reader openUnregistered(URI uri, int page) throws IOException {
         return new Reader(
                 Storage.reader(uri, this, page, false),
@@ -101,6 +106,11 @@ public class VarintCodedSequenceColumn extends AbstractObjectColumn<VarintCodedS
             return VarintCodedSequenceColumn.this;
         }
 
+        @Override
+        public boolean isDirect() {
+            return storage.isDirect();
+        }
+
         @Override
         public void skip(long positions) throws IOException {
             for (int i = 0; i < positions; i++) {

code/libraries/language-processing/java/nu/marginalia/language/sentence/SentenceExtractor.java

@@ -155,8 +155,15 @@ public class SentenceExtractor {
     public List<DocumentSentence> extractSentencesFromString(String text, EnumSet<HtmlTag> htmlTags) {
         String[] sentences;
 
-        // Normalize spaces
+        // Safety net against malformed data DOS attacks,
+        // found 5+ MB <p>-tags in the wild that just break
+        // the sentence extractor causing it to stall forever.
+        if (text.length() > 50_000) {
+            // 50k chars can hold a small novel, let alone single html tags
+            text = text.substring(0, 50_000);
+        }
 
+        // Normalize spaces
         text = normalizeSpaces(text);
 
         // Split into sentences

code/libraries/message-queue/java/nu/marginalia/actor/prototype/RecordActorPrototype.java

@@ -5,9 +5,7 @@ import nu.marginalia.actor.state.*;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 
-import java.util.ArrayList;
+import java.util.*;
-import java.util.Arrays;
-import java.util.List;
 
 public abstract class RecordActorPrototype implements ActorPrototype {
 
@@ -118,7 +116,7 @@ public abstract class RecordActorPrototype implements ActorPrototype {
         }
 
         private String functionName(Class<? extends ActorStep> functionClass) {
-            return functionClass.getSimpleName().toUpperCase();
+            return ActorStep.functionName(functionClass);
         }
 
         private ActorStep constructState(String message) throws ReflectiveOperationException {
@@ -145,4 +143,43 @@ public abstract class RecordActorPrototype implements ActorPrototype {
         }
     }
 
+    /** Get a list of JSON prototypes for each actor step declared by this actor */
+    @SuppressWarnings("unchecked")
+    public Map<String, String> getMessagePrototypes() {
+        Map<String, String> messagePrototypes = new HashMap<>();
+
+        for (var clazz : getClass().getDeclaredClasses()) {
+            if (!clazz.isRecord() || !ActorStep.class.isAssignableFrom(clazz))
+                continue;
+
+            StringJoiner sj = new StringJoiner(",\n\t", "{\n\t", "\n}");
+
+            renderToJsonPrototype(sj, (Class<? extends Record>) clazz);
+
+            messagePrototypes.put(ActorStep.functionName((Class<? extends ActorStep>) clazz), sj.toString());
+        }
+
+        return messagePrototypes;
+    }
+
+    @SuppressWarnings("unchecked")
+    private void renderToJsonPrototype(StringJoiner sj, Class<? extends Record> recordType) {
+        for (var field : recordType.getDeclaredFields()) {
+            String typeName = field.getType().getSimpleName();
+
+            if ("List".equals(typeName)) {
+                sj.add(String.format("\"%s\": [ ]", field.getName()));
+            }
+            else if (field.getType().isRecord()) {
+                var innerSj = new StringJoiner(",", "{", "}");
+                renderToJsonPrototype(innerSj, (Class<? extends Record>) field.getType());
+                sj.add(String.format("\"%s\": %s", field.getName(), sj));
+            }
+            else {
+                sj.add(String.format("\"%s\": \"%s\"", field.getName(), typeName));
+            }
+        }
+
+    }
+
 }

code/libraries/message-queue/java/nu/marginalia/actor/state/ActorStep.java

@@ -1,3 +1,7 @@
 package nu.marginalia.actor.state;
 
-public interface ActorStep {}
+public interface ActorStep {
+    static String functionName(Class<? extends ActorStep> type) {
+        return type.getSimpleName().toUpperCase();
+    }
+}

code/processes/converting-process/build.gradle

@@ -62,6 +62,7 @@ dependencies {
     implementation libs.jwarc
 
     implementation libs.jsoup
+    implementation libs.pdfbox
 
     implementation libs.guava
     implementation dependencies.create(libs.guice.get()) {
@@ -87,6 +88,8 @@ dependencies {
     implementation libs.commons.compress
     implementation libs.sqlite
 
+    implementation libs.bundles.httpcomponents
+
     testImplementation libs.bundles.slf4j.test
     testImplementation libs.bundles.junit
     testImplementation libs.mockito

code/processes/converting-process/ft-keyword-extraction/java/nu/marginalia/keyword/DocumentPositionMapper.java

@@ -152,7 +152,10 @@ public class DocumentPositionMapper {
     }
 
     boolean matchesWordPattern(String s) {
-        // this function is an unrolled version of the regexp [\da-zA-Z]{1,15}([.\-_/:+*][\da-zA-Z]{1,10}){0,4}
+        if (s.length() > 48)
+            return false;
+
+        // this function is an unrolled version of the regexp [\da-zA-Z]{1,15}([.\-_/:+*][\da-zA-Z]{1,10}){0,8}
 
         String wordPartSeparator = ".-_/:+*";
 
@@ -169,7 +172,7 @@ public class DocumentPositionMapper {
         if (i == 0)
             return false;
 
-        for (int j = 0; j < 5; j++) {
+        for (int j = 0; j < 8; j++) {
             if (i == s.length()) return true;
 
             if (wordPartSeparator.indexOf(s.charAt(i)) < 0) {

code/processes/converting-process/ft-keyword-extraction/test/nu/marginalia/keyword/DocumentPositionMapperTest.java

@@ -30,9 +30,11 @@ class DocumentPositionMapperTest {
         Assertions.assertFalse(positionMapper.matchesWordPattern("1234567890abcdef"));
 
         Assertions.assertTrue(positionMapper.matchesWordPattern("test-test-test-test-test"));
-        Assertions.assertFalse(positionMapper.matchesWordPattern("test-test-test-test-test-test"));
+        Assertions.assertFalse(positionMapper.matchesWordPattern("test-test-test-test-test-test-test-test-test"));
         Assertions.assertTrue(positionMapper.matchesWordPattern("192.168.1.100/24"));
         Assertions.assertTrue(positionMapper.matchesWordPattern("std::vector"));
+        Assertions.assertTrue(positionMapper.matchesWordPattern("std::vector::push_back"));
+
         Assertions.assertTrue(positionMapper.matchesWordPattern("c++"));
         Assertions.assertTrue(positionMapper.matchesWordPattern("m*a*s*h"));
         Assertions.assertFalse(positionMapper.matchesWordPattern("Stulpnagelstrasse"));

code/processes/converting-process/java/nu/marginalia/converting/ConverterMain.java

@@ -12,7 +12,6 @@ import nu.marginalia.converting.sideload.SideloadSourceFactory;
 import nu.marginalia.converting.writer.ConverterBatchWritableIf;
 import nu.marginalia.converting.writer.ConverterBatchWriter;
 import nu.marginalia.converting.writer.ConverterWriter;
-import nu.marginalia.io.CrawledDomainReader;
 import nu.marginalia.io.SerializableCrawlDataStream;
 import nu.marginalia.mq.MessageQueueFactory;
 import nu.marginalia.mqapi.converting.ConvertRequest;
@@ -36,6 +35,7 @@ import java.io.IOException;
 import java.nio.file.Files;
 import java.nio.file.Path;
 import java.sql.SQLException;
+import java.util.ArrayList;
 import java.util.Collection;
 import java.util.List;
 import java.util.Optional;
@@ -51,6 +51,7 @@ public class ConverterMain extends ProcessMainClass {
     private final ProcessHeartbeat heartbeat;
     private final FileStorageService fileStorageService;
     private final SideloadSourceFactory sideloadSourceFactory;
+    private static final int SIDELOAD_THRESHOLD = Integer.getInteger("converter.sideloadThreshold", 10_000);
 
     public static void main(String... args) throws Exception {
 
@@ -201,12 +202,26 @@ public class ConverterMain extends ProcessMainClass {
             processedDomains.set(batchingWorkLog.size());
             heartbeat.setProgress(processedDomains.get() / (double) totalDomains);
 
-            for (var domain : WorkLog.iterableMap(crawlDir.getLogFile(),
+            logger.info("Processing small items");
+
+            // We separate the large and small domains to reduce the number of critical sections,
+            // as the large domains have a separate processing track that doesn't store everything
+            // in memory
+
+            final List<Path> bigTasks = new ArrayList<>();
+
+            // First process the small items
+            for (var dataPath : WorkLog.iterableMap(crawlDir.getLogFile(),
                     new CrawlDataLocator(crawlDir.getDir(), batchingWorkLog)))
             {
+                if (SerializableCrawlDataStream.getSizeHint(dataPath) >= SIDELOAD_THRESHOLD) {
+                    bigTasks.add(dataPath);
+                    continue;
+                }
+
                 pool.submit(() -> {
-                    try {
+                    try (var dataStream = SerializableCrawlDataStream.openDataStream(dataPath)) {
-                        ConverterBatchWritableIf writable = processor.createWritable(domain);
+                        ConverterBatchWritableIf writable = processor.fullProcessing(dataStream) ;
                         converterWriter.accept(writable);
                     }
                     catch (Exception ex) {
@@ -225,10 +240,39 @@ public class ConverterMain extends ProcessMainClass {
             do {
                 System.out.println("Waiting for pool to terminate... " + pool.getActiveCount() + " remaining");
             } while (!pool.awaitTermination(60, TimeUnit.SECONDS));
+
+            logger.info("Processing large items");
+
+            try (var hb = heartbeat.createAdHocTaskHeartbeat("Large Domains")) {
+                int bigTaskIdx = 0;
+                // Next the big items domain-by-domain
+                for (var dataPath : bigTasks) {
+                    hb.progress(dataPath.toFile().getName(), bigTaskIdx++, bigTasks.size());
+
+                    try {
+                        // SerializableCrawlDataStream is autocloseable, we can't try-with-resources because then it will be
+                        // closed before it's consumed by the converterWriter.  Instead, the converterWriter guarantees it
+                        // will close it after it's consumed.
+
+                        var stream = SerializableCrawlDataStream.openDataStream(dataPath);
+                        ConverterBatchWritableIf writable = processor.simpleProcessing(stream, SerializableCrawlDataStream.getSizeHint(dataPath));
+
+                        converterWriter.accept(writable);
+                    }
+                    catch (Exception ex) {
+                        logger.info("Error in processing", ex);
+                    }
+                    finally {
+                        heartbeat.setProgress(processedDomains.incrementAndGet() / (double) totalDomains);
+                    }
+                }
+            }
+
+            logger.info("Processing complete");
         }
     }
 
-    private static class CrawlDataLocator implements Function<WorkLogEntry, Optional<SerializableCrawlDataStream>> {
+    private static class CrawlDataLocator implements Function<WorkLogEntry, Optional<Path>> {
 
         private final Path crawlRootDir;
         private final BatchingWorkLog batchingWorkLog;
@@ -239,7 +283,7 @@ public class ConverterMain extends ProcessMainClass {
         }
 
         @Override
-        public Optional<SerializableCrawlDataStream> apply(WorkLogEntry entry) {
+        public Optional<Path> apply(WorkLogEntry entry) {
             if (batchingWorkLog.isItemProcessed(entry.id())) {
                 return Optional.empty();
             }
@@ -252,7 +296,7 @@ public class ConverterMain extends ProcessMainClass {
             }
 
             try {
-                return Optional.of(CrawledDomainReader.createDataStream(path));
+                return Optional.of(path);
             }
             catch (Exception ex) {
                 return Optional.empty();

code/processes/converting-process/java/nu/marginalia/converting/model/ProcessedDocumentDetails.java

@@ -1,8 +1,8 @@
 package nu.marginalia.converting.model;
 
+import nu.marginalia.model.DocumentFormat;
 import nu.marginalia.model.EdgeUrl;
 import nu.marginalia.model.crawl.HtmlFeature;
-import nu.marginalia.model.html.HtmlStandard;
 import nu.marginalia.model.idx.DocumentMetadata;
 
 import javax.annotation.Nullable;
@@ -21,7 +21,7 @@ public class ProcessedDocumentDetails {
     public long hashCode;
 
     public Set<HtmlFeature> features;
-    public HtmlStandard standard;
+    public DocumentFormat format;
 
     public List<EdgeUrl> linksInternal;
     public List<EdgeUrl> linksExternal;
@@ -30,6 +30,6 @@ public class ProcessedDocumentDetails {
     public GeneratorType generator;
 
     public String toString() {
-        return "ProcessedDocumentDetails(title=" + this.title + ", description=" + this.description + ", pubYear=" + this.pubYear + ", length=" + this.length + ", quality=" + this.quality + ", hashCode=" + this.hashCode + ", features=" + this.features + ", standard=" + this.standard + ", linksInternal=" + this.linksInternal + ", linksExternal=" + this.linksExternal + ", metadata=" + this.metadata + ", generator=" + this.generator + ")";
+        return "ProcessedDocumentDetails(title=" + this.title + ", description=" + this.description + ", pubYear=" + this.pubYear + ", length=" + this.length + ", quality=" + this.quality + ", hashCode=" + this.hashCode + ", features=" + this.features + ", standard=" + this.format + ", linksInternal=" + this.linksInternal + ", linksExternal=" + this.linksExternal + ", metadata=" + this.metadata + ", generator=" + this.generator + ")";
     }
 }

code/processes/converting-process/java/nu/marginalia/converting/processor/DocumentProcessor.java

@@ -7,6 +7,7 @@ import nu.marginalia.converting.model.DisqualifiedException;
 import nu.marginalia.converting.model.ProcessedDocument;
 import nu.marginalia.converting.processor.plugin.AbstractDocumentProcessorPlugin;
 import nu.marginalia.converting.processor.plugin.HtmlDocumentProcessorPlugin;
+import nu.marginalia.converting.processor.plugin.PdfDocumentProcessorPlugin;
 import nu.marginalia.converting.processor.plugin.PlainTextDocumentProcessorPlugin;
 import nu.marginalia.keyword.LinkTexts;
 import nu.marginalia.model.EdgeDomain;
@@ -19,6 +20,7 @@ import nu.marginalia.model.idx.WordFlags;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 
+import java.io.IOException;
 import java.net.URISyntaxException;
 import java.util.ArrayList;
 import java.util.List;
@@ -32,7 +34,8 @@ public class DocumentProcessor {
     private static final Set<String> acceptedContentTypes = Set.of("application/xhtml+xml",
             "application/xhtml",
             "text/html",
-            "text/plain");
+            "text/plain",
+            "application/pdf");
 
 
     private final List<AbstractDocumentProcessorPlugin> processorPlugins = new ArrayList<>();
@@ -41,12 +44,14 @@ public class DocumentProcessor {
     @Inject
     public DocumentProcessor(HtmlDocumentProcessorPlugin htmlDocumentProcessorPlugin,
                              PlainTextDocumentProcessorPlugin plainTextDocumentProcessorPlugin,
+                             PdfDocumentProcessorPlugin pdfDocumentProcessorPlugin,
                              AnchorTextKeywords anchorTextKeywords)
     {
         this.anchorTextKeywords = anchorTextKeywords;
 
         processorPlugins.add(htmlDocumentProcessorPlugin);
         processorPlugins.add(plainTextDocumentProcessorPlugin);
+        processorPlugins.add(pdfDocumentProcessorPlugin);
     }
 
     public ProcessedDocument process(CrawledDocument crawledDocument,
@@ -91,7 +96,7 @@ public class DocumentProcessor {
                                  DocumentClass documentClass,
                                  DocumentDecorator documentDecorator,
                                  DomainLinks externalDomainLinks,
-                                 ProcessedDocument ret) throws URISyntaxException, DisqualifiedException
+                                 ProcessedDocument ret) throws URISyntaxException, IOException, DisqualifiedException
     {
 
         var crawlerStatus = CrawlerDocumentStatus.valueOf(crawledDocument.crawlerStatus);
@@ -109,7 +114,7 @@ public class DocumentProcessor {
 
         ret.state = crawlerStatusToUrlState(crawledDocument.crawlerStatus, crawledDocument.httpStatus);
 
-        final var plugin = findPlugin(crawledDocument);
+        AbstractDocumentProcessorPlugin plugin = findPlugin(crawledDocument);
 
         EdgeUrl url = new EdgeUrl(crawledDocument.url);
         LinkTexts linkTexts = anchorTextKeywords.getAnchorTextKeywords(externalDomainLinks, url);

code/processes/converting-process/java/nu/marginalia/converting/processor/DomainProcessor.java

@@ -32,7 +32,6 @@ import java.util.*;
 import java.util.regex.Pattern;
 
 public class DomainProcessor {
-    private static final int SIDELOAD_THRESHOLD = Integer.getInteger("converter.sideloadThreshold", 10_000);
     private final DocumentProcessor documentProcessor;
     private final SiteWords siteWords;
     private final AnchorTagsSource anchorTagsSource;
@@ -54,21 +53,9 @@ public class DomainProcessor {
         geoIpDictionary.waitReady();
     }
 
-    public ConverterBatchWritableIf createWritable(SerializableCrawlDataStream domain) {
+    public SimpleProcessing simpleProcessing(SerializableCrawlDataStream dataStream, int sizeHint, Collection<String> extraKeywords) {
-        final int sizeHint = domain.sizeHint();
-
-        if (sizeHint > SIDELOAD_THRESHOLD) {
-            // If the file is too big, we run a processing mode that doesn't
-            // require loading the entire dataset into RAM
-            return sideloadProcessing(domain, sizeHint);
-        }
-
-        return fullProcessing(domain);
-    }
-
-    public SideloadProcessing sideloadProcessing(SerializableCrawlDataStream dataStream, int sizeHint, Collection<String> extraKeywords) {
         try {
-            return new SideloadProcessing(dataStream, sizeHint, extraKeywords);
+            return new SimpleProcessing(dataStream, sizeHint, extraKeywords);
         }
         catch (Exception ex) {
             logger.warn("Failed to process domain sideload", ex);
@@ -76,9 +63,9 @@ public class DomainProcessor {
         }
     }
 
-    public SideloadProcessing sideloadProcessing(SerializableCrawlDataStream dataStream, int sizeHint) {
+    public SimpleProcessing simpleProcessing(SerializableCrawlDataStream dataStream, int sizeHint) {
         try {
-            return new SideloadProcessing(dataStream, sizeHint);
+            return new SimpleProcessing(dataStream, sizeHint);
         }
         catch (Exception ex) {
             logger.warn("Failed to process domain sideload", ex);
@@ -86,22 +73,84 @@ public class DomainProcessor {
         }
     }
 
-    public class SideloadProcessing implements ConverterBatchWritableIf, SideloadSource {
+    @Nullable
+    public ProcessedDomain fullProcessing(SerializableCrawlDataStream dataStream) {
+        try {
+            if (!dataStream.hasNext()) {
+                return null;
+            }
+
+            List<ProcessedDocument> docs = new ArrayList<>();
+            Set<String> processedUrls = new HashSet<>();
+
+            if (!(dataStream.next() instanceof CrawledDomain crawledDomain)) {
+                throw new IllegalStateException("First record must be a domain, was " + dataStream.next().getClass().getSimpleName());
+            }
+
+            DomainLinks externalDomainLinks = anchorTagsSource.getAnchorTags(crawledDomain.getDomain());
+            DocumentDecorator documentDecorator = new DocumentDecorator();
+
+            // Process Domain Record
+
+            ProcessedDomain ret = new ProcessedDomain();
+            processDomain(crawledDomain, ret, documentDecorator);
+            ret.documents = docs;
+
+            // Process Documents
+
+            try (var deduplicator = new LshDocumentDeduplicator()) {
+                while (dataStream.hasNext()) {
+                    if (!(dataStream.next() instanceof CrawledDocument doc))
+                        continue;
+                    if (doc.url == null)
+                        continue;
+                    if (doc.documentBodyBytes.length == 0)
+                        continue;
+                    if (!processedUrls.add(doc.url))
+                        continue;
+
+                    try {
+                        var processedDoc = documentProcessor.process(doc, ret.domain, externalDomainLinks, documentDecorator);
+                        deduplicator.markIfDuplicate(processedDoc);
+                        docs.add(processedDoc);
+                    } catch (Exception ex) {
+                        logger.warn("Failed to process " + doc.url, ex);
+                    }
+                }
+            }
+
+            // Add late keywords and features from domain-level information
+
+            calculateStatistics(ret, externalDomainLinks);
+
+            return ret;
+        }
+        catch (Exception ex) {
+            logger.warn("Failed to process domain", ex);
+            return null;
+        }
+    }
+
+    /** The simple processing track processes documents individually, and does not perform any domain-level analysis.
+     *  This is needed to process extremely large domains, which would otherwise eat up too much RAM.
+     */
+    public class SimpleProcessing implements ConverterBatchWritableIf, SideloadSource {
         private final SerializableCrawlDataStream dataStream;
         private final ProcessedDomain domain;
         private final DocumentDecorator documentDecorator;
         private final Set<String> processedUrls = new HashSet<>();
         private final DomainLinks externalDomainLinks;
         private final LshDocumentDeduplicator deduplicator = new LshDocumentDeduplicator();
+
         private static final ProcessingIterator.Factory iteratorFactory = ProcessingIterator.factory(8,
                 Integer.getInteger("java.util.concurrent.ForkJoinPool.common.parallelism", Runtime.getRuntime().availableProcessors())
         );
 
-        SideloadProcessing(SerializableCrawlDataStream dataStream, int sizeHint) throws IOException {
+        SimpleProcessing(SerializableCrawlDataStream dataStream, int sizeHint) throws IOException {
             this(dataStream, sizeHint, List.of());
         }
 
-        SideloadProcessing(SerializableCrawlDataStream dataStream, int sizeHint, Collection<String> extraKeywords) throws IOException {
+        SimpleProcessing(SerializableCrawlDataStream dataStream, int sizeHint, Collection<String> extraKeywords) throws IOException {
             this.dataStream = dataStream;
 
             if (!dataStream.hasNext() || !(dataStream.next() instanceof CrawledDomain crawledDomain))
@@ -128,6 +177,7 @@ public class DomainProcessor {
         @Override
         public Iterator<ProcessedDocument> getDocumentsStream() {
             return iteratorFactory.create((taskConsumer) -> {
+
                 while (dataStream.hasNext())
                 {
                     if (!(dataStream.next() instanceof CrawledDocument doc))
@@ -172,65 +222,6 @@ public class DomainProcessor {
         }
     }
 
-
-    @Nullable
-    public ProcessedDomain fullProcessing(SerializableCrawlDataStream dataStream) {
-        try {
-            if (!dataStream.hasNext()) {
-                return null;
-            }
-
-            List<ProcessedDocument> docs = new ArrayList<>();
-            Set<String> processedUrls = new HashSet<>();
-
-            if (!(dataStream.next() instanceof CrawledDomain crawledDomain)) {
-                throw new IllegalStateException("First record must be a domain, was " + dataStream.next().getClass().getSimpleName());
-            }
-
-            DomainLinks externalDomainLinks = anchorTagsSource.getAnchorTags(crawledDomain.getDomain());
-            DocumentDecorator documentDecorator = new DocumentDecorator();
-
-            // Process Domain Record
-
-            ProcessedDomain ret = new ProcessedDomain();
-            processDomain(crawledDomain, ret, documentDecorator);
-            ret.documents = docs;
-
-            // Process Documents
-
-            try (var deduplicator = new LshDocumentDeduplicator()) {
-                while (dataStream.hasNext()) {
-                    if (!(dataStream.next() instanceof CrawledDocument doc))
-                        continue;
-                    if (doc.url == null)
-                        continue;
-                    if (doc.documentBody.isBlank())
-                        continue;
-                    if (!processedUrls.add(doc.url))
-                        continue;
-
-                    try {
-                        var processedDoc = documentProcessor.process(doc, ret.domain, externalDomainLinks, documentDecorator);
-                        deduplicator.markIfDuplicate(processedDoc);
-                        docs.add(processedDoc);
-                    } catch (Exception ex) {
-                        logger.warn("Failed to process " + doc.url, ex);
-                    }
-                }
-            }
-
-            // Add late keywords and features from domain-level information
-
-            calculateStatistics(ret, externalDomainLinks);
-
-            return ret;
-        }
-        catch (Exception ex) {
-            logger.warn("Failed to process domain", ex);
-            return null;
-        }
-    }
-
     private void processDomain(CrawledDomain crawledDomain,
                                           ProcessedDomain domain,
                                           DocumentDecorator decorator)

code/processes/converting-process/java/nu/marginalia/converting/processor/classifier/adblock/AdblockSimulator.java

@@ -116,7 +116,7 @@ public class AdblockSimulator {
 
 
     // Refrain from cleaning up this code, it's very hot code and needs to be fast.
-    // This version is about 100x faster than the a "clean" first stab implementation.
+    // This version is about 100x faster than a "clean" first stab implementation.
 
     class RuleVisitor implements NodeFilter {
         public boolean sawAds;

code/processes/converting-process/java/nu/marginalia/converting/processor/logic/DocumentGeneratorExtractor.java

@@ -23,7 +23,7 @@ public class DocumentGeneratorExtractor {
 
         var tags = doc.select("meta[name=generator]");
 
-        if (tags.size() == 0) {
+        if (tags.isEmpty()) {
             // Some sites have a comment in the head instead of a meta tag
             return fingerprintServerTech(doc, responseHeaders);
         }

code/processes/converting-process/java/nu/marginalia/converting/processor/logic/DocumentValuator.java

@@ -2,9 +2,9 @@ package nu.marginalia.converting.processor.logic;
 
 import crawlercommons.utils.Strings;
 import nu.marginalia.converting.model.DisqualifiedException;
+import nu.marginalia.model.DocumentFormat;
 import nu.marginalia.model.crawl.HtmlFeature;
 import nu.marginalia.model.crawldata.CrawledDocument;
-import nu.marginalia.model.html.HtmlStandard;
 import org.jetbrains.annotations.NotNull;
 import org.jsoup.nodes.Document;
 import org.jsoup.nodes.Element;
@@ -17,14 +17,14 @@ import java.util.Set;
 public class DocumentValuator {
 
     public double getQuality(CrawledDocument crawledDocument,
-                             HtmlStandard htmlStandard,
+                             DocumentFormat htmlStandard,
                              Document parsedDocument,
                              int textLength) throws DisqualifiedException {
 
         double scriptPenalty = getScriptPenalty(parsedDocument);
         double chatGptPenalty = getChatGptContentFarmPenalty(parsedDocument);
 
-        int rawLength = crawledDocument.documentBody.length();
+        int rawLength = crawledDocument.documentBodyBytes.length;
 
         if (textLength == 0) {
             throw new DisqualifiedException(DisqualifiedException.DisqualificationReason.LENGTH);

code/processes/converting-process/java/nu/marginalia/converting/processor/logic/FeatureExtractor.java

@@ -218,7 +218,10 @@ public class FeatureExtractor {
             }
         }
 
-        if (features.contains(HtmlFeature.JS) && adblockSimulator.hasAds(doc.clone())) {
+        if (features.contains(HtmlFeature.JS)
+            // remove while disabled to get rid of expensive clone() call:
+            // adblockSimulator.hasAds(doc.clone())
+            ) {
             features.add(HtmlFeature.ADVERTISEMENT);
         }
 

code/processes/converting-process/java/nu/marginalia/converting/processor/logic/HtmlStandardExtractor.java

@@ -1,7 +1,7 @@
 package nu.marginalia.converting.processor.logic;
 
 import com.google.common.base.Strings;
-import nu.marginalia.model.html.HtmlStandard;
+import nu.marginalia.model.DocumentFormat;
 import org.jsoup.nodes.Document;
 import org.jsoup.nodes.DocumentType;
 import org.slf4j.Logger;
@@ -12,54 +12,54 @@ public class HtmlStandardExtractor {
 
     private static final Logger logger = LoggerFactory.getLogger(HtmlStandardExtractor.class);
 
-    public static HtmlStandard parseDocType(DocumentType docType) {
+    public static DocumentFormat parseDocType(DocumentType docType) {
         if (null == docType) {
-            return HtmlStandard.UNKNOWN;
+            return DocumentFormat.UNKNOWN;
         }
 
         String publicId = docType.publicId();
         if (Strings.isNullOrEmpty(publicId))
-            return HtmlStandard.HTML5;
+            return DocumentFormat.HTML5;
 
         publicId = publicId.toUpperCase();
         if (publicId.startsWith("-//SOFTQUAD SOFTWARE//DTD") && publicId.contains("HTML 4")) {
-            return HtmlStandard.HTML4;
+            return DocumentFormat.HTML4;
         }
         if (publicId.startsWith("-//SOFTQUAD SOFTWARE//DTD") && publicId.contains("HTML 3")) {
-            return HtmlStandard.HTML123;
+            return DocumentFormat.HTML123;
         }
         if (publicId.startsWith("-//INTERNET/RFC XXXX//EN"))
-            return HtmlStandard.HTML123;
+            return DocumentFormat.HTML123;
         if (publicId.startsWith("-//NETSCAPE COMM. CORP"))
-            return HtmlStandard.HTML123;
+            return DocumentFormat.HTML123;
         if (publicId.startsWith("-//SQ//DTD HTML 2"))
-            return HtmlStandard.HTML123;
+            return DocumentFormat.HTML123;
         if (publicId.startsWith("-//SOFTQUAD//DTD HTML 2"))
-            return HtmlStandard.HTML123;
+            return DocumentFormat.HTML123;
         if (publicId.startsWith("-//W3O//DTD W3 HTML 2"))
-            return HtmlStandard.HTML123;
+            return DocumentFormat.HTML123;
         if (publicId.startsWith("-//IETF//DTD HTML 2"))
-            return HtmlStandard.HTML123;
+            return DocumentFormat.HTML123;
         if (publicId.startsWith("-//IETF//DTD HTML//EN"))
-            return HtmlStandard.HTML123;
+            return DocumentFormat.HTML123;
         if (publicId.startsWith("-/W3C//DTD HTML 3"))
-            return HtmlStandard.HTML123;
+            return DocumentFormat.HTML123;
         if (publicId.startsWith("-/W3C/DTD HTML 3"))
-            return HtmlStandard.HTML123;
+            return DocumentFormat.HTML123;
         if (publicId.startsWith("-//IETF//DTD HTML 3"))
-            return HtmlStandard.HTML123;
+            return DocumentFormat.HTML123;
         if (publicId.startsWith("-//W3C//DTD XHTML"))
-            return HtmlStandard.XHTML;
+            return DocumentFormat.XHTML;
         if (publicId.startsWith("ISO/IEC 15445:2000//DTD"))
-            return HtmlStandard.XHTML;
+            return DocumentFormat.XHTML;
         if (publicId.startsWith("-//W3C//DTD HTML"))
-            return HtmlStandard.HTML4;
+            return DocumentFormat.HTML4;
 
         logger.debug("Unknown publicID standard {}", publicId);
-        return HtmlStandard.UNKNOWN;
+        return DocumentFormat.UNKNOWN;
     }
 
-    public static HtmlStandard sniffHtmlStandard(Document parsed) {
+    public static DocumentFormat sniffHtmlStandard(Document parsed) {
         int html4Attributes = 0;
         int html5Attributes = 0;
 
@@ -73,11 +73,11 @@ public class HtmlStandardExtractor {
             html4Attributes++;
         }
         if (html5Attributes > 0) {
-            return HtmlStandard.HTML5;
+            return DocumentFormat.HTML5;
         }
         if (html4Attributes > 0) {
-            return HtmlStandard.HTML4;
+            return DocumentFormat.HTML4;
         }
-        return HtmlStandard.HTML123;
+        return DocumentFormat.HTML123;
     }
 }

code/processes/converting-process/java/nu/marginalia/converting/processor/plugin/AbstractDocumentProcessorPlugin.java

@@ -7,13 +7,14 @@ import nu.marginalia.keyword.LinkTexts;
 import nu.marginalia.keyword.model.DocumentKeywordsBuilder;
 import nu.marginalia.language.filter.LanguageFilter;
 import nu.marginalia.language.model.DocumentLanguageData;
+import nu.marginalia.model.DocumentFormat;
 import nu.marginalia.model.EdgeUrl;
 import nu.marginalia.model.crawl.HtmlFeature;
 import nu.marginalia.model.crawl.PubDate;
 import nu.marginalia.model.crawldata.CrawledDocument;
-import nu.marginalia.model.html.HtmlStandard;
 
 import javax.annotation.Nullable;
+import java.io.IOException;
 import java.net.URISyntaxException;
 import java.util.HashSet;
 import java.util.List;
@@ -25,7 +26,7 @@ public abstract class AbstractDocumentProcessorPlugin {
         this.languageFilter = languageFilter;
     }
 
-    public abstract DetailsWithWords createDetails(CrawledDocument crawledDocument, LinkTexts linkTexts, DocumentClass documentClass) throws DisqualifiedException, URISyntaxException;
+    public abstract DetailsWithWords createDetails(CrawledDocument crawledDocument, LinkTexts linkTexts, DocumentClass documentClass) throws DisqualifiedException, URISyntaxException, IOException;
     public abstract boolean isApplicable(CrawledDocument doc);
 
     protected void checkDocumentLanguage(DocumentLanguageData dld) throws DisqualifiedException {
@@ -72,7 +73,7 @@ public abstract class AbstractDocumentProcessorPlugin {
             return this;
         }
 
-        public MetaTagsBuilder addFormat(HtmlStandard standard) {
+        public MetaTagsBuilder addFormat(DocumentFormat standard) {
 
             add("format", standard);
 
@@ -86,6 +87,7 @@ public abstract class AbstractDocumentProcessorPlugin {
 
             return this;
         }
+
         public MetaTagsBuilder addPubDate(PubDate pubDate) {
 
             if (pubDate.year() > 1900) {

code/processes/converting-process/java/nu/marginalia/converting/processor/plugin/HtmlDocumentProcessorPlugin.java

@@ -6,6 +6,7 @@ import nu.marginalia.converting.model.DisqualifiedException;
 import nu.marginalia.converting.model.DocumentHeaders;
 import nu.marginalia.converting.model.GeneratorType;
 import nu.marginalia.converting.model.ProcessedDocumentDetails;
+import nu.marginalia.converting.processor.AcceptableAds;
 import nu.marginalia.converting.processor.DocumentClass;
 import nu.marginalia.converting.processor.MetaRobotsTag;
 import nu.marginalia.converting.processor.logic.*;
@@ -24,19 +25,19 @@ import nu.marginalia.language.model.DocumentLanguageData;
 import nu.marginalia.language.sentence.ThreadLocalSentenceExtractorProvider;
 import nu.marginalia.link_parser.FeedExtractor;
 import nu.marginalia.link_parser.LinkParser;
+import nu.marginalia.model.DocumentFormat;
 import nu.marginalia.model.EdgeDomain;
 import nu.marginalia.model.EdgeUrl;
 import nu.marginalia.model.crawl.HtmlFeature;
 import nu.marginalia.model.crawl.PubDate;
 import nu.marginalia.model.crawldata.CrawledDocument;
-import nu.marginalia.model.html.HtmlStandard;
 import nu.marginalia.model.idx.DocumentFlags;
 import nu.marginalia.model.idx.DocumentMetadata;
-import org.jsoup.Jsoup;
 import org.jsoup.nodes.Document;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 
+import java.io.IOException;
 import java.net.URISyntaxException;
 import java.util.EnumSet;
 import java.util.HashSet;
@@ -51,7 +52,6 @@ public class HtmlDocumentProcessorPlugin extends AbstractDocumentProcessorPlugin
     private final double minDocumentQuality;
 
     private final FeatureExtractor featureExtractor;
-    private final TitleExtractor titleExtractor;
     private final DocumentKeywordExtractor keywordExtractor;
     private final PubDateSniffer pubDateSniffer;
 
@@ -74,7 +74,6 @@ public class HtmlDocumentProcessorPlugin extends AbstractDocumentProcessorPlugin
             @Named("min-document-quality") Double minDocumentQuality,
             LanguageFilter languageFilter,
             FeatureExtractor featureExtractor,
-            TitleExtractor titleExtractor,
             DocumentKeywordExtractor keywordExtractor,
             PubDateSniffer pubDateSniffer,
             DocumentLengthLogic documentLengthLogic,
@@ -89,7 +88,6 @@ public class HtmlDocumentProcessorPlugin extends AbstractDocumentProcessorPlugin
         this.minDocumentQuality = minDocumentQuality;
         this.featureExtractor = featureExtractor;
 
-        this.titleExtractor = titleExtractor;
         this.keywordExtractor = keywordExtractor;
         this.pubDateSniffer = pubDateSniffer;
         this.metaRobotsTag = metaRobotsTag;
@@ -108,19 +106,17 @@ public class HtmlDocumentProcessorPlugin extends AbstractDocumentProcessorPlugin
     public DetailsWithWords createDetails(CrawledDocument crawledDocument,
                                           LinkTexts linkTexts,
                                           DocumentClass documentClass)
-            throws DisqualifiedException, URISyntaxException {
+            throws DisqualifiedException, URISyntaxException, IOException {
 
-        String documentBody = crawledDocument.documentBody;
+        if (languageFilter.isBlockedUnicodeRange(crawledDocument.documentBody(512))) {
-
-        if (languageFilter.isBlockedUnicodeRange(documentBody)) {
             throw new DisqualifiedException(DisqualificationReason.LANGUAGE);
         }
 
-        if (documentBody.length() > MAX_DOCUMENT_LENGTH_BYTES) { // 128kb
+        Document doc = crawledDocument.parseBody();
-            documentBody = documentBody.substring(0, MAX_DOCUMENT_LENGTH_BYTES);
-        }
 
-        Document doc = Jsoup.parse(documentBody);
+        if (AcceptableAds.hasAcceptableAdsTag(doc)) {
+            throw new DisqualifiedException(DisqualifiedException.DisqualificationReason.ACCEPTABLE_ADS);
+        }
 
         if (!metaRobotsTag.allowIndexingByMetaTag(doc)) {
             throw new DisqualifiedException(DisqualificationReason.FORBIDDEN);
@@ -138,33 +134,34 @@ public class HtmlDocumentProcessorPlugin extends AbstractDocumentProcessorPlugin
         }
 
         var prunedDoc = specialization.prune(doc);
+
+
+        final int length = getLength(doc);
+        final DocumentFormat format = getDocumentFormat(doc);
+        final double quality = documentValuator.getQuality(crawledDocument, format, doc, length);
+
+        if (isDisqualified(documentClass, url, quality, doc.title())) {
+            throw new DisqualifiedException(DisqualificationReason.QUALITY);
+        }
+
         DocumentLanguageData dld = sentenceExtractorProvider.get().extractSentences(prunedDoc);
 
         checkDocumentLanguage(dld);
+        documentLengthLogic.validateLength(dld, specialization.lengthModifier() * documentClass.lengthLimitModifier());
 
         var ret = new ProcessedDocumentDetails();
 
-        final int length = getLength(doc);
-        final HtmlStandard standard = getHtmlStandard(doc);
-        final double quality = documentValuator.getQuality(crawledDocument, standard, doc, length);
-
         ret.length = length;
-        ret.standard = standard;
+        ret.format = format;
         ret.title = specialization.getTitle(doc, dld, crawledDocument.url);
 
-        documentLengthLogic.validateLength(dld, specialization.lengthModifier() * documentClass.lengthLimitModifier());
-
         final Set<HtmlFeature> features = featureExtractor.getFeatures(url, doc, documentHeaders, dld);
 
         ret.features = features;
         ret.quality = documentValuator.adjustQuality(quality, features);
         ret.hashCode = dld.localitySensitiveHashCode();
 
-        if (isDisqualified(documentClass, url, quality, ret.title)) {
+        PubDate pubDate = pubDateSniffer.getPubDate(documentHeaders, url, doc, format, true);
-            throw new DisqualifiedException(DisqualificationReason.QUALITY);
-        }
-
-        PubDate pubDate = pubDateSniffer.getPubDate(documentHeaders, url, doc, standard, true);
 
         EnumSet<DocumentFlags> documentFlags = documentFlags(features, generatorParts.type());
 
@@ -183,7 +180,7 @@ public class HtmlDocumentProcessorPlugin extends AbstractDocumentProcessorPlugin
                 .addPubDate(pubDate)
                 .addUrl(url)
                 .addFeatures(features)
-                .addFormat(standard)
+                .addFormat(format)
                 .addGenerator(generatorParts.keywords())
                 .build();
 
@@ -319,12 +316,12 @@ public class HtmlDocumentProcessorPlugin extends AbstractDocumentProcessorPlugin
         return linkTerms;
     }
 
-    private HtmlStandard getHtmlStandard(Document doc) {
+    private DocumentFormat getDocumentFormat(Document doc) {
-        HtmlStandard htmlStandard = HtmlStandardExtractor.parseDocType(doc.documentType());
+        DocumentFormat format = HtmlStandardExtractor.parseDocType(doc.documentType());
-        if (HtmlStandard.UNKNOWN.equals(htmlStandard)) {
+        if (DocumentFormat.UNKNOWN.equals(format)) {
             return HtmlStandardExtractor.sniffHtmlStandard(doc);
         }
-        return htmlStandard;
+        return format;
     }
 
     private int getLength(Document doc) {

code/processes/converting-process/java/nu/marginalia/converting/processor/plugin/PdfDocumentProcessorPlugin.java

@@ -0,0 +1,286 @@
+package nu.marginalia.converting.processor.plugin;
+
+import com.google.inject.Inject;
+import com.google.inject.name.Named;
+import nu.marginalia.converting.model.DisqualifiedException;
+import nu.marginalia.converting.model.ProcessedDocumentDetails;
+import nu.marginalia.converting.processor.DocumentClass;
+import nu.marginalia.converting.processor.logic.DocumentLengthLogic;
+import nu.marginalia.converting.processor.plugin.specialization.DefaultSpecialization;
+import nu.marginalia.keyword.DocumentKeywordExtractor;
+import nu.marginalia.keyword.LinkTexts;
+import nu.marginalia.keyword.model.DocumentKeywordsBuilder;
+import nu.marginalia.language.filter.LanguageFilter;
+import nu.marginalia.language.model.DocumentLanguageData;
+import nu.marginalia.language.sentence.ThreadLocalSentenceExtractorProvider;
+import nu.marginalia.model.DocumentFormat;
+import nu.marginalia.model.EdgeUrl;
+import nu.marginalia.model.crawl.HtmlFeature;
+import nu.marginalia.model.crawl.PubDate;
+import nu.marginalia.model.crawldata.CrawledDocument;
+import nu.marginalia.model.idx.DocumentFlags;
+import nu.marginalia.model.idx.DocumentMetadata;
+import org.apache.commons.lang3.StringUtils;
+import org.apache.pdfbox.Loader;
+import org.apache.pdfbox.text.HeadingAwarePDFTextStripper;
+import org.jsoup.Jsoup;
+import org.jsoup.nodes.Document;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+import java.io.IOException;
+import java.net.URISyntaxException;
+import java.time.LocalDate;
+import java.util.*;
+
+
+public class PdfDocumentProcessorPlugin extends AbstractDocumentProcessorPlugin {
+
+    private final int maxTitleLength;
+    private final DocumentKeywordExtractor keywordExtractor;
+    private final ThreadLocalSentenceExtractorProvider sentenceExtractorProvider;
+    private final DocumentLengthLogic documentLengthLogic;
+    private final DefaultSpecialization defaultSpecialization;
+
+    private static final Logger logger = LoggerFactory.getLogger(PdfDocumentProcessorPlugin.class);
+
+    @Inject
+    public PdfDocumentProcessorPlugin(@Named("max-title-length") Integer maxTitleLength,
+                                      LanguageFilter languageFilter,
+                                      ThreadLocalSentenceExtractorProvider sentenceExtractorProvider,
+                                      DocumentKeywordExtractor keywordExtractor,
+                                      DocumentLengthLogic documentLengthLogic,
+                                      DefaultSpecialization defaultSpecialization)
+
+    {
+        super(languageFilter);
+        this.sentenceExtractorProvider = sentenceExtractorProvider;
+        this.documentLengthLogic = documentLengthLogic;
+        this.maxTitleLength = maxTitleLength;
+        this.keywordExtractor = keywordExtractor;
+        this.defaultSpecialization = defaultSpecialization;
+    }
+
+    @Override
+    public boolean isApplicable(CrawledDocument doc) {
+        String contentType = doc.contentType.toLowerCase();
+
+        if (contentType.equals("application/pdf"))
+            return true;
+        if (contentType.startsWith("application/pdf;")) // charset=blabla
+            return true;
+
+        return false;
+    }
+
+    @Override
+    public DetailsWithWords createDetails(CrawledDocument crawledDocument,
+                                          LinkTexts linkTexts,
+                                          DocumentClass documentClass)
+            throws DisqualifiedException, URISyntaxException, IOException {
+
+        String documentBody = crawledDocument.documentBody();
+
+        if (languageFilter.isBlockedUnicodeRange(documentBody)) {
+            throw new DisqualifiedException(DisqualifiedException.DisqualificationReason.LANGUAGE);
+        }
+
+        final EdgeUrl url = new EdgeUrl(crawledDocument.url);
+
+
+        Document doc;
+        try {
+            doc = convertPdfToHtml(crawledDocument.documentBodyBytes);
+        } catch (IOException e) {
+            logger.error("Failed to convert PDF file {} - {}", url, e.getMessage());
+            throw new DisqualifiedException(DisqualifiedException.DisqualificationReason.ERROR);
+        }
+
+        DocumentLanguageData dld = sentenceExtractorProvider.get().extractSentences(doc);
+
+        checkDocumentLanguage(dld);
+
+        documentLengthLogic.validateLength(dld, 1.0);
+
+        var ret = new ProcessedDocumentDetails();
+
+        ret.length = documentBody.length();
+
+        ret.format = DocumentFormat.PDF;
+        ret.title = StringUtils.truncate(defaultSpecialization.getTitle(doc, dld, url.toString()), maxTitleLength);
+
+        ret.quality = -5;
+
+        ret.features = Set.of(HtmlFeature.PDF);
+        ret.description = getDescription(doc);
+        ret.hashCode = dld.localitySensitiveHashCode();
+
+        final PubDate pubDate = new PubDate(LocalDate.ofYearDay(1993, 1));
+
+        EnumSet<DocumentFlags> documentFlags = EnumSet.of(DocumentFlags.PdfFile);
+
+        ret.metadata = new DocumentMetadata(
+                documentLengthLogic.getEncodedAverageLength(dld),
+                pubDate.yearByte(),
+                (int) -ret.quality,
+                documentFlags);
+
+        DocumentKeywordsBuilder words = keywordExtractor.extractKeywords(dld, linkTexts, url);
+
+        var tagWords = new MetaTagsBuilder()
+                .addPubDate(pubDate)
+                .addUrl(url)
+                .addFeatures(ret.features)
+                .addFormat(ret.format)
+                .build();
+
+        words.addAllSyntheticTerms(tagWords);
+
+        if (pubDate.hasYear()) {
+            ret.pubYear = pubDate.year();
+        }
+
+        /* These are assumed to be populated */
+        ret.linksInternal = new ArrayList<>();
+        ret.linksExternal = new ArrayList<>();
+
+        return new DetailsWithWords(ret, words);
+    }
+
+    private String getDescription(Document doc) {
+        int cnt = 0;
+        boolean useNext = false;
+        for (var ptag : doc.getElementsByTag("p")) {
+            String text = ptag.text();
+
+            // Many academic documents have an abstract at the start of the document,
+            // which makes a nice summary.  Though they tend to bleed into the text,
+            // so we check for the word "Abstract" at the start of the paragraph.
+
+            if (text.startsWith("Abstract ")) {
+                return StringUtils.abbreviate(text.substring("Abstract ".length()), "...", 255);
+            }
+            else if (text.equals("Abstract")) {
+                useNext = true;
+            }
+            else if (useNext) {
+                return StringUtils.abbreviate(text, "...", 255);
+            }
+
+            if (++cnt > 15) { // Don't scan the entire document
+                break;
+            }
+        }
+
+        // Fall back to the default specialization
+        return defaultSpecialization.getSummary(doc, Set.of());
+
+    }
+
+    /** Convert the provided PDF bytes into a HTML rendering that can be fed
+     * to the HTML processor.
+     */
+    Document convertPdfToHtml(byte[] pdfBytes) throws IOException {
+        try (var doc = Loader.loadPDF(pdfBytes)) {
+            String docMetaTitle = Objects.requireNonNullElse(doc.getDocumentInformation().getTitle(), "");
+
+            var stripper = new HeadingAwarePDFTextStripper();
+            stripper.setStartPage(1);
+            stripper.setSortByPosition(true);
+            stripper.setWordSeparator(" ");
+
+            // Increase the tolerance for line spacing to deal better with paragraphs.
+            stripper.setDropThreshold(5f);
+
+            stripper.setPageStart("<div>");
+            stripper.setParagraphStart("<p>");
+            stripper.setParagraphEnd("</p>\n");
+            stripper.setPageEnd("</div>\n");
+            stripper.setHeadingStart("<h1>");
+            stripper.setHeadingEnd("</h1>\n");
+            stripper.setLineSeparator("\n");
+
+            String text = stripper.getText(doc);
+
+            StringBuilder htmlBuilder = new StringBuilder(text.length() + 1024);
+            htmlBuilder.append("<html><body>")
+                    .append(text)
+                    .append("</body></html>");
+
+            var parsed = Jsoup.parse(htmlBuilder.toString());
+
+            repairDOM(parsed);
+
+            for (var heading : parsed.getElementsByTag("h1")) {
+                String headingText = heading.text();
+                if (headingText.length() > 2) {
+                    parsed.title(headingText);
+                    break;
+                }
+            }
+
+
+            if (parsed.title().isEmpty()) {
+                // Prefer setting the title to the first paragraph in the
+                // document, as this is almost always correct.  Otherwise,
+                // we fall back on the metadata title, which is almost always
+                // useless
+
+                var firstP = parsed.getElementsByTag("p").first();
+                if (firstP != null) parsed.title(firstP.text());
+                else parsed.title(docMetaTitle);
+            }
+            return parsed;
+        }
+
+
+    }
+
+    /** Repair the DOM to remove some common issues with PDF conversion,
+     * including empty paragraphs, and multiline headers that are split into multiple
+     * conescutive h1 tags.
+     */
+    private void repairDOM(Document parsed) {
+
+        // <p><h1>...</h1></p> -> <h1>...</h1>
+        parsed.getElementsByTag("h1").forEach(h1 -> {
+            var parent = h1.parent();
+            if (parent == null || !"p".equals(parent.tagName())) {
+                return;
+            }
+
+            if (parent.childrenSize() == 1) {
+                parent.replaceWith(h1);
+            }
+        });
+
+        // Remove empty <p> tags
+        parsed.getElementsByTag("p").forEach(p -> {
+            if (p.childrenSize() == 0 && !p.hasText()) {
+                p.remove();
+            }
+        });
+
+        // <h1>...</h1><h1>...</h1> -> <h1>...</h1>
+        parsed.getElementsByTag("h1").forEach(h1 -> {
+            var nextSibling = h1.nextElementSibling();
+            if (nextSibling == null || !"h1".equals(nextSibling.tagName())) {
+                return; // Short-circuit to avoid unnecessary work
+            }
+
+            StringJoiner joiner = new StringJoiner(" ");
+            joiner.add(h1.text());
+
+            for (var sibling : h1.nextElementSiblings()) {
+                if (!"h1".equals(sibling.tagName()))
+                    break;
+                joiner.add(sibling.text());
+                sibling.remove();
+            }
+
+            h1.text(joiner.toString());
+        });
+
+    }
+
+}

code/processes/converting-process/java/nu/marginalia/converting/processor/plugin/PlainTextDocumentProcessorPlugin.java

@@ -13,10 +13,10 @@ import nu.marginalia.keyword.LinkTexts;
 import nu.marginalia.keyword.model.DocumentKeywordsBuilder;
 import nu.marginalia.language.filter.LanguageFilter;
 import nu.marginalia.language.sentence.ThreadLocalSentenceExtractorProvider;
+import nu.marginalia.model.DocumentFormat;
 import nu.marginalia.model.EdgeUrl;
 import nu.marginalia.model.crawl.PubDate;
 import nu.marginalia.model.crawldata.CrawledDocument;
-import nu.marginalia.model.html.HtmlStandard;
 import nu.marginalia.model.idx.DocumentFlags;
 import nu.marginalia.model.idx.DocumentMetadata;
 import org.apache.commons.lang3.StringUtils;
@@ -71,7 +71,7 @@ public class PlainTextDocumentProcessorPlugin extends AbstractDocumentProcessorP
                                           DocumentClass documentClass)
             throws DisqualifiedException, URISyntaxException {
 
-        String documentBody = crawledDocument.documentBody;
+        String documentBody = crawledDocument.documentBody();
 
         if (languageFilter.isBlockedUnicodeRange(documentBody)) {
             throw new DisqualifiedException(DisqualifiedException.DisqualificationReason.LANGUAGE);
@@ -91,7 +91,7 @@ public class PlainTextDocumentProcessorPlugin extends AbstractDocumentProcessorP
 
         ret.length = documentBody.length();
 
-        ret.standard = HtmlStandard.PLAIN;
+        ret.format = DocumentFormat.PLAIN;
         ret.title = StringUtils.truncate(plainTextLogic.getTitle(url, firstFewLines), maxTitleLength);
 
         ret.quality = -1;
@@ -113,7 +113,7 @@ public class PlainTextDocumentProcessorPlugin extends AbstractDocumentProcessorP
                 .addPubDate(pubDate)
                 .addUrl(url)
                 .addFeatures(ret.features)
-                .addFormat(ret.standard)
+                .addFormat(ret.format)
                 .build();
 
         words.addAllSyntheticTerms(tagWords);