(domainstatedb) Allow creation of disconnected db

This is required for executor services that do not have crawl data to still be able to initialize.

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

@@ -22,6 +22,7 @@ public class DbDomainQueries {
     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();
 
@@ -59,6 +60,34 @@ public class DbDomainQueries {
         }
     }
 
+
+    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());
+        }
+    }
+
     public OptionalInt tryGetDomainId(EdgeDomain domain) {
 
         Integer maybeId = domainIdCache.getIfPresent(domain);
@@ -145,4 +174,6 @@ public class DbDomainQueries {
             return nodeAffinity > 0;
         }
     }
+
+    public record DomainIdWithNode (int domainId, int nodeAffinity) { }
 }

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/processes/crawling-process/java/nu/marginalia/crawl/DomainStateDb.java

@@ -1,5 +1,8 @@
 package nu.marginalia.crawl;
 
+import com.google.inject.Inject;
+import nu.marginalia.storage.FileStorageService;
+import nu.marginalia.storage.model.FileStorageType;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 
@@ -63,7 +66,29 @@ public class DomainStateDb implements AutoCloseable {
 
     public record FaviconRecord(String contentType, byte[] imageData) {}
 
-    public DomainStateDb(Path filename) throws SQLException {
+    @Inject
+    public DomainStateDb(FileStorageService fileStorageService) throws SQLException {
+        this(findFilename(fileStorageService));
+    }
+
+    private static Path findFilename(FileStorageService fileStorageService) throws SQLException {
+        var fsId = fileStorageService.getOnlyActiveFileStorage(FileStorageType.CRAWL_DATA);
+
+        if (fsId.isPresent()) {
+            var fs = fileStorageService.getStorage(fsId.get());
+            return fs.asPath().resolve("domainstate.db");
+        }
+        else {
+            return null;
+        }
+    }
+
+    public DomainStateDb(@Nullable Path filename) throws SQLException {
+        if (null == filename) {
+            connection = null;
+            return;
+        }
+
         String sqliteDbString = "jdbc:sqlite:" + filename.toString();
         connection = DriverManager.getConnection(sqliteDbString);
 
@@ -90,11 +115,18 @@ public class DomainStateDb implements AutoCloseable {
 
     @Override
     public void close() throws SQLException {
-        connection.close();
+        if (connection != null) {
+            connection.close();
+        }
     }
 
+    public boolean isAvailable() {
+        return connection != null;
+    }
 
     public void saveIcon(String domain, FaviconRecord faviconRecord) {
+        if (connection == null) throw new IllegalStateException("No connection to domainstate db");
+
         try (var stmt = connection.prepareStatement("""
                 INSERT OR REPLACE INTO favicon (domain, contentType, icon)
                        VALUES(?, ?, ?)
@@ -110,6 +142,9 @@ public class DomainStateDb implements AutoCloseable {
     }
 
     public Optional<FaviconRecord> getIcon(String domain) {
+        if (connection == null)
+            return Optional.empty();
+
         try (var stmt = connection.prepareStatement("SELECT contentType, icon FROM favicon WHERE DOMAIN = ?")) {
             stmt.setString(1, domain);
             var rs = stmt.executeQuery();
@@ -130,6 +165,8 @@ public class DomainStateDb implements AutoCloseable {
     }
 
     public void save(SummaryRecord record) {
+        if (connection == null) throw new IllegalStateException("No connection to domainstate db");
+
         try (var stmt = connection.prepareStatement("""
                 INSERT OR REPLACE INTO summary (domain, lastUpdatedEpochMs, state, stateDesc, feedUrl)
                 VALUES (?, ?, ?, ?, ?)
@@ -146,6 +183,9 @@ public class DomainStateDb implements AutoCloseable {
     }
 
     public Optional<SummaryRecord> get(String domainName) {
+        if (connection == null)
+            return Optional.empty();
+
         try (var stmt = connection.prepareStatement("""
                 SELECT domain, lastUpdatedEpochMs, state, stateDesc, feedUrl
                 FROM summary

code/services-application/search-service/build.gradle

@@ -41,6 +41,7 @@ dependencies {
 
     implementation project(':code:functions:live-capture:api')
     implementation project(':code:functions:math:api')
+    implementation project(':code:functions:favicon:api')
     implementation project(':code:functions:domain-info:api')
     implementation project(':code:functions:search-query:api')
 

code/services-application/search-service/java/nu/marginalia/search/SearchService.java

@@ -3,10 +3,14 @@ package nu.marginalia.search;
 import com.google.inject.Inject;
 import io.jooby.Context;
 import io.jooby.Jooby;
+import io.jooby.MediaType;
 import io.jooby.StatusCode;
 import io.prometheus.client.Counter;
 import io.prometheus.client.Histogram;
 import nu.marginalia.WebsiteUrl;
+import nu.marginalia.api.favicon.FaviconClient;
+import nu.marginalia.db.DbDomainQueries;
+import nu.marginalia.model.EdgeDomain;
 import nu.marginalia.search.svc.*;
 import nu.marginalia.service.discovery.property.ServicePartition;
 import nu.marginalia.service.server.BaseServiceParams;
@@ -15,11 +19,14 @@ import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 
 import java.util.List;
+import java.util.NoSuchElementException;
 
 public class SearchService extends JoobyService {
 
     private final WebsiteUrl websiteUrl;
     private final SearchSiteSubscriptionService siteSubscriptionService;
+    private final FaviconClient faviconClient;
+    private final DbDomainQueries domainQueries;
 
     private static final Logger logger = LoggerFactory.getLogger(SearchService.class);
     private static final Histogram wmsa_search_service_request_time = Histogram.build()
@@ -43,6 +50,8 @@ public class SearchService extends JoobyService {
                          SearchSiteInfoService siteInfoService,
                          SearchCrosstalkService crosstalkService,
                          SearchBrowseService searchBrowseService,
+                         FaviconClient faviconClient,
+                         DbDomainQueries domainQueries,
                          SearchQueryService searchQueryService)
     throws Exception {
         super(params,
@@ -58,6 +67,8 @@ public class SearchService extends JoobyService {
         this.websiteUrl = websiteUrl;
 
         this.siteSubscriptionService = siteSubscriptionService;
+        this.faviconClient = faviconClient;
+        this.domainQueries = domainQueries;
     }
 
     @Override
@@ -71,6 +82,31 @@ public class SearchService extends JoobyService {
         jooby.get("/site/https://*", this::handleSiteUrlRedirect);
         jooby.get("/site/http://*", this::handleSiteUrlRedirect);
 
+        jooby.get("/site/{domain}/favicon", ctx -> {
+            String domain = ctx.path("domain").value();
+            logger.info("Finding icon for domain {}", domain);
+            domainQueries.getDomainId(new EdgeDomain(domain));
+            try {
+                DbDomainQueries.DomainIdWithNode domainIdWithNode = domainQueries.getDomainIdWithNode(new EdgeDomain(domain));
+                var faviconMaybe = faviconClient.getFavicon(domain, domainIdWithNode.nodeAffinity());
+
+                if (faviconMaybe.isEmpty()) {
+                    ctx.setResponseCode(404);
+                    return "";
+                } else {
+                    var favicon = faviconMaybe.get();
+
+                    ctx.responseStream(MediaType.valueOf(favicon.contentType()), consumer -> {
+                        consumer.write(favicon.bytes());
+                    });
+                }
+            }
+            catch (NoSuchElementException ex) {
+                ctx.setResponseCode(404);
+            }
+            return "";
+        });
+
         jooby.before((Context ctx) -> {
             ctx.setAttribute(startTimeAttribute, System.nanoTime());
         });

code/services-application/search-service/resources/jte/serp/part/footerHowto.jte

@@ -26,15 +26,15 @@
             It operates a bit like a clock, starting at the top and working its way around clockwise.</p>
 
         <div class="flex gap-4 place-items-middle">
-            @template.serp.part.matchogram(mask = 90)
+            @template.serp.part.matchogram(mask = 90, domain = "example.com")
             <div>This is by the beginning</div>
         </div>
         <div class="flex gap-4 place-items-middle">
-            @template.serp.part.matchogram(mask = 90L<<26)
+            @template.serp.part.matchogram(mask = 90L<<26, domain = "example.com")
             <div>This is in the middle</div>
         </div>
         <div class="flex gap-4 place-items-middle">
-            @template.serp.part.matchogram(mask = 5L<<48)
+            @template.serp.part.matchogram(mask = 5L<<48, domain = "example.com")
             <div>This is toward the end</div>
         </div>
 

code/services-application/search-service/resources/jte/serp/part/matchogram.jte

@@ -1,11 +1,13 @@
 @import java.util.stream.IntStream
 
 @param long mask
+@param String domain
 
-<svg width="40" height="40">
+<svg width="40" height="40"
+     style="background-image: url('/site/${domain}/favicon'); background-repeat: no-repeat; background-size: 16px 16px; background-position: center; ">
     <circle
-            cx="18"
-            cy="18"
+            cx="20"
+            cy="20"
             r="16"
             fill="none"
             stroke="#eee"
@@ -13,10 +15,10 @@
     />
 @for (int bit : IntStream.range(0, 56).filter(bit -> (mask & (1L << bit)) != 0).toArray())
             <line
-                    x1="${18 + 15*Math.sin(2 * Math.PI * bit / 56.)}"
-                    y1="${18 - 15*Math.cos(2 * Math.PI * bit / 56.)}"
-                    x2="${18 + 17*Math.sin(2 * Math.PI * bit / 56.)}"
-                    y2="${18 - 17*Math.cos(2 * Math.PI * bit / 56.)}"
+                    x1="${20 + 15*Math.sin(2 * Math.PI * bit / 56.)}"
+                    y1="${20 - 15*Math.cos(2 * Math.PI * bit / 56.)}"
+                    x2="${20 + 17*Math.sin(2 * Math.PI * bit / 56.)}"
+                    y2="${20 - 17*Math.cos(2 * Math.PI * bit / 56.)}"
                     stroke="#444"
                     stroke-width="2"
             />

code/services-application/search-service/resources/jte/serp/part/result.jte

@@ -12,7 +12,7 @@
             <div class="flex flex-col grow" >
                 <div class="flex flex-row space-x-2 place-items-center">
                     <div class="flex-0" title="Match density">
-                        @template.serp.part.matchogram(mask = result.first.positionsMask)
+                        @template.serp.part.matchogram(mask = result.first.positionsMask, domain=result.getFirst().url.domain.toString())
                     </div>
                     <div class="flex grow justify-between items-start">
                         <div class="flex-1">

code/services-application/search-service/resources/jte/siteinfo/view/overview.jte

@@ -9,8 +9,8 @@
 
 <div class="flex-1 p-4 space-y-4 mx-auto w-full md:w-auto">
     <div class="flex border dark:border-gray-600 rounded bg-white dark:bg-gray-800 flex-col space-y-4 pb-4 overflow-hidden md:max-w-lg" >
-        <div class="flex place-items-baseline space-x-2 p-2 text-md border-b dark:border-gray-600 bg-margeblue text-white">
-            <i class="fa fa-globe"></i>
+        <div class="flex place-items-center space-x-2 p-2 text-md border-b dark:border-gray-600 bg-margeblue text-white">
+            <img src="/site/${siteInfo.domain()}/favicon" style="width: 16px; height: 16px; vertical-align: center">
             <span>${siteInfo.domain()}</span>
             <div class="grow">
             </div>

code/services-core/executor-service/build.gradle

@@ -42,6 +42,8 @@ dependencies {
     implementation project(':code:libraries:message-queue')
 
     implementation project(':code:functions:link-graph:api')
+    implementation project(':code:functions:favicon')
+    implementation project(':code:functions:favicon:api')
 
     implementation project(':code:processes:crawling-process:model')
     implementation project(':code:processes:crawling-process:model')

code/services-core/executor-service/java/nu/marginalia/executor/ExecutorSvc.java

@@ -2,6 +2,7 @@ package nu.marginalia.executor;
 
 import com.google.inject.Inject;
 import nu.marginalia.execution.*;
+import nu.marginalia.functions.favicon.FaviconGrpcService;
 import nu.marginalia.service.discovery.property.ServicePartition;
 import nu.marginalia.service.server.BaseServiceParams;
 import nu.marginalia.service.server.SparkService;
@@ -24,6 +25,7 @@ public class ExecutorSvc extends SparkService {
                        ExecutorCrawlGrpcService executorCrawlGrpcService,
                        ExecutorSideloadGrpcService executorSideloadGrpcService,
                        ExecutorExportGrpcService executorExportGrpcService,
+                       FaviconGrpcService faviconGrpcService,
                        ExecutionInit executionInit,
                        ExecutorFileTransferService fileTransferService) throws Exception {
         super(params,
@@ -31,7 +33,8 @@ public class ExecutorSvc extends SparkService {
                 List.of(executorGrpcService,
                         executorCrawlGrpcService,
                         executorSideloadGrpcService,
-                        executorExportGrpcService)
+                        executorExportGrpcService,
+                        faviconGrpcService)
             );
 
         this.executionInit = executionInit;

settings.gradle

@@ -16,7 +16,8 @@ include 'code:services-application:status-service'
 
 include 'code:functions:math'
 include 'code:functions:math:api'
-
+include 'code:functions:favicon'
+include 'code:functions:favicon:api'
 include 'code:functions:domain-info'
 include 'code:functions:domain-info:api'