1
1
mirror of https://github.com/MarginaliaSearch/MarginaliaSearch.git synced 2025-10-06 07:32:38 +02:00

Compare commits

...

23 Commits

Author SHA1 Message Date
Viktor Lofgren
6e47e58e0e (sample-actor) Add progress tracking to sample export actor 2025-05-05 13:04:14 +02:00
Viktor Lofgren
9dc43d8b4a (sample-actor) Update the actor export sample actor to not generate empty files when the filter is not applicable. 2025-05-05 12:56:12 +02:00
Viktor Lofgren
83967e3305 (sample-actor) Update the actor export sample actor to not generate empty files when the filter is not applicable. 2025-05-05 12:50:21 +02:00
Viktor Lofgren
4db980a291 (jooby-service) Set an upper limit on the number of worker threads 2025-05-05 12:40:31 +02:00
Viktor Lofgren
089b177868 (deploy) Executor partition 4. 2025-05-05 12:21:27 +02:00
Viktor Lofgren
9c8e9a68d5 (deploy) Executor partition 4. 2025-05-05 12:00:05 +02:00
Viktor Lofgren
413d5cc788 (url, minor) Fix typo in test 2025-05-04 16:28:30 +02:00
Viktor Lofgren
58539b92ac (search) Don't show addresses with URLencoding in the UI 2025-05-04 16:26:39 +02:00
Viktor Lofgren
fe72f16df1 (url) Add additional tests for parameter handling 2025-05-04 16:23:39 +02:00
Viktor Lofgren
b49a244a2e (url) Fix encoding handling of query parameters 2025-05-04 16:18:47 +02:00
Viktor Lofgren
3f0b4c010f (deploy) Fix deploy script to be aware of the status service 2025-05-04 16:14:07 +02:00
Viktor Lofgren
c6e0cd93f7 (status) Fix status service to poll the new domain 2025-05-04 16:11:08 +02:00
Viktor Lofgren
80a7ccb080 Trigger redeploy of qs, search and api 2025-05-04 16:07:28 +02:00
Viktor Lofgren
54dec347c4 (url) Fix urlencoding issues with certain symbols
Optimize the code by adding a simple heuristic for guessing whether we need to repair the URI before we pass it to Java's parser.
2025-05-04 13:39:39 +02:00
Viktor Lofgren
d6ee3f0785 (url) Fix urlencoding issues with certain symbols
The urlencoding logic would consider the need to urlencode on an element basis, which is incorrect.  Even if we urlencode on an element basis, we should either urlencode or not urlencode, never a mix of the two.
2025-05-04 13:08:49 +02:00
Viktor Lofgren
8be88afcf3 (url) Fix urlencoding issues with certain symbols
We also need to apply the fix when performing toString() on the EdgeUrl, the URI class will URLDecode the input.

The change also alters the parseURI method to only run the URLEncode-fixer during parsing if URI doesn't throw an exception.  This bad path is obviously going to be slower, but realistically, most URLs are valid, so it's probably a significant optimization to do it like this.
2025-05-04 12:58:13 +02:00
Viktor Lofgren
0e3c00d3e1 (url) Fix urlencoding issues with certain symbols
Minor fix of issue where url sanitizer would strip some trailing slashes.
2025-05-03 23:58:28 +02:00
Viktor Lofgren
4279a7f1aa (url) Fix urlencoding issues with certain symbols
Minor fix with previously urlencoded codepoints, we need to account for the fact that they are encoded in hexadecimal.
2025-05-03 23:51:39 +02:00
Viktor Lofgren
251006d4f9 (url) Fix urlencoding issues with certain symbols
Problems primarily cropped up with sideloaded wikipedia articles, though the search engine has been returning inconsistently URLEncoded search results for a while, though browsers and servers have seemingly magically fixed the issues in many scenarios.

This addresses Issue #195 and Issue #131.
2025-05-03 23:48:45 +02:00
Viktor Lofgren
c3e99dc12a (service) Limit logging from ad hoc task heartbeats
Certain usage patterns of the ad hoc task heartbeats would lead to an incredible amount of log noise, as it would log each update.

Limit log updates to increments of 10% to avoid this problem.
2025-05-03 12:39:58 +02:00
Viktor
aaaa2de022 Merge pull request #196 from MarginaliaSearch/filter-export-sample-data
Add the ability to filter sample data based on content type
2025-05-02 13:23:49 +02:00
Viktor Lofgren
fc1388422a (actor) Add the ability to filter sample data based on content type
This will help in extracting relevant test sets for PDF processing.
2025-05-02 13:09:22 +02:00
Viktor Lofgren
b07080db16 (crawler) Don't retry requests when encountering UnknownHostException 2025-05-01 16:07:34 +02:00
20 changed files with 470 additions and 126 deletions

View File

@@ -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.URI;
import java.net.URISyntaxException;
import java.net.URL;
import java.net.*;
import java.nio.charset.StandardCharsets;
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(param);
sb.append('?').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);
}
}

View File

@@ -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 {
System.out.println(new EdgeUrl("https://memex.marginalia.nu/index.php?id=1").toString());
System.out.println(new EdgeUrl("https://memex.marginalia.nu/showthread.php?id=1&count=5&tracking=123").toString());
}
@Test
void urlencodeFixer() throws URISyntaxException {
System.out.println(EdgeUrl.urlencodeFixer("https://www.example.com/#heredoc"));
System.out.println(EdgeUrl.urlencodeFixer("https://www.example.com/%-sign"));
System.out.println(EdgeUrl.urlencodeFixer("https://www.example.com/%22-sign"));
System.out.println(EdgeUrl.urlencodeFixer("https://www.example.com/\n \"huh\""));
void testUriFromString() throws URISyntaxException {
// We test these URLs several times as we perform URLEncode-fixing both when parsing the URL and when
// converting it back to a string, we want to ensure there is no changes along the way.
Assertions.assertEquals("/", EdgeUriFactory.parseURILenient("https://www.example.com/").getPath());
Assertions.assertEquals("https://www.example.com/", EdgeUriFactory.parseURILenient("https://www.example.com/").toString());
Assertions.assertEquals("https://www.example.com/", new EdgeUrl("https://www.example.com/").toString());
Assertions.assertEquals("/", EdgeUriFactory.parseURILenient("https://www.example.com/#heredoc").getPath());
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"));
System.out.println(new EdgeUrl("https://search.marginalia.nu/?t=123"));
System.out.println(new EdgeUrl("https://search.marginalia.nu/?v=123"));
System.out.println(new EdgeUrl("https://search.marginalia.nu/?m=123"));
System.out.println(new EdgeUrl("https://search.marginalia.nu/?follow=123"));
Assertions.assertEquals("id=123", new EdgeUrl("https://search.marginalia.nu/?id=123").param);
Assertions.assertEquals("https://search.marginalia.nu/?id=123", new EdgeUrl("https://search.marginalia.nu/?id=123").toString());
Assertions.assertEquals("t=123", new EdgeUrl("https://search.marginalia.nu/?t=123").param);
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);
}
}

View File

@@ -59,17 +59,14 @@ 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);
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 */
@Override

View File

@@ -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() {

View File

@@ -122,6 +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);

View File

@@ -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());
}

View File

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

View File

@@ -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 Run(FileStorageId crawlId, FileStorageId destId, int size, String name, long msgId) implements ActorStep {
public Run(FileStorageId crawlId, FileStorageId destId, int size, String name) {
this(crawlId, destId, size, name, -1);
public record Export(FileStorageId crawlId, int size, String ctFilter, String name) 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, String ctFilter) {
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));
yield new Run(crawlId, destId, size, name, newMsgId);
long newMsgId = exportTasksOutbox.sendAsync(ExportTaskRequest.sampleData(crawlId, destId, ctFilter, size, name));
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

View File

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

View File

@@ -51,6 +51,7 @@ import javax.net.ssl.SSLException;
import java.io.IOException;
import java.net.SocketTimeoutException;
import java.net.URISyntaxException;
import java.net.UnknownHostException;
import java.security.NoSuchAlgorithmException;
import java.time.Duration;
import java.time.Instant;
@@ -635,14 +636,12 @@ public class HttpFetcherImpl implements HttpFetcher, HttpRequestRetryStrategy {
@Override
public boolean retryRequest(HttpRequest request, IOException exception, int executionCount, HttpContext context) {
if (exception instanceof SocketTimeoutException) { // Timeouts are not recoverable
return false;
}
if (exception instanceof SSLException) { // SSL exceptions are unlikely to be recoverable
return false;
}
return executionCount <= 3;
return switch (exception) {
case SocketTimeoutException ste -> false;
case SSLException ssle -> false;
case UnknownHostException uhe -> false;
default -> executionCount <= 3;
};
}
@Override

View File

@@ -53,6 +53,8 @@ dependencies {
implementation libs.commons.compress
implementation libs.commons.codec
implementation libs.jsoup
implementation libs.slop
implementation libs.jwarc

View File

@@ -1,13 +1,18 @@
package nu.marginalia.extractor;
import com.google.inject.Inject;
import nu.marginalia.process.control.ProcessHeartbeat;
import nu.marginalia.process.log.WorkLog;
import nu.marginalia.process.log.WorkLogEntry;
import nu.marginalia.slop.SlopCrawlDataRecord;
import nu.marginalia.slop.SlopTablePacker;
import nu.marginalia.storage.FileStorageService;
import nu.marginalia.storage.model.FileStorage;
import nu.marginalia.storage.model.FileStorageId;
import org.apache.commons.compress.archivers.tar.TarArchiveOutputStream;
import org.apache.commons.compress.utils.IOUtils;
import org.apache.commons.io.FileUtils;
import org.apache.commons.lang3.StringUtils;
import java.io.IOException;
import java.nio.file.Files;
@@ -19,15 +24,19 @@ import java.sql.SQLException;
import java.util.ArrayList;
import java.util.Collections;
import java.util.List;
import java.util.NoSuchElementException;
public class SampleDataExporter {
private final FileStorageService storageService;
private final ProcessHeartbeat processHeartbeat;
@Inject
public SampleDataExporter(FileStorageService storageService) {
public SampleDataExporter(FileStorageService storageService, ProcessHeartbeat processHeartbeat) {
this.storageService = storageService;
this.processHeartbeat = processHeartbeat;
}
public void export(FileStorageId crawlId, FileStorageId destId, int size, String name) throws SQLException, IOException {
public void export(FileStorageId crawlId, FileStorageId destId, int size, String ctFilter, String name) throws SQLException, IOException {
FileStorage destStorage = storageService.getStorage(destId);
Path inputDir = storageService.getStorage(crawlId).asPath();
@@ -54,11 +63,6 @@ public class SampleDataExporter {
Path newCrawlerLogFile = Files.createTempFile(destStorage.asPath(), "crawler", ".log",
PosixFilePermissions.asFileAttribute(PosixFilePermissions.fromString("rw-r--r--")));
try (var bw = Files.newBufferedWriter(newCrawlerLogFile, StandardOpenOption.CREATE, StandardOpenOption.TRUNCATE_EXISTING)) {
for (var item : entriesAll) {
bw.write(item.id() + " " + item.ts() + " " + item.relPath() + " " + item.cnt() + "\n");
}
}
Path newManifestJsonFile = Files.createTempFile(destStorage.asPath(), "manifest", ".json",
PosixFilePermissions.asFileAttribute(PosixFilePermissions.fromString("rw-r--r--")));
@@ -67,12 +71,34 @@ public class SampleDataExporter {
var tmpTarFile = Files.createTempFile(destStorage.asPath(), "data", ".tar",
PosixFilePermissions.asFileAttribute(PosixFilePermissions.fromString("rw-r--r--")));
try (var stream = new TarArchiveOutputStream(Files.newOutputStream(tmpTarFile, StandardOpenOption.CREATE, StandardOpenOption.TRUNCATE_EXISTING))) {
for (var item : entriesAll) {
try (var stream = new TarArchiveOutputStream(Files.newOutputStream(tmpTarFile, StandardOpenOption.CREATE, StandardOpenOption.TRUNCATE_EXISTING));
var logWriter = Files.newBufferedWriter(newCrawlerLogFile, StandardOpenOption.CREATE, StandardOpenOption.TRUNCATE_EXISTING);
var hb = processHeartbeat.createAdHocTaskHeartbeat("Generating Sample")
) {
for (var item : hb.wrap("Scanning", entriesAll)) {
Path crawlDataPath = inputDir.resolve(item.relPath());
if (!Files.exists(crawlDataPath)) continue;
if (StringUtils.isBlank(ctFilter)) {
addFileToTar(stream, crawlDataPath, item.relPath());
logWriter.write(item.id() + " " + item.ts() + " " + item.relPath() + " " + item.cnt() + "\n");
}
else /* filter != null */ {
Path filteredData = null;
try {
filteredData = filterEntries(crawlDataPath, ctFilter);
addFileToTar(stream, filteredData, item.relPath());
logWriter.write(item.id() + " " + item.ts() + " " + item.relPath() + " " + item.cnt() + "\n");
}
catch (NoSuchElementException ex) {
// Ignore
}
finally {
if (filteredData != null) {
Files.deleteIfExists(filteredData);
}
}
}
}
addFileToTar(stream, newCrawlerLogFile, "crawler.log");
@@ -86,6 +112,50 @@ public class SampleDataExporter {
Files.move(tmpTarFile, destStorage.asPath().resolve("crawl-data.tar"), StandardCopyOption.ATOMIC_MOVE, StandardCopyOption.REPLACE_EXISTING);
}
/** Filters the entries in the crawl data file based on the content type. */
private Path filterEntries(Path crawlDataPath, String contentTypeFilter) throws IOException, NoSuchElementException {
Path tempDir = crawlDataPath.resolveSibling(crawlDataPath.getFileName() + ".filtered");
Path tempFile = crawlDataPath.resolveSibling(crawlDataPath.getFileName() + ".filtered.slop.zip");
Files.createDirectory(tempDir);
try (var writer = new SlopCrawlDataRecord.Writer(tempDir);
var reader = new SlopCrawlDataRecord.FilteringReader(crawlDataPath) {
@Override
public boolean filter(String url, int status, String contentType) {
if (contentTypeFilter.equals(contentType))
return true;
else if (contentType.startsWith("x-marginalia/"))
// This is a metadata entry, typically domain or redirect information
// let's keep those to not confuse the consumer of the data, which might
// expect at least the domain summary
return true;
return false;
}
}
) {
boolean wroteEntry = false;
while (reader.hasRemaining()) {
var entry = reader.get();
writer.write(entry);
wroteEntry = wroteEntry || contentTypeFilter.equals(entry.contentType());
}
if (!wroteEntry) {
throw new NoSuchElementException("No relevant entries");
}
SlopTablePacker.packToSlopZip(tempDir, tempFile);
}
finally {
FileUtils.deleteDirectory(tempDir.toFile());
}
return tempFile;
}
private void addFileToTar(TarArchiveOutputStream outputStream, Path file, String fileName) throws IOException {
var entry = outputStream.createArchiveEntry(file.toFile(), fileName);
entry.setSize(Files.size(file));

View File

@@ -92,7 +92,7 @@ public class ExportTasksMain extends ProcessMainClass {
termFrequencyExporter.export(request.crawlId, request.destId);
break;
case SAMPLE_DATA:
sampleDataExporter.export(request.crawlId, request.destId, request.size, request.name);
sampleDataExporter.export(request.crawlId, request.destId, request.size, request.ctFilter, request.name);
break;
case ADJACENCIES:
websiteAdjacenciesCalculator.export();

View File

@@ -16,6 +16,7 @@ public class ExportTaskRequest {
public FileStorageId destId;
public int size;
public String name;
public String ctFilter;
public ExportTaskRequest(Task task) {
this.task = task;
@@ -42,12 +43,13 @@ public class ExportTaskRequest {
return request;
}
public static ExportTaskRequest sampleData(FileStorageId crawlId, FileStorageId destId, int size, String name) {
public static ExportTaskRequest sampleData(FileStorageId crawlId, FileStorageId destId, String ctFilter, int size, String name) {
ExportTaskRequest request = new ExportTaskRequest(Task.SAMPLE_DATA);
request.crawlId = crawlId;
request.destId = destId;
request.size = size;
request.name = name;
request.ctFilter = ctFilter;
return request;
}

View File

@@ -180,7 +180,7 @@ public class UrlDetails implements Comparable<UrlDetails> {
* semantically meaningful codepoints into entity codes */
public String displayUrl() {
StringBuilder sb = new StringBuilder();
String urlStr = url.toString();
String urlStr = url.toDisplayString();
for (int i = 0; i < urlStr.length(); i++) {
char c = urlStr.charAt(i);

View File

@@ -20,6 +20,6 @@ public class StatusModule extends AbstractModule {
bind(String.class)
.annotatedWith(Names.named("searchEngineTestQuery"))
.toInstance(System.getProperty("status-service.public-query",
"https://search.marginalia.nu/search?query=plato&ref=marginalia-automatic-metrics"));
"https://marginalia-search.com/search?query=plato&ref=marginalia-automatic-metrics"));
}
}

View File

@@ -321,9 +321,10 @@ public class ControlNodeActionsService {
private Object exportSampleData(Request req, Response rsp) {
FileStorageId source = parseSourceFileStorageId(req.queryParams("source"));
int size = Integer.parseInt(req.queryParams("size"));
String ctFilter = req.queryParams("ctFilter");
String name = req.queryParams("name");
exportClient.exportSampleData(Integer.parseInt(req.params("id")), source, size, name);
exportClient.exportSampleData(Integer.parseInt(req.params("id")), source, size, ctFilter, name);
return "";
}

View File

@@ -35,6 +35,11 @@
<div><input type="text" name="size" id="size" pattern="\d+" /></div>
<small class="text-muted">How many domains to include in the sample set</small>
</div>
<div class="mb-3">
<label for="ctFilter">Content Type Filter</label>
<div><input type="text" name="ctFilter" id="ctFilter" /></div>
<small class="text-muted">If set, includes only documents with the specified content type value</small>
</div>
<div class="mb-3">
<label for="name">Name</label>
<div><input type="text" name="name" id="name" /></div>

View File

@@ -4,3 +4,6 @@
2025-01-07: Deploy executor.
2025-04-24: Deploy executor.
2025-04-24: Deploy assistant.
2025-05-04: Deploy qs, search and api-services.
2025-05-05: Deploy executor partition 4.
2025-05-05: Deploy control.

View File

@@ -314,6 +314,13 @@ if __name__ == '__main__':
deploy_tier=0,
groups={"all", "core"}
),
'status': ServiceConfig(
gradle_target=':code:services-application:status-service:docker',
docker_name='status-service',
instances=None,
deploy_tier=4,
groups={"all"}
),
'query': ServiceConfig(
gradle_target=':code:services-core:query-service:docker',
docker_name='query-service',