(search) Make it clearer that it's a search engine

Pdf support in crawler

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());
-                     var os = new BufferedOutputStream(Files.newOutputStream(Path.of(tarFileName), StandardOpenOption.CREATE))) {
-                    is.transferTo(os);
+                HttpURLConnection urlConnection = (HttpURLConnection) new URI(downloadURI).toURL().openConnection();
+
+                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/processes/crawling-process/java/nu/marginalia/crawl/CrawlerMain.java

@@ -495,7 +495,7 @@ public class CrawlerMain extends ProcessMainClass {
                     // (mostly a case when migrating from legacy->warc)
                     reference.delete();
 
-                    // Convert the WARC file to Parquet
+                    // Convert the WARC file to Slop
                     SlopCrawlDataRecord
                             .convertWarc(domain, userAgent, newWarcFile, slopFile);
 

code/processes/crawling-process/java/nu/marginalia/crawl/fetcher/warc/WarcRecorder.java

@@ -41,7 +41,7 @@ public class WarcRecorder implements AutoCloseable {
     static final int MAX_TIME = 30_000;
 
     /** Maximum (decompressed) size we'll save */
-    static final int MAX_SIZE = Integer.getInteger("crawler.maxFetchSize", 10 * 1024 * 1024);
+    static final int MAX_SIZE = Integer.getInteger("crawler.maxFetchSize", 32 * 1024 * 1024);
 
     private final WarcWriter writer;
     private final Path warcFile;

code/processes/crawling-process/java/nu/marginalia/crawl/logic/DomainLocks.java

@@ -43,18 +43,18 @@ public class DomainLocks {
             return new Semaphore(16);
         if (topDomain.equals("blogspot.com"))
             return new Semaphore(8);
-
+        if (topDomain.equals("tumblr.com"))
+            return new Semaphore(8);
         if (topDomain.equals("neocities.org"))
-            return new Semaphore(4);
+            return new Semaphore(8);
         if (topDomain.equals("github.io"))
-            return new Semaphore(4);
+            return new Semaphore(8);
 
+        // Substack really dislikes broad-scale crawlers, so we need to be careful
+        // to not get blocked.
         if (topDomain.equals("substack.com")) {
             return new Semaphore(1);
         }
-        if (topDomain.endsWith(".edu")) {
-            return new Semaphore(1);
-        }
 
         return new Semaphore(2);
     }

code/processes/crawling-process/model/java/nu/marginalia/ContentTypes.java

@@ -6,6 +6,7 @@ public class ContentTypes {
     public static final Set<String> acceptedContentTypes = Set.of("application/xhtml+xml",
             "application/xhtml",
             "text/html",
+            "application/pdf",
             "image/x-icon",
             "text/plain");
 
@@ -19,4 +20,9 @@ public class ContentTypes {
         return false;
     }
 
+    public static boolean isBinary(String contentTypeHeader) {
+        String lcHeader = contentTypeHeader.toLowerCase();
+        return lcHeader.startsWith("application/pdf");
+    }
+
 }

code/processes/crawling-process/model/java/nu/marginalia/io/crawldata/format/SlopSerializableCrawlDataStream.java

@@ -37,8 +37,12 @@ public class SlopSerializableCrawlDataStream implements AutoCloseable, Serializa
             public boolean filter(String url, int status, String contentType) {
                 String ctLc = contentType.toLowerCase();
 
+                // Permit all plain text content types
                 if (ctLc.startsWith("text/"))
                     return true;
+                // PDF
+                else if (ctLc.startsWith("application/pdf"))
+                    return true;
                 else if (ctLc.startsWith("x-marginalia/"))
                     return true;
 

code/processes/crawling-process/model/java/nu/marginalia/model/body/ContentTypeLogic.java

@@ -10,7 +10,7 @@ import java.util.regex.Pattern;
 
 public class ContentTypeLogic {
 
-    private static final Predicate<String> probableHtmlPattern = Pattern.compile("^.*\\.(htm|html|php|txt|md)$").asMatchPredicate();
+    private static final Predicate<String> probableGoodPattern = Pattern.compile("^.*\\.(htm|html|php|txt|md|pdf)$").asMatchPredicate();
     private static final Predicate<String> probableBinaryPattern = Pattern.compile("^.*\\.[a-z]+$").asMatchPredicate();
     private static final Set<String> blockedContentTypes = Set.of("text/css", "text/javascript");
     private static final List<String> acceptedContentTypePrefixes = List.of(
@@ -22,6 +22,7 @@ public class ContentTypeLogic {
             "application/rss+xml",
             "application/x-rss+xml",
             "application/rdf+xml",
+            "application/pdf",
             "x-rss+xml"
     );
     private boolean allowAllContentTypes = false;
@@ -34,7 +35,7 @@ public class ContentTypeLogic {
     public boolean isUrlLikeBinary(EdgeUrl url) {
         String pathLowerCase = url.path.toLowerCase();
 
-        if (probableHtmlPattern.test(pathLowerCase))
+        if (probableGoodPattern.test(pathLowerCase))
             return false;
 
         return probableBinaryPattern.test(pathLowerCase);

code/processes/crawling-process/model/java/nu/marginalia/slop/SlopCrawlDataRecord.java

@@ -216,6 +216,11 @@ public record SlopCrawlDataRecord(String domain,
             return false;
         }
 
+        // If the format is binary, we don't want to translate it if the response is truncated
+        if (response.truncated() != WarcTruncationReason.NOT_TRUNCATED && ContentTypes.isBinary(contentType)) {
+            return false;
+        }
+
         return true;
     }
 

code/processes/crawling-process/test/nu/marginalia/crawl/fetcher/HttpFetcherImplFetchTest.java

@@ -40,6 +40,8 @@ class HttpFetcherImplFetchTest {
     private static EdgeUrl badHttpStatusUrl;
     private static EdgeUrl keepAliveUrl;
 
+    private static EdgeUrl pdfUrl;
+
     @BeforeAll
     public static void setupAll() throws URISyntaxException {
         wireMockServer =
@@ -133,6 +135,13 @@ class HttpFetcherImplFetchTest {
                         ));
 
 
+        pdfUrl = new EdgeUrl("http://localhost:18089/test.pdf");
+        wireMockServer.stubFor(WireMock.get(WireMock.urlEqualTo(pdfUrl.path))
+                .willReturn(WireMock.aResponse()
+                        .withHeader("Content-Type", "application/pdf")
+                        .withStatus(200)
+                        .withBody("Hello World")));
+
         wireMockServer.start();
 
     }
@@ -352,6 +361,14 @@ class HttpFetcherImplFetchTest {
         Assertions.assertTrue(result.isOk());
     }
 
+    @Test
+    public void testPdf() {
+        var result = fetcher.fetchContent(pdfUrl, warcRecorder, new DomainCookies(), new CrawlDelayTimer(1000), ContentTags.empty(), HttpFetcher.ProbeType.FULL);
+
+        Assertions.assertInstanceOf(HttpFetchResult.ResultOk.class, result);
+        Assertions.assertTrue(result.isOk());
+    }
+
 
     private List<WarcRecord> getWarcRecords() throws IOException {
         List<WarcRecord> records = new ArrayList<>();

code/processes/crawling-process/test/nu/marginalia/crawl/retreival/fetcher/WarcRecorderTest.java

@@ -4,9 +4,9 @@ import nu.marginalia.UserAgent;
 import nu.marginalia.crawl.fetcher.ContentTags;
 import nu.marginalia.crawl.fetcher.DomainCookies;
 import nu.marginalia.crawl.fetcher.warc.WarcRecorder;
+import nu.marginalia.io.SerializableCrawlDataStream;
 import nu.marginalia.model.EdgeUrl;
-import nu.marginalia.parquet.crawldata.CrawledDocumentParquetRecordFileReader;
-import nu.marginalia.parquet.crawldata.CrawledDocumentParquetRecordFileWriter;
+import nu.marginalia.slop.SlopCrawlDataRecord;
 import org.apache.hc.client5.http.classic.HttpClient;
 import org.apache.hc.client5.http.classic.methods.HttpGet;
 import org.apache.hc.client5.http.impl.classic.HttpClients;
@@ -24,13 +24,14 @@ import java.nio.file.Files;
 import java.nio.file.Path;
 import java.security.NoSuchAlgorithmException;
 import java.util.HashMap;
+import java.util.List;
 import java.util.Map;
 
 import static org.junit.jupiter.api.Assertions.assertEquals;
 
 class WarcRecorderTest {
     Path fileNameWarc;
-    Path fileNameParquet;
+    Path fileNameSlop;
     WarcRecorder client;
 
     HttpClient httpClient;
@@ -39,7 +40,7 @@ class WarcRecorderTest {
         httpClient = HttpClients.createDefault();
 
         fileNameWarc = Files.createTempFile("test", ".warc");
-        fileNameParquet = Files.createTempFile("test", ".parquet");
+        fileNameSlop = Files.createTempFile("test", ".slop.zip");
 
         client = new WarcRecorder(fileNameWarc);
     }
@@ -159,17 +160,28 @@ class WarcRecorderTest {
 
         client.fetch(httpClient, new DomainCookies(), request3);
 
-        CrawledDocumentParquetRecordFileWriter.convertWarc(
+        HttpGet request4 = new HttpGet("https://downloads.marginalia.nu/test.pdf");
+        request4.addHeader("User-agent", "test.marginalia.nu");
+        request4.addHeader("Accept-Encoding", "gzip");
+
+        client.fetch(httpClient, new DomainCookies(), request4);
+
+        SlopCrawlDataRecord.convertWarc(
                 "www.marginalia.nu",
                 new UserAgent("test", "test"),
                 fileNameWarc,
-                fileNameParquet);
+                fileNameSlop);
 
-        var urls = CrawledDocumentParquetRecordFileReader.stream(fileNameParquet).map(doc -> doc.url).toList();
-        assertEquals(2, urls.size());
+        List<String> urls;
+        try (var stream = SerializableCrawlDataStream.openDataStream(fileNameSlop)) {
+            urls = stream.docsAsList().stream().map(doc -> doc.url.toString()).toList();
+        }
+
+        assertEquals(3, urls.size());
         assertEquals("https://www.marginalia.nu/", urls.get(0));
         assertEquals("https://www.marginalia.nu/log/", urls.get(1));
         // sanic.jpg gets filtered out for its bad mime type
+        assertEquals("https://downloads.marginalia.nu/test.pdf", urls.get(2));
 
     }
 

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

@@ -13,7 +13,7 @@
                    class="shadow-inner flex-1 dark:bg-black dark:text-gray-100 bg-gray-50 border dark:border-gray-600 border-gray-300 text-gray-900 text-sm rounded-sm block w-full p-2.5"
                    value="${query}"
                    autofocus
-                   placeholder="Search..."
+                   placeholder="Search the web!"
                    autocomplete="off"
                    name="query"
                    id="searchInput" />
@@ -21,7 +21,7 @@
             <input type="text"
                    class="shadow-inner flex-1 dark:bg-black dark:text-gray-100 bg-gray-50 border dark:border-gray-600 border-gray-300 text-gray-900 text-sm rounded-sm block w-full p-2.5"
                    value="${query}"
-                   placeholder="Search..."
+                   placeholder="Search the web!"
                    autocomplete="off"
                    name="query"
                    id="searchInput" />

code/services-core/control-service/resources/templates/control/node/actions/partial-download-sample-data.hdb

@@ -24,25 +24,25 @@ This is a sample of real crawl data.  It is intended for demo, testing and devel
     <tr>
         <td><input id="sample-s" value="sample-s" name="sample" class="form-check-input" type="radio"></td>
         <td><label for="sample-s">Small</label></td>
-        <td>1000 Domains. About 2 GB. </td>
+        <td>1000 Domains. About 1 GB. </td>
     </tr>
 
     <tr>
         <td><input id="sample-m" value="sample-m" name="sample" class="form-check-input" type="radio"></td>
         <td><label for="sample-m">Medium</label></td>
-        <td>2000 Domains. About 6 GB. Recommended.</td>
+        <td>2000 Domains. About 2 GB. Recommended.</td>
     </tr>
 
     <tr>
         <td><input id="sample-l" value="sample-l" name="sample" class="form-check-input" type="radio"></td>
         <td><label for="sample-l">Large</label></td>
-        <td>5000 Domains.  About 20 GB.</td>
+        <td>5000 Domains.  About 7 GB.</td>
     </tr>
 
     <tr>
         <td><input id="sample-xl" value="sample-xl" name="sample" class="form-check-input" type="radio"></td>
         <td><label for="sample-xl">Huge</label></td>
-        <td>50,000 Domains.  Around 180 GB.  Primarily intended for pre-production like testing environments.
+        <td>50,000 Domains.  Around 80 GB.  Primarily intended for pre-production like testing environments.
             Expect hours of processing time. </td>
     </tr>
 </table>