(sample) Fix bug where slop files would not be saved despite containing data

We now sometimes get 206 when good.

code/processes/crawling-process/java/nu/marginalia/crawl/CrawlerMain.java

@@ -448,13 +448,7 @@ public class CrawlerMain extends ProcessMainClass {
             // We don't have a lock, so we can't run this task
             // we return to avoid blocking the pool for too long
             if (lock.isEmpty()) {
-                if (retryQueue.remainingCapacity() > 0) {
+                pendingCrawlTasks.remove(domain);
-                    // Sleep a moment to avoid busy looping via the retry queue
-                    // in the case when few tasks remain and almost all are ineligible for
-                    // immediate restart
-                    Thread.sleep(5);
-                }
-
                 retryQueue.put(this);
                 return;
             }

code/processes/crawling-process/java/nu/marginalia/crawl/retreival/revisit/CrawlerRevisitor.java

@@ -74,7 +74,7 @@ public class CrawlerRevisitor {
 
             // If the reference document is empty or the HTTP status is not 200, we'll skip it since it's
             // unlikely to produce anything meaningful for us.
-            if (doc.httpStatus != 200)
+            if (doc.httpStatus != 200 && doc.httpStatus != 206)
                 continue;
             if (!doc.hasBody())
                 continue;

code/processes/crawling-process/java/nu/marginalia/crawl/retreival/revisit/DocumentWithReference.java

@@ -58,7 +58,7 @@ public record DocumentWithReference(
         if (null == doc)
             return ContentTags.empty();
 
-        if (doc.documentBodyBytes.length == 0 || doc.httpStatus != 200)
+        if (doc.documentBodyBytes.length == 0 || (doc.httpStatus != 200 && doc.httpStatus != 206))
             return ContentTags.empty();
 
         String lastmod = doc.getLastModified();

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

@@ -1,5 +1,7 @@
 package nu.marginalia;
 
+import org.apache.commons.lang3.StringUtils;
+
 import java.util.Set;
 
 public class ContentTypes {
@@ -11,9 +13,9 @@ public class ContentTypes {
             "text/plain");
 
     public static boolean isAccepted(String contentTypeHeader) {
-        String lcHeader = contentTypeHeader.toLowerCase();
+        String lcHeader = StringUtils.substringBefore(contentTypeHeader.toLowerCase(), ';');
         for (var type : acceptedContentTypes) {
-            if (lcHeader.startsWith(type)) {
+            if (lcHeader.equals(type)) {
                 return true;
             }
         }
@@ -21,7 +23,7 @@ public class ContentTypes {
     }
 
     public static boolean isBinary(String contentTypeHeader) {
-        String lcHeader = contentTypeHeader.toLowerCase();
+        String lcHeader = StringUtils.substringBefore(contentTypeHeader.toLowerCase(), ';');
         return lcHeader.startsWith("application/pdf");
     }
 

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

@@ -277,7 +277,8 @@ public record SlopCrawlDataRecord(String domain,
         try (var table = new SlopTable(path)) {
             ShortColumn.Reader statusReader = statusColumn.open(table);
             while (statusReader.hasRemaining()) {
-                if (statusReader.get() == 200) {
+                int status = statusReader.get();
+                if (status == 200 || status == 206) {
                     cnt++;
                 }
             }

code/processes/export-task-process/java/nu/marginalia/extractor/SampleDataExporter.java

@@ -21,10 +21,7 @@ import java.nio.file.StandardCopyOption;
 import java.nio.file.StandardOpenOption;
 import java.nio.file.attribute.PosixFilePermissions;
 import java.sql.SQLException;
-import java.util.ArrayList;
+import java.util.*;
-import java.util.Collections;
-import java.util.List;
-import java.util.NoSuchElementException;
 
 public class SampleDataExporter {
     private final FileStorageService storageService;
@@ -127,14 +124,8 @@ public class SampleDataExporter {
              var reader = new SlopCrawlDataRecord.FilteringReader(crawlDataPath) {
                  @Override
                  public boolean filter(String url, int status, String contentType) {
-                     if (contentTypeFilter.equals(contentType))
+                     return Objects.equals(StringUtils.substringBefore(contentType, ';'), contentTypeFilter)
-                         return true;
+                                || contentType.startsWith("x-marginalia/"); // metadata records
-                     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;
                  }
              }
         ) {
@@ -143,7 +134,7 @@ public class SampleDataExporter {
                 var entry = reader.get();
                 writer.write(entry);
 
-                wroteEntry = wroteEntry || contentTypeFilter.equals(entry.contentType());
+                wroteEntry = wroteEntry || Objects.equals(StringUtils.substringBefore(entry.contentType(), ';'), contentTypeFilter);
             }
 
             if (!wroteEntry) {