(converter) Increase work buffer for converter

Conversion on index node  7 in production is crashing ostensibly because this buffer is too small.

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

@@ -112,14 +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();
-    }
-
     /** 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/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/index/index-forward/java/nu/marginalia/index/forward/construction/ForwardIndexConverter.java

@@ -84,7 +84,7 @@ public class ForwardIndexConverter {
 
             LongArray docFileData = LongArrayFactory.mmapForWritingConfined(outputFileDocsData, ForwardIndexParameters.ENTRY_SIZE * docsFileId.size());
 
-            ByteBuffer workArea = ByteBuffer.allocate(65536);
+            ByteBuffer workArea = ByteBuffer.allocate(1024*1024*100);
             for (var instance : journal.pages()) {
                 try (var slopTable = new SlopTable(instance.baseDir(), instance.page()))
                 {

code/processes/converting-process/java/org/apache/pdfbox/text/HeadingAwarePDFTextStripper.java

@@ -2002,12 +2002,11 @@ public class HeadingAwarePDFTextStripper extends LegacyPDFStreamEngine
         float minFontWeight = Integer.MAX_VALUE;
         for (var word : line)
         {
-            int i = 0;
             for (var textPosition : word.getTextPositions())
             {
-                if (word.text.charAt(i++) == ' ') {
-                    continue;
-                }
+                // Skip empty text positions as they may have a different font
+                if (word.text.isBlank()) continue;
+
                 var font = textPosition.getFont();
                 if (font == null) continue;
                 var descriptor = font.getFontDescriptor();

code/services-application/search-service-legacy/java/nu/marginalia/search/results/UrlDeduplicator.java

@@ -61,7 +61,7 @@ public class UrlDeduplicator {
 
     private boolean limitResultsPerDomain(DecoratedSearchResultItem details) {
         final var domain = details.getUrl().getDomain();
-        final String key = domain.getDomainKey();
+        final String key = domain.toString();
 
         return keyCount.adjustOrPutValue(key, 1, 1) <= resultsPerKey;
     }

code/services-application/search-service/java/nu/marginalia/search/results/UrlDeduplicator.java

@@ -25,6 +25,7 @@ public class UrlDeduplicator {
     }
 
     public boolean shouldRemove(DecoratedSearchResultItem details) {
+
         if (!deduplicateOnSuperficialHash(details))
             return true;
         if (!deduplicateOnLSH(details))
@@ -61,7 +62,7 @@ public class UrlDeduplicator {
 
     private boolean limitResultsPerDomain(DecoratedSearchResultItem details) {
         final var domain = details.getUrl().getDomain();
-        final String key = domain.getDomainKey();
+        final String key = domain.toString();
 
         return keyCount.adjustOrPutValue(key, 1, 1) <= resultsPerKey;
     }

deploy.txt

@@ -7,4 +7,5 @@
 2025-05-04:  Deploy qs, search and api-services.
 2025-05-05:  Deploy executor partition 4.
 2025-05-05:  Deploy control.
-2025-05-08:  Deploy assistant.
+2025-05-08:  Deploy assistant.
+2025-05-17:  Redeploy all.