Merge pull request DSpace#10152 from DSpace/backport-9893-to-main

[Port main] Fix full-text indexing for files over the character limit
atmire · Dec 19, 2024 · 5610412 · 5610412
2 parents 97dd1e0 + fc4cf8f
commit 5610412
Showing 1 changed file with 15 additions and 16 deletions.
diff --git a/dspace-api/src/main/java/org/dspace/discovery/indexobject/IndexFactoryImpl.java b/dspace-api/src/main/java/org/dspace/discovery/indexobject/IndexFactoryImpl.java
@@ -118,20 +118,10 @@ protected void writeDocument(SolrInputDocument doc, FullTextContentStreams strea
                 ParseContext tikaContext = new ParseContext();
 
                 // Use Apache Tika to parse the full text stream(s)
+                boolean extractionSucceeded = false;
                 try (InputStream fullTextStreams = streams.getStream()) {
                     tikaParser.parse(fullTextStreams, tikaHandler, tikaMetadata, tikaContext);
-
-                    // Write Tika metadata to "tika_meta_*" fields.
-                    // This metadata is not very useful right now,
-                    // but we'll keep it just in case it becomes more useful.
-                    for (String name : tikaMetadata.names()) {
-                        for (String value : tikaMetadata.getValues(name)) {
-                            doc.addField("tika_meta_" + name, value);
-                        }
-                    }
-
-                    // Save (parsed) full text to "fulltext" field
-                    doc.addField("fulltext", tikaHandler.toString());
+                    extractionSucceeded = true;
                 } catch (SAXException saxe) {
                     // Check if this SAXException is just a notice that this file was longer than the character limit.
                     // Unfortunately there is not a unique, public exception type to catch here. This error is thrown
@@ -141,18 +131,27 @@ protected void writeDocument(SolrInputDocument doc, FullTextContentStreams strea
                         // log that we only indexed up to that configured limit
                         log.info("Full text is larger than the configured limit (discovery.solr.fulltext.charLimit)."
                                 + " Only the first {} characters were indexed.", charLimit);
+                        extractionSucceeded = true;
                     } else {
                         log.error("Tika parsing error. Could not index full text.", saxe);
                         throw new IOException("Tika parsing error. Could not index full text.", saxe);
                     }
                 } catch (TikaException | IOException ex) {
                     log.error("Tika parsing error. Could not index full text.", ex);
                     throw new IOException("Tika parsing error. Could not index full text.", ex);
-                } finally {
-                    // Add document to index
-                    solr.add(doc);
                 }
-                return;
+                if (extractionSucceeded) {
+                    // Write Tika metadata to "tika_meta_*" fields.
+                    // This metadata is not very useful right now,
+                    // but we'll keep it just in case it becomes more useful.
+                    for (String name : tikaMetadata.names()) {
+                        for (String value : tikaMetadata.getValues(name)) {
+                            doc.addField("tika_meta_" + name, value);
+                        }
+                    }
+                    // Save (parsed) full text to "fulltext" field
+                    doc.addField("fulltext", tikaHandler.toString());
+                }
             }
             // Add document to index
             solr.add(doc);