From: Dominik Stadler <centic@apache.org>
Date: Wed, 6 Dec 2023 19:50:02 +0000 (+0000)
Subject: Make sure OLE2ScratchpadExtractorFacory is sorted first
X-Git-Tag: REL_5_3_0~156
X-Git-Url: https://source.dussan.org/?a=commitdiff_plain;h=c8c8130ae352d8647b03faba492732883471e8bf;p=poi.git

Make sure OLE2ScratchpadExtractorFacory is sorted first

Otherwise order of found extractors would depend on jar-loading order
and thus might have unexpected side-effects and missing features in
text-extraction.

git-svn-id: https://svn.apache.org/repos/asf/poi/trunk@1914407 13f79535-47bb-0310-9956-ffa450edef68
---

diff --git a/poi/src/main/java/org/apache/poi/extractor/ExtractorFactory.java b/poi/src/main/java/org/apache/poi/extractor/ExtractorFactory.java
index f784842274..ea3ee82f9a 100644
--- a/poi/src/main/java/org/apache/poi/extractor/ExtractorFactory.java
+++ b/poi/src/main/java/org/apache/poi/extractor/ExtractorFactory.java
@@ -88,6 +88,21 @@ public final class ExtractorFactory {
     private ExtractorFactory() {
         ClassLoader cl = ExtractorFactory.class.getClassLoader();
         ServiceLoader.load(ExtractorProvider.class, cl).forEach(provider::add);
+
+        // loading of service-files is non-deterministic as it depends on order of loaded jars
+        // however we would like to "prefer" one Factory, so let's make sure the more
+        // powerful "ScratchpadProvider" is sorted first
+        provider.sort((o1, o2) -> {
+            if (o1.getClass() != o2.getClass()) {
+                if (o1.getClass().getSimpleName().equals("OLE2ScratchpadExtractorFactory")) {
+                    return -1;
+                } else if (o2.getClass().getSimpleName().equals("OLE2ScratchpadExtractorFactory")) {
+                    return 1;
+                }
+            }
+
+            return o1.getClass().getName().compareTo(o2.getClass().getName());
+        });
     }
 
     /**
diff --git a/poi/src/main/java/org/apache/poi/extractor/ExtractorProvider.java b/poi/src/main/java/org/apache/poi/extractor/ExtractorProvider.java
index 5669cd9fe7..777cd38709 100644
--- a/poi/src/main/java/org/apache/poi/extractor/ExtractorProvider.java
+++ b/poi/src/main/java/org/apache/poi/extractor/ExtractorProvider.java
@@ -70,7 +70,8 @@ public interface ExtractorProvider {
      * @param dirs a list to be filled with directory references holding embedded
      * @param nonPOIFS a list to be filled with streams which aren't based on POIFS entries
      *
-     * @throws IOException when the format specific extraction fails because of invalid entires
+     * @throws IOException when the format specific extraction fails because of invalid entries
+     * @throws java.lang.IllegalArgumentException if implementations do not overwrite this method
      */
     default void identifyEmbeddedResources(POIOLE2TextExtractor ext, List<Entry> dirs, List<InputStream> nonPOIFS) throws IOException {
         throw new IllegalArgumentException("Error checking for Scratchpad embedded resources");