Update tika to 2.9.2

Signed-off-by: Sandesh Kumar <sandeshkr419@gmail.com>
opensearch-project · Apr 2, 2024 · 86977e8 · 86977e8
1 parent 16c6977
commit 86977e8
Show file tree

Hide file tree

Showing 43 changed files with 28 additions and 30 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -57,7 +57,10 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 - Bump `org.bouncycastle:bcpkix-jdk15to18` to `org.bouncycastle:bcpkix-jdk18on` version 1.77 ([#12317](https://github.com/opensearch-project/OpenSearch/pull/12317))
 - Bump Jackson version from 2.16.1 to 2.16.2 ([#12611](https://github.com/opensearch-project/OpenSearch/pull/12611))
 - Bump `aws-sdk-java` from 2.20.55 to 2.20.86 ([#12251](https://github.com/opensearch-project/OpenSearch/pull/12251))
-- Bump `org.apache.commons:commons-compress` from 1.24.0 to 1.26.0 ([#12604](https://github.com/opensearch-project/OpenSearch/pull/12604))
+- Bump `org.apache.commons:commons-compress` from 1.24.0 to 1.26.1 ([#12627](https://github.com/opensearch-project/OpenSearch/pull/12627))
+- Bump `org.apache.commons:commonscodec` from 1.15 to 1.16.1 ([#12627](https://github.com/opensearch-project/OpenSearch/pull/12627))
+- Bump `org.apache.commons:commonslang` from 3.13.0 to 3.14.0 ([#12627](https://github.com/opensearch-project/OpenSearch/pull/12627))
+- Bump Apache Tika from 2.6.0 to 2.9.2 ([#12627](https://github.com/opensearch-project/OpenSearch/pull/12627))
 
 ### Changed
 - [CCR] Add getHistoryOperationsFromTranslog method to fetch the history snapshot from translogs ([#3948](https://github.com/opensearch-project/OpenSearch/pull/3948))

diff --git a/plugins/ingest-attachment/build.gradle b/plugins/ingest-attachment/build.gradle
@@ -38,10 +38,10 @@ opensearchplugin {
 }
 
 versions << [
-  'tika'  : '2.9.1',
-  'pdfbox': '2.0.27',
-  'poi'   : '5.2.3',
-  'mime4j': '0.8.8'
+  'tika'  : '2.9.2',
+  'pdfbox': '2.0.31',
+  'poi'   : '5.2.5',
+  'mime4j': '0.8.11'
 ]
 
 dependencies {

diff --git a/plugins/ingest-attachment/licenses/apache-mime4j-core-0.8.11.jar.sha1 b/plugins/ingest-attachment/licenses/apache-mime4j-core-0.8.11.jar.sha1
@@ -0,0 +1 @@
+6d1eb5f7b84eaa9d38fca13b761f01c693aef3da
diff --git a/plugins/ingest-attachment/licenses/apache-mime4j-core-0.8.8.jar.sha1 b/plugins/ingest-attachment/licenses/apache-mime4j-core-0.8.8.jar.sha1
diff --git a/plugins/ingest-attachment/licenses/apache-mime4j-dom-0.8.11.jar.sha1 b/plugins/ingest-attachment/licenses/apache-mime4j-dom-0.8.11.jar.sha1
@@ -0,0 +1 @@
+f0d42ab9a5832b5f5d05afc004b31245b838e0fc
diff --git a/plugins/ingest-attachment/licenses/apache-mime4j-dom-0.8.8.jar.sha1 b/plugins/ingest-attachment/licenses/apache-mime4j-dom-0.8.8.jar.sha1
diff --git a/plugins/ingest-attachment/licenses/fontbox-2.0.27.jar.sha1 b/plugins/ingest-attachment/licenses/fontbox-2.0.27.jar.sha1
diff --git a/plugins/ingest-attachment/licenses/fontbox-2.0.31.jar.sha1 b/plugins/ingest-attachment/licenses/fontbox-2.0.31.jar.sha1
@@ -0,0 +1 @@
+96999ecdb7324bf718b88724818fa62f81286c36
diff --git a/plugins/ingest-attachment/licenses/pdfbox-2.0.27.jar.sha1 b/plugins/ingest-attachment/licenses/pdfbox-2.0.27.jar.sha1
diff --git a/plugins/ingest-attachment/licenses/pdfbox-2.0.31.jar.sha1 b/plugins/ingest-attachment/licenses/pdfbox-2.0.31.jar.sha1
@@ -0,0 +1 @@
+29b25053099bc30784a766ccb821417e06f4b8a1
diff --git a/plugins/ingest-attachment/licenses/poi-5.2.3.jar.sha1 b/plugins/ingest-attachment/licenses/poi-5.2.3.jar.sha1
diff --git a/plugins/ingest-attachment/licenses/poi-5.2.5.jar.sha1 b/plugins/ingest-attachment/licenses/poi-5.2.5.jar.sha1
@@ -0,0 +1 @@
+7e00f6b2f76375fe89022d5a7db8acb71cbd55f5
diff --git a/plugins/ingest-attachment/licenses/poi-ooxml-5.2.3.jar.sha1 b/plugins/ingest-attachment/licenses/poi-ooxml-5.2.3.jar.sha1
diff --git a/plugins/ingest-attachment/licenses/poi-ooxml-5.2.5.jar.sha1 b/plugins/ingest-attachment/licenses/poi-ooxml-5.2.5.jar.sha1
@@ -0,0 +1 @@
+df9f2c52371eeba24db8ea8cafa77285c3cc0742
diff --git a/plugins/ingest-attachment/licenses/poi-ooxml-lite-5.2.3.jar.sha1 b/plugins/ingest-attachment/licenses/poi-ooxml-lite-5.2.3.jar.sha1
diff --git a/plugins/ingest-attachment/licenses/poi-ooxml-lite-5.2.5.jar.sha1 b/plugins/ingest-attachment/licenses/poi-ooxml-lite-5.2.5.jar.sha1
@@ -0,0 +1 @@
+eaa61452d8f0d13080fbb4757a392f09f90e4c49
diff --git a/plugins/ingest-attachment/licenses/poi-scratchpad-5.2.3.jar.sha1 b/plugins/ingest-attachment/licenses/poi-scratchpad-5.2.3.jar.sha1
diff --git a/plugins/ingest-attachment/licenses/poi-scratchpad-5.2.5.jar.sha1 b/plugins/ingest-attachment/licenses/poi-scratchpad-5.2.5.jar.sha1
@@ -0,0 +1 @@
+fc600cf765a49d73935a6e48a5b84f4abcdd0518
diff --git a/plugins/ingest-attachment/licenses/tika-core-2.9.1.jar.sha1 b/plugins/ingest-attachment/licenses/tika-core-2.9.1.jar.sha1
diff --git a/plugins/ingest-attachment/licenses/tika-core-2.9.2.jar.sha1 b/plugins/ingest-attachment/licenses/tika-core-2.9.2.jar.sha1
@@ -0,0 +1 @@
+796a21391780339e3d4862626339b49df170024e
diff --git a/plugins/ingest-attachment/licenses/tika-langdetect-optimaize-2.9.1.jar.sha1 b/plugins/ingest-attachment/licenses/tika-langdetect-optimaize-2.9.1.jar.sha1
diff --git a/plugins/ingest-attachment/licenses/tika-langdetect-optimaize-2.9.2.jar.sha1 b/plugins/ingest-attachment/licenses/tika-langdetect-optimaize-2.9.2.jar.sha1
@@ -0,0 +1 @@
+7a48a287e464b456a85c79f318d7bad7db201518
diff --git a/plugins/ingest-attachment/licenses/tika-parser-apple-module-2.9.1.jar.sha1 b/plugins/ingest-attachment/licenses/tika-parser-apple-module-2.9.1.jar.sha1
diff --git a/plugins/ingest-attachment/licenses/tika-parser-apple-module-2.9.2.jar.sha1 b/plugins/ingest-attachment/licenses/tika-parser-apple-module-2.9.2.jar.sha1
@@ -0,0 +1 @@
+758dac27c246c51b019562bab7e266d2da6a6e01
diff --git a/plugins/ingest-attachment/licenses/tika-parser-html-module-2.9.1.jar.sha1 b/plugins/ingest-attachment/licenses/tika-parser-html-module-2.9.1.jar.sha1
diff --git a/plugins/ingest-attachment/licenses/tika-parser-html-module-2.9.2.jar.sha1 b/plugins/ingest-attachment/licenses/tika-parser-html-module-2.9.2.jar.sha1
@@ -0,0 +1 @@
+47f6a4c46b92616d14e82cd7ad4d05cb43077b83
diff --git a/plugins/ingest-attachment/licenses/tika-parser-microsoft-module-2.9.1.jar.sha1 b/plugins/ingest-attachment/licenses/tika-parser-microsoft-module-2.9.1.jar.sha1
diff --git a/plugins/ingest-attachment/licenses/tika-parser-microsoft-module-2.9.2.jar.sha1 b/plugins/ingest-attachment/licenses/tika-parser-microsoft-module-2.9.2.jar.sha1
@@ -0,0 +1 @@
+235a20823c02c699ce3d57f3d6b9550db05d91a9
diff --git a/plugins/ingest-attachment/licenses/tika-parser-miscoffice-module-2.9.1.jar.sha1 b/plugins/ingest-attachment/licenses/tika-parser-miscoffice-module-2.9.1.jar.sha1
diff --git a/plugins/ingest-attachment/licenses/tika-parser-miscoffice-module-2.9.2.jar.sha1 b/plugins/ingest-attachment/licenses/tika-parser-miscoffice-module-2.9.2.jar.sha1
@@ -0,0 +1 @@
+7688a4220d07c32b505230479f957cd495c0bef2
diff --git a/plugins/ingest-attachment/licenses/tika-parser-pdf-module-2.9.1.jar.sha1 b/plugins/ingest-attachment/licenses/tika-parser-pdf-module-2.9.1.jar.sha1
diff --git a/plugins/ingest-attachment/licenses/tika-parser-pdf-module-2.9.2.jar.sha1 b/plugins/ingest-attachment/licenses/tika-parser-pdf-module-2.9.2.jar.sha1
@@ -0,0 +1 @@
+4d0f0e3f6eff184040402094f4fabbb3c5c7d09f
diff --git a/plugins/ingest-attachment/licenses/tika-parser-text-module-2.9.1.jar.sha1 b/plugins/ingest-attachment/licenses/tika-parser-text-module-2.9.1.jar.sha1
diff --git a/plugins/ingest-attachment/licenses/tika-parser-text-module-2.9.2.jar.sha1 b/plugins/ingest-attachment/licenses/tika-parser-text-module-2.9.2.jar.sha1
@@ -0,0 +1 @@
+b3a93e538ba6cb4066aba96d629febf181ec9f92
diff --git a/plugins/ingest-attachment/licenses/tika-parser-xml-module-2.9.1.jar.sha1 b/plugins/ingest-attachment/licenses/tika-parser-xml-module-2.9.1.jar.sha1
diff --git a/plugins/ingest-attachment/licenses/tika-parser-xml-module-2.9.2.jar.sha1 b/plugins/ingest-attachment/licenses/tika-parser-xml-module-2.9.2.jar.sha1
@@ -0,0 +1 @@
+ff707716c0c4748ffeb21996aefa8d269b3eab5b
diff --git a/plugins/ingest-attachment/licenses/tika-parser-xmp-commons-2.9.1.jar.sha1 b/plugins/ingest-attachment/licenses/tika-parser-xmp-commons-2.9.1.jar.sha1
diff --git a/plugins/ingest-attachment/licenses/tika-parser-xmp-commons-2.9.2.jar.sha1 b/plugins/ingest-attachment/licenses/tika-parser-xmp-commons-2.9.2.jar.sha1
@@ -0,0 +1 @@
+69104107ff85194df5acf682178128771863e442
diff --git a/plugins/ingest-attachment/licenses/tika-parser-zip-commons-2.9.1.jar.sha1 b/plugins/ingest-attachment/licenses/tika-parser-zip-commons-2.9.1.jar.sha1
diff --git a/plugins/ingest-attachment/licenses/tika-parser-zip-commons-2.9.2.jar.sha1 b/plugins/ingest-attachment/licenses/tika-parser-zip-commons-2.9.2.jar.sha1
@@ -0,0 +1 @@
+2fcea85a56f93a5c0cb81f3d6dd8673f3d81c598
diff --git a/plugins/ingest-attachment/licenses/tika-parsers-standard-package-2.9.1.jar.sha1 b/plugins/ingest-attachment/licenses/tika-parsers-standard-package-2.9.1.jar.sha1
diff --git a/plugins/ingest-attachment/licenses/tika-parsers-standard-package-2.9.2.jar.sha1 b/plugins/ingest-attachment/licenses/tika-parsers-standard-package-2.9.2.jar.sha1
@@ -0,0 +1 @@
+c8408deb51fa617ef4e912b4d161712e695d3a29
diff --git a/plugins/ingest-attachment/src/test/java/org/opensearch/ingest/attachment/TikaDocTests.java b/plugins/ingest-attachment/src/test/java/org/opensearch/ingest/attachment/TikaDocTests.java
@@ -65,11 +65,6 @@ public void testFiles() throws Exception {
 
         try (DirectoryStream<Path> stream = Files.newDirectoryStream(tmp)) {
             for (Path doc : stream) {
-                // Testing change with skipping - will remove in final code
-                if (doc.toString().endsWith(".pages") || doc.toString().endsWith(".key")) {
-                    logger.debug("skipping: {}", doc);
-                    continue;
-                }
                 logger.debug("parsing: {}", doc);
                 assertParseable(doc);
             }