From 4da546cc291a9994ec5c1d2de87d4dcb79555e2d Mon Sep 17 00:00:00 2001
From: Grant Ingersoll <gsingers@apache.org>
Date: Sun, 22 Nov 2009 16:18:49 +0000
Subject: [PATCH] SOLR-1567: Upgrade to Tika 0.5

git-svn-id: https://svn.apache.org/repos/asf/lucene/solr/trunk@883095 13f79535-47bb-0310-9956-ffa450edef68
---
 contrib/extraction/CHANGES.txt                       |  8 ++++++--
 contrib/extraction/lib/bcmail-jdk14-136.jar          |  2 --
 contrib/extraction/lib/bcprov-jdk14-136.jar          |  2 --
 contrib/extraction/lib/commons-codec-1.3.jar         |  2 --
 contrib/extraction/lib/commons-io-1.4.jar            |  2 --
 contrib/extraction/lib/commons-lang-2.1.jar          |  2 --
 contrib/extraction/lib/fontbox-0.1.0.jar             |  2 --
 contrib/extraction/lib/fontbox-0.8.0-incubator.jar   |  2 ++
 .../lib/geronimo-stax-api_1.0_spec-1.0.1.jar         |  2 ++
 .../lib/geronimo-stax-api_1.0_spec-1.0.jar           |  2 --
 contrib/extraction/lib/icu4j-3.8.jar                 |  2 --
 contrib/extraction/lib/jempbox-0.2.0.jar             |  2 --
 contrib/extraction/lib/jempbox-0.8.0-incubator.jar   |  2 ++
 .../lib/metadata-extractor-2.4.0-beta-1.jar          |  2 ++
 contrib/extraction/lib/nekohtml-1.9.9.jar            |  2 --
 contrib/extraction/lib/pdfbox-0.7.3.jar              |  2 --
 contrib/extraction/lib/pdfbox-0.8.0-incubating.jar   |  2 ++
 contrib/extraction/lib/poi-3.5-FINAL.jar             |  2 ++
 contrib/extraction/lib/poi-3.5-beta6.jar             |  2 --
 contrib/extraction/lib/poi-ooxml-3.5-FINAL.jar       |  2 ++
 contrib/extraction/lib/poi-ooxml-3.5-beta6.jar       |  2 --
 contrib/extraction/lib/poi-scratchpad-3.5-FINAL.jar  |  2 ++
 contrib/extraction/lib/poi-scratchpad-3.5-beta6.jar  |  2 --
 contrib/extraction/lib/tagsoup-1.2.jar               |  2 ++
 contrib/extraction/lib/tika-core-0.4.jar             |  2 --
 contrib/extraction/lib/tika-core-0.5.jar             |  2 ++
 contrib/extraction/lib/tika-parsers-0.4.jar          |  2 --
 contrib/extraction/lib/tika-parsers-0.5.jar          |  2 ++
 .../handler/extraction/ExtractingRequestHandler.java | 12 ++----------
 29 files changed, 30 insertions(+), 44 deletions(-)
 delete mode 100644 contrib/extraction/lib/bcmail-jdk14-136.jar
 delete mode 100644 contrib/extraction/lib/bcprov-jdk14-136.jar
 delete mode 100644 contrib/extraction/lib/commons-codec-1.3.jar
 delete mode 100644 contrib/extraction/lib/commons-io-1.4.jar
 delete mode 100644 contrib/extraction/lib/commons-lang-2.1.jar
 delete mode 100644 contrib/extraction/lib/fontbox-0.1.0.jar
 create mode 100644 contrib/extraction/lib/fontbox-0.8.0-incubator.jar
 create mode 100644 contrib/extraction/lib/geronimo-stax-api_1.0_spec-1.0.1.jar
 delete mode 100644 contrib/extraction/lib/geronimo-stax-api_1.0_spec-1.0.jar
 delete mode 100644 contrib/extraction/lib/icu4j-3.8.jar
 delete mode 100644 contrib/extraction/lib/jempbox-0.2.0.jar
 create mode 100644 contrib/extraction/lib/jempbox-0.8.0-incubator.jar
 create mode 100644 contrib/extraction/lib/metadata-extractor-2.4.0-beta-1.jar
 delete mode 100644 contrib/extraction/lib/nekohtml-1.9.9.jar
 delete mode 100644 contrib/extraction/lib/pdfbox-0.7.3.jar
 create mode 100644 contrib/extraction/lib/pdfbox-0.8.0-incubating.jar
 create mode 100644 contrib/extraction/lib/poi-3.5-FINAL.jar
 delete mode 100644 contrib/extraction/lib/poi-3.5-beta6.jar
 create mode 100644 contrib/extraction/lib/poi-ooxml-3.5-FINAL.jar
 delete mode 100644 contrib/extraction/lib/poi-ooxml-3.5-beta6.jar
 create mode 100644 contrib/extraction/lib/poi-scratchpad-3.5-FINAL.jar
 delete mode 100644 contrib/extraction/lib/poi-scratchpad-3.5-beta6.jar
 create mode 100644 contrib/extraction/lib/tagsoup-1.2.jar
 delete mode 100644 contrib/extraction/lib/tika-core-0.4.jar
 create mode 100644 contrib/extraction/lib/tika-core-0.5.jar
 delete mode 100644 contrib/extraction/lib/tika-parsers-0.4.jar
 create mode 100644 contrib/extraction/lib/tika-parsers-0.5.jar

diff --git a/contrib/extraction/CHANGES.txt b/contrib/extraction/CHANGES.txt
index 5b2729bd3dd..1609b2da955 100644
--- a/contrib/extraction/CHANGES.txt
+++ b/contrib/extraction/CHANGES.txt
@@ -17,8 +17,12 @@ You will need Solr up and running.  Then, simply add the extraction JAR file, pl
 to your Solr Home lib directory.  See http://wiki.apache.org/solr/ExtractingRequestHandler for more details on hooking it in
  and configuring.
 
-
 $Id:$
+
+================== Release 1.5-dev ==================
+
+* SOLR-1567: Upgrade to Tika 0.5, which upgrades many of the underlying libraries (PDFBox, for example) too (gsingers)
+
 ================== Release 1.4.0 ==================
 
 1. SOLR-284:  Added in support for extraction. (Eric Pugh, Chris Harris, gsingers)
@@ -34,4 +38,4 @@ $Id:$
     for discussion on language detection.
     See http://www.apache.org/dist/lucene/tika/CHANGES-0.4.txt. (gsingers)
 
-6. SOLR-1274: Added text serialization output for extractOnly (Peter Wolanin, gsingers)    
\ No newline at end of file
+6. SOLR-1274: Added text serialization output for extractOnly (Peter Wolanin, gsingers)    
diff --git a/contrib/extraction/lib/bcmail-jdk14-136.jar b/contrib/extraction/lib/bcmail-jdk14-136.jar
deleted file mode 100644
index 17c87e6170e..00000000000
--- a/contrib/extraction/lib/bcmail-jdk14-136.jar
+++ /dev/null
@@ -1,2 +0,0 @@
-AnyObjectId[3f78d2a6b7154d80e06bb96441dcf4faa60518c5] was removed in git history.
-Apache SVN contains full history.
\ No newline at end of file
diff --git a/contrib/extraction/lib/bcprov-jdk14-136.jar b/contrib/extraction/lib/bcprov-jdk14-136.jar
deleted file mode 100644
index 8e97af41f0c..00000000000
--- a/contrib/extraction/lib/bcprov-jdk14-136.jar
+++ /dev/null
@@ -1,2 +0,0 @@
-AnyObjectId[c1cf5b3c233fabfc9765d0cb641dcb7e903ec179] was removed in git history.
-Apache SVN contains full history.
\ No newline at end of file
diff --git a/contrib/extraction/lib/commons-codec-1.3.jar b/contrib/extraction/lib/commons-codec-1.3.jar
deleted file mode 100644
index 41a0921bf9f..00000000000
--- a/contrib/extraction/lib/commons-codec-1.3.jar
+++ /dev/null
@@ -1,2 +0,0 @@
-AnyObjectId[957b6752af9a60c1bb2a4f65db0e90e5ce00f521] was removed in git history.
-Apache SVN contains full history.
\ No newline at end of file
diff --git a/contrib/extraction/lib/commons-io-1.4.jar b/contrib/extraction/lib/commons-io-1.4.jar
deleted file mode 100644
index 6052eb538ab..00000000000
--- a/contrib/extraction/lib/commons-io-1.4.jar
+++ /dev/null
@@ -1,2 +0,0 @@
-AnyObjectId[133dc6cb35f5ca2c5920fd0933a557c2def88680] was removed in git history.
-Apache SVN contains full history.
\ No newline at end of file
diff --git a/contrib/extraction/lib/commons-lang-2.1.jar b/contrib/extraction/lib/commons-lang-2.1.jar
deleted file mode 100644
index c56ef0cba61..00000000000
--- a/contrib/extraction/lib/commons-lang-2.1.jar
+++ /dev/null
@@ -1,2 +0,0 @@
-AnyObjectId[87b80ab5db1729662ccf3439e147430a28c36d03] was removed in git history.
-Apache SVN contains full history.
\ No newline at end of file
diff --git a/contrib/extraction/lib/fontbox-0.1.0.jar b/contrib/extraction/lib/fontbox-0.1.0.jar
deleted file mode 100644
index 91ca9cca647..00000000000
--- a/contrib/extraction/lib/fontbox-0.1.0.jar
+++ /dev/null
@@ -1,2 +0,0 @@
-AnyObjectId[acb1d7b1d4e00241adba2188235bc85e51ab010c] was removed in git history.
-Apache SVN contains full history.
\ No newline at end of file
diff --git a/contrib/extraction/lib/fontbox-0.8.0-incubator.jar b/contrib/extraction/lib/fontbox-0.8.0-incubator.jar
new file mode 100644
index 00000000000..e0b3470ce1a
--- /dev/null
+++ b/contrib/extraction/lib/fontbox-0.8.0-incubator.jar
@@ -0,0 +1,2 @@
+AnyObjectId[91a496ac1164c08522c1e622fc39b8e991dd2d0b] was removed in git history.
+Apache SVN contains full history.
\ No newline at end of file
diff --git a/contrib/extraction/lib/geronimo-stax-api_1.0_spec-1.0.1.jar b/contrib/extraction/lib/geronimo-stax-api_1.0_spec-1.0.1.jar
new file mode 100644
index 00000000000..77da03b5bd6
--- /dev/null
+++ b/contrib/extraction/lib/geronimo-stax-api_1.0_spec-1.0.1.jar
@@ -0,0 +1,2 @@
+AnyObjectId[ab1ee3ba605df11b3075677c808d092845dad123] was removed in git history.
+Apache SVN contains full history.
\ No newline at end of file
diff --git a/contrib/extraction/lib/geronimo-stax-api_1.0_spec-1.0.jar b/contrib/extraction/lib/geronimo-stax-api_1.0_spec-1.0.jar
deleted file mode 100644
index a560a688df5..00000000000
--- a/contrib/extraction/lib/geronimo-stax-api_1.0_spec-1.0.jar
+++ /dev/null
@@ -1,2 +0,0 @@
-AnyObjectId[0d6d37423b2e2b53aba04bb09845233502619cb6] was removed in git history.
-Apache SVN contains full history.
\ No newline at end of file
diff --git a/contrib/extraction/lib/icu4j-3.8.jar b/contrib/extraction/lib/icu4j-3.8.jar
deleted file mode 100644
index 61523bbc913..00000000000
--- a/contrib/extraction/lib/icu4j-3.8.jar
+++ /dev/null
@@ -1,2 +0,0 @@
-AnyObjectId[26654862dc6a6ed6b9c8c4766660150d6dd0efd6] was removed in git history.
-Apache SVN contains full history.
\ No newline at end of file
diff --git a/contrib/extraction/lib/jempbox-0.2.0.jar b/contrib/extraction/lib/jempbox-0.2.0.jar
deleted file mode 100644
index 76a0bd2eaf0..00000000000
--- a/contrib/extraction/lib/jempbox-0.2.0.jar
+++ /dev/null
@@ -1,2 +0,0 @@
-AnyObjectId[953613cba49cc76b0562116af797c68ccf2a0a52] was removed in git history.
-Apache SVN contains full history.
\ No newline at end of file
diff --git a/contrib/extraction/lib/jempbox-0.8.0-incubator.jar b/contrib/extraction/lib/jempbox-0.8.0-incubator.jar
new file mode 100644
index 00000000000..f01f9529a03
--- /dev/null
+++ b/contrib/extraction/lib/jempbox-0.8.0-incubator.jar
@@ -0,0 +1,2 @@
+AnyObjectId[adcead7737700efc2c77f7f16a83ce0c0547381a] was removed in git history.
+Apache SVN contains full history.
\ No newline at end of file
diff --git a/contrib/extraction/lib/metadata-extractor-2.4.0-beta-1.jar b/contrib/extraction/lib/metadata-extractor-2.4.0-beta-1.jar
new file mode 100644
index 00000000000..a396b86ca0b
--- /dev/null
+++ b/contrib/extraction/lib/metadata-extractor-2.4.0-beta-1.jar
@@ -0,0 +1,2 @@
+AnyObjectId[3720d649dd56d96f9351435dea7c2c921a0be050] was removed in git history.
+Apache SVN contains full history.
\ No newline at end of file
diff --git a/contrib/extraction/lib/nekohtml-1.9.9.jar b/contrib/extraction/lib/nekohtml-1.9.9.jar
deleted file mode 100644
index 00195ba2b18..00000000000
--- a/contrib/extraction/lib/nekohtml-1.9.9.jar
+++ /dev/null
@@ -1,2 +0,0 @@
-AnyObjectId[2e0627173abfce31e41a5c6b9b06f0d6a911ae6a] was removed in git history.
-Apache SVN contains full history.
\ No newline at end of file
diff --git a/contrib/extraction/lib/pdfbox-0.7.3.jar b/contrib/extraction/lib/pdfbox-0.7.3.jar
deleted file mode 100644
index 88d36fee995..00000000000
--- a/contrib/extraction/lib/pdfbox-0.7.3.jar
+++ /dev/null
@@ -1,2 +0,0 @@
-AnyObjectId[9bfe670f69cbf240764b6afd74c166afc7d62256] was removed in git history.
-Apache SVN contains full history.
\ No newline at end of file
diff --git a/contrib/extraction/lib/pdfbox-0.8.0-incubating.jar b/contrib/extraction/lib/pdfbox-0.8.0-incubating.jar
new file mode 100644
index 00000000000..f0bffd53e8c
--- /dev/null
+++ b/contrib/extraction/lib/pdfbox-0.8.0-incubating.jar
@@ -0,0 +1,2 @@
+AnyObjectId[637324e4666d5fbf2fe29cb8151a790b1ccabcec] was removed in git history.
+Apache SVN contains full history.
\ No newline at end of file
diff --git a/contrib/extraction/lib/poi-3.5-FINAL.jar b/contrib/extraction/lib/poi-3.5-FINAL.jar
new file mode 100644
index 00000000000..928ae65dc1a
--- /dev/null
+++ b/contrib/extraction/lib/poi-3.5-FINAL.jar
@@ -0,0 +1,2 @@
+AnyObjectId[7c5c5343469ce7f60f5bff6db1f3c2de51933b70] was removed in git history.
+Apache SVN contains full history.
\ No newline at end of file
diff --git a/contrib/extraction/lib/poi-3.5-beta6.jar b/contrib/extraction/lib/poi-3.5-beta6.jar
deleted file mode 100644
index d847acd2d71..00000000000
--- a/contrib/extraction/lib/poi-3.5-beta6.jar
+++ /dev/null
@@ -1,2 +0,0 @@
-AnyObjectId[3281c6e43309fc39dda416d47f632e10f2367c1b] was removed in git history.
-Apache SVN contains full history.
\ No newline at end of file
diff --git a/contrib/extraction/lib/poi-ooxml-3.5-FINAL.jar b/contrib/extraction/lib/poi-ooxml-3.5-FINAL.jar
new file mode 100644
index 00000000000..d7d72b4045f
--- /dev/null
+++ b/contrib/extraction/lib/poi-ooxml-3.5-FINAL.jar
@@ -0,0 +1,2 @@
+AnyObjectId[2fe4979a8d8ee91a5b517a84c405fc1fcf1560cb] was removed in git history.
+Apache SVN contains full history.
\ No newline at end of file
diff --git a/contrib/extraction/lib/poi-ooxml-3.5-beta6.jar b/contrib/extraction/lib/poi-ooxml-3.5-beta6.jar
deleted file mode 100644
index 92354c8c8ea..00000000000
--- a/contrib/extraction/lib/poi-ooxml-3.5-beta6.jar
+++ /dev/null
@@ -1,2 +0,0 @@
-AnyObjectId[206d9920d36cbe51e1cd7c801294734ae1401505] was removed in git history.
-Apache SVN contains full history.
\ No newline at end of file
diff --git a/contrib/extraction/lib/poi-scratchpad-3.5-FINAL.jar b/contrib/extraction/lib/poi-scratchpad-3.5-FINAL.jar
new file mode 100644
index 00000000000..05896f003e9
--- /dev/null
+++ b/contrib/extraction/lib/poi-scratchpad-3.5-FINAL.jar
@@ -0,0 +1,2 @@
+AnyObjectId[b220b35818d54fa7c6efae42821754c54c358293] was removed in git history.
+Apache SVN contains full history.
\ No newline at end of file
diff --git a/contrib/extraction/lib/poi-scratchpad-3.5-beta6.jar b/contrib/extraction/lib/poi-scratchpad-3.5-beta6.jar
deleted file mode 100644
index cd9ed02aa04..00000000000
--- a/contrib/extraction/lib/poi-scratchpad-3.5-beta6.jar
+++ /dev/null
@@ -1,2 +0,0 @@
-AnyObjectId[c1a27b5fef037f0d81a115a8ef377b95356bbbed] was removed in git history.
-Apache SVN contains full history.
\ No newline at end of file
diff --git a/contrib/extraction/lib/tagsoup-1.2.jar b/contrib/extraction/lib/tagsoup-1.2.jar
new file mode 100644
index 00000000000..95267dcb1d0
--- /dev/null
+++ b/contrib/extraction/lib/tagsoup-1.2.jar
@@ -0,0 +1,2 @@
+AnyObjectId[af27803ec117e6ec643b8522e266481253b35fe3] was removed in git history.
+Apache SVN contains full history.
\ No newline at end of file
diff --git a/contrib/extraction/lib/tika-core-0.4.jar b/contrib/extraction/lib/tika-core-0.4.jar
deleted file mode 100644
index 6f8fb822caa..00000000000
--- a/contrib/extraction/lib/tika-core-0.4.jar
+++ /dev/null
@@ -1,2 +0,0 @@
-AnyObjectId[32b269484b5fa5f7e695e66346df089cfb740780] was removed in git history.
-Apache SVN contains full history.
\ No newline at end of file
diff --git a/contrib/extraction/lib/tika-core-0.5.jar b/contrib/extraction/lib/tika-core-0.5.jar
new file mode 100644
index 00000000000..cd5e66b2169
--- /dev/null
+++ b/contrib/extraction/lib/tika-core-0.5.jar
@@ -0,0 +1,2 @@
+AnyObjectId[1adcb48b593b05ebdd77e99d79e9600f24cac99c] was removed in git history.
+Apache SVN contains full history.
\ No newline at end of file
diff --git a/contrib/extraction/lib/tika-parsers-0.4.jar b/contrib/extraction/lib/tika-parsers-0.4.jar
deleted file mode 100644
index aae708c2528..00000000000
--- a/contrib/extraction/lib/tika-parsers-0.4.jar
+++ /dev/null
@@ -1,2 +0,0 @@
-AnyObjectId[8451a7cc91b2c3c25429446d0d3adf89af2e31c8] was removed in git history.
-Apache SVN contains full history.
\ No newline at end of file
diff --git a/contrib/extraction/lib/tika-parsers-0.5.jar b/contrib/extraction/lib/tika-parsers-0.5.jar
new file mode 100644
index 00000000000..0b7e57a13b8
--- /dev/null
+++ b/contrib/extraction/lib/tika-parsers-0.5.jar
@@ -0,0 +1,2 @@
+AnyObjectId[0928f3cdf4b2d077b1f1f874c2d15c796e9d73c4] was removed in git history.
+Apache SVN contains full history.
\ No newline at end of file
diff --git a/contrib/extraction/src/main/java/org/apache/solr/handler/extraction/ExtractingRequestHandler.java b/contrib/extraction/src/main/java/org/apache/solr/handler/extraction/ExtractingRequestHandler.java
index 0d870366d31..32cb7b3a13c 100644
--- a/contrib/extraction/src/main/java/org/apache/solr/handler/extraction/ExtractingRequestHandler.java
+++ b/contrib/extraction/src/main/java/org/apache/solr/handler/extraction/ExtractingRequestHandler.java
@@ -76,11 +76,7 @@ public class ExtractingRequestHandler extends ContentStreamHandlerBase implement
           throw new SolrException(ErrorCode.SERVER_ERROR, e);
         }
       } else {
-        try {
-          config = TikaConfig.getDefaultConfig();
-        } catch (TikaException e) {
-          throw new SolrException(ErrorCode.SERVER_ERROR, e);
-        }
+        config = TikaConfig.getDefaultConfig();
       }
       NamedList configDateFormats = (NamedList) initArgs.get(DATE_FORMATS);
       if (configDateFormats != null && configDateFormats.size() > 0) {
@@ -92,11 +88,7 @@ public class ExtractingRequestHandler extends ContentStreamHandlerBase implement
         }
       }
     } else {
-      try {
-        config = TikaConfig.getDefaultConfig();
-      } catch (TikaException e) {
-        throw new SolrException(ErrorCode.SERVER_ERROR, e);
-      }
+      config = TikaConfig.getDefaultConfig();
     }
     factory = createFactory();
   }