diff --git a/contrib/extraction/CHANGES.txt b/contrib/extraction/CHANGES.txt index 4b398795eb6..4b949e6623a 100644 --- a/contrib/extraction/CHANGES.txt +++ b/contrib/extraction/CHANGES.txt @@ -24,4 +24,6 @@ $Id:$ 1. SOLR-284: Added in support for extraction. (Eric Pugh, Chris Harris, gsingers) -2. SOLR-284: Removed "silent success" key generation (gsingers) \ No newline at end of file +2. SOLR-284: Removed "silent success" key generation (gsingers) + +3. SOLR-1075: Upgrade to Tika 0.3. See http://www.apache.org/dist/lucene/tika/CHANGES-0.3.txt (gsingers) \ 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 new file mode 100644 index 00000000000..41a0921bf9f --- /dev/null +++ b/contrib/extraction/lib/commons-codec-1.3.jar @@ -0,0 +1,2 @@ +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 new file mode 100644 index 00000000000..6052eb538ab --- /dev/null +++ b/contrib/extraction/lib/commons-io-1.4.jar @@ -0,0 +1,2 @@ +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 new file mode 100644 index 00000000000..c56ef0cba61 --- /dev/null +++ b/contrib/extraction/lib/commons-lang-2.1.jar @@ -0,0 +1,2 @@ +AnyObjectId[87b80ab5db1729662ccf3439e147430a28c36d03] was removed in git history. +Apache SVN contains full history. \ No newline at end of file diff --git a/contrib/extraction/lib/commons-lang-2.4.jar b/contrib/extraction/lib/commons-lang-2.4.jar deleted file mode 100644 index 63550c90560..00000000000 --- a/contrib/extraction/lib/commons-lang-2.4.jar +++ /dev/null @@ -1,2 +0,0 @@ -AnyObjectId[ce0ca22c8d29a9be736d775fe50bfdc6ce770186] was removed in git history. -Apache SVN contains full history. \ No newline at end of file diff --git a/contrib/extraction/lib/dom4j-1.6.1.jar b/contrib/extraction/lib/dom4j-1.6.1.jar new file mode 100644 index 00000000000..cf7601ea8ff --- /dev/null +++ b/contrib/extraction/lib/dom4j-1.6.1.jar @@ -0,0 +1,2 @@ +AnyObjectId[c8c4dbb92d6c23a7fbb2813eb721eb4cce91750c] was removed in git history. +Apache SVN contains full history. \ No newline at end of file diff --git a/contrib/extraction/lib/icu4j-3.4.4.jar b/contrib/extraction/lib/icu4j-3.4.4.jar deleted file mode 100644 index 9ca4579e0ad..00000000000 --- a/contrib/extraction/lib/icu4j-3.4.4.jar +++ /dev/null @@ -1,2 +0,0 @@ -AnyObjectId[f5e8c167e7f7f3d078407859cb50b8abf23c697e] 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 new file mode 100644 index 00000000000..61523bbc913 --- /dev/null +++ b/contrib/extraction/lib/icu4j-3.8.jar @@ -0,0 +1,2 @@ +AnyObjectId[26654862dc6a6ed6b9c8c4766660150d6dd0efd6] was removed in git history. +Apache SVN contains full history. \ No newline at end of file diff --git a/contrib/extraction/lib/nekohtml-1.9.7.jar b/contrib/extraction/lib/nekohtml-1.9.7.jar deleted file mode 100644 index 179e5e6f5a8..00000000000 --- a/contrib/extraction/lib/nekohtml-1.9.7.jar +++ /dev/null @@ -1,2 +0,0 @@ -AnyObjectId[037b4fe2743eb161eec649f6fa5fa4725585b518] 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 new file mode 100644 index 00000000000..00195ba2b18 --- /dev/null +++ b/contrib/extraction/lib/nekohtml-1.9.9.jar @@ -0,0 +1,2 @@ +AnyObjectId[2e0627173abfce31e41a5c6b9b06f0d6a911ae6a] was removed in git history. +Apache SVN contains full history. \ No newline at end of file diff --git a/contrib/extraction/lib/ooxml-schemas-1.0.jar b/contrib/extraction/lib/ooxml-schemas-1.0.jar new file mode 100644 index 00000000000..b7d8a76a2d6 --- /dev/null +++ b/contrib/extraction/lib/ooxml-schemas-1.0.jar @@ -0,0 +1,2 @@ +AnyObjectId[9283af18b3b82015044831f72f420d707db6d8d8] was removed in git history. +Apache SVN contains full history. \ No newline at end of file diff --git a/contrib/extraction/lib/poi-3.1-FINAL.jar b/contrib/extraction/lib/poi-3.1-FINAL.jar deleted file mode 100644 index c528a8c6306..00000000000 --- a/contrib/extraction/lib/poi-3.1-FINAL.jar +++ /dev/null @@ -1,2 +0,0 @@ -AnyObjectId[9e472a1610fa5d6736ecd56aec663623170003a3] 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-beta5.jar b/contrib/extraction/lib/poi-3.5-beta5.jar new file mode 100644 index 00000000000..ca9c3ec8b93 --- /dev/null +++ b/contrib/extraction/lib/poi-3.5-beta5.jar @@ -0,0 +1,2 @@ +AnyObjectId[0e3279b961c07d4d4ed909dade956aacfbf1e785] 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-beta5.jar b/contrib/extraction/lib/poi-ooxml-3.5-beta5.jar new file mode 100644 index 00000000000..9a6071c458e --- /dev/null +++ b/contrib/extraction/lib/poi-ooxml-3.5-beta5.jar @@ -0,0 +1,2 @@ +AnyObjectId[74bb8d3fa6185ca3619504f9efc5404534137a7a] was removed in git history. +Apache SVN contains full history. \ No newline at end of file diff --git a/contrib/extraction/lib/poi-scratchpad-3.1-FINAL.jar b/contrib/extraction/lib/poi-scratchpad-3.1-FINAL.jar deleted file mode 100644 index 35cd05c41e1..00000000000 --- a/contrib/extraction/lib/poi-scratchpad-3.1-FINAL.jar +++ /dev/null @@ -1,2 +0,0 @@ -AnyObjectId[58a33ac11683bec703fadffdbb263036146d7a74] 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-beta5.jar b/contrib/extraction/lib/poi-scratchpad-3.5-beta5.jar new file mode 100644 index 00000000000..b25ec0fa647 --- /dev/null +++ b/contrib/extraction/lib/poi-scratchpad-3.5-beta5.jar @@ -0,0 +1,2 @@ +AnyObjectId[6877c97853cafab42e5c0a88f3b39b6cd31165ca] was removed in git history. +Apache SVN contains full history. \ No newline at end of file diff --git a/contrib/extraction/lib/tika-0.2.jar b/contrib/extraction/lib/tika-0.2.jar deleted file mode 100644 index 7a5227130db..00000000000 --- a/contrib/extraction/lib/tika-0.2.jar +++ /dev/null @@ -1,2 +0,0 @@ -AnyObjectId[65882f20fd59a46c577fbdfd3ddb63f4d49cb71c] was removed in git history. -Apache SVN contains full history. \ No newline at end of file diff --git a/contrib/extraction/lib/tika-0.3.jar b/contrib/extraction/lib/tika-0.3.jar new file mode 100644 index 00000000000..b668cdddf59 --- /dev/null +++ b/contrib/extraction/lib/tika-0.3.jar @@ -0,0 +1,2 @@ +AnyObjectId[4dd90139eda215b4489ae99c1eb6cb97b953c884] was removed in git history. +Apache SVN contains full history. \ No newline at end of file diff --git a/contrib/extraction/lib/xml-apis-1.0.b2.jar b/contrib/extraction/lib/xml-apis-1.0.b2.jar new file mode 100644 index 00000000000..65531da804a --- /dev/null +++ b/contrib/extraction/lib/xml-apis-1.0.b2.jar @@ -0,0 +1,2 @@ +AnyObjectId[ad33a5afa6ddae02f3ed0b42b1c7fcbf22a7d2ab] was removed in git history. +Apache SVN contains full history. \ No newline at end of file diff --git a/contrib/extraction/lib/xml-apis-1.3.03.jar b/contrib/extraction/lib/xml-apis-1.3.03.jar deleted file mode 100644 index 11897087c76..00000000000 --- a/contrib/extraction/lib/xml-apis-1.3.03.jar +++ /dev/null @@ -1,2 +0,0 @@ -AnyObjectId[b338fb66932a763d6939dc93f27ed985ca5d1ebb] was removed in git history. -Apache SVN contains full history. \ No newline at end of file diff --git a/contrib/extraction/lib/xmlbeans-2.3.0.jar b/contrib/extraction/lib/xmlbeans-2.3.0.jar new file mode 100644 index 00000000000..36e6d816b99 --- /dev/null +++ b/contrib/extraction/lib/xmlbeans-2.3.0.jar @@ -0,0 +1,2 @@ +AnyObjectId[ccd8163421ba8d0361315fb947f2432f1e6d7a83] was removed in git history. +Apache SVN contains full history. \ No newline at end of file diff --git a/contrib/extraction/src/test/java/org/apache/solr/handler/ExtractingRequestHandlerTest.java b/contrib/extraction/src/test/java/org/apache/solr/handler/ExtractingRequestHandlerTest.java index d8e74a628b3..ba2e17e5b25 100644 --- a/contrib/extraction/src/test/java/org/apache/solr/handler/ExtractingRequestHandlerTest.java +++ b/contrib/extraction/src/test/java/org/apache/solr/handler/ExtractingRequestHandlerTest.java @@ -62,6 +62,7 @@ public class ExtractingRequestHandlerTest extends AbstractSolrTestCase { "ext.map.Author", "extractedAuthor", "ext.literal.id", "three", "ext.def.fl", "extractedContent", + "ext.map.language", "extractedLanguage", "ext.map.Last-Modified", "extractedDate" ); assertQ(req("stream_name:version_control.xml"), "//*[@numFound='0']"); @@ -81,6 +82,7 @@ public class ExtractingRequestHandlerTest extends AbstractSolrTestCase { "ext.map.Author", "extractedAuthor", "ext.def.fl", "extractedContent", "ext.literal.id", "one", + "ext.map.language", "extractedLanguage", "ext.literal.extractionLiteralMV", "one", "ext.literal.extractionLiteralMV", "two", "ext.map.Last-Modified", "extractedDate" @@ -99,6 +101,7 @@ public class ExtractingRequestHandlerTest extends AbstractSolrTestCase { "ext.map.Author", "extractedAuthor", "ext.def.fl", "extractedContent", "ext.literal.id", "two", + "ext.map.language", "extractedLanguage", "ext.literal.extractionLiteral", "one", "ext.literal.extractionLiteral", "two", "ext.map.Last-Modified", "extractedDate" @@ -113,6 +116,7 @@ public class ExtractingRequestHandlerTest extends AbstractSolrTestCase { "ext.map.Author", "extractedAuthor", "ext.def.fl", "extractedContent", "ext.literal.id", "three", + "ext.map.language", "extractedLanguage", "ext.literal.extractionLiteral", "one", "ext.map.Last-Modified", "extractedDate" );