You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@tika.apache.org by ni...@apache.org on 2012/05/17 22:05:23 UTC
svn commit: r1339833 [2/2] - in /tika/trunk/tika-parsers/src:
main/java/org/apache/tika/parser/asm/ main/java/org/apache/tika/parser/feed/
main/java/org/apache/tika/parser/font/ main/java/org/apache/tika/parser/html/
main/java/org/apache/tika/parser/im...
Modified: tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/iwork/IWorkParserTest.java
URL: http://svn.apache.org/viewvc/tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/iwork/IWorkParserTest.java?rev=1339833&r1=1339832&r2=1339833&view=diff
==============================================================================
--- tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/iwork/IWorkParserTest.java (original)
+++ tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/iwork/IWorkParserTest.java Thu May 17 20:05:20 2012
@@ -17,10 +17,13 @@
package org.apache.tika.parser.iwork;
import java.io.InputStream;
+import java.util.Arrays;
+import java.util.List;
import junit.framework.TestCase;
import org.apache.tika.metadata.Metadata;
+import org.apache.tika.metadata.TikaCoreProperties;
import org.apache.tika.parser.AutoDetectParser;
import org.apache.tika.parser.ParseContext;
import org.apache.tika.parser.Parser;
@@ -48,13 +51,22 @@ public class IWorkParserTest extends Tes
ContentHandler handler = new BodyContentHandler();
iWorkParser.parse(input, handler, metadata, parseContext);
- assertEquals(6, metadata.size());
+ // Make sure enough keys came through
+ // (Exact numbers will vary based on composites)
+ assertTrue("Insufficient metadata found " + metadata.size(), metadata.size() >= 6);
+ List<String> metadataKeys = Arrays.asList(metadata.names());
+ assertTrue("Metadata not found in " + metadataKeys, metadataKeys.contains(Metadata.CONTENT_TYPE));
+ assertTrue("Metadata not found in " + metadataKeys, metadataKeys.contains(Metadata.SLIDE_COUNT.getName()));
+ assertTrue("Metadata not found in " + metadataKeys, metadataKeys.contains(Metadata.AUTHOR));
+ assertTrue("Metadata not found in " + metadataKeys, metadataKeys.contains(Metadata.TITLE));
+
+ // Check the metadata values
assertEquals("application/vnd.apple.keynote", metadata.get(Metadata.CONTENT_TYPE));
assertEquals("3", metadata.get(Metadata.SLIDE_COUNT));
assertEquals("1024", metadata.get(KeynoteContentHandler.PRESENTATION_WIDTH));
assertEquals("768", metadata.get(KeynoteContentHandler.PRESENTATION_HEIGHT));
assertEquals("Tika user", metadata.get(Metadata.AUTHOR));
- assertEquals("Apache tika", metadata.get(Metadata.TITLE));
+ assertEquals("Apache tika", metadata.get(TikaCoreProperties.TITLE));
String content = handler.toString();
assertTrue(content.contains("A sample presentation"));
@@ -82,13 +94,24 @@ public class IWorkParserTest extends Tes
ContentHandler handler = new BodyContentHandler();
iWorkParser.parse(input, handler, metadata, parseContext);
- assertEquals(51, metadata.size());
+ // Make sure enough keys came through
+ // (Exact numbers will vary based on composites)
+ assertTrue("Insufficient metadata found " + metadata.size(), metadata.size() >= 50);
+ List<String> metadataKeys = Arrays.asList(metadata.names());
+ assertTrue("Metadata not found in " + metadataKeys, metadataKeys.contains(Metadata.CONTENT_TYPE));
+ assertTrue("Metadata not found in " + metadataKeys, metadataKeys.contains(Metadata.PAGE_COUNT.getName()));
+ assertTrue("Metadata not found in " + metadataKeys, metadataKeys.contains(Metadata.AUTHOR));
+ assertTrue("Metadata not found in " + metadataKeys, metadataKeys.contains(Metadata.TITLE));
+ assertTrue("Metadata not found in " + metadataKeys, metadataKeys.contains(Metadata.LAST_MODIFIED.getName()));
+ assertTrue("Metadata not found in " + metadataKeys, metadataKeys.contains(Metadata.LANGUAGE));
+
+ // Check the metadata values
assertEquals("application/vnd.apple.pages", metadata.get(Metadata.CONTENT_TYPE));
assertEquals("Tika user", metadata.get(Metadata.AUTHOR));
- assertEquals("Apache tika", metadata.get(Metadata.TITLE));
+ assertEquals("Apache tika", metadata.get(TikaCoreProperties.TITLE));
assertEquals("2010-05-09T21:34:38+0200", metadata.get(Metadata.CREATION_DATE));
assertEquals("2010-05-09T23:50:36+0200", metadata.get(Metadata.LAST_MODIFIED));
- assertEquals("en", metadata.get(Metadata.LANGUAGE));
+ assertEquals("en", metadata.get(TikaCoreProperties.LANGUAGE));
assertEquals("2", metadata.get(Metadata.PAGE_COUNT));
String content = handler.toString();
@@ -119,13 +142,24 @@ public class IWorkParserTest extends Tes
iWorkParser.parse(input, handler, metadata, parseContext);
- String content = handler.toString();
- assertEquals(9, metadata.size());
+ // Make sure enough keys came through
+ // (Exact numbers will vary based on composites)
+ assertTrue("Insufficient metadata found " + metadata.size(), metadata.size() >= 8);
+ List<String> metadataKeys = Arrays.asList(metadata.names());
+ assertTrue("Metadata not found in " + metadataKeys, metadataKeys.contains(Metadata.CONTENT_TYPE));
+ assertTrue("Metadata not found in " + metadataKeys, metadataKeys.contains(Metadata.PAGE_COUNT.getName()));
+ assertTrue("Metadata not found in " + metadataKeys, metadataKeys.contains(Metadata.AUTHOR));
+ assertTrue("Metadata not found in " + metadataKeys, metadataKeys.contains(Metadata.COMMENT));
+ assertTrue("Metadata not found in " + metadataKeys, metadataKeys.contains(Metadata.TITLE));
+ assertTrue("Metadata not found in " + metadataKeys, metadataKeys.contains(TikaCoreProperties.TITLE.getName()));
+
+ // Check the metadata values
assertEquals("2", metadata.get(Metadata.PAGE_COUNT));
assertEquals("Tika User", metadata.get(Metadata.AUTHOR));
- assertEquals("Account checking", metadata.get(Metadata.TITLE));
+ assertEquals("Account checking", metadata.get(TikaCoreProperties.TITLE));
assertEquals("a comment", metadata.get(Metadata.COMMENT));
+ String content = handler.toString();
assertTrue(content.contains("Category"));
assertTrue(content.contains("Home"));
assertTrue(content.contains("-226"));
Modified: tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/jpeg/JpegParserTest.java
URL: http://svn.apache.org/viewvc/tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/jpeg/JpegParserTest.java?rev=1339833&r1=1339832&r2=1339833&view=diff
==============================================================================
--- tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/jpeg/JpegParserTest.java (original)
+++ tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/jpeg/JpegParserTest.java Thu May 17 20:05:20 2012
@@ -24,6 +24,7 @@ import junit.framework.TestCase;
import org.apache.tika.metadata.Metadata;
import org.apache.tika.metadata.TIFF;
+import org.apache.tika.metadata.TikaCoreProperties;
import org.apache.tika.parser.ParseContext;
import org.apache.tika.parser.Parser;
import org.xml.sax.helpers.DefaultHandler;
@@ -65,8 +66,8 @@ public class JpegParserTest extends Test
// Common tags
//assertEquals("2009-10-02T23:02:49", metadata.get(Metadata.LAST_MODIFIED));
assertEquals("Date/Time Original for when the photo was taken, unspecified time zone",
- "2009-08-11T09:09:45", metadata.get(Metadata.DATE));
- List<String> keywords = Arrays.asList(metadata.getValues(Metadata.SUBJECT));
+ "2009-08-11T09:09:45", metadata.get(TikaCoreProperties.DATE));
+ List<String> keywords = Arrays.asList(metadata.getValues(TikaCoreProperties.SUBJECT));
assertTrue("'canon-55-250' expected in " + keywords, keywords.contains("canon-55-250"));
assertTrue("'moscow-birds' expected in " + keywords, keywords.contains("moscow-birds"));
assertTrue("'serbor' expected in " + keywords, keywords.contains("serbor"));
@@ -108,7 +109,7 @@ public class JpegParserTest extends Test
// Common tags
assertEquals("Date/Time Original for when the photo was taken, unspecified time zone",
- "2009-08-11T09:09:45", metadata.get(Metadata.DATE));
+ "2009-08-11T09:09:45", metadata.get(TikaCoreProperties.DATE));
assertEquals("This image has different Date/Time than Date/Time Original, so it is probably modification date",
"2009-10-02T23:02:49", metadata.get(Metadata.LAST_MODIFIED));
assertEquals("Date/Time Original should be stored in EXIF field too",
@@ -141,17 +142,17 @@ public class JpegParserTest extends Test
parser.parse(stream, new DefaultHandler(), metadata, new ParseContext());
// embedded comments with non-ascii characters
- assertEquals("Tosteberga \u00C4ngar", metadata.get(Metadata.TITLE));
- assertEquals("Bird site in north eastern Sk\u00E5ne, Sweden.\n(new line)", metadata.get(Metadata.DESCRIPTION));
+ assertEquals("Tosteberga \u00C4ngar", metadata.get(TikaCoreProperties.TITLE));
+ assertEquals("Bird site in north eastern Sk\u00E5ne, Sweden.\n(new line)", metadata.get(TikaCoreProperties.DESCRIPTION));
assertEquals("Some Tourist", metadata.get(Metadata.AUTHOR));
- assertEquals("Some Tourist", metadata.get(Metadata.CREATOR)); // Dublin Core
+ assertEquals("Some Tourist", metadata.get(TikaCoreProperties.CREATOR)); // Dublin Core
// xmp handles spaces in keywords, returns "bird watching, nature reserve, coast, grazelands"
// but we have to replace them with underscore
List<String> keywords = Arrays.asList(metadata.getValues(Metadata.KEYWORDS));
assertTrue(keywords.contains("coast"));
assertTrue(keywords.contains("bird watching"));
- assertEquals(keywords, Arrays.asList(metadata.getValues(Metadata.SUBJECT)));
+ assertEquals(keywords, Arrays.asList(metadata.getValues(TikaCoreProperties.SUBJECT)));
// Core EXIF/TIFF tags
assertEquals("103", metadata.get(Metadata.IMAGE_WIDTH));
@@ -180,10 +181,10 @@ public class JpegParserTest extends Test
parser.parse(stream, new DefaultHandler(), metadata, new ParseContext());
// embedded comments with non-ascii characters
- assertEquals("Tosteberga \u00C4ngar", metadata.get(Metadata.TITLE));
- assertEquals("Bird site in north eastern Sk\u00E5ne, Sweden.\n(new line)", metadata.get(Metadata.DESCRIPTION));
- assertEquals("Some Tourist", metadata.get(Metadata.CREATOR));
- List<String> subject = Arrays.asList(metadata.getValues(Metadata.SUBJECT));
+ assertEquals("Tosteberga \u00C4ngar", metadata.get(TikaCoreProperties.TITLE));
+ assertEquals("Bird site in north eastern Sk\u00E5ne, Sweden.\n(new line)", metadata.get(TikaCoreProperties.DESCRIPTION));
+ assertEquals("Some Tourist", metadata.get(TikaCoreProperties.CREATOR));
+ List<String> subject = Arrays.asList(metadata.getValues(TikaCoreProperties.SUBJECT));
assertTrue("got " + subject, subject.contains("bird watching"));
}
@@ -196,10 +197,10 @@ public class JpegParserTest extends Test
// XnViewMp's default comment dialog has only comment, not headline.
// Comment is embedded only if "Write comments in XMP" is enabled in settings
- assertEquals("Bird site in north eastern Sk\u00E5ne, Sweden.\n(new line)", metadata.get(Metadata.DESCRIPTION));
+ assertEquals("Bird site in north eastern Sk\u00E5ne, Sweden.\n(new line)", metadata.get(TikaCoreProperties.DESCRIPTION));
// xmp handles spaces in keywords, returns "bird watching, nature reserve, coast, grazelands"
// but we have to replace them with underscore
- String[] subject = metadata.getValues(Metadata.SUBJECT);
+ String[] subject = metadata.getValues(TikaCoreProperties.SUBJECT);
List<String> keywords = Arrays.asList(subject);
assertTrue("'coast'" + " not in " + keywords, keywords.contains("coast"));
assertTrue("'nature reserve'" + " not in " + keywords, keywords.contains("nature reserve"));
@@ -212,8 +213,8 @@ public class JpegParserTest extends Test
getClass().getResourceAsStream("/test-documents/testJPEG_oddTagComponent.jpg");
parser.parse(stream, new DefaultHandler(), metadata, new ParseContext());
- assertEquals(null, metadata.get(Metadata.TITLE));
- assertEquals(null, metadata.get(Metadata.DESCRIPTION));
+ assertEquals(null, metadata.get(TikaCoreProperties.TITLE));
+ assertEquals(null, metadata.get(TikaCoreProperties.DESCRIPTION));
assertEquals("251", metadata.get(Metadata.IMAGE_WIDTH));
assertEquals("384", metadata.get(Metadata.IMAGE_LENGTH));
}
Modified: tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/mail/RFC822ParserTest.java
URL: http://svn.apache.org/viewvc/tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/mail/RFC822ParserTest.java?rev=1339833&r1=1339832&r2=1339833&view=diff
==============================================================================
--- tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/mail/RFC822ParserTest.java (original)
+++ tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/mail/RFC822ParserTest.java Thu May 17 20:05:20 2012
@@ -31,6 +31,7 @@ import junit.framework.TestCase;
import org.apache.james.mime4j.stream.MimeConfig;
import org.apache.tika.exception.TikaException;
import org.apache.tika.metadata.Metadata;
+import org.apache.tika.metadata.TikaCoreProperties;
import org.apache.tika.parser.ParseContext;
import org.apache.tika.parser.Parser;
import org.apache.tika.sax.BodyContentHandler;
@@ -59,7 +60,7 @@ public class RFC822ParserTest extends Te
verify(handler).endDocument();
//note no leading spaces, and no quotes
assertEquals("Julien Nioche (JIRA) <ji...@apache.org>", metadata.get(Metadata.AUTHOR));
- assertEquals("[jira] Commented: (TIKA-461) RFC822 messages not parsed", metadata.get(Metadata.SUBJECT));
+ assertEquals("[jira] Commented: (TIKA-461) RFC822 messages not parsed", metadata.get(TikaCoreProperties.SUBJECT));
} catch (Exception e) {
fail("Exception thrown: " + e.getMessage());
}
@@ -147,7 +148,7 @@ public class RFC822ParserTest extends Te
//tests correct decoding of internationalized headers, both
//quoted-printable (Q) and Base64 (B).
assertEquals("Keld J\u00F8rn Simonsen <ke...@dkuug.dk>", metadata.get(Metadata.AUTHOR));
- assertEquals("If you can read this you understand the example.", metadata.get(Metadata.SUBJECT));
+ assertEquals("If you can read this you understand the example.", metadata.get(TikaCoreProperties.SUBJECT));
} catch (Exception e) {
fail("Exception thrown: " + e.getMessage());
}
@@ -165,7 +166,7 @@ public class RFC822ParserTest extends Te
parser.parse(stream, handler, metadata, new ParseContext());
assertEquals("Saved by Windows Internet Explorer 7", metadata.get(Metadata.AUTHOR));
- assertEquals("Air Permit Programs | Air & Radiation | US EPA", metadata.get(Metadata.SUBJECT));
+ assertEquals("Air Permit Programs | Air & Radiation | US EPA", metadata.get(TikaCoreProperties.SUBJECT));
}
/**
@@ -220,7 +221,7 @@ public class RFC822ParserTest extends Te
assertEquals(true, metadata.isMultiValued(Metadata.MESSAGE_TO));
assertEquals("abc", metadata.getValues(Metadata.MESSAGE_TO)[0]);
assertEquals("def", metadata.getValues(Metadata.MESSAGE_TO)[1]);
- assertEquals("abcd", metadata.get(Metadata.SUBJECT));
+ assertEquals("abcd", metadata.get(TikaCoreProperties.SUBJECT));
assertTrue(handler.toString().contains("bar biz bat"));
}
Modified: tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/mbox/MboxParserTest.java
URL: http://svn.apache.org/viewvc/tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/mbox/MboxParserTest.java?rev=1339833&r1=1339832&r2=1339833&view=diff
==============================================================================
--- tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/mbox/MboxParserTest.java (original)
+++ tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/mbox/MboxParserTest.java Thu May 17 20:05:20 2012
@@ -27,6 +27,7 @@ import java.io.InputStream;
import junit.framework.TestCase;
import org.apache.tika.metadata.Metadata;
+import org.apache.tika.metadata.TikaCoreProperties;
import org.apache.tika.parser.ParseContext;
import org.apache.tika.parser.Parser;
import org.apache.tika.sax.XHTMLContentHandler;
@@ -69,14 +70,14 @@ public class MboxParserTest extends Test
verify(handler).characters(new String("Test content").toCharArray(), 0, 12);
verify(handler).endDocument();
- assertEquals("subject", metadata.get(Metadata.TITLE));
- assertEquals("subject", metadata.get(Metadata.SUBJECT));
+ assertEquals("subject", metadata.get(TikaCoreProperties.TITLE));
+ assertEquals("subject", metadata.get(TikaCoreProperties.SUBJECT));
assertEquals("<au...@domain.com>", metadata.get(Metadata.AUTHOR));
- assertEquals("<au...@domain.com>", metadata.get(Metadata.CREATOR));
+ assertEquals("<au...@domain.com>", metadata.get(TikaCoreProperties.CREATOR));
assertEquals(null, metadata.get(Metadata.MESSAGE_RECIPIENT_ADDRESS));
assertEquals("<na...@domain.com>", metadata.get("MboxParser-return-path"));
assertEquals("Should be ISO date in UTC, converted from 'Tue, 9 Jun 2009 23:58:45 -0400'",
- "2009-06-10T03:58:45Z", metadata.get(Metadata.DATE));
+ "2009-06-10T03:58:45Z", metadata.get(TikaCoreProperties.DATE));
} catch (Exception e) {
fail("Exception thrown: " + e.getMessage());
}
@@ -133,10 +134,10 @@ public class MboxParserTest extends Test
try {
parser.parse(stream, handler, metadata, new ParseContext());
- assertEquals("Re: question about when shuffle/sort start working", metadata.get(Metadata.TITLE));
- assertEquals("Re: question about when shuffle/sort start working", metadata.get(Metadata.SUBJECT));
+ assertEquals("Re: question about when shuffle/sort start working", metadata.get(TikaCoreProperties.TITLE));
+ assertEquals("Re: question about when shuffle/sort start working", metadata.get(TikaCoreProperties.SUBJECT));
assertEquals("Jothi Padmanabhan <jo...@yahoo-inc.com>", metadata.get(Metadata.AUTHOR));
- assertEquals("Jothi Padmanabhan <jo...@yahoo-inc.com>", metadata.get(Metadata.CREATOR));
+ assertEquals("Jothi Padmanabhan <jo...@yahoo-inc.com>", metadata.get(TikaCoreProperties.CREATOR));
assertEquals("core-user@hadoop.apache.org", metadata.get(Metadata.MESSAGE_RECIPIENT_ADDRESS));
verify(handler).startDocument();
Modified: tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/microsoft/ExcelParserTest.java
URL: http://svn.apache.org/viewvc/tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/microsoft/ExcelParserTest.java?rev=1339833&r1=1339832&r2=1339833&view=diff
==============================================================================
--- tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/microsoft/ExcelParserTest.java (original)
+++ tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/microsoft/ExcelParserTest.java Thu May 17 20:05:20 2012
@@ -24,6 +24,7 @@ import junit.framework.TestCase;
import org.apache.tika.detect.DefaultDetector;
import org.apache.tika.detect.Detector;
import org.apache.tika.metadata.Metadata;
+import org.apache.tika.metadata.TikaCoreProperties;
import org.apache.tika.mime.MediaType;
import org.apache.tika.parser.AutoDetectParser;
import org.apache.tika.parser.ParseContext;
@@ -45,7 +46,7 @@ public class ExcelParserTest extends Tes
assertEquals(
"application/vnd.ms-excel",
metadata.get(Metadata.CONTENT_TYPE));
- assertEquals("Simple Excel document", metadata.get(Metadata.TITLE));
+ assertEquals("Simple Excel document", metadata.get(TikaCoreProperties.TITLE));
assertEquals("Keith Bennett", metadata.get(Metadata.AUTHOR));
// Mon Oct 01 17:13:56 BST 2007
Modified: tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/microsoft/OutlookParserTest.java
URL: http://svn.apache.org/viewvc/tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/microsoft/OutlookParserTest.java?rev=1339833&r1=1339832&r2=1339833&view=diff
==============================================================================
--- tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/microsoft/OutlookParserTest.java (original)
+++ tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/microsoft/OutlookParserTest.java Thu May 17 20:05:20 2012
@@ -29,6 +29,7 @@ import javax.xml.transform.stream.Stream
import junit.framework.TestCase;
import org.apache.tika.metadata.Metadata;
+import org.apache.tika.metadata.TikaCoreProperties;
import org.apache.tika.parser.AutoDetectParser;
import org.apache.tika.parser.ParseContext;
import org.apache.tika.parser.Parser;
@@ -58,7 +59,7 @@ public class OutlookParserTest extends T
metadata.get(Metadata.CONTENT_TYPE));
assertEquals(
"Microsoft Outlook Express 6",
- metadata.get(Metadata.TITLE));
+ metadata.get(TikaCoreProperties.TITLE));
assertEquals(
"Nouvel utilisateur de Outlook Express",
metadata.get(Metadata.MESSAGE_RECIPIENT_ADDRESS));
@@ -69,7 +70,7 @@ public class OutlookParserTest extends T
// Stored as Thu, 5 Apr 2007 09:26:06 -0700
assertEquals(
"2007-04-05T16:26:06Z",
- metadata.get(Metadata.DATE));
+ metadata.get(TikaCoreProperties.DATE));
String content = handler.toString();
assertTrue(content.contains(""));
@@ -131,7 +132,7 @@ public class OutlookParserTest extends T
metadata.get(Metadata.CONTENT_TYPE));
assertEquals(
"Welcome to Microsoft Office Outlook 2003",
- metadata.get(Metadata.TITLE));
+ metadata.get(TikaCoreProperties.TITLE));
String content = handler.toString();
assertTrue(content.contains("Outlook 2003"));
Modified: tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/microsoft/PowerPointParserTest.java
URL: http://svn.apache.org/viewvc/tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/microsoft/PowerPointParserTest.java?rev=1339833&r1=1339832&r2=1339833&view=diff
==============================================================================
--- tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/microsoft/PowerPointParserTest.java (original)
+++ tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/microsoft/PowerPointParserTest.java Thu May 17 20:05:20 2012
@@ -21,6 +21,7 @@ import java.util.Locale;
import org.apache.tika.TikaTest;
import org.apache.tika.metadata.Metadata;
+import org.apache.tika.metadata.TikaCoreProperties;
import org.apache.tika.parser.ParseContext;
import org.apache.tika.sax.BodyContentHandler;
import org.xml.sax.ContentHandler;
@@ -38,7 +39,7 @@ public class PowerPointParserTest extend
assertEquals(
"application/vnd.ms-powerpoint",
metadata.get(Metadata.CONTENT_TYPE));
- assertEquals("Sample Powerpoint Slide", metadata.get(Metadata.TITLE));
+ assertEquals("Sample Powerpoint Slide", metadata.get(TikaCoreProperties.TITLE));
assertEquals("Keith Bennett", metadata.get(Metadata.AUTHOR));
String content = handler.toString();
assertTrue(content.contains("Sample Powerpoint Slide"));
@@ -104,7 +105,7 @@ public class PowerPointParserTest extend
assertContains("Subject is here", content);
assertEquals("Subject is here",
- metadata.get(Metadata.SUBJECT));
+ metadata.get(TikaCoreProperties.SUBJECT));
assertContains("Suddenly some Japanese text:", content);
// Special version of (GHQ)
@@ -203,7 +204,7 @@ public class PowerPointParserTest extend
assertEquals("2011-08-22T13:30:53Z", metadata.get(Metadata.CREATION_DATE));
assertEquals("1", metadata.get(Metadata.SLIDE_COUNT));
assertEquals("3", metadata.get(Metadata.WORD_COUNT));
- assertEquals("Test extraction properties pptx", metadata.get(Metadata.TITLE));
+ assertEquals("Test extraction properties pptx", metadata.get(TikaCoreProperties.TITLE));
assertEquals("true", metadata.get("custom:myCustomBoolean"));
assertEquals("3", metadata.get("custom:myCustomNumber"));
assertEquals("MyStringValue", metadata.get("custom:MyCustomString"));
Modified: tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/microsoft/ProjectParserTest.java
URL: http://svn.apache.org/viewvc/tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/microsoft/ProjectParserTest.java?rev=1339833&r1=1339832&r2=1339833&view=diff
==============================================================================
--- tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/microsoft/ProjectParserTest.java (original)
+++ tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/microsoft/ProjectParserTest.java Thu May 17 20:05:20 2012
@@ -19,6 +19,7 @@ package org.apache.tika.parser.microsoft
import java.io.InputStream;
import org.apache.tika.metadata.Metadata;
+import org.apache.tika.metadata.TikaCoreProperties;
import org.apache.tika.parser.ParseContext;
import org.apache.tika.sax.BodyContentHandler;
import org.xml.sax.ContentHandler;
@@ -61,8 +62,8 @@ public class ProjectParserTest extends T
"application/vnd.ms-project",
metadata.get(Metadata.CONTENT_TYPE));
- assertEquals("The quick brown fox jumps over the lazy dog", metadata.get(Metadata.TITLE));
- assertEquals("Gym class featuring a brown fox and lazy dog", metadata.get(Metadata.SUBJECT));
+ assertEquals("The quick brown fox jumps over the lazy dog", metadata.get(TikaCoreProperties.TITLE));
+ assertEquals("Gym class featuring a brown fox and lazy dog", metadata.get(TikaCoreProperties.SUBJECT));
assertEquals("Nevin Nollop", metadata.get(Metadata.AUTHOR));
assertEquals("", metadata.get(Metadata.LAST_AUTHOR));
assertEquals("Pangram, fox, dog", metadata.get(Metadata.KEYWORDS));
Modified: tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/microsoft/PublisherParserTest.java
URL: http://svn.apache.org/viewvc/tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/microsoft/PublisherParserTest.java?rev=1339833&r1=1339832&r2=1339833&view=diff
==============================================================================
--- tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/microsoft/PublisherParserTest.java (original)
+++ tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/microsoft/PublisherParserTest.java Thu May 17 20:05:20 2012
@@ -19,6 +19,7 @@ package org.apache.tika.parser.microsoft
import java.io.InputStream;
import org.apache.tika.metadata.Metadata;
+import org.apache.tika.metadata.TikaCoreProperties;
import org.apache.tika.parser.ParseContext;
import org.apache.tika.sax.BodyContentHandler;
import org.xml.sax.ContentHandler;
@@ -38,7 +39,7 @@ public class PublisherParserTest extends
assertEquals(
"application/x-mspublisher",
metadata.get(Metadata.CONTENT_TYPE));
- assertEquals(null, metadata.get(Metadata.TITLE));
+ assertEquals(null, metadata.get(TikaCoreProperties.TITLE));
assertEquals("Nick Burch", metadata.get(Metadata.AUTHOR));
String content = handler.toString();
assertTrue(content.contains("0123456789"));
Modified: tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/microsoft/TNEFParserTest.java
URL: http://svn.apache.org/viewvc/tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/microsoft/TNEFParserTest.java?rev=1339833&r1=1339832&r2=1339833&view=diff
==============================================================================
--- tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/microsoft/TNEFParserTest.java (original)
+++ tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/microsoft/TNEFParserTest.java Thu May 17 20:05:20 2012
@@ -22,6 +22,7 @@ import org.apache.tika.extractor.Contain
import org.apache.tika.extractor.ParserContainerExtractor;
import org.apache.tika.io.TikaInputStream;
import org.apache.tika.metadata.Metadata;
+import org.apache.tika.metadata.TikaCoreProperties;
import org.apache.tika.mime.MediaType;
import org.apache.tika.parser.ParseContext;
import org.apache.tika.sax.BodyContentHandler;
@@ -54,7 +55,7 @@ public class TNEFParserTest extends Abst
TNEFParser tnef = new TNEFParser();
tnef.parse(stream, handler, metadata, new ParseContext());
- assertEquals("This is a test message", metadata.get(Metadata.SUBJECT));
+ assertEquals("This is a test message", metadata.get(TikaCoreProperties.SUBJECT));
}
/**
Modified: tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/microsoft/VisioParserTest.java
URL: http://svn.apache.org/viewvc/tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/microsoft/VisioParserTest.java?rev=1339833&r1=1339832&r2=1339833&view=diff
==============================================================================
--- tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/microsoft/VisioParserTest.java (original)
+++ tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/microsoft/VisioParserTest.java Thu May 17 20:05:20 2012
@@ -19,6 +19,7 @@ package org.apache.tika.parser.microsoft
import java.io.InputStream;
import org.apache.tika.metadata.Metadata;
+import org.apache.tika.metadata.TikaCoreProperties;
import org.apache.tika.parser.ParseContext;
import org.apache.tika.sax.BodyContentHandler;
import org.xml.sax.ContentHandler;
@@ -38,7 +39,7 @@ public class VisioParserTest extends Tes
assertEquals(
"application/vnd.visio",
metadata.get(Metadata.CONTENT_TYPE));
- assertEquals("", metadata.get(Metadata.TITLE));
+ assertEquals("", metadata.get(TikaCoreProperties.TITLE));
assertEquals("Hogwarts", metadata.get(Metadata.AUTHOR));
String content = handler.toString();
assertTrue(content.contains("Some random text, on a page"));
Modified: tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/microsoft/WordParserTest.java
URL: http://svn.apache.org/viewvc/tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/microsoft/WordParserTest.java?rev=1339833&r1=1339832&r2=1339833&view=diff
==============================================================================
--- tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/microsoft/WordParserTest.java (original)
+++ tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/microsoft/WordParserTest.java Thu May 17 20:05:20 2012
@@ -27,6 +27,7 @@ import javax.xml.transform.stream.Stream
import org.apache.tika.TikaTest;
import org.apache.tika.metadata.Metadata;
+import org.apache.tika.metadata.TikaCoreProperties;
import org.apache.tika.parser.ParseContext;
import org.apache.tika.parser.microsoft.ooxml.OOXMLParserTest;
import org.apache.tika.sax.BodyContentHandler;
@@ -45,7 +46,7 @@ public class WordParserTest extends Tika
assertEquals(
"application/msword",
metadata.get(Metadata.CONTENT_TYPE));
- assertEquals("Sample Word Document", metadata.get(Metadata.TITLE));
+ assertEquals("Sample Word Document", metadata.get(TikaCoreProperties.TITLE));
assertEquals("Keith Bennett", metadata.get(Metadata.AUTHOR));
assertTrue(handler.toString().contains("Sample Word Document"));
} finally {
@@ -114,7 +115,7 @@ public class WordParserTest extends Tika
assertEquals(
"application/msword",
metadata.get(Metadata.CONTENT_TYPE));
- assertEquals("Sample Word Document", metadata.get(Metadata.TITLE));
+ assertEquals("Sample Word Document", metadata.get(TikaCoreProperties.TITLE));
assertEquals("Keith Bennett", metadata.get(Metadata.AUTHOR));
assertTrue(xml.contains("Sample Word Document"));
@@ -176,8 +177,8 @@ public class WordParserTest extends Tika
assertEquals(
"application/msword",
metadata.get(Metadata.CONTENT_TYPE));
- assertEquals("The quick brown fox jumps over the lazy dog", metadata.get(Metadata.TITLE));
- assertEquals("Gym class featuring a brown fox and lazy dog", metadata.get(Metadata.SUBJECT));
+ assertEquals("The quick brown fox jumps over the lazy dog", metadata.get(TikaCoreProperties.TITLE));
+ assertEquals("Gym class featuring a brown fox and lazy dog", metadata.get(TikaCoreProperties.SUBJECT));
assertEquals("Nevin Nollop", metadata.get(Metadata.AUTHOR));
assertTrue(handler.toString().contains("The quick brown fox jumps over the lazy dog"));
} finally {
@@ -241,7 +242,7 @@ public class WordParserTest extends Tika
assertContains("Subject is here", content);
assertEquals("Subject is here",
- metadata.get(Metadata.SUBJECT));
+ metadata.get(TikaCoreProperties.SUBJECT));
assertContains("Suddenly some Japanese text:", content);
// Special version of (GHQ)
@@ -278,11 +279,11 @@ public class WordParserTest extends Tika
assertEquals("Microsoft Office Word",metadata.get(Metadata.APPLICATION_NAME));
assertEquals("1", metadata.get(Metadata.PAGE_COUNT));
assertEquals("2", metadata.get(Metadata.WORD_COUNT));
- assertEquals("My Title", metadata.get(Metadata.TITLE));
+ assertEquals("My Title", metadata.get(TikaCoreProperties.TITLE));
assertEquals("My Keyword", metadata.get(Metadata.KEYWORDS));
assertEquals("Normal.dotm", metadata.get(Metadata.TEMPLATE));
assertEquals("My Comments", metadata.get(Metadata.COMMENTS));
- assertEquals("My subject", metadata.get(Metadata.SUBJECT));
+ assertEquals("My subject", metadata.get(TikaCoreProperties.SUBJECT));
assertEquals("EDF-DIT", metadata.get(Metadata.COMPANY));
assertEquals("MyStringValue", metadata.get("custom:MyCustomString"));
assertEquals("2010-12-30T23:00:00Z", metadata.get("custom:MyCustomDate"));
Modified: tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/microsoft/ooxml/OOXMLParserTest.java
URL: http://svn.apache.org/viewvc/tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/microsoft/ooxml/OOXMLParserTest.java?rev=1339833&r1=1339832&r2=1339833&view=diff
==============================================================================
--- tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/microsoft/ooxml/OOXMLParserTest.java (original)
+++ tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/microsoft/ooxml/OOXMLParserTest.java Thu May 17 20:05:20 2012
@@ -28,6 +28,7 @@ import javax.xml.transform.stream.Stream
import org.apache.tika.TikaTest;
import org.apache.tika.io.TikaInputStream;
import org.apache.tika.metadata.Metadata;
+import org.apache.tika.metadata.TikaCoreProperties;
import org.apache.tika.metadata.TikaMetadataKeys;
import org.apache.tika.parser.AutoDetectParser;
import org.apache.tika.parser.ParseContext;
@@ -57,7 +58,7 @@ public class OOXMLParserTest extends Tik
assertEquals(
"application/vnd.openxmlformats-officedocument.spreadsheetml.sheet",
metadata.get(Metadata.CONTENT_TYPE));
- assertEquals("Simple Excel document", metadata.get(Metadata.TITLE));
+ assertEquals("Simple Excel document", metadata.get(TikaCoreProperties.TITLE));
assertEquals("Keith Bennett", metadata.get(Metadata.AUTHOR));
String content = handler.toString();
assertTrue(content.contains("Sample Excel Worksheet"));
@@ -184,7 +185,7 @@ public class OOXMLParserTest extends Tik
"Mime-type checking for " + filename,
mimeTypes[i],
metadata.get(Metadata.CONTENT_TYPE));
- assertEquals("Attachment Test", metadata.get(Metadata.TITLE));
+ assertEquals("Attachment Test", metadata.get(TikaCoreProperties.TITLE));
assertEquals("Rajiv", metadata.get(Metadata.AUTHOR));
String content = handler.toString();
@@ -273,7 +274,7 @@ public class OOXMLParserTest extends Tik
assertEquals(
"application/vnd.openxmlformats-officedocument.wordprocessingml.document",
metadata.get(Metadata.CONTENT_TYPE));
- assertEquals("Sample Word Document", metadata.get(Metadata.TITLE));
+ assertEquals("Sample Word Document", metadata.get(TikaCoreProperties.TITLE));
assertEquals("Keith Bennett", metadata.get(Metadata.AUTHOR));
assertTrue(handler.toString().contains("Sample Word Document"));
} finally {
@@ -344,7 +345,7 @@ public class OOXMLParserTest extends Tik
assertEquals(
"application/vnd.openxmlformats-officedocument.wordprocessingml.document",
metadata.get(Metadata.CONTENT_TYPE));
- assertEquals("Sample Word Document", metadata.get(Metadata.TITLE));
+ assertEquals("Sample Word Document", metadata.get(TikaCoreProperties.TITLE));
assertEquals("Keith Bennett", metadata.get(Metadata.AUTHOR));
assertTrue(xml.contains("Sample Word Document"));
@@ -557,7 +558,7 @@ public class OOXMLParserTest extends Tik
assertContains("Subject is here", content);
assertEquals("Subject is here",
- metadata.get(Metadata.SUBJECT));
+ metadata.get(TikaCoreProperties.SUBJECT));
assertContains("Suddenly some Japanese text:", content);
// Special version of (GHQ)
@@ -625,7 +626,7 @@ public class OOXMLParserTest extends Tik
assertContains("Subject is here", content);
assertEquals("Subject is here",
- metadata.get(Metadata.SUBJECT));
+ metadata.get(TikaCoreProperties.SUBJECT));
assertContains("Suddenly some Japanese text:", content);
// Special version of (GHQ)
@@ -729,7 +730,7 @@ public class OOXMLParserTest extends Tik
metadata.get(Metadata.CONTENT_TYPE));
assertEquals(null, metadata.get(Metadata.AUTHOR));
assertEquals(null, metadata.get(Metadata.LAST_AUTHOR));
- assertEquals("2006-09-12T15:06:44Z", metadata.get(Metadata.DATE));
+ assertEquals("2006-09-12T15:06:44Z", metadata.get(TikaCoreProperties.DATE));
assertEquals("2006-09-12T15:06:44Z", metadata.get(Metadata.CREATION_DATE));
assertEquals("2011-08-22T14:24:38Z", metadata.get(Metadata.LAST_MODIFIED));
assertEquals("Microsoft Excel", metadata.get(Metadata.APPLICATION_NAME));
@@ -758,17 +759,17 @@ public class OOXMLParserTest extends Tik
metadata.get(Metadata.CONTENT_TYPE));
assertEquals("EJ04325S", metadata.get(Metadata.AUTHOR));
assertEquals("Etienne Jouvin", metadata.get(Metadata.LAST_AUTHOR));
- assertEquals("2011-07-29T16:52:00Z", metadata.get(Metadata.DATE));
+ assertEquals("2011-07-29T16:52:00Z", metadata.get(TikaCoreProperties.DATE));
assertEquals("2011-07-29T16:52:00Z", metadata.get(Metadata.CREATION_DATE));
assertEquals("2012-01-03T22:14:00Z", metadata.get(Metadata.LAST_MODIFIED));
assertEquals("Microsoft Office Word",metadata.get(Metadata.APPLICATION_NAME));
assertEquals("1", metadata.get(Metadata.PAGE_COUNT));
assertEquals("2", metadata.get(Metadata.WORD_COUNT));
- assertEquals("My Title", metadata.get(Metadata.TITLE));
+ assertEquals("My Title", metadata.get(TikaCoreProperties.TITLE));
assertEquals("My Keyword", metadata.get(Metadata.KEYWORDS));
assertEquals("Normal.dotm", metadata.get(Metadata.TEMPLATE));
- assertEquals("My subject", metadata.get(Metadata.SUBJECT));
- assertEquals("EDF-DIT", metadata.get(Metadata.PUBLISHER));
+ assertEquals("My subject", metadata.get(TikaCoreProperties.SUBJECT));
+ assertEquals("EDF-DIT", metadata.get(TikaCoreProperties.PUBLISHER));
assertEquals("true", metadata.get("custom:myCustomBoolean"));
assertEquals("3", metadata.get("custom:myCustomNumber"));
assertEquals("MyStringValue", metadata.get("custom:MyCustomString"));
@@ -794,12 +795,12 @@ public class OOXMLParserTest extends Tik
metadata.get(Metadata.CONTENT_TYPE));
assertEquals("JOUVIN ETIENNE", metadata.get(Metadata.AUTHOR));
assertEquals("EJ04325S", metadata.get(Metadata.LAST_AUTHOR));
- assertEquals("2011-08-22T13:30:53Z", metadata.get(Metadata.DATE));
+ assertEquals("2011-08-22T13:30:53Z", metadata.get(TikaCoreProperties.DATE));
assertEquals("2011-08-22T13:30:53Z", metadata.get(Metadata.CREATION_DATE));
assertEquals("2011-08-22T13:32:49Z", metadata.get(Metadata.LAST_MODIFIED));
assertEquals("1", metadata.get(Metadata.SLIDE_COUNT));
assertEquals("3", metadata.get(Metadata.WORD_COUNT));
- assertEquals("Test extraction properties pptx", metadata.get(Metadata.TITLE));
+ assertEquals("Test extraction properties pptx", metadata.get(TikaCoreProperties.TITLE));
assertEquals("true", metadata.get("custom:myCustomBoolean"));
assertEquals("3", metadata.get("custom:myCustomNumber"));
assertEquals("MyStringValue", metadata.get("custom:MyCustomString"));
Modified: tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/mp3/Mp3ParserTest.java
URL: http://svn.apache.org/viewvc/tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/mp3/Mp3ParserTest.java?rev=1339833&r1=1339832&r2=1339833&view=diff
==============================================================================
--- tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/mp3/Mp3ParserTest.java (original)
+++ tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/mp3/Mp3ParserTest.java Thu May 17 20:05:20 2012
@@ -22,6 +22,7 @@ import java.io.InputStream;
import junit.framework.TestCase;
import org.apache.tika.metadata.Metadata;
+import org.apache.tika.metadata.TikaCoreProperties;
import org.apache.tika.metadata.XMPDM;
import org.apache.tika.parser.AutoDetectParser;
import org.apache.tika.parser.ParseContext;
@@ -51,7 +52,7 @@ public class Mp3ParserTest extends TestC
}
assertEquals("audio/mpeg", metadata.get(Metadata.CONTENT_TYPE));
- assertEquals("Test Title", metadata.get(Metadata.TITLE));
+ assertEquals("Test Title", metadata.get(TikaCoreProperties.TITLE));
assertEquals("Test Artist", metadata.get(Metadata.AUTHOR));
String content = handler.toString();
@@ -86,7 +87,7 @@ public class Mp3ParserTest extends TestC
// Check core properties
assertEquals("audio/mpeg", metadata.get(Metadata.CONTENT_TYPE));
- assertEquals("Test Title", metadata.get(Metadata.TITLE));
+ assertEquals("Test Title", metadata.get(TikaCoreProperties.TITLE));
assertEquals("Test Artist", metadata.get(Metadata.AUTHOR));
// Check the textual contents
@@ -135,7 +136,7 @@ public class Mp3ParserTest extends TestC
}
assertEquals("audio/mpeg", metadata.get(Metadata.CONTENT_TYPE));
- assertEquals("Test Title", metadata.get(Metadata.TITLE));
+ assertEquals("Test Title", metadata.get(TikaCoreProperties.TITLE));
assertEquals("Test Artist", metadata.get(Metadata.AUTHOR));
String content = handler.toString();
@@ -169,7 +170,7 @@ public class Mp3ParserTest extends TestC
}
assertEquals("audio/mpeg", metadata.get(Metadata.CONTENT_TYPE));
- assertEquals("Test Title", metadata.get(Metadata.TITLE));
+ assertEquals("Test Title", metadata.get(TikaCoreProperties.TITLE));
assertEquals("Test Artist", metadata.get(Metadata.AUTHOR));
String content = handler.toString();
@@ -203,7 +204,7 @@ public class Mp3ParserTest extends TestC
}
assertEquals("audio/mpeg", metadata.get(Metadata.CONTENT_TYPE));
- assertEquals("Une chason en Fran\u00e7ais", metadata.get(Metadata.TITLE));
+ assertEquals("Une chason en Fran\u00e7ais", metadata.get(TikaCoreProperties.TITLE));
assertEquals("Test Artist \u2468\u2460", metadata.get(Metadata.AUTHOR));
assertEquals("Test Artist \u2468\u2460", metadata.get(XMPDM.ARTIST));
assertEquals("Test Album \u2460\u2468", metadata.get(XMPDM.ALBUM));
@@ -241,7 +242,7 @@ public class Mp3ParserTest extends TestC
}
assertEquals("audio/mpeg", metadata.get(Metadata.CONTENT_TYPE));
- assertEquals("Test Title", metadata.get(Metadata.TITLE));
+ assertEquals("Test Title", metadata.get(TikaCoreProperties.TITLE));
assertEquals("Test Artist", metadata.get(Metadata.AUTHOR));
String content = handler.toString();
@@ -306,7 +307,7 @@ public class Mp3ParserTest extends TestC
}
assertEquals("audio/mpeg", metadata.get(Metadata.CONTENT_TYPE));
- assertEquals("Plus loin vers l'ouest", metadata.get(Metadata.TITLE));
+ assertEquals("Plus loin vers l'ouest", metadata.get(TikaCoreProperties.TITLE));
assertEquals("Merzhin", metadata.get(Metadata.AUTHOR));
String content = handler.toString();
@@ -341,7 +342,7 @@ public class Mp3ParserTest extends TestC
// Check we coud get the headers from the start
assertEquals("audio/mpeg", metadata.get(Metadata.CONTENT_TYPE));
- assertEquals("Girl you have no faith in medicine", metadata.get(Metadata.TITLE));
+ assertEquals("Girl you have no faith in medicine", metadata.get(TikaCoreProperties.TITLE));
assertEquals("The White Stripes", metadata.get(Metadata.AUTHOR));
String content = handler.toString();
Modified: tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/mp4/MP4ParserTest.java
URL: http://svn.apache.org/viewvc/tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/mp4/MP4ParserTest.java?rev=1339833&r1=1339832&r2=1339833&view=diff
==============================================================================
--- tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/mp4/MP4ParserTest.java (original)
+++ tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/mp4/MP4ParserTest.java Thu May 17 20:05:20 2012
@@ -22,6 +22,7 @@ import junit.framework.TestCase;
import org.apache.tika.io.TikaInputStream;
import org.apache.tika.metadata.Metadata;
+import org.apache.tika.metadata.TikaCoreProperties;
import org.apache.tika.metadata.XMPDM;
import org.apache.tika.parser.AutoDetectParser;
import org.apache.tika.parser.ParseContext;
@@ -52,10 +53,10 @@ public class MP4ParserTest extends TestC
// Check core properties
assertEquals("audio/mp4", metadata.get(Metadata.CONTENT_TYPE));
- assertEquals("Test Title", metadata.get(Metadata.TITLE));
+ assertEquals("Test Title", metadata.get(TikaCoreProperties.TITLE));
assertEquals("Test Artist", metadata.get(Metadata.AUTHOR));
assertEquals("2012-01-28T18:39:18Z", metadata.get(Metadata.CREATION_DATE));
- assertEquals("2012-01-28T18:40:25Z", metadata.get(Metadata.MODIFIED));
+ assertEquals("2012-01-28T18:40:25Z", metadata.get(TikaCoreProperties.MODIFIED));
// Check the textual contents
String content = handler.toString();
Modified: tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/netcdf/NetCDFParserTest.java
URL: http://svn.apache.org/viewvc/tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/netcdf/NetCDFParserTest.java?rev=1339833&r1=1339832&r2=1339833&view=diff
==============================================================================
--- tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/netcdf/NetCDFParserTest.java (original)
+++ tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/netcdf/NetCDFParserTest.java Thu May 17 20:05:20 2012
@@ -21,6 +21,7 @@ import java.io.InputStream;
//TIKA imports
import org.apache.tika.metadata.Metadata;
+import org.apache.tika.metadata.TikaCoreProperties;
import org.apache.tika.parser.ParseContext;
import org.apache.tika.parser.Parser;
import org.apache.tika.sax.BodyContentHandler;
@@ -52,7 +53,7 @@ public class NetCDFParserTest extends Te
stream.close();
}
- assertEquals(metadata.get(Metadata.TITLE),
+ assertEquals(metadata.get(TikaCoreProperties.TITLE),
"model output prepared for IPCC AR4");
assertEquals(metadata.get(Metadata.CONTACT), "ccsm@ucar.edu");
assertEquals(metadata.get(Metadata.PROJECT_ID),
Modified: tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/odf/ODFParserTest.java
URL: http://svn.apache.org/viewvc/tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/odf/ODFParserTest.java?rev=1339833&r1=1339832&r2=1339833&view=diff
==============================================================================
--- tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/odf/ODFParserTest.java (original)
+++ tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/odf/ODFParserTest.java Thu May 17 20:05:20 2012
@@ -21,6 +21,7 @@ import java.io.InputStream;
import org.apache.tika.TikaTest;
import org.apache.tika.metadata.Metadata;
import org.apache.tika.metadata.Office;
+import org.apache.tika.metadata.TikaCoreProperties;
import org.apache.tika.parser.AutoDetectParser;
import org.apache.tika.parser.ParseContext;
import org.apache.tika.parser.Parser;
@@ -77,14 +78,18 @@ public class ODFParserTest extends TikaT
assertEquals(
"application/vnd.oasis.opendocument.text",
metadata.get(Metadata.CONTENT_TYPE));
- assertEquals("2007-09-14T11:07:10", metadata.get(Metadata.DATE));
- assertEquals("2007-09-14T11:06:08", metadata.get(Metadata.CREATION_DATE));
assertEquals("en-US", metadata.get(Metadata.LANGUAGE));
assertEquals("PT1M7S", metadata.get(Metadata.EDIT_TIME));
assertEquals(
"NeoOffice/2.2$Unix OpenOffice.org_project/680m18$Build-9161",
metadata.get("generator"));
+ // Check date metadata, both old-style and new-style
+ assertEquals("2007-09-14T11:07:10", metadata.get(TikaCoreProperties.DATE));
+ assertEquals("2007-09-14T11:06:08", metadata.get(TikaCoreProperties.CREATION_DATE));
+ assertEquals("2007-09-14T11:07:10", metadata.get(Metadata.DATE));
+ assertEquals("2007-09-14T11:06:08", metadata.get(Metadata.CREATION_DATE));
+
// Check the document statistics
assertEquals("1", metadata.get(Office.PAGE_COUNT));
assertEquals("1", metadata.get(Office.PARAGRAPH_COUNT));
@@ -143,10 +148,10 @@ public class ODFParserTest extends TikaT
assertEquals(
"application/vnd.oasis.opendocument.formula",
metadata.get(Metadata.CONTENT_TYPE));
- assertEquals(null, metadata.get(Metadata.DATE));
+ assertEquals(null, metadata.get(TikaCoreProperties.DATE));
assertEquals("2006-01-27T11:55:22", metadata.get(Metadata.CREATION_DATE));
- assertEquals("The quick brown fox jumps over the lazy dog", metadata.get(Metadata.TITLE));
- assertEquals("Gym class featuring a brown fox and lazy dog", metadata.get(Metadata.SUBJECT));
+ assertEquals("The quick brown fox jumps over the lazy dog", metadata.get(TikaCoreProperties.TITLE));
+ assertEquals("Gym class featuring a brown fox and lazy dog", metadata.get(TikaCoreProperties.SUBJECT));
assertEquals("PT0S", metadata.get(Metadata.EDIT_TIME));
assertEquals("1", metadata.get("editing-cycles"));
assertEquals(
@@ -198,12 +203,12 @@ public class ODFParserTest extends TikaT
assertEquals(
"application/vnd.oasis.opendocument.text",
metadata.get(Metadata.CONTENT_TYPE));
- assertEquals("2009-10-05T21:22:38", metadata.get(Metadata.DATE));
+ assertEquals("2009-10-05T21:22:38", metadata.get(TikaCoreProperties.DATE));
assertEquals("2009-10-05T19:04:01", metadata.get(Metadata.CREATION_DATE));
- assertEquals("Apache Tika", metadata.get(Metadata.TITLE));
- assertEquals("Test document", metadata.get(Metadata.SUBJECT));
- assertEquals("A rather complex document", metadata.get(Metadata.DESCRIPTION));
- assertEquals("Bart Hanssens", metadata.get(Metadata.CREATOR));
+ assertEquals("Apache Tika", metadata.get(TikaCoreProperties.TITLE));
+ assertEquals("Test document", metadata.get(TikaCoreProperties.SUBJECT));
+ assertEquals("A rather complex document", metadata.get(TikaCoreProperties.DESCRIPTION));
+ assertEquals("Bart Hanssens", metadata.get(TikaCoreProperties.CREATOR));
assertEquals("Bart Hanssens", metadata.get("initial-creator"));
assertEquals("2", metadata.get("editing-cycles"));
assertEquals("PT02H03M24S", metadata.get(Metadata.EDIT_TIME));
Modified: tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/pdf/PDFParserTest.java
URL: http://svn.apache.org/viewvc/tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/pdf/PDFParserTest.java?rev=1339833&r1=1339832&r2=1339833&view=diff
==============================================================================
--- tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/pdf/PDFParserTest.java (original)
+++ tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/pdf/PDFParserTest.java Thu May 17 20:05:20 2012
@@ -26,6 +26,7 @@ import javax.xml.transform.stream.Stream
import org.apache.tika.TikaTest;
import org.apache.tika.metadata.Metadata;
+import org.apache.tika.metadata.TikaCoreProperties;
import org.apache.tika.parser.AutoDetectParser;
import org.apache.tika.parser.ParseContext;
import org.apache.tika.parser.Parser;
@@ -53,7 +54,7 @@ public class PDFParserTest extends TikaT
assertEquals("application/pdf", metadata.get(Metadata.CONTENT_TYPE));
assertEquals("Bertrand Delacr\u00e9taz", metadata.get(Metadata.AUTHOR));
- assertEquals("Apache Tika - Apache Tika", metadata.get(Metadata.TITLE));
+ assertEquals("Apache Tika - Apache Tika", metadata.get(TikaCoreProperties.TITLE));
// Can't reliably test dates yet - see TIKA-451
// assertEquals("Sat Sep 15 10:02:31 BST 2007", metadata.get(Metadata.CREATION_DATE));
@@ -87,7 +88,7 @@ public class PDFParserTest extends TikaT
assertEquals("application/pdf", metadata.get(Metadata.CONTENT_TYPE));
assertEquals("Document author", metadata.get(Metadata.AUTHOR));
- assertEquals("Document title", metadata.get(Metadata.TITLE));
+ assertEquals("Document title", metadata.get(TikaCoreProperties.TITLE));
assertEquals("Custom Value", metadata.get("Custom Property"));
assertEquals("Array Entry 1", metadata.get("Custom Array"));
@@ -120,8 +121,8 @@ public class PDFParserTest extends TikaT
assertEquals("application/pdf", metadata.get(Metadata.CONTENT_TYPE));
assertEquals("The Bank of England", metadata.get(Metadata.AUTHOR));
- assertEquals("Speeches by Andrew G Haldane", metadata.get(Metadata.SUBJECT));
- assertEquals("Rethinking the Financial Network, Speech by Andrew G Haldane, Executive Director, Financial Stability delivered at the Financial Student Association, Amsterdam on 28 April 2009", metadata.get(Metadata.TITLE));
+ assertEquals("Speeches by Andrew G Haldane", metadata.get(TikaCoreProperties.SUBJECT));
+ assertEquals("Rethinking the Financial Network, Speech by Andrew G Haldane, Executive Director, Financial Stability delivered at the Financial Student Association, Amsterdam on 28 April 2009", metadata.get(TikaCoreProperties.TITLE));
String content = handler.toString();
assertTrue(content.contains("RETHINKING THE FINANCIAL NETWORK"));
@@ -150,8 +151,8 @@ public class PDFParserTest extends TikaT
assertEquals("application/pdf", metadata.get(Metadata.CONTENT_TYPE));
assertEquals("The Bank of England", metadata.get(Metadata.AUTHOR));
- assertEquals("Speeches by Andrew G Haldane", metadata.get(Metadata.SUBJECT));
- assertEquals("Rethinking the Financial Network, Speech by Andrew G Haldane, Executive Director, Financial Stability delivered at the Financial Student Association, Amsterdam on 28 April 2009", metadata.get(Metadata.TITLE));
+ assertEquals("Speeches by Andrew G Haldane", metadata.get(TikaCoreProperties.SUBJECT));
+ assertEquals("Rethinking the Financial Network, Speech by Andrew G Haldane, Executive Director, Financial Stability delivered at the Financial Student Association, Amsterdam on 28 April 2009", metadata.get(TikaCoreProperties.TITLE));
assertTrue(content.contains("RETHINKING THE FINANCIAL NETWORK"));
assertTrue(content.contains("On 16 November 2002"));
@@ -233,7 +234,7 @@ public class PDFParserTest extends TikaT
assertContains("Subject is here", content);
assertEquals("Subject is here",
- metadata.get(Metadata.SUBJECT));
+ metadata.get(TikaCoreProperties.SUBJECT));
assertContains("Suddenly some Japanese text:", content);
// Special version of (GHQ)
Modified: tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/prt/PRTParserTest.java
URL: http://svn.apache.org/viewvc/tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/prt/PRTParserTest.java?rev=1339833&r1=1339832&r2=1339833&view=diff
==============================================================================
--- tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/prt/PRTParserTest.java (original)
+++ tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/prt/PRTParserTest.java Thu May 17 20:05:20 2012
@@ -20,6 +20,7 @@ import java.io.InputStream;
import org.apache.tika.TikaTest;
import org.apache.tika.metadata.Metadata;
+import org.apache.tika.metadata.TikaCoreProperties;
import org.apache.tika.sax.BodyContentHandler;
import org.xml.sax.ContentHandler;
@@ -38,11 +39,11 @@ public class PRTParserTest extends TikaT
// This file has a date
assertEquals("2011-06-20T16:54:00",
- metadata.get(Metadata.DATE));
+ metadata.get(TikaCoreProperties.DATE));
assertEquals("2011-06-20T16:54:00",
metadata.get(Metadata.CREATION_DATE));
// But no description
- assertEquals(null, metadata.get(Metadata.DESCRIPTION));
+ assertEquals(null, metadata.get(TikaCoreProperties.DESCRIPTION));
String contents = handler.toString();
@@ -80,7 +81,7 @@ public class PRTParserTest extends TikaT
assertEquals("1997-04-01T08:59:00",
metadata.get(Metadata.CREATION_DATE));
assertEquals("TIKA TEST PART DESCRIPTION INFORMATION\r\n",
- metadata.get(Metadata.DESCRIPTION));
+ metadata.get(TikaCoreProperties.DESCRIPTION));
String contents = handler.toString();
Modified: tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/rtf/RTFParserTest.java
URL: http://svn.apache.org/viewvc/tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/rtf/RTFParserTest.java?rev=1339833&r1=1339832&r2=1339833&view=diff
==============================================================================
--- tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/rtf/RTFParserTest.java (original)
+++ tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/rtf/RTFParserTest.java Thu May 17 20:05:20 2012
@@ -30,6 +30,7 @@ import org.apache.tika.Tika;
import org.apache.tika.TikaTest;
import org.apache.tika.io.TikaInputStream;
import org.apache.tika.metadata.Metadata;
+import org.apache.tika.metadata.TikaCoreProperties;
import org.apache.tika.parser.ParseContext;
import org.apache.tika.sax.WriteOutContentHandler;
@@ -132,7 +133,7 @@ public class RTFParserTest extends TikaT
// Verify title, since it was also encoded with MS932:
Result r = getResult("testRTF-ms932.rtf");
- assertEquals("\u30bf\u30a4\u30c8\u30eb", r.metadata.get(Metadata.TITLE));
+ assertEquals("\u30bf\u30a4\u30c8\u30eb", r.metadata.get(TikaCoreProperties.TITLE));
}
public void testUmlautSpacesExtraction() throws Exception {
@@ -153,7 +154,7 @@ public class RTFParserTest extends TikaT
// Verify title -- this title uses upr escape inside
// title info field:
assertEquals("\u30be\u30eb\u30b2\u3068\u5c3e\u5d0e\u3001\u6de1\u3005\u3068\u6700\u671f\u3000",
- r.metadata.get(Metadata.TITLE));
+ r.metadata.get(TikaCoreProperties.TITLE));
assertEquals("VMazel", r.metadata.get(Metadata.AUTHOR));
assertEquals("StarWriter", r.metadata.get(Metadata.COMMENT));
assertContains("1.", content);
@@ -267,7 +268,7 @@ public class RTFParserTest extends TikaT
assertContains("Subject is here", content);
assertEquals("Subject is here",
- r.metadata.get(Metadata.SUBJECT));
+ r.metadata.get(TikaCoreProperties.SUBJECT));
assertContains("Suddenly some Japanese text:", content);
// Special version of (GHQ)
Modified: tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/txt/TXTParserTest.java
URL: http://svn.apache.org/viewvc/tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/txt/TXTParserTest.java?rev=1339833&r1=1339832&r2=1339833&view=diff
==============================================================================
--- tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/txt/TXTParserTest.java (original)
+++ tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/txt/TXTParserTest.java Thu May 17 20:05:20 2012
@@ -20,6 +20,7 @@ import java.io.ByteArrayInputStream;
import java.io.StringWriter;
import org.apache.tika.metadata.Metadata;
+import org.apache.tika.metadata.TikaCoreProperties;
import org.apache.tika.parser.ParseContext;
import org.apache.tika.parser.Parser;
import org.apache.tika.sax.BodyContentHandler;
@@ -51,7 +52,7 @@ public class TXTParserTest extends TestC
// TIKA-501: Remove language detection from TXTParser
assertNull(metadata.get(Metadata.CONTENT_LANGUAGE));
- assertNull(metadata.get(Metadata.LANGUAGE));
+ assertNull(metadata.get(TikaCoreProperties.LANGUAGE));
assertTrue(content.contains("Hello"));
assertTrue(content.contains("World"));
@@ -169,13 +170,13 @@ public class TXTParserTest extends TestC
final String test = "Simple Content";
Metadata metadata = new Metadata();
- metadata.set(Metadata.LANGUAGE, "en");
+ metadata.set(TikaCoreProperties.LANGUAGE, "en");
parser.parse(
new ByteArrayInputStream(test.getBytes("UTF-8")),
new BodyContentHandler(), metadata, new ParseContext());
- assertEquals("en", metadata.get(Metadata.LANGUAGE));
+ assertEquals("en", metadata.get(TikaCoreProperties.LANGUAGE));
}
public void testCP866() throws Exception {
Modified: tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/xml/DcXMLParserTest.java
URL: http://svn.apache.org/viewvc/tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/xml/DcXMLParserTest.java?rev=1339833&r1=1339832&r2=1339833&view=diff
==============================================================================
--- tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/xml/DcXMLParserTest.java (original)
+++ tika/trunk/tika-parsers/src/test/java/org/apache/tika/parser/xml/DcXMLParserTest.java Thu May 17 20:05:20 2012
@@ -21,6 +21,7 @@ import java.io.InputStream;
import junit.framework.TestCase;
import org.apache.tika.metadata.Metadata;
+import org.apache.tika.metadata.TikaCoreProperties;
import org.apache.tika.sax.BodyContentHandler;
import org.xml.sax.ContentHandler;
import org.xml.sax.helpers.DefaultHandler;
@@ -38,34 +39,34 @@ public class DcXMLParserTest extends Tes
assertEquals(
"application/xml",
metadata.get(Metadata.CONTENT_TYPE));
- assertEquals("Tika test document", metadata.get(Metadata.TITLE));
- assertEquals("Rida Benjelloun", metadata.get(Metadata.CREATOR));
+ assertEquals("Tika test document", metadata.get(TikaCoreProperties.TITLE));
+ assertEquals("Rida Benjelloun", metadata.get(TikaCoreProperties.CREATOR));
// The file contains 5 dc:subject tags, which come through as
// a multi-valued Tika Metadata entry in file order
- assertEquals(true, metadata.isMultiValued(Metadata.SUBJECT));
- assertEquals(5, metadata.getValues(Metadata.SUBJECT).length);
- assertEquals("Java", metadata.getValues(Metadata.SUBJECT)[0]);
- assertEquals("XML", metadata.getValues(Metadata.SUBJECT)[1]);
- assertEquals("XSLT", metadata.getValues(Metadata.SUBJECT)[2]);
- assertEquals("JDOM", metadata.getValues(Metadata.SUBJECT)[3]);
- assertEquals("Indexation", metadata.getValues(Metadata.SUBJECT)[4]);
+ assertEquals(true, metadata.isMultiValued(TikaCoreProperties.SUBJECT));
+ assertEquals(5, metadata.getValues(TikaCoreProperties.SUBJECT).length);
+ assertEquals("Java", metadata.getValues(TikaCoreProperties.SUBJECT)[0]);
+ assertEquals("XML", metadata.getValues(TikaCoreProperties.SUBJECT)[1]);
+ assertEquals("XSLT", metadata.getValues(TikaCoreProperties.SUBJECT)[2]);
+ assertEquals("JDOM", metadata.getValues(TikaCoreProperties.SUBJECT)[3]);
+ assertEquals("Indexation", metadata.getValues(TikaCoreProperties.SUBJECT)[4]);
assertEquals(
"Framework d\'indexation des documents XML, HTML, PDF etc..",
- metadata.get(Metadata.DESCRIPTION));
+ metadata.get(TikaCoreProperties.DESCRIPTION));
assertEquals(
"http://www.apache.org",
- metadata.get(Metadata.IDENTIFIER));
- assertEquals("test", metadata.get(Metadata.TYPE));
- assertEquals("application/msword", metadata.get(Metadata.FORMAT));
- assertEquals("Fr", metadata.get(Metadata.LANGUAGE));
- assertTrue(metadata.get(Metadata.RIGHTS).contains("testing chars"));
+ metadata.get(TikaCoreProperties.IDENTIFIER));
+ assertEquals("test", metadata.get(TikaCoreProperties.TYPE));
+ assertEquals("application/msword", metadata.get(TikaCoreProperties.FORMAT));
+ assertEquals("Fr", metadata.get(TikaCoreProperties.LANGUAGE));
+ assertTrue(metadata.get(TikaCoreProperties.RIGHTS).contains("testing chars"));
String content = handler.toString();
assertTrue(content.contains("Tika test document"));
- assertEquals("2000-12-01T00:00:00.000Z", metadata.get(Metadata.DATE));
+ assertEquals("2000-12-01T00:00:00.000Z", metadata.get(TikaCoreProperties.DATE));
} finally {
input.close();
}
@@ -78,7 +79,7 @@ public class DcXMLParserTest extends Tes
new DcXMLParser().parse(input, new DefaultHandler(), metadata);
final String expected = "Archim\u00E8de et Lius \u00E0 Ch\u00E2teauneuf testing chars en \u00E9t\u00E9";
- assertEquals(expected,metadata.get(Metadata.RIGHTS));
+ assertEquals(expected,metadata.get(TikaCoreProperties.RIGHTS));
} finally {
input.close();
}