svn commit: r567112 - in /incubator/tika/trunk/src/main/java/org/apache/tika: parser/ParserFactory.java utils/MimeTypesUtils.java

[jukka]

Author: jukka
Date: Fri Aug 17 12:11:25 2007
New Revision: 567112

URL: http://svn.apache.org/viewvc?view=rev&rev=567112
Log:
TIKA-8: Replaced the jmimeinfo dependency with a trivial mime type detector.

Modified:
    incubator/tika/trunk/src/main/java/org/apache/tika/parser/ParserFactory.java
    incubator/tika/trunk/src/main/java/org/apache/tika/utils/MimeTypesUtils.java

Modified: 
incubator/tika/trunk/src/main/java/org/apache/tika/parser/ParserFactory.java
URL: 
http://svn.apache.org/viewvc/incubator/tika/trunk/src/main/java/org/apache/tika/parser/ParserFactory.java?view=diff&rev=567112&r1=567111&r2=567112
==============================================================================
--- 
incubator/tika/trunk/src/main/java/org/apache/tika/parser/ParserFactory.java 
(original)
+++ 
incubator/tika/trunk/src/main/java/org/apache/tika/parser/ParserFactory.java 
Fri Aug 17 12:11:25 2007
@@ -24,7 +24,6 @@
 import org.apache.tika.config.ParserConfig;
 import org.apache.tika.exception.LiusException;
 import org.apache.tika.utils.MimeTypesUtils;
-import net.hedges.mimeinfo.MimeInfoException;
 
 import org.apache.log4j.Logger;
 
@@ -41,7 +40,7 @@
      * Build parser from file and Lius config object
      */
     public static Parser getParser(File file, LiusConfig tc)
-            throws MimeInfoException, IOException, LiusException {
+            throws IOException, LiusException {
         String mimeType = MimeTypesUtils.getMimeType(file);
         ParserConfig pc = tc.getParserConfig(mimeType);
         String className = pc.getParserClass();
@@ -76,7 +75,7 @@
      * Build parser from string file path and Lius config object
      */
     public static Parser getParser(String str, LiusConfig tc)
-            throws MimeInfoException, IOException, LiusException {
+            throws IOException, LiusException {
         return getParser(new File(str), tc);
     }
 
@@ -84,7 +83,7 @@
      * Build parser from string file path and Lius config file path
      */
     public static Parser getParser(String str, String tcPath)
-            throws MimeInfoException, IOException, LiusException {
+            throws IOException, LiusException {
         LiusConfig tc = LiusConfig.getInstance(tcPath);
         return getParser(new File(str), tc);
     }
@@ -93,7 +92,7 @@
      * Build parser from file and Lius config file path
      */
     public static Parser getParser(File file, String tcPath)
-            throws MimeInfoException, IOException, LiusException {
+            throws IOException, LiusException {
         LiusConfig tc = LiusConfig.getInstance(tcPath);
         return getParser(file, tc);
     }

Modified: 
incubator/tika/trunk/src/main/java/org/apache/tika/utils/MimeTypesUtils.java
URL: 
http://svn.apache.org/viewvc/incubator/tika/trunk/src/main/java/org/apache/tika/utils/MimeTypesUtils.java?view=diff&rev=567112&r1=567111&r2=567112
==============================================================================
--- 
incubator/tika/trunk/src/main/java/org/apache/tika/utils/MimeTypesUtils.java 
(original)
+++ 
incubator/tika/trunk/src/main/java/org/apache/tika/utils/MimeTypesUtils.java 
Fri Aug 17 12:11:25 2007
@@ -17,10 +17,6 @@
 package org.apache.tika.utils;
 
 import java.io.File;
-import java.io.IOException;
-
-import net.hedges.mimeinfo.MimeInfo;
-import net.hedges.mimeinfo.MimeInfoException;
 
 /**
  * Detect mime type from file
@@ -29,10 +25,32 @@
  */
 public class MimeTypesUtils {
 
-    public static String getMimeType(File file) throws MimeInfoException,
-            IOException {
-        MimeInfo mimeInfo = new MimeInfo();
-        return mimeInfo.getMimeType(file);
+    public static String getMimeType(File file) {
+        // FIXME: See TIKA-8
+        String name = file.getName().toLowerCase();
+        if (name.endsWith(".txt")) {
+            return "text/plain";
+        } else if (name.endsWith(".pdf")) {
+            return "application/pdf";
+        } else if (name.endsWith(".htm")) {
+            return "text/html";
+        } else if (name.endsWith(".html")) {
+            return "text/html";
+        } else if (name.endsWith(".xhtml")) {
+            return "application/xhtml+xml";
+        } else if (name.endsWith(".xml")) {
+            return "application/xml";
+        } else if (name.endsWith(".doc")) {
+            return "application/msword";
+        } else if (name.endsWith(".ppt")) {
+            return "application/vnd.ms-powerpoint";
+        } else if (name.endsWith(".xls")) {
+            return "application/vnd.ms-excel";
+        } else if (name.endsWith(".zip")) {
+            return "application/zip";
+        } else {
+            return "application/octet-stream";
+        }
     }
 
 }