From a1b2c9a67d6ffaeb08f42b5bada6c6088bba4956 Mon Sep 17 00:00:00 2001
From: Michael Peter Christen <mc@yacy.net>
Date: Thu, 16 Aug 2012 17:49:35 +0200
Subject: [PATCH] doctype2mime fix, influences metadata conversion between old
 metadata and solr

---
 defaults/httpd.mime                              |  1 +
 source/de/anomic/crawler/retrieval/Response.java | 16 +++++++++++-----
 .../net/yacy/cora/document/Classification.java   |  2 +-
 3 files changed, 13 insertions(+), 6 deletions(-)

diff --git a/defaults/httpd.mime b/defaults/httpd.mime
index 2fc577369..9b763ccb5 100644
--- a/defaults/httpd.mime
+++ b/defaults/httpd.mime
@@ -47,6 +47,7 @@ kml     = application/vnd.google-earth.kml+xml
 lha     = application/x-lzh
 lzh     = application/x-lzh
 m4v     = video/x-m4v
+md5     = text/plain
 mf      = application/octet-stream
 mov     = video/quicktime
 mm      = application/freemind
diff --git a/source/de/anomic/crawler/retrieval/Response.java b/source/de/anomic/crawler/retrieval/Response.java
index b9c75c845..46e674bee 100644
--- a/source/de/anomic/crawler/retrieval/Response.java
+++ b/source/de/anomic/crawler/retrieval/Response.java
@@ -118,16 +118,22 @@ public class Response {
     }
 
     public static String doctype2mime(String ext, char doctype) {
-    	String mime = Classification.ext2mime(ext);
-    	int p = mime.indexOf('/');
-    	if (p < 0) return mime;
-    	if (doctype == DT_TEXT) return "text" + mime.substring(p);
+        if (doctype == DT_PDFPS) return "application/pdf";
+        if (doctype == DT_HTML) return "text/html";
+        if (doctype == DT_DOC) return "application/msword";
+        if (doctype == DT_FLASH) return "application/x-shockwave-flash";
+        if (doctype == DT_SHARE) return "text/plain";
+        if (doctype == DT_BINARY) return "application/octet-stream";
+        String mime = Classification.ext2mime(ext);
+        int p = mime.indexOf('/');
+        if (p < 0) return mime;
+        if (doctype == DT_TEXT) return "text" + mime.substring(p);
     	if (doctype == DT_IMAGE) return "image" + mime.substring(p);
     	if (doctype == DT_AUDIO) return "audio" + mime.substring(p);
     	if (doctype == DT_MOVIE) return "video" + mime.substring(p);
     	return mime;
     }
-    
+
     public static final int QUEUE_STATE_FRESH             = 0;
     public static final int QUEUE_STATE_PARSING           = 1;
     public static final int QUEUE_STATE_CONDENSING        = 2;
diff --git a/source/net/yacy/cora/document/Classification.java b/source/net/yacy/cora/document/Classification.java
index 13afe7018..7afbf7272 100644
--- a/source/net/yacy/cora/document/Classification.java
+++ b/source/net/yacy/cora/document/Classification.java
@@ -184,7 +184,7 @@ public class Classification {
     }
 
     public static String ext2mime(final String ext) {
-        return mimeTable.getProperty(ext, "application/" + ext);
+        return mimeTable.getProperty(ext, "application/" + (ext == null || ext.length() == 0 ? "octet-stream" : ext));
     }
 
     public static String ext2mime(final String ext, final String dfltMime) {