From fe2792e9ce88b6e2ba9c7472b28ee0e942c3259a Mon Sep 17 00:00:00 2001
From: lotus <lotus@6c8d7289-2bf4-0310-a012-ef5d649a1542>
Date: Wed, 1 Oct 2008 17:47:11 +0000
Subject: [PATCH] use accept-language header instead of user agent for language
 detection

git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@5235 6c8d7289-2bf4-0310-a012-ef5d649a1542
---
 htroot/yacysearch.java                | 2 +-
 source/de/anomic/http/httpHeader.java | 6 ++++++
 source/de/anomic/tools/iso639.java    | 1 +
 3 files changed, 8 insertions(+), 1 deletion(-)

diff --git a/htroot/yacysearch.java b/htroot/yacysearch.java
index 3a8080236..f1d09f3d2 100644
--- a/htroot/yacysearch.java
+++ b/htroot/yacysearch.java
@@ -212,7 +212,7 @@ public class yacysearch {
             if (language.startsWith("lang_")) language = language.substring(5);
             if (!iso639.exists(language)) {
                 // find out language of the user by reading of the user-agent string
-                String agent = header.get("User-Agent");
+                String agent = header.get(httpRequestHeader.ACCEPT_LANGUAGE);
                 if (agent == null) agent = System.getProperty("user.language");
                 language = (agent == null) ? "en" : iso639.userAgentLanguageDetection(agent);
                 if (language == null) language = "en";
diff --git a/source/de/anomic/http/httpHeader.java b/source/de/anomic/http/httpHeader.java
index f7f0c67f8..578a3b926 100644
--- a/source/de/anomic/http/httpHeader.java
+++ b/source/de/anomic/http/httpHeader.java
@@ -76,6 +76,12 @@ public class httpHeader extends TreeMap<String, String> implements Map<String, S
     // TODO: sort these header properties into request and response properties (some are both)
     public static final String HOST = "Host";
     public static final String USER_AGENT = "User-Agent";
+    
+    public static final String ACCEPT = "Accept";
+    public static final String ACCEPT_LANGUAGE = "Accept-Language";
+    public static final String ACCEPT_ENCODING = "Accept-Encoding";
+    public static final String ACCEPT_CHARSET = "Accept-Charset";
+    
 
     public static final String CONTENT_LENGTH = "Content-Length";
     public static final String CONTENT_TYPE = "Content-Type";
diff --git a/source/de/anomic/tools/iso639.java b/source/de/anomic/tools/iso639.java
index b72992a15..f3d77a5a0 100755
--- a/source/de/anomic/tools/iso639.java
+++ b/source/de/anomic/tools/iso639.java
@@ -202,6 +202,7 @@ public class iso639 {
     public static final String userAgentLanguageDetection(String userAgent) {
         if (userAgent == null || userAgent.length() < 2) return null;
         userAgent = userAgent.toLowerCase();
+        if (mapping.containsKey(userAgent.substring(0, 2))) return userAgent.substring(0, 2);
         if (userAgent.length() == 2 && mapping.containsKey(userAgent)) return userAgent;
         if (userAgent.length() == 5 && mapping.containsKey(userAgent.substring(0, 2))) return userAgent.substring(0, 2);
         int p = 2;