From 78e889d9842d4bc99e2cd3e8b7ed736c5537e1f4 Mon Sep 17 00:00:00 2001
From: Jean-Francois Dockes <jfd@recoll.org>
Date: Wed, 23 Jan 2013 12:04:02 +0100
Subject: [PATCH] use the "charset" extended attribute for text files if it is
 set

---
 src/internfile/mh_text.cpp | 15 ++++++++++++---
 src/internfile/mh_text.h   |  3 ++-
 2 files changed, 14 insertions(+), 4 deletions(-)

diff --git a/src/internfile/mh_text.cpp b/src/internfile/mh_text.cpp
index 03a2461b..5f5fd023 100644
--- a/src/internfile/mh_text.cpp
+++ b/src/internfile/mh_text.cpp
@@ -33,6 +33,7 @@ using namespace std;
 #include "readfile.h"
 #include "md5.h"
 #include "rclconfig.h"
+#include "pxattr.h"
 
 const int MB = 1024*1024;
 const int KB = 1024;
@@ -53,6 +54,10 @@ bool MimeHandlerText::set_document_file(const string &fn)
         return false;
     }
 
+    // Check for charset defined in extended attribute as per:
+    // http://freedesktop.org/wiki/CommonExtendedAttributes
+    pxattr::get(m_fn, "charset", &m_charsetfromxattr);
+
     // Max file size parameter: texts over this size are not indexed
     int maxmbs = 20;
     m_config->getConfParam("textfilemaxmbs", &maxmbs);
@@ -115,14 +120,18 @@ bool MimeHandlerText::next_document()
     if (m_havedoc == false)
 	return false;
 
-    // We transcode even if defcharset is supposedly already utf-8:
-    // this validates the encoding.
-    m_metaData[cstr_dj_keyorigcharset] = m_dfltInputCharset;
+    if (m_charsetfromxattr.empty())
+	m_metaData[cstr_dj_keyorigcharset] = m_dfltInputCharset;
+    else 
+	m_metaData[cstr_dj_keyorigcharset] = m_charsetfromxattr;
+
     m_metaData[cstr_dj_keymt] = cstr_textplain;
 
     size_t srclen = m_text.length();
     m_metaData[cstr_dj_keycontent].swap(m_text);
 
+    // We transcode even if defcharset is supposedly already utf-8:
+    // this validates the encoding.
     // txtdcode() truncates the text if transcoding fails
     (void)txtdcode("mh_text");
 
diff --git a/src/internfile/mh_text.h b/src/internfile/mh_text.h
index cb97aef5..800822fe 100644
--- a/src/internfile/mh_text.h
+++ b/src/internfile/mh_text.h
@@ -56,7 +56,8 @@ private:
     string m_fn;
     off_t  m_offs; // Offset of next read in file if we're paging
     size_t m_pagesz;
-    
+    string m_charsetfromxattr; 
+
     bool readnext();
 };