From e636fc143efca40a36e0925ac995d9a068ffebfa Mon Sep 17 00:00:00 2001
From: Wolfgang Sourdeau <wsourdeau@inverse.ca>
Date: Tue, 20 Nov 2007 22:18:19 +0000
Subject: [PATCH] Monotone-Parent: 3188fd0271db42fb785221461e9a3655109f23a8
 Monotone-Revision: 962a83d2b39bca57d666c58151e93ea62a7cd0d3

Monotone-Author: wsourdeau@inverse.ca
Monotone-Date: 2007-11-20T22:18:19
Monotone-Branch: ca.inverse.sogo
---
 ChangeLog                                  |  7 ++
 UI/MailPartViewers/GNUmakefile             |  2 +-
 UI/MailPartViewers/UIxMailPartHTMLViewer.m | 81 +++++++++++++++++++++-
 3 files changed, 88 insertions(+), 2 deletions(-)

diff --git a/ChangeLog b/ChangeLog
index 1a4b1c0ac..2f24b0ec7 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,5 +1,12 @@
 2007-11-20  Wolfgang Sourdeau  <wsourdeau@inverse.ca>
 
+	* UI/MailPartViewers/UIxMailPartHTMLViewer.m
+	([_UIxHTMLMailContentHandler
+	-setContentEncoding:newContentEncoding]): new method that
+	specifies an xml charset to return to the parser.
+	([UIxMailPartHTMLViewer -flatContentAsString]): specifies the
+	charset to the content handler before parsing.
+
 	* SoObjects/Appointments/SOGoAppointmentObject.m: don't take an
 	empty c_name as a valid response when looking up an event. Create
 	one in that case instead.
diff --git a/UI/MailPartViewers/GNUmakefile b/UI/MailPartViewers/GNUmakefile
index a86ea4560..8bbcc1c87 100644
--- a/UI/MailPartViewers/GNUmakefile
+++ b/UI/MailPartViewers/GNUmakefile
@@ -40,7 +40,7 @@ MailPartViewers_LOCALIZED_RESOURCE_FILES += \
 
 # make
 
-ADDITIONAL_INCLUDE_DIRS += -I../../SOPE/
+ADDITIONAL_INCLUDE_DIRS += $(shell xml2-config --cflags)
 ADDITIONAL_LIB_DIRS += -L../../SOPE/GDLContentStore/obj/
 
 -include GNUmakefile.preamble
diff --git a/UI/MailPartViewers/UIxMailPartHTMLViewer.m b/UI/MailPartViewers/UIxMailPartHTMLViewer.m
index 6be09e6dd..c93c83b26 100644
--- a/UI/MailPartViewers/UIxMailPartHTMLViewer.m
+++ b/UI/MailPartViewers/UIxMailPartHTMLViewer.m
@@ -22,6 +22,9 @@
 
 #import <Foundation/NSArray.h>
 #import <Foundation/NSDictionary.h>
+#import <Foundation/NSKeyValueCoding.h>
+#import <Foundation/NSValue.h>
+
 #import <SaxObjC/SaxAttributes.h>
 #import <SaxObjC/SaxContentHandler.h>
 #import <SaxObjC/SaxLexicalHandler.h>
@@ -30,6 +33,8 @@
 #import <NGExtensions/NSString+misc.h>
 #import <NGObjWeb/SoObjects.h>
 
+#include <libxml/encoding.h>
+
 #import "UIxMailPartHTMLViewer.h"
 
 #if 0
@@ -49,6 +54,7 @@
   BOOL inCSSDeclaration;
   BOOL hasEmbeddedCSS;
   NSMutableArray *crumb;
+  xmlCharEncoding contentEncoding;
 }
 
 - (NSString *) result;
@@ -65,6 +71,7 @@
       css = nil;
       result = nil;
       attachmentIds = nil;
+      contentEncoding = XML_CHAR_ENCODING_UTF8;
     }
 
   return self;
@@ -78,6 +85,16 @@
   [super dealloc];
 }
 
+- (void) setContentEncoding: (xmlCharEncoding) newContentEncoding
+{
+  contentEncoding = newContentEncoding;
+}
+
+- (xmlCharEncoding) contentEncoding
+{
+  return contentEncoding;
+}
+
 - (void) setAttachmentIds: (NSDictionary *) newAttachmentIds
 {
   attachmentIds = newAttachmentIds;
@@ -463,17 +480,79 @@
   return attachmentIds;
 }
 
+- (xmlCharEncoding) _xmlCharsetForCharset: (NSString *) charset
+{
+  struct { NSString *name; xmlCharEncoding encoding; } xmlEncodings[] = {
+    { @"us-ascii", XML_CHAR_ENCODING_NONE},
+    { @"utf-8", XML_CHAR_ENCODING_UTF8},
+    { @"utf-16le", XML_CHAR_ENCODING_UTF16LE},
+    { @"utf-16be",  XML_CHAR_ENCODING_UTF16BE},
+    { @"ucs-4le", XML_CHAR_ENCODING_UCS4LE},
+    { @"ucs-4be", XML_CHAR_ENCODING_UCS4BE},
+    { @"ebcdic", XML_CHAR_ENCODING_EBCDIC},
+//     { @"iso-10646" , XML_CHAR_ENCODING_UCS4_2143},
+//     {  , XML_CHAR_ENCODING_UCS4_3412},
+//     { @"ucs-2", XML_CHAR_ENCODING_UCS2},
+    { @"iso8859_1", XML_CHAR_ENCODING_8859_1},
+    { @"iso-8859-1", XML_CHAR_ENCODING_8859_1},
+    { @"iso-8859-2",  XML_CHAR_ENCODING_8859_2},
+    { @"iso-8859-3", XML_CHAR_ENCODING_8859_3},
+    { @"iso-8859-4", XML_CHAR_ENCODING_8859_4},
+    { @"iso-8859-5", XML_CHAR_ENCODING_8859_5},
+    { @"iso-8859-6", XML_CHAR_ENCODING_8859_6},
+    { @"iso-8859-7", XML_CHAR_ENCODING_8859_7},
+    { @"iso-8859-8", XML_CHAR_ENCODING_8859_8},
+    { @"iso-8859-9", XML_CHAR_ENCODING_8859_9},
+    { @"iso-2022-jp", XML_CHAR_ENCODING_2022_JP},
+//     { @"iso-2022-jp", XML_CHAR_ENCODING_SHIFT_JIS},
+    { @"euc-jp", XML_CHAR_ENCODING_EUC_JP},
+    { @"us-ascii", XML_CHAR_ENCODING_ASCII}};
+  unsigned count;
+  xmlCharEncoding encoding;
+
+  encoding = XML_CHAR_ENCODING_NONE;
+  count = 0;
+
+  while (encoding == XML_CHAR_ENCODING_NONE
+	 && count < (sizeof (xmlEncodings) / sizeof (xmlEncodings[0])))
+    if ([charset isEqualToString: xmlEncodings[count].name])
+      encoding = xmlEncodings[count].encoding;
+    else
+      count++;
+
+  if (encoding == XML_CHAR_ENCODING_NONE)
+    encoding = XML_CHAR_ENCODING_8859_1;
+
+  return encoding;
+}
+
+- (xmlCharEncoding) _xmlCharEncoding
+{
+
+  NSString *charset;
+
+  charset = [[bodyInfo objectForKey:@"parameterList"]
+	      objectForKey: @"charset"];
+  if (![charset length])
+    charset = @"us-ascii";
+  
+  return [self _xmlCharsetForCharset: [charset lowercaseString]];
+}
+
 - (void) _parseContent
 {
-  id <NSObject, SaxXMLReader> parser;
+  NSObject <SaxXMLReader> *parser;
   NSData *preparsedContent;
 
   preparsedContent = [super decodedFlatContent];
   parser = [[SaxXMLReaderFactory standardXMLReaderFactory]
              createXMLReaderForMimeType: @"text/html"];
+  [parser setValue: [NSNumber numberWithBool: NO]
+	  forKey: @"encodeEntities"];
 
   handler = [_UIxHTMLMailContentHandler new];
   [handler setAttachmentIds: [self _attachmentIds]];
+  [handler setContentEncoding: [self _xmlCharEncoding]];
   [parser setContentHandler: handler];
   [parser parseFromSource: preparsedContent];
 }