From b6da0ae136d962d49a9f9bfc37ef232adc7460ba Mon Sep 17 00:00:00 2001
From: Matt Farina <matt@mattfarina.com>
Date: Wed, 29 May 2013 17:13:24 -0400
Subject: Moved mbstring encoding to be used before iconv. Tests passing on PHP
 5.4.

---
 src/HTML5/Parser/UTF8Utils.php | 31 +++++++++++++++++++++----------
 1 file changed, 21 insertions(+), 10 deletions(-)

(limited to 'src/HTML5/Parser')

diff --git a/src/HTML5/Parser/UTF8Utils.php b/src/HTML5/Parser/UTF8Utils.php
index 022d628..974a670 100644
--- a/src/HTML5/Parser/UTF8Utils.php
+++ b/src/HTML5/Parser/UTF8Utils.php
@@ -86,7 +86,27 @@ class UTF8Utils {
     could not be converted to Unicode characters must be converted
     to U+FFFD REPLACEMENT CHARACTER code points. */
 
-    if (function_exists('iconv') && $encoding != 'auto') {
+    // mb_convert_encoding is chosen over iconv because of a bug. The best
+    // details for the bug are on http://us1.php.net/manual/en/function.iconv.php#108643
+    // which contains links to the actual but reports as well as work around
+    // details.
+    if (function_exists('mb_convert_encoding')) {
+      // mb library has the following behaviors:
+      // - UTF-16 surrogates result in FALSE.
+      // - Overlongs and outside Plane 16 result in empty strings.
+      
+      // Before we run mb_convert_encoding we need to tell it what to do with
+      // characters it does not know. This could be different than the parent
+      // application executing this library so we store the value, change it
+      // to our needs, and then change it back when we are done. This feels
+      // a little excessive and it would be great if there was a better way.
+      $save = ini_get('mbstring.substitute_character');
+      ini_set('mbstring.substitute_character', "none");
+      $data = mb_convert_encoding($data, 'UTF-8', $encoding);
+      ini_set('mbstring.substitute_character', $save);
+    }
+    // @todo Get iconv running in at least some environments if that is possible.
+    elseif (function_exists('iconv') && $encoding != 'auto') {
       // fprintf(STDOUT, "iconv found\n");
       // iconv has the following behaviors:
       // - Overlong representations are ignored.
@@ -94,15 +114,6 @@ class UTF8Utils {
       // - Incomplete sequences generate a warning.
       $data = @iconv($encoding, 'UTF-8//IGNORE', $data);
     }
-    // MPB: Testing the newer mb_convert_encoding(). This might need
-    // to be removed again.
-    elseif (function_exists('mb_convert_encoding')) {
-      fprintf(STDOUT, "MB found\n");
-      // mb library has the following behaviors:
-      // - UTF-16 surrogates result in FALSE.
-      // - Overlongs and outside Plane 16 result in empty strings.
-      $data = mb_convert_encoding($data, 'UTF-8', $encoding);
-    }
     else {
       // we can make a conforming native implementation
       throw new Exception('Not implemented, please install mbstring or iconv');
-- 
cgit v1.2.3