patch from Dodji Seketeli about UTF16 BOM when using the push XML parser.

* parserInternals.c: patch from Dodji Seketeli about UTF16 BOM
  when using the push XML parser.
* result/utf16bom.xml result/noent/utf16bom.xml test/utf16bom.xml:
  added the test to the regression suite.
Daniel

ChangeLog

@@ -1,3 +1,10 @@
+Wed Jul 16 23:15:53 CEST 2003 Daniel Veillard <daniel@veillard.com>
+
+	* parserInternals.c: patch from Dodji Seketeli about UTF16 BOM
+	  when using the push XML parser.
+	* result/utf16bom.xml result/noent/utf16bom.xml test/utf16bom.xml:
+	  added the test to the regression suite.
+
 Tue Jul 15 22:03:13 CEST 2003 Daniel Veillard <daniel@veillard.com>
 
 	* globals.c: add xmlThrDefMutex = NULL in xmlCleanupGlobals() 

parserInternals.c

@@ -1621,6 +1621,23 @@ xmlSwitchEncoding(xmlParserCtxtPtr ctxt, xmlCharEncoding enc)
 		ctxt->input->cur += 3;
 	    }
 	    return(0);
+    case XML_CHAR_ENCODING_UTF16LE:
+    case XML_CHAR_ENCODING_UTF16BE:
+        /*The raw input characters are encoded
+         *in UTF-16. As we expect this function
+         *to be called after xmlCharEncInFunc, we expect
+         *ctxt->input->cur to contain UTF-8 encoded characters.
+         *So the raw UTF16 Byte Order Mark
+         *has also been converted into
+         *an UTF-8 BOM. Let's skip that BOM.
+         */
+        if ((ctxt->input != NULL) &&
+            (ctxt->input->cur[0] == 0xEF) &&
+            (ctxt->input->cur[1] == 0xBB) &&
+            (ctxt->input->cur[2] == 0xBF)) {
+            ctxt->input->cur += 3;
+        }
+	break ;
 	default:
 	    break;
     }