msxml3: Switch parser encoding manually when it won't be able to detect it (UTF-16 case).

dlls/msxml3/saxreader.c

@@ -2144,7 +2144,7 @@ static HRESULT internal_parseBuffer(saxreader *This, const char *buffer, int siz
     HRESULT hr;
 
     hr = SAXLocator_create(This, &locator, vbInterface);
-    if(FAILED(hr))
+    if (FAILED(hr))
         return hr;
 
     if (size >= 4)
@@ -2163,22 +2163,44 @@ static HRESULT internal_parseBuffer(saxreader *This, const char *buffer, int siz
         }
     }
 
+    /* if libxml2 detection failed try to guess */
+    if (encoding == XML_CHAR_ENCODING_NONE)
+    {
+        const WCHAR *ptr = (WCHAR*)buffer;
+        /* xml declaration with possibly specfied encoding will be still handled by parser */
+        if ((size >= 2) && *ptr == '<' && ptr[1] != '?')
+        {
+            enc_name = (xmlChar*)xmlGetCharEncodingName(XML_CHAR_ENCODING_UTF16LE);
+            encoding = XML_CHAR_ENCODING_UTF16LE;
+        }
+    }
+    else if (encoding == XML_CHAR_ENCODING_UTF8)
+        enc_name = (xmlChar*)xmlGetCharEncodingName(encoding);
+    else
+        enc_name = NULL;
+
     locator->pParserCtxt = xmlCreateMemoryParserCtxt(buffer, size);
-    if(!locator->pParserCtxt)
+    if (!locator->pParserCtxt)
     {
         ISAXLocator_Release(&locator->ISAXLocator_iface);
         return E_FAIL;
     }
 
-    if (encoding == XML_CHAR_ENCODING_UTF8)
+    if (enc_name)
+    {
         locator->pParserCtxt->encoding = xmlStrdup(enc_name);
+        if (encoding == XML_CHAR_ENCODING_UTF16LE) {
+            TRACE("switching to %s\n", enc_name);
+            xmlSwitchEncoding(locator->pParserCtxt, encoding);
+        }
+    }
 
     xmlFree(locator->pParserCtxt->sax);
     locator->pParserCtxt->sax = &locator->saxreader->sax;
     locator->pParserCtxt->userData = locator;
 
     This->isParsing = TRUE;
-    if(xmlParseDocument(locator->pParserCtxt)==-1 && locator->ret==S_OK)
+    if(xmlParseDocument(locator->pParserCtxt) == -1 && locator->ret == S_OK)
         hr = E_FAIL;
     else
         hr = locator->ret;

dlls/msxml3/tests/saxreader.c

@@ -2317,18 +2317,20 @@ static const struct enc_test_entry_t encoding_test_data[] = {
     { 0 }
 };
 
-static void test_encoding(void)
+static void test_saxreader_encoding(void)
 {
     const struct enc_test_entry_t *entry = encoding_test_data;
     static const WCHAR testXmlW[] = {'t','e','s','t','.','x','m','l',0};
     static const CHAR testXmlA[] = "test.xml";
-    ISAXXMLReader *reader;
-    DWORD written;
-    HANDLE file;
-    HRESULT hr;
 
     while (entry->guid)
     {
+        ISAXXMLReader *reader;
+        VARIANT input;
+        DWORD written;
+        HANDLE file;
+        HRESULT hr;
+
         hr = CoCreateInstance(entry->guid, NULL, CLSCTX_INPROC_SERVER, &IID_ISAXXMLReader, (void**)&reader);
         if (hr != S_OK)
         {
@@ -2349,8 +2351,16 @@ static void test_encoding(void)
             ok(hr == entry->hr, "Expected 0x%08x, got 0x%08x. CLSID %s\n", entry->hr, hr, entry->clsid);
 
         DeleteFileA(testXmlA);
+
+        /* try BSTR input with no BOM or '<?xml' instruction */
+        V_VT(&input) = VT_BSTR;
+        V_BSTR(&input) = _bstr_("<element></element>");
+        hr = ISAXXMLReader_parse(reader, input);
+        EXPECT_HR(hr, S_OK);
+
         ISAXXMLReader_Release(reader);
 
+        free_bstrs();
         entry++;
     }
 }
@@ -4474,7 +4484,7 @@ START_TEST(saxreader)
     test_saxreader();
     test_saxreader_properties();
     test_saxreader_features();
-    test_encoding();
+    test_saxreader_encoding();
     test_dispex();
 
     /* MXXMLWriter tests */