[PATCH v5 2/3] mshtml: Try to guess the script encoding when there's no BOM.

[Gabriel Ivăncescu]

From: Gabriel Ivăncescu

Signed-off-by: Gabriel Ivăncescu 
---
 dlls/mshtml/script.c    | 25 +++++++++++++++----------
 dlls/mshtml/tests/dom.c |  1 -
 2 files changed, 15 insertions(+), 11 deletions(-)

diff --git a/dlls/mshtml/script.c b/dlls/mshtml/script.c
index 9d663ee1cf7..065c3568410 100644
--- a/dlls/mshtml/script.c
+++ b/dlls/mshtml/script.c
@@ -963,6 +963,7 @@ typedef struct {

static HRESULT get_binding_text(ScriptBSC *bsc, WCHAR **ret)
 {
+    binding_bom_t bom;
     UINT cp = CP_UTF8;
     WCHAR *text;

@@ -975,8 +976,19 @@ static HRESULT get_binding_text(ScriptBSC *bsc, WCHAR **ret)
         return S_OK;
     }

-    switch(bsc->bsc.bom) {
-    case BOM_UTF16:
+    bom = bsc->bsc.bom;
+    if(bom == BOM_NONE) {
+        /* FIXME: Try to use charset from HTTP headers first */
+
+        /* IE guesses the encoding here using heuristics. Since valid script must start with an
+         * ASCII char (keyword, comment slash, string literal char, etc) that should be enough. */
+        if(bsc->bsc.read > sizeof(WCHAR) && *(WCHAR*)bsc->buf < 128)
+            bom = BOM_UTF16;
+        else
+            cp = get_document_charset(bsc->script_elem->element.node.doc);
+    }
+
+    if(bom == BOM_UTF16) {
         if(bsc->bsc.read % sizeof(WCHAR)) {
             FIXME("The buffer is not a valid utf16 string\n");
             return E_FAIL;
@@ -988,13 +1000,7 @@ static HRESULT get_binding_text(ScriptBSC *bsc, WCHAR **ret)

memcpy(text, bsc->buf, bsc->bsc.read);
         text[bsc->bsc.read/sizeof(WCHAR)] = 0;
-        break;
-
-    default:
-        /* FIXME: Try to use charset from HTTP headers first */
-        cp = get_document_charset(bsc->script_elem->element.node.doc);
-        /* fall through */
-    case BOM_UTF8: {
+    }else {
         DWORD len;

len = MultiByteToWideChar(cp, 0, bsc->buf, bsc->bsc.read, NULL, 0);
@@ -1005,7 +1011,6 @@ static HRESULT get_binding_text(ScriptBSC *bsc, WCHAR **ret)
         MultiByteToWideChar(cp, 0, bsc->buf, bsc->bsc.read, text, len);
         text[len] = 0;
     }
-    }

*ret = text;
     return S_OK;
diff --git a/dlls/mshtml/tests/dom.c b/dlls/mshtml/tests/dom.c
index b09c5ef971f..cd03db02df8 100644
--- a/dlls/mshtml/tests/dom.c
+++ b/dlls/mshtml/tests/dom.c
@@ -12173,7 +12173,6 @@ static void test_default_content_charset(void)

bstr = SysAllocString(L"wineTestProp");
             hres = IHTMLDocument2_GetIDsOfNames(doc, &IID_NULL, &bstr, 1, 0, &dispid);
-            todo_wine_if((doc_charset == utf8 || doc_charset == utf16) && rsrc_charset == utf16)
             ok(hres == S_OK, "[%s:%s] GetIDsOfNames(wineTestProp) returned: %08lx\n", charsets[doc_charset], charsets[rsrc_charset], hres);
             SysFreeString(bstr);