[PATCH 1/3] msxml3/tests: Add a test for saving XML with not default encoding.

[Dmitry Timoshkov]

Signed-off-by: Dmitry Timoshkov <dmitry(a)baikal.ru>
---
 dlls/msxml3/tests/domdoc.c | 51 ++++++++++++++++++++++++++++++++++++++
 1 file changed, 51 insertions(+)

diff --git a/dlls/msxml3/tests/domdoc.c b/dlls/msxml3/tests/domdoc.c
index e94a2c88a81..6ba88cf0bf7 100644
--- a/dlls/msxml3/tests/domdoc.c
+++ b/dlls/msxml3/tests/domdoc.c
@@ -8522,10 +8522,20 @@ static void test_events(void)
 
 static void test_createProcessingInstruction(void)
 {
+    static const WCHAR xml1[] = L"<?xml version=\"1.0\"?>\r\n<test/>\r\n";
+    static const char xml2[] = "<?xml version=\"1.0\" encoding=\"iso-8859-1\"?>\r\n<test/>\r\n";
+    static const char xml2_vista[] = "<?xml version=\"1.0\" encoding=\"iso-8859-1\"?>\r\n";
     IXMLDOMProcessingInstruction *pi;
     IXMLDOMDocument *doc;
+    IXMLDOMNode *node;
+    IXMLDOMElement *element;
     WCHAR buff[10];
+    BSTR xml;
+    VARIANT var;
     HRESULT hr;
+    IStream *stream;
+    HGLOBAL global;
+    char *p;
 
     doc = create_document(&IID_IXMLDOMDocument);
 
@@ -8538,6 +8548,47 @@ static void test_createProcessingInstruction(void)
 
     IXMLDOMProcessingInstruction_Release(pi);
     IXMLDOMDocument_Release(doc);
+
+    doc = create_document(&IID_IXMLDOMDocument);
+
+    hr = IXMLDOMDocument_createProcessingInstruction(doc, _bstr_("xml"), _bstr_("version=\"1.0\" encoding=\"iso-8859-1\""), &pi);
+    ok(hr == S_OK, "got 0x%08x\n", hr);
+    hr = IXMLDOMProcessingInstruction_QueryInterface(pi, &IID_IXMLDOMNode, (void **)&node);
+    ok(hr == S_OK, "got 0x%08x\n", hr);
+    hr = IXMLDOMDocument_appendChild(doc, node, NULL);
+    ok(hr == S_OK, "got 0x%08x\n", hr);
+    IXMLDOMNode_Release(node);
+    IXMLDOMProcessingInstruction_Release(pi);
+
+    hr = IXMLDOMDocument_createElement(doc, _bstr_("test"), &element);
+    ok(hr == S_OK, "got 0x%08x\n", hr);
+    hr = IXMLDOMDocument_appendChild(doc, (IXMLDOMNode *)element, NULL);
+    ok(hr == S_OK, "got 0x%08x\n", hr);
+    IXMLDOMElement_Release(element);
+
+    hr = IXMLDOMDocument_get_xml(doc, &xml);
+    ok(hr == S_OK, "got 0x%08x\n", hr);
+todo_wine
+    ok(!wcscmp(xml, xml1), "got %s\n", wine_dbgstr_w(xml));
+    SysFreeString(xml);
+
+    hr = CreateStreamOnHGlobal(NULL, TRUE, &stream);
+    ok(hr == S_OK, "got 0x%08x\n", hr);
+    V_VT(&var) = VT_UNKNOWN;
+    V_UNKNOWN(&var) = (IUnknown*)stream;
+    hr = IXMLDOMDocument_save(doc, var);
+    ok(hr == S_OK, "got 0x%08x\n", hr);
+
+    hr = GetHGlobalFromStream(stream, &global);
+    ok(hr == S_OK, "got 0x%08x\n", hr);
+    p = GlobalLock(global);
+    p[GlobalSize(global)] = 0;
+todo_wine
+    ok(!strcmp(p, xml2) || broken(!strcmp(p, xml2_vista)), "got %s\n", wine_dbgstr_a(p));
+    GlobalUnlock(global);
+    IStream_Release(stream);
+
+    IXMLDOMDocument_Release(doc);
 }
 
 static void test_put_nodeTypedValue(void)
-- 
2.30.2

[Dmitry Timoshkov]

What's the status of the first two patches in the series?

-- 
Dmitry.

[Nikolay Sivov]

On 3/24/21 10:42 AM, Dmitry Timoshkov wrote:
> What's the status of the first two patches in the series?

Encoding you used seems to be a reason for broken Vista case. Since this
does not test anything encoding-specific, utf-8 will make it work
without broken() cases. Test itself belongs to test_save().

Regarding 2/3, the reason to unlink originally was to fix duplicated
prologs, occasionally appearing in output. I don't remember details now,
it's possible it was about first loading document, and then inserting
prolog PI. This is again a case where implementing own dumping logic
will help, both with prolog and newlines.

[Dmitry Timoshkov]

Nikolay Sivov <nsivov(a)codeweavers.com> wrote:
> On 3/24/21 10:42 AM, Dmitry Timoshkov wrote:
> > What's the status of the first two patches in the series?
>
> Encoding you used seems to be a reason for broken Vista case. Since this
> does not test anything encoding-specific, utf-8 will make it work
> without broken() cases. Test itself belongs to test_save().

Which encoding would you suggest to use for the test then?
> Regarding 2/3, the reason to unlink originally was to fix duplicated
> prologs, occasionally appearing in output. I don't remember details now,
> it's possible it was about first loading document, and then inserting
> prolog PI.

I guess since the patch in 2/3 doesn't break existing tests then the problem
you mentioned might be fixed already, so it might be safe to assume that
the patch is correct.
> This is again a case where implementing own dumping logic
> will help, both with prolog and newlines.

Then again, the patch could be a light alternative to a full rewrite,
especially if nobody is planning to work on this in near future.

-- 
Dmitry.