June 2022 - Wine-Devel - list.winehq.org

[PATCH v2 3/6] uxtheme: Test DrawThemeBackgroundEx() alpha channel handling.

by Zhiyi Zhang

From: Zhiyi Zhang <zzhang(a)codeweavers.com> Signed-off-by: Zhiyi Zhang <zzhang(a)codeweavers.com> --- dlls/uxtheme/tests/system.c | 133 ++++++++++++++++++++++++++++++++++-- 1 file changed, 129 insertions(+), 4 deletions(-) diff --git a/dlls/uxtheme/tests/system.c b/dlls/uxtheme/tests/system.c index aa491ffb2d2..e9191de1507 100644 --- a/dlls/uxtheme/tests/system.c +++ b/dlls/uxtheme/tests/system.c @@ -2265,10 +2265,139 @@ static void test_EnableThemeDialogTexture(void) static void test_DrawThemeBackgroundEx(void) { + static const int width = 10, height = 10; + static const RECT rect = {0, 0, 10, 10}; + HBITMAP bitmap, old_bitmap; + BOOL transparent, found; + BITMAPINFO bitmap_info; + int i, glyph_type; + BYTE *bits, *ptr; + HTHEME htheme; void *proc; + HDC mem_dc; + HRESULT hr; + HWND hwnd; proc = GetProcAddress(GetModuleHandleA("uxtheme.dll"), MAKEINTRESOURCEA(47)); ok(proc == (void *)pDrawThemeBackgroundEx, "Expected DrawThemeBackgroundEx() at ordinal 47.\n"); + + hwnd = CreateWindowA(WC_STATICA, "", WS_POPUP, 0, 0, 1, 1, 0, 0, 0, NULL); + ok(hwnd != NULL, "CreateWindowA failed, error %#lx.\n", GetLastError()); + htheme = OpenThemeData(hwnd, L"Spin"); + if (!htheme) + { + skip("Theming is inactive.\n"); + DestroyWindow(hwnd); + return; + } + + bitmap_info.bmiHeader.biSize = sizeof(bitmap_info.bmiHeader); + bitmap_info.bmiHeader.biWidth = width; + bitmap_info.bmiHeader.biHeight = height; + bitmap_info.bmiHeader.biPlanes = 1; + bitmap_info.bmiHeader.biBitCount = 32; + bitmap_info.bmiHeader.biCompression = BI_RGB; + bitmap = CreateDIBSection(0, &bitmap_info, DIB_RGB_COLORS, (void **)&bits, 0, 0); + mem_dc = CreateCompatibleDC(NULL); + old_bitmap = SelectObject(mem_dc, bitmap); + + /* Drawing opaque background with transparent glyphs should discard the alpha values from the glyphs */ + transparent = IsThemeBackgroundPartiallyTransparent(htheme, SPNP_UP, UPS_NORMAL); + ok(!transparent, "Expected spin button background opaque.\n"); + hr = GetThemeBool(htheme, SPNP_UP, UPS_NORMAL, TMT_TRANSPARENT, &transparent); + ok(hr == E_PROP_ID_UNSUPPORTED, "Got unexpected hr %#lx.\n", hr); + transparent = FALSE; + hr = GetThemeBool(htheme, SPNP_UP, UPS_NORMAL, TMT_GLYPHTRANSPARENT, &transparent); + ok(hr == S_OK, "Got unexpected hr %#lx.\n", hr); + ok(transparent, "Expected spin button glyph transparent.\n"); + + memset(bits, 0xa5, width * height * sizeof(int)); + hr = DrawThemeBackgroundEx(htheme, mem_dc, SPNP_UP, UPS_NORMAL, &rect, NULL); + ok(hr == S_OK, "Got unexpected hr %#lx.\n", hr); + ptr = bits; + for (i = 0; i < width * height; ++i) + { + if (ptr[3] != 0xff) + break; + + ptr += 4; + } + todo_wine + ok(i == width * height || broken(ptr[3] == 0) /* Spin button glyphs on XP don't use alpha */, + "Unexpected alpha value %#x at (%d,%d).\n", ptr[3], i % height, i / height); + + /* Drawing transparent background without glyphs should keep the alpha values */ + CloseThemeData(htheme); + htheme = OpenThemeData(hwnd, L"Scrollbar"); + transparent = IsThemeBackgroundPartiallyTransparent(htheme, SBP_SIZEBOX, SZB_RIGHTALIGN); + ok(transparent, "Expected scrollbar sizebox transparent.\n"); + transparent = FALSE; + hr = GetThemeBool(htheme, SBP_SIZEBOX, SZB_RIGHTALIGN, TMT_TRANSPARENT, &transparent); + ok(hr == S_OK, "Got unexpected hr %#lx.\n", hr); + ok(transparent, "Expected scrollbar sizebox transparent.\n"); + hr = GetThemeEnumValue(htheme, SBP_SIZEBOX, SZB_RIGHTALIGN, TMT_GLYPHTYPE, &glyph_type); + ok(hr == E_PROP_ID_UNSUPPORTED, "Got unexpected hr %#lx.\n", hr); + + memset(bits, 0xa5, width * height * sizeof(int)); + hr = DrawThemeBackgroundEx(htheme, mem_dc, SBP_SIZEBOX, SZB_RIGHTALIGN, &rect, NULL); + ok(hr == S_OK, "Got unexpected hr %#lx.\n", hr); + found = FALSE; + ptr = bits; + for (i = 0; i < width * height; ++i) + { + if (ptr[3] == 0xa5) + { + found = TRUE; + break; + } + + ptr += 4; + } + ok(found, "Expected alpha values found.\n"); + + /* Drawing transparent background with transparent glyphs should keep alpha values */ + CloseThemeData(htheme); + htheme = OpenThemeData(hwnd, L"Header"); + if (IsThemePartDefined(htheme, HP_HEADERDROPDOWN, 0)) + { + transparent = IsThemeBackgroundPartiallyTransparent(htheme, HP_HEADERDROPDOWN, HDDS_NORMAL); + ok(transparent, "Expected header dropdown transparent.\n"); + transparent = FALSE; + hr = GetThemeBool(htheme, HP_HEADERDROPDOWN, HDDS_NORMAL, TMT_TRANSPARENT, &transparent); + ok(hr == S_OK, "Got unexpected hr %#lx.\n", hr); + ok(transparent, "Expected header dropdown background transparent.\n"); + transparent = FALSE; + hr = GetThemeBool(htheme, HP_HEADERDROPDOWN, HDDS_NORMAL, TMT_GLYPHTRANSPARENT, &transparent); + ok(hr == S_OK, "Got unexpected hr %#lx.\n", hr); + ok(transparent, "Expected header dropdown glyph transparent.\n"); + + memset(bits, 0xa5, width * height * sizeof(int)); + hr = DrawThemeBackgroundEx(htheme, mem_dc, HP_HEADERDROPDOWN, HDDS_NORMAL, &rect, NULL); + ok(hr == S_OK, "Got unexpected hr %#lx.\n", hr); + found = FALSE; + ptr = bits; + for (i = 0; i < width * height; ++i) + { + if (ptr[3] == 0xa5) + { + found = TRUE; + break; + } + + ptr += 4; + } + ok(found, "Expected alpha values found.\n"); + } + else + { + skip("Failed to get header dropdown parts.\n"); + } + + SelectObject(mem_dc, old_bitmap); + DeleteDC(mem_dc); + DeleteObject(bitmap); + CloseThemeData(htheme); + DestroyWindow(hwnd); } START_TEST(system) @@ -2279,10 +2408,6 @@ START_TEST(system) init_funcs(); init_msg_sequences(sequences, NUM_MSG_SEQUENCES); - /* No real functional theme API tests will be done (yet). The current tests - * only show input/return behaviour - */ - test_IsThemed(); test_IsThemePartDefined(); test_GetWindowTheme(); -- GitLab https://gitlab.winehq.org/wine/wine/-/merge_requests/317

June 27, 2022

[PATCH v2 2/6] uxtheme: Place DrawThemeBackgroundEx() at ordinal 47.

by Zhiyi Zhang

From: Zhiyi Zhang <zzhang(a)codeweavers.com> Qt5 applications call DrawThemeBackgroundEx() by ordinal. Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=53066 Signed-off-by: Zhiyi Zhang <zzhang(a)codeweavers.com> --- dlls/uxtheme/tests/system.c | 1 - dlls/uxtheme/uxtheme.spec | 2 +- 2 files changed, 1 insertion(+), 2 deletions(-) diff --git a/dlls/uxtheme/tests/system.c b/dlls/uxtheme/tests/system.c index f169231b214..aa491ffb2d2 100644 --- a/dlls/uxtheme/tests/system.c +++ b/dlls/uxtheme/tests/system.c @@ -2268,7 +2268,6 @@ static void test_DrawThemeBackgroundEx(void) void *proc; proc = GetProcAddress(GetModuleHandleA("uxtheme.dll"), MAKEINTRESOURCEA(47)); - todo_wine ok(proc == (void *)pDrawThemeBackgroundEx, "Expected DrawThemeBackgroundEx() at ordinal 47.\n"); } diff --git a/dlls/uxtheme/uxtheme.spec b/dlls/uxtheme/uxtheme.spec index 140388245f2..c60254b677d 100644 --- a/dlls/uxtheme/uxtheme.spec +++ b/dlls/uxtheme/uxtheme.spec @@ -36,6 +36,7 @@ 44 stub -noname ClassicSystemParametersInfoA 45 stub -noname ClassicSystemParametersInfoW 46 stub -noname ClassicAdjustWindowRectEx +47 stdcall DrawThemeBackgroundEx(ptr ptr long long ptr ptr) 48 stub -noname GetThemeParseErrorInfo 60 stub -noname CreateThemeDataFromObjects 61 stdcall OpenThemeDataEx(ptr wstr long) @@ -53,7 +54,6 @@ @ stdcall BufferedPaintUnInit() @ stdcall CloseThemeData(ptr) @ stdcall DrawThemeBackground(ptr ptr long long ptr ptr) -@ stdcall DrawThemeBackgroundEx(ptr ptr long long ptr ptr) @ stdcall DrawThemeEdge(ptr ptr long long ptr long long ptr) @ stdcall DrawThemeIcon(ptr ptr long long ptr ptr long) @ stdcall DrawThemeParentBackground(ptr ptr ptr) -- GitLab https://gitlab.winehq.org/wine/wine/-/merge_requests/317

June 27, 2022

[PATCH v2 1/6] uxtheme/tests: Test that DrawThemeBackgroundEx() is at ordinal 47.

by Zhiyi Zhang

From: Zhiyi Zhang <zzhang(a)codeweavers.com> Signed-off-by: Zhiyi Zhang <zzhang(a)codeweavers.com> --- dlls/uxtheme/tests/system.c | 12 ++++++++++++ 1 file changed, 12 insertions(+) diff --git a/dlls/uxtheme/tests/system.c b/dlls/uxtheme/tests/system.c index 1f04ac879ec..f169231b214 100644 --- a/dlls/uxtheme/tests/system.c +++ b/dlls/uxtheme/tests/system.c @@ -40,6 +40,7 @@ static HTHEME (WINAPI * pOpenThemeDataEx)(HWND, LPCWSTR, DWORD); static HTHEME (WINAPI *pOpenThemeDataForDpi)(HWND, LPCWSTR, UINT); static HPAINTBUFFER (WINAPI *pBeginBufferedPaint)(HDC, const RECT *, BP_BUFFERFORMAT, BP_PAINTPARAMS *, HDC *); static HRESULT (WINAPI *pBufferedPaintClear)(HPAINTBUFFER, const RECT *); +static HRESULT (WINAPI *pDrawThemeBackgroundEx)(HTHEME, HDC, int, int, const RECT *, const DTBGOPTS *); static HRESULT (WINAPI *pEndBufferedPaint)(HPAINTBUFFER, BOOL); static HRESULT (WINAPI *pGetBufferedPaintBits)(HPAINTBUFFER, RGBQUAD **, int *); static HDC (WINAPI *pGetBufferedPaintDC)(HPAINTBUFFER); @@ -80,6 +81,7 @@ static void init_funcs(void) GET_PROC(uxtheme, BeginBufferedPaint) GET_PROC(uxtheme, BufferedPaintClear) GET_PROC(uxtheme, EndBufferedPaint) + GET_PROC(uxtheme, DrawThemeBackgroundEx) GET_PROC(uxtheme, GetBufferedPaintBits) GET_PROC(uxtheme, GetBufferedPaintDC) GET_PROC(uxtheme, GetBufferedPaintTargetDC) @@ -2261,6 +2263,15 @@ static void test_EnableThemeDialogTexture(void) UnregisterClassA(cls.lpszClassName, GetModuleHandleA(NULL)); } +static void test_DrawThemeBackgroundEx(void) +{ + void *proc; + + proc = GetProcAddress(GetModuleHandleA("uxtheme.dll"), MAKEINTRESOURCEA(47)); + todo_wine + ok(proc == (void *)pDrawThemeBackgroundEx, "Expected DrawThemeBackgroundEx() at ordinal 47.\n"); +} + START_TEST(system) { ULONG_PTR ctx_cookie; @@ -2287,6 +2298,7 @@ START_TEST(system) test_GetThemeIntList(); test_GetThemeTransitionDuration(); test_DrawThemeParentBackground(); + test_DrawThemeBackgroundEx(); if (load_v6_module(&ctx_cookie, &ctx)) { -- GitLab https://gitlab.winehq.org/wine/wine/-/merge_requests/317

June 27, 2022

[PATCH v2 0/6] MR317: uxtheme/tests: Test that DrawThemeBackgroundEx() is at ordinal 47.

by Zhiyi Zhang (＠zhiyi)

Signed-off-by: Zhiyi Zhang <zzhang(a)codeweavers.com> -- v2: uxtheme: Handle BT_NONE in GetThemeBackgroundRegion(). uxtheme/tests: Add GetThemeBackgroundRegion() tests. uxtheme: Do not transfer glyph alpha values when drawing opaque background. uxtheme: Test DrawThemeBackgroundEx() alpha channel handling. uxtheme: Place DrawThemeBackgroundEx() at ordinal 47. uxtheme/tests: Test that DrawThemeBackgroundEx() is at ordinal 47. https://gitlab.winehq.org/wine/wine/-/merge_requests/317

June 27, 2022

Re: [PATCH v6 0/1] MR296: nsiproxy.sys: Add static ARP entries which are always present on Windows. - approved

by Alexandre Julliard (＠julliard)

This merge request was approved by Alexandre Julliard. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/296

June 27, 2022

Re: [PATCH v3 0/5] MR307: dnsapi: Fix wow64 translation. - approved

by Alexandre Julliard (＠julliard)

This merge request was approved by Alexandre Julliard. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/307

June 27, 2022

Re: Interest in integrating with HarfBuzz shaper?

by Nikolay Sivov

On 6/24/22 20:29, Behdad Esfahbod wrote: > Hi, > > I've been looking at the internal Wine implementation of usp10.dll, and > wondering if there is an interest to integrate with HarfBuzz instead / > optionally. Hi, Behdad. Uniscribe implementation hasn't been touched in a while. DirectWrite implementation currently supports all types of GSUB/GPOS lookups, uses structure and code taken from harfbuzz wrt to feature masks, loop organization, sorting, etc. Currently it only implements Arabic shaping. My longer term plan was to move all that to a separate textshaping.dll that modern Windows releases are using, and link to it from both gdi32 and dwrite, after removing existing uniscribe code. We don't need textshaping.dll interface to be compatible with Windows, because it's undocumented and isn't used by applications directly. > > The main hindrance is that HarfBuzz does NOT currently provide separate > substitution vs positioning steps ala ScriptShape() and ScriptPosition(). > But I've been thinking about how to make that possibly happen, at least > enough for Wine's needs. Yes, I believe we talked about that some time (few years?) ago. Uniscribe's ScriptShape() and dwrite's GetGlyphs() are using very similar interface, and we'll need DWRITE_SHAPING_TEXT_PROPERTIES and DWRITE_SHAPING_GLYPH_PROPERTIES to be compatible with what Windows returns. When I was working on this I found what some of the "reserved" fields mean in SHAPING_GLYPH_PROPERTIES, you can find structure definition in dlls/dwrite/dwrite_private.h:struct shaping_glyph_properties. I'm talking about last two fields "components" and "lig_component", that are important for GPOS handling. Another field that harfbuzz does not provide I think is "DWRITE_SHAPING_GLYPH_PROPERTIES.justification" mask, which I believe is only useful for Arabic shaping, and roughly corresponds to character classes. So technically, if we get correct ScriptShape()-style output from harfbuzz, including correct clustermap in WCHAR terms, we don't really need harfbuzz to produce ScriptPosition() output, because we have that implemented, and afaik it does not require script-specific handling, except maybe for features that have to be enabled. Another consideration is that we'll likely have to import harfbuzz code, and build it as a static win32 library. That's not a decision I can make. I think the only build requirement would be a recent enough C++ compiler? And maybe ragel, at some point you had precompiled header in the tree. We'll have to ask Alexandre's opinion here. We could also benefit from harfbuzz subsetter, and use it for fontsub and for Direct2D XPS printing. > > Regards, > > behdad > http://behdad.org/ >

June 27, 2022

Re: [PATCH v5 4/7] mfplat/buffer: Deduplicate common code between Lock2D() and Lock2DSize().

by Nikolay Sivov

On 6/24/22 10:48, Giovanni Mascellani wrote: > > -static HRESULT WINAPI dxgi_surface_buffer_Lock2D(IMF2DBuffer2 *iface, BYTE **scanline0, LONG *pitch) > +static HRESULT dxgi_surface_buffer_lock2d(struct buffer *buffer, MF2DBuffer_LockFlags flags, > + BYTE **scanline0, LONG *pitch, BYTE **buffer_start, DWORD *buffer_length) > { > - struct buffer *buffer = impl_from_IMF2DBuffer2(iface); > HRESULT hr = S_OK; > > - TRACE("%p, %p, %p.\n", iface, scanline0, pitch); > - > - if (!scanline0 || !pitch) > - return E_POINTER; > - > - EnterCriticalSection(&buffer->cs); > - > if (buffer->_2d.linear_buffer) > hr = MF_E_UNEXPECTED; > else if (!buffer->_2d.locks++) > @@ -1051,6 +1042,10 @@ static HRESULT WINAPI dxgi_surface_buffer_Lock2D(IMF2DBuffer2 *iface, BYTE **sca > { > *scanline0 = buffer->dxgi_surface.map_desc.pData; > *pitch = buffer->dxgi_surface.map_desc.RowPitch; > + if (buffer_start) > + *buffer_start = *scanline0; > + if (buffer_length) > + *buffer_length = buffer->dxgi_surface.map_desc.RowPitch * buffer->_2d.height; > } > > LeaveCriticalSection(&buffer->cs); > @@ -1058,6 +1053,25 @@ static HRESULT WINAPI dxgi_surface_buffer_Lock2D(IMF2DBuffer2 *iface, BYTE **sca > return hr; > } You forgot to remove LeaveCriticalSection() here. That's actually visible in the test output when running on Wine. For consistency between d3d9 and dxgi cases, let's pick one suffix - _lock2d() or _lock().

June 27, 2022

Re: [PATCH v5 3/7] mfplat/buffer: Fix the image copy function for IMC2/IMC4 buffers with odd height.

by Nikolay Sivov

Signed-off-by: Nikolay Sivov <nsivov(a)codeweavers.com>

June 27, 2022

Re: [PATCH v5 2/7] mfplat/tests: Use a more unique pattern when testing image formats.

by Nikolay Sivov

Signed-off-by: Nikolay Sivov <nsivov(a)codeweavers.com>

June 27, 2022