Re: [ft-devel] PDF text rendering error

This is compilable via command line?

Johnson Y. Yan wrote:

> Toshiya-san，

> please check again, thanks.

> Johnson Y. Yan

> 发件人： suzuki toshiya

> 发送时间： 2012-06-13 00:18:46

> 收件人： Johnson Y. Yan

> 抄送： freetype-devel

> 主题： Re: [ft-devel] PDF text rendering error

> Hmm. Can you post the C source code as an attachment?

> Yet I've not checked, but I'm not sure your code is compilable

> because no header files are included. If you make the compiler

> include the header files by compiler flags written in project file,

> please post project file too.

> Regards,

> mpsuzuki

> Johnson Y. Yan wrote:

>> Toshiya-san，

>> thanks for your patience,

>> I build freetype 2.4.9 with vc6，I merged your patch and make sure this is a tricky font.

>> I can debug into：

>> tt_check_trickyness( FT_Face face )

>> {

>> if ( !face )

>> return FALSE;

>> /* For first, check the face name for quick check. */

>> if ( face->family_name &&

>> tt_check_trickyness_family( face->family_name ) )

>> return TRUE;

>> /* Type42 fonts may lack `name' tables, we thus try to identify */

>> /* tricky fonts by checking the checksums of Type42-persistent */

>> /* sfnt tables (`cvt', `fpgm', and `prep'). */

>> if ( tt_check_trickyness_sfnt_ids( (TT_Face)face ) )

>> return TRUE;

>> return FALSE;

>> }

>> and tt_check_trickyness_sfnt_ids return TRUE;

>> but render result still error(I sent in 11/6/2012 ).I don't know what reason is.

>> please see my test code, you can use the embeded font 1_BKLGIB+FA.ttf i sent yesterday to test:

>> extern "C" int g_bRecordBI = 0;

>> DWORD gid = 2134;

>> void CTestDlg::OnTestFreetype()

>> {

>> UpdateData();

>> CModuleMgr mm;

>> CString csFilter(_T("OpenType (*.ttf,*.ttc,*.otf)|*.ttf;*.ttc;*.otf|"

>> "Type1 (*.pfa,*.pfb)|*.pfa;*.pfb|"

>> "CFF (*.cff)|*.cff|"

>> "All files (*.*)|*.*||"));

>> CFileDialog fd(TRUE, NULL, NULL, OFN_FILEMUSTEXIST|OFN_LONGNAMES|OFN_PATHMUSTEXIST, csFilter, this);

>> if (fd.DoModal() == IDOK){

>> FT_Library library;

>> FT_Init_FreeType(&library);

>> CString csFile = fd.GetPathName();

>> IFileRead *pFileRead = CreateFileRead((LPCTSTR)csFile);

>> DWORD dwFileSize = (DWORD)pFileRead->GetSize();

>> BYTE* pBuffer = Alloc(BYTE, dwFileSize);

>> pFileRead->ReadBlock(pBuffer, 0, dwFileSize);

>> pFileRead->Release();

>> FT_Face face;

>> int error = FT_New_Memory_Face(library, pBuffer, dwFileSize, 0, &face);

>> ASSERT(error == 0);

>> error = FT_Set_Pixel_Sizes(face, 0, 64);

>> ASSERT(error == 0);

>> FT_Matrix matrix;

>> matrix.xx = (long)(32.f /64 * 65536);//22215;//32.f /64 * 65536;

>> matrix.xy = 0;

>> matrix.yx = 0;

>> matrix.yy = (long)(32.f /64 * 65536);//23616;//32.f /64 * 65536;

>> FT_Set_Transform(face, &matrix, NULL);

>> ASSERT(error == 0);

>> int load_flags = FT_LOAD_NO_BITMAP ;

>> DEBUG1("CTestDlg::OnTestFreetype -- Start Load Glyph ----------------------------------");

>> g_bRecordBI = 1; // Change flag

>> error = FT_Load_Glyph(face, gid, load_flags);

>> g_bRecordBI = 0;

>> DEBUG1("CTestDlg::OnTestFreetype -- End Load Glyph ----------------------------------");

>> ASSERT(error == 0);

>> int render_mode = FT_RENDER_MODE_NORMAL;

>> error = FT_Render_Glyph(face, render_mode);

>> ASSERT(error == 0);

>> FT_GlyphSlot slot = face->glyph;

>> int left = slot->bitmap_left;

>> int top = slot->bitmap_top;

>> int pixel_width = slot->bitmap.width;

>> int pixel_height = slot->bitmap.rows;

>> int stride = slot->bitmap.pitch;

>> int type = slot->bitmap.pixel_mode;

>> BYTE*buffer = slot->bitmap.buffer;

>> CFX_DIBitmap bitmap;

>> BOOL ret = bitmap.Create(pixel_width, pixel_height, FXDIB_8bppRgb);

>> ASSERT(ret);

>> int row, col;

>> for (row = 0; row < pixel_height; row ++) {

>> BYTE* pDestBuf = (BYTE*)bitmap.GetScanline(row);

>> BYTE* pSrcBuf= buffer+row*stride;

>> for (col = 0; col < pixel_width; col ++) {

>> *pDestBuf++ = ~*pSrcBuf++;

>> }

>> CDC *pDC = GetDC();

>> CFX_WindowsDevice wd(pDC->GetSafeHdc());

>> wd.SetDIBits(&bitmap, 0, 0);

>> ReleaseDC(pDC);

>> FT_Done_Face(face);

>> FT_Done_FreeType(library);

>> if (pBuffer) Free(pBuffer);

>> }

>> Johnson Y. Yan

>> 发件人： suzuki toshiya

>> 发送时间： 2012-06-12 18:33:24

>> 收件人： Johnson Y. Yan

>> 抄送： freetype-devel

>> 主题： Re: [ft-devel] PDF text rendering error

>> OK, please let us work with same testing program.

>> I guess you're working for the proprietary software

>> and you cannot disclose the source to me (and I think

>> I don't have appropriate developer environment even

>> if you can disclose the source).

>> Thus, I will make a patch generating many trace

>> messages, and a small CUI program just checking

>> if the font is found to be tricky. Is it acceptable?

>> Please check if you can build FT2 library with

>> debug/trace functions, while I'm writing a patch

>> for debugging.

>> Regards,

>> mpsuzuki

>> Johnson Y. Yan wrote:

>>> Toshiya-san,

>>> I apply your patch, but still render error, : (.

>>> If the way i call the function is wrong?

>>> FT_Load_Glyph(FtFace, gid, FT_LOAD_NO_BITMAP).

>>>

>>> Johnson Y. Yan

>>>

>>> 发件人： suzuki toshiya

>>> 发送时间： 2012-06-11 16:38:11

>>> 收件人： Johnson Y. Yan

>>> 抄送： freetype-devel

>>> 主题： Re: [ft-devel] PDF text rendering error

>>>

>>> Attached is my proposal to fix this issue, please try.

>>> In my case (Debian GNU/Linux xpdf-3.02-11), using the

>>> patched freetype library by LD_PRELOAD, xpdf shows

>>> readable result (without patch, the result is unreadable).

>>> Regards,

>>> mpsuzuki

>>> suzuki toshiya wrote:

>>>> Johnson Y. Yan wrote (2012/06/11 16:03):

>>>>> Werner and Toshiya-san,

>>>>>

>>>>> I put the font as a tricky font to deal with.

>>>>> static const char trick_names[TRICK_NAMES_COUNT]

>>>>> [TRICK_NAMES_MAX_CHARACTERS + 1] =

>>>>> {

>>>>> "DFKaiSho-SB", /* dfkaisb.ttf */

>>>>> "DFKaiShu",

>>>>> "DFKai-SB", /* kaiu.ttf */

>>>>> "HuaTianKaiTi?", /* htkt2.ttf */

>>>>> "HuaTianSongTi?", /* htst3.ttf */

>>>>> "MingLiU", /* mingliu.ttf & mingliu.ttc */

>>>>> "PMingLiU", /* mingliu.ttc */

>>>>> "MingLi43", /* mingli.ttf */

>>>>> "BKLGIB+FA" ---------------------------------------- this is the

>>>>> embeded font in th document.

>>>>> };

>>>>> it still render error. .

>>>> Including the randomization prefix to the tricky font name list is bad

>>>> idea.

>>>>

>>>>> I opened the embeded font use font lab, found the gid = 2134 is like

>>>>> this, please see the attach bitmap,

>>>>> some of lines form a 0 area areas, then fill the area(have no area),

>>>>> Does freetype made special processing?

>>>>>

>>>>> Johnson Y. Yan

>>>>>

>>>>> 发件人： Werner LEMBERG

>>>>> 发送时间： 2012-06-11 12:55:42

>>>>> 收件人： yinsen_yan

>>>>> 抄送： freetype-devel; mpsuzuki

>>>>> 主题： Re: PDF text rendering error

>>>>>

>>>>>> Do you mean the font must be rendered use

>>>>>> TT_CONFIG_OPTION_BYTECODE_INTERPRETER?

>>>>> No, the FreeType *library* must be built with it: Edit ftoption.h to

>>>>> set this flag, then recompile.

>>>>> Additionally, hinting must be enabled for this font, this is,

>>>>> FT_LOAD_NO_SCALE doesn't work here.

>>>>> Normally, those steps are (more or less) handled automatically by

>>>>> FreeType, since we maintain a list of `tricky' font like the one you

>>>>> have encountered in the PDF; see file `ttobjs.c'. The subsetted font

>>>>> in question, however, which identifies itself as

>>>>> BKLGIB+FA-RoundGothicB

>>>>> ＦＡ丸ゴシックＢ and FA-RoundGothicB are trademarks of NEC Office

>>>>> Systems,Ltd.

>>>>> isn't in this list; for the NEC FA fonts (and others), a second

>>>>> mechanism is used which doesn't work with subsetted fonts.

>>>>> Toshiya-san, any ideas?

>>>>> Werner

>>>> _______________________________________________

>>>> Freetype-devel mailing list

>>>> address@hidden

>>>> https://lists.nongnu.org/mailman/listinfo/freetype-devel

From:	Johnson Y. Yan
Subject:	Re: [ft-devel] PDF text rendering error
Date:	Wed, 13 Jun 2012 15:02:03 +0800