| Summary: | copy and pasting asian text problems | ||
|---|---|---|---|
| Product: | Ghostscript | Reporter: | Jack Moffitt <jack> |
| Component: | PDF Writer | Assignee: | Igor Melichev <igor.melichev> |
| Status: | NOTIFIED FIXED | ||
| Severity: | normal | ||
| Priority: | P2 | ||
| Version: | master | ||
| Hardware: | All | ||
| OS: | All | ||
| Customer: | 242 | Word Size: | --- |
|
Description
Jack Moffitt
2005-02-24 09:44:32 UTC
Created attachment 1221 [details]
text.ps
The problem happens due to "PScript5.dll Version 5.2", which originally created the test file. In the table GlyphNames2Unicode (which is formed in the document sections G2UBegin - G2UEnd) it specifies character codes instead glyph names (which should be CIDs with a CID font). Hoewver in the test document CIDs are not equal to character codes - see the embedded CMap named WinCharSetFFFF-H2. Since GlyphNames2Unicode is an undocumented feature, which PScript5.dll uses to communicate with Adobe Distiller, we guess that Microsoft named it inaccurately, so that our old reconstruction of its semantics appears not enough correct. We could patch Ghostscript gdevpdtc.c ln 388 with unicode_char = subfont->procs.decode_glyph(subfont, chr); But we do not know what consequences it will cause in general and we have no technology for testing possible regressions. For now, we'll apply the patch, and will see the consequences. The fix explained in the last comment uppears insufficient. Here is a complete one : Patch to HEAD : http://ghostscript.com/pipermail/gs-cvs/2005-March/005263.html Patch to GS_8_1X : http://ghostscript.com/pipermail/gs-cvs/2005-March/005264.html Note that the test document prints each character several times to emulate an enboldness. Ghostscript/pdfwrite has no workaround against it. Therefore copy&paste multiplies text in 3 times. That is not a part of the current bug. |