10 Replies Latest reply: May 12, 2014 9:40 AM by Test Screen Name

Is it a bug of Distiller?

PigPigPig May 10, 2014 10:42 AM

I printed a Microsoft Word file as a PDF file by Distiller with an embed SakkalMajalla truetype font. I want to extract unicode texts from the PDF file. I found ToUnicode misses part of mapping. For example, CID 06B4 doen't have any mapping. I guess 06b4 should be mapped to U+0644. There are some substitutions in SakkalMajalla. So uni0644.medi (U+FEE0) is replaced by liga.0758.medi.alt1 (U+10354). Why can't Distiller deal with the situation? How can I recover missed mapping from PDF objects except ToUnicode? Thanks

/GS1 gs

/TT1 1 Tf

24 0 0 24 513.84 764.1203 Tm

0 g

0 Tc

0 Tw

<0284>Tj

.495 .5925 TD

<0551>Tj

-.1675 -.5925 TD

<06b4>Tj

.4 .4225 TD

<0551>Tj

-.12 -.4225 TD

<024f>Tj

/TT2 1 Tf

12 0 0 12 506.58 764.1203 Tm

( )Tj

/CIDInit /ProcSet findresource begin 12 dict begin begincmap /CIDSystemInfo <<

/Registry (JJEELB+TT1+0) /Ordering (T42UV) /Supplement 0 >> def

/CMapName /JJEELB+TT1+0 def

/CMapType 2 def

1 begincodespacerange <024f> <0551> endcodespacerange

3 beginbfchar

<024f> <0639>

<0284> <0649>

<0551> <064E>

endbfchar

endcmap CMapName currentdict /CMap defineresource pop end end

1. Re: Is it a bug of Distiller?

MichaelKazlow May 11, 2014 1:26 PM (in response to PigPigPig)

Moved to https://forums.adobe.com/community/acrobat/creating__editing_%26_exporting_pdfs
Like Show 0 Likes (0)

Actions
2. Re: Is it a bug of Distiller?

Bill@VT May 11, 2014 6:14 PM (in response to PigPigPig)

Are the fonts embedded in the PDF? When you extract text, you generally have to have the fonts on your system. However, if you created the PDF from WORD, why not just go to the WORD document. That is much more effective as editing and extraction issues with PDFs is not normally recommended. Have you tried to save the PDF as a DOCX file and see if that gives you want you want?
Like Show 0 Likes (0)

Actions
3. Re: Is it a bug of Distiller?

PigPigPig May 11, 2014 6:54 PM (in response to Bill@VT)

I am sure I have the font in my system and the font is embedded in the PDF. I had tried to save the PDF as a DOCX file and it is different from the PDF. Some characters miss. I am a software engineer. I found parts of Arabic PDF files can't be correctly extracted. I tried to debug them and reproduce the bug. I guess its cmap is wrong or is incomplete so that we can't correctly extract its texts. I don't know why Adobe Acrobat generates an incomplete cmap.
Like Show 0 Likes (0)

Actions
4. Re: Is it a bug of Distiller?

PigPigPig May 12, 2014 2:35 AM (in response to PigPigPig)

Shall I move the question to PDF Language and Specifications?
Like Show 0 Likes (0)

Actions
5. Re: Is it a bug of Distiller?

Test Screen Name May 12, 2014 5:06 AM (in response to PigPigPig)

I cannot find liga.0758.medi.alt1 (U+10354) in any Unicode chart or discussion. These do not seem to be part of the Unicode standard.
Like Show 0 Likes (0)

Actions
6. Re: Is it a bug of Distiller?

PigPigPig May 12, 2014 8:11 AM (in response to Test Screen Name)

The light yellow part is liga.0758.medi.alt1. It is not a standard unicode. However it is a substitution of standard unicode 0644. I am not an expert on font. I can't explain more.
Like Show 0 Likes (0)

Actions
7. Re: Is it a bug of Distiller?

Test Screen Name May 12, 2014 8:19 AM (in response to PigPigPig)

It may not be working because it is not standard Unicode. Perhaps Distiller has no ability or wish to map to private Unicode ranges, because this has little value in text extraction. I believe it uses only standard Unicode characters, so the choice of font for text extraction is not crucial.
Like Show 0 Likes (0)

Actions
8. Re: Is it a bug of Distiller?

PigPigPig May 12, 2014 8:54 AM (in response to Test Screen Name)

I am not familiar with opentype or truetype font. Maybe what you said is a reason. However standard unicode FEE0 is also a substitution of standard unicode 0644. Why does adobe acrobat choose U+10354 instead of U+FEE0, if Distiller has no ability to map to private Unicode ranges?
Like Show 0 Likes (0)

Actions
9. Re: Is it a bug of Distiller?

PigPigPig May 12, 2014 9:11 AM (in response to Test Screen Name)

I found not only Sakkal Majalla truetype font but also Arabic Typesetting truetype font defines glyphs in private unicode ranges. I guess a lot of fonts use private unicode ranges. If Distiller can't deal with the problem, it should be a big problem in extracting texts from PDF files.
Like Show 0 Likes (0)

Actions
10. Re: Is it a bug of Distiller?

Test Screen Name May 12, 2014 9:40 AM (in response to PigPigPig)

It seems to me that, yes, that is a problem. Text extraction is just a question of getting out Unicode values. If fonts use private ranges I don't see that it can work.
Like Show 0 Likes (0)

Actions