Implement font switching

[wyatt-wong]

I have a document with an image and some text description below. When I replace the text and write to a new PDF file, I found the replaced text becomes a square with an X inside which looks like below, why ?

The original text is "To," and the replaced text is "Ti,". Notice not only "i" was incorrectly changed to a square with an X, the comma that follows the text is also missing.

However, when I used pypdf to read the generated PDF file and then use reader.pages[0].extract_text(), I saw the text was properly replaced as "Ti," without any problems.

[hoehermann]

Thank you for the report. When rendering the text in the document, PDF viewers do not rely on the fonts installed on the computer, but rather use the fonts embedded in the PDF file itself. It is very likely that the font that is being used in that particular location does not contain an "i" glyph because it has been removed for reducing the file size.

The reason that it still works for the text extraction is that the PDF document knows there should be an "i" there, just the PDF viewer cannot render the glyph because it is not available in the font.

It is a very common problem and I try to cover that with warnings, but there are many edge cases and it looks like I missed yours. I should probably add a visual example to the readme.

[wyatt-wong]

I originally replace a whole more text instead of just one character but the same results happen with a square and an x inside. I have use Adobe Acrobat to view the PDF file with the same result.

[hoehermann]

I expanded the readme with a visual explanation of what might be going on. Does that explain your issue?

If not, can you send me the PDF to github@hehoe.de and the replacement you want to perform? Then I can have a more detailed look into the matter.

[wyatt-wong]

If not, can you send me the PDF to github@hehoe.de and the replacement you want do perform? Then I can have a more detailed look into the matter.

Sure.

[hoehermann]

I received your file. Thank you. I identified the problem: Whatever software created that PDF document split one single font into 21 separate resources. The alphabet is distributed all over the place:

• /Ty1 '/AAAAAD+TimesLTMM' -Saf
• /Ty2 '/AAAAAE+TimesLTMM_134_267' t
• /Ty3 '/AAAAAF+TimesLTMM_201_68' ()019o
• /Ty4 '/AAAAAG+TimesLTMM_201_134' Ph
• /Ty5 '/AAAAAH+TimesLTMM_1_1000' 2
• /Ty6 '/AAAAAI+TimesLTMM' (),.BEFKLPRSVYaefgiklorsvwyz
• /Ty7 '/AAAAAJ+TimesLTMM_1_134' mp
• /Ty8 '/AAAAAK+TimesLTMM_1_534' 'Ct
• /Ty9 '/AAAAAL+TimesLTMM_1_267' Wh
• /Ty10 '/AAAAAM+TimesLTMM_1_68' 134d
• /Ty11 '/AAAAAN+TimesLTMM_1_800' t
• /Ty12 '/AAAAAO+TimesLTMM_1_401' h
• /Ty13 '/AAAAAP+TimesLTMM_1_134' r
• /Ty14 '/AAAAAQ+TimesLTMM' :osw
• /Ty15 '/AAAAAR+TimesLTMM_1_401' M
• /Ty16 '/AAAAAS+TimesLTMM_1_467' n
• /Ty17 '/AAAAAT+TimesLTMM_1_867' T
• /Ty18 '/AAAAAU+TimesLTMM_1_201' u
• /Ty19 '/AAAAAV+TimesLTMM_1_1000' DHN
• /Ty20 '/AAAAAW+TimesLTMM_1_1000' 2
• /Ty21 '/AAAAAX+TimesLTMM_1_600' n

This means I cannot simply remove some text and then insert "hello". Before insertion, I must switch to font /Ty12, insert the "h", then switch to font /Ty6, insert "ello". Currently, font switching is not implemented at all. In your case, it looks doable, but I cannot say if I can implement it this year. 😕

[wyatt-wong]

If you can share how I can perform it manually, it will be fine. After all, I only need to do it once. But according to the list, there are only 21 characters from Ty1 to Ty21

[hoehermann]

according to the list, there are only 21 characters from Ty1 to Ty21

The list is supposed to be read the other way around: If you want to insert text, you can use font /Ty6 for most of the letters ( (),.BEFKLPRSVYaefgiklorsvwyz to be specific). But if you want to insert any other letter, you need to switch to one of the other fonts.

In this tool pypdf_strreplace, there currently is no way to add a PDF operation. It can only delete or change existing operations. I have an idea how that could work, but I am not sure. That is why I cannot tell you quickly how to do it.

For a manual approach of editing the document one time, I recommend using an established tool:

• Adobe Pro
• https://code-industry.net/masterpdfeditor/
• https://www.qoppa.com/pdfstudio/
• https://github.com/JakubMelka/PDF4QT ← this one is free and open, and sometimes it even works for editing