'add_redact_annot' asking for 'need font file or buffer' for an existing font in PDF

[santanuOUP]

Description of the bug

Trying to replace one text in a PDF using the following methods:
add_redact_annot
apply_redactions

Text replacement is done but the replaced text is not showing just like the searched text. So, wanted to apply the same font but PyMuPDF is throwing exception saying "need font file or buffer". Font for searched text was 'Verdana' so want to apply the same font for the replaced text.

Here is the log for the operation:

2024-02-29 11:42:34,188 Filename to process: Sample
2024-02-29 11:42:34,194 fonts: [(7, 'ttf', 'TrueType', 'FAAAAH+TimesNewRomanPSMT', 'FAAAAH', 'WinAnsiEncoding'), (11, 'ttf', 'TrueType', 'FAAABB+Verdana', 'FAAABB', 'WinAnsiEncoding'), (16, 'ttf', 'TrueType', 'FAAABG+Verdana-Bold', 'FAAABG', 'WinAnsiEncoding')]
2024-02-29 11:42:34,204 hits: [Rect(170.0, 252.3948974609375, 222.64393615722656, 265.7635498046875)]
2024-02-29 11:42:34,209 search_text: LocalPath
2024-02-29 11:42:34,209 search_text_size: 11.0 search_text_font: Verdana
2024-02-29 11:42:34,213 Exception raised: need font file or buffer

Here is my code:

try:
                # Open a document
                search_text = "LocalPath"
                replace_text = "better text"
                global search_text_size
                global search_text_font 
                with fitz.open(os.path.join("asset", file)) as doc:  
                    for page in doc:                  
                        logger.info(f'fonts: {page.get_fonts()}')
                        hits = page.search_for(search_text)  # list of rectangles where to replace
                        logger.info(f'hits: {hits}')
                        page_blocks = page.get_text("dict")["blocks"]
                        for block in page_blocks:
                            if "lines" in block.keys():
                                spans = block['lines']
                                for span in spans:
                                    data = span['spans']                                    
                                    for lines in data:
                                        if search_text == lines['text']:
                                            search_text_size = lines['size']
                                            search_text_font = lines['font']
                                            logger.info(f'search_text_size: {search_text_size} search_text_font: {search_text_font}')
                                                    
                        for rect in hits:
                            page.add_redact_annot(rect, replace_text, fontname = 'Verdana')  

                        page.apply_redactions(images=fitz.PDF_REDACT_IMAGE_NONE)  # don't touch images
                    doc.save("replaced.pdf", garbage=3, deflate=True)     
                
            except Exception as e:
                logger.error(f'Exception raised: {str(e)}')

Trying for long time, need help!

How to reproduce the bug

1. Run the same code with a PDF file
2. Try to replace one existing text and apply the same font like existing font to the replaced text

PyMuPDF version

1.23.25

Operating system

Windows

Python version

3.8

[JorjMcKie]

This doesn't work like that!
If you want a font different from the Base14 fonts, you must use 2 or 3 steps:

1. Extract the text information of the to-be-deleted text. Use the "dict" option with the rect as the clip.
2. Erase text via apply_redactions.
3. Provide your desired font in the usual way: font file or buffer, page.insert_font() etc. Then insert new text via è.g. page.insert_textbox() providing information previously extracted in step 1.

[santanuOUP]

Hi @JorjMcKie ,

Thanks for your reply.

As you said I have implemented following:

page.get_text("dict", rect)
page.add_redact_annot(rect)
page.apply_redactions(images=fitz.PDF_REDACT_IMAGE_NONE)
page.insert_font(search_text_font)
page.insert_textbox(rect, replace_text)

My desired font is : 'CIDFont+F2' which is not a Base14 font, I think. But it is an existing font within the PDF and used by other texts also, so why it needs to be included again ?
page.insert_font throwing exception 'need font file or buffer'.

[JorjMcKie]

• For methods Page.insert_font() and Page.insert_textbox() the fontname parameter is not "CIDFont+F2", but the reference name (second-to-last sub-item in Page.get_fonts() items).
• So at most you can try to extract the font buffer of such a file and use it to re-insert the font made from it. This however only has a chance to work if the font is actually embedded via a font file. May be not the case in that example ... Type 3 fonts are examples for this situation.
• Fonts in PDFs usually are not complete but subset fonts, i.e. only containing those glyphs of characters that were actually used here. New text has a high chance to contain characters beyond that range. So the new text would contain spaces or "tofus" for missing glyphs.
• In addition, embedded font files often are also often technically damaged: if you try to re-insert their extracted binary then this will raise exceptions.