font-family in insert_html does not work

[crazymxm]

Description of the bug

font-family in insert_html does not work.

How to reproduce the bug

The following code specifies the font by style, but it doesn't seem to be taking effect.

html = "<div> <span style = 'font-size: 6pt; color: #231f20; font-family: \"Y001C\";'> W </span>  </div> "
page.insert_htmlbox(rect, html, css="body {margin:0}")

Y001C is a font embedded in the source pdf ( attached file), It is used to draw some special icons. For example, the W character is an upward triangle indicator in this font.

EOS_760D.pdf

PyMuPDF version

No response

Operating system

No response

Python version

No response

[JorjMcKie]

If those two code lines is all you do to identify the desired font, then of course it cannot work. Your full code properly identifying the font is required to reproduce your problem.

[JorjMcKie]

On another note:
Not every font - even when extractable - behaves as expected / desired, especially if it is a subset font.
Which is the case here too.
So even when you do everything right: that font is a subset which is not usable for rewriting.

Going to close this for now.
Please open a Discussions item for any questions or comments you may have.

[crazymxm]

@JorjMcKie
Yes , the tested font is a subset font.
I think the problems only arise when there are characters that are not included in the subset font， Is my understanding correct?
so I only tested the character ( W ) that used the "Y001C" font in the source document . But strangely it didn't become an upward arrow like before.

This is my test code， I just tested outputting the char (W) with the font on the original PDF and saved it as a new file.
Thank you.

doc = fitz.open(input_pdf_path)
page = doc.load_page(0)
rect = fitz.Rect([0, 0, 500, 200])
html = "<div> <span style = 'font-size: 6pt; color: #231f20; font-family: \"Y001C\";'> W </span>  </div> "
page.insert_htmlbox(rect, html, css="body {margin:0}")
doc.save(output_pdf_path,garbage=8, clean=True,deflate=True,deflate_fonts=True)

[JorjMcKie]

... only arise when there are characters that are not included in the subset font ...

This assumption is incorrect! The glyphs in a font file obviously are accessed by some way of addressing (e.g. the index number in some array). The algorithm for computing the array index depends on the font type and whatever else (don't know the details here).
You cannot assume that the index number generated by finding the glyph for Unicode "W" in the original font file remains the same after the subset.
If you want to write "W" again using the subset font, the index number generated now need no longer be point to the original glyph or even be valid.

[crazymxm]

This assumption is incorrect! The glyphs in a font file obviously are accessed by some way of addressing (e.g. the index number in some array). The algorithm for computing the array index depends on the font type and whatever else (don't know the details here). You cannot assume that the index number generated by finding the glyph for Unicode "W" in the original font file remains the same after the subset. ....

Your answer cleared up my confusion, thank you very much

[JorjMcKie]

You are welcome.
It is always better to somehow get hold of and use the original (non-subset) font (or even a similar one) and accept some file size increase. When done in this way, you can still execute .subset_fonts() again ...

[crazymxm]

@JorjMcKie
Thank you very much for your reply. It might bother you a bit, Please let me continue discussing the issue of subset fonts.
I used the following code to get the unicode mapping of the font in the original pdf.

When I tested it before, I read the first page of the PDF from file A, and then write a character on the page using the same font through insert_Html, and saved it as B. I checked the fonts of file A and B (following code), and their unicode mappings are unchanged. The unicode of the test character 'W' (unicode 0057) is in the CMap mapping table. Strangely, the font did not take effect.

import fitz  # PyMuPDF
import pymupdf

def get_tounicode_mapping(doc: fitz.Document, pno: int, font: str):
    for font_tuple in doc.get_page_fonts(pno):  # Could also be done using page.loadPage(pno); page.getFontList()
        print(font_tuple)

        if font == font_tuple[3] or (font_tuple[3].find("+")>0 and  (font_split := font_tuple[3].split("+", maxsplit=1)) and font == font_split[1]):
            # or condition to cope with "AAAAAA+"-style prefixes in the font names
            for line in doc.xref_object(font_tuple[0]).splitlines():
                line = line.strip()
                if line.startswith("/ToUnicode"):
                    stream_id = int(line.split()[1])
                    old_stream_decoded = doc.xref_stream(stream_id).decode()
                    print(old_stream_decoded)

if __name__ == "__main__":

    input_pdf = "testpdf/EOS_760D.pdf"
    output_pdf = "testpdf/pymupdf_test.pdf"
    dest_lang = 'es'

    doc = fitz.open(input_pdf)
    page = doc.load_page(0)

    rect = fitz.Rect([0, 0, 500, 200])

    get_tounicode_mapping(doc, 0, "Y001C")

    html = "<div> <span style = 'font-size: 6pt; color: #231f20; font-family: \"BHDMVB+Y001C\";'>W</span>   <span style = 'font-size: 6pt; color: #231f20; font-family: \"Y001C\";'>W</span> </div> "
    print(html)
    page.insert_htmlbox(rect, html, css="body {margin:0}")

    doc.save(output_pdf, garbage=8, clean=True, deflate=True, deflate_fonts=True)

    print("New ")
    newDoc = fitz.open(output_pdf)
    get_tounicode_mapping(newDoc, 0, "Y001C")