Embed Font in existing PDF

PyMuPDF
1

Hi there,

I do have a legacy 3rd party tool, generating a pdf with a lot of graphics and texts.
This tool can not embed fonts.
This Tool somehow removes spaces from font names.

I use “JetBrains Mono” for some texts, and want to embed this font into the pdf
So, I managed to add this font to the pdf using pymupdf.

so output from pdffonts looks like:

name                                 type              encoding         emb sub uni object ID

JetBrainsMono                        TrueType          WinAnsi          no  no  no      50  0
Courier-Bold                         Type 1            WinAnsi          no  no  no      52  0
fstroke3                             TrueType          WinAnsi          no  no  no      53  0
JetBrains Mono Regular               CID Type 0C (OT)  Identity-H       yes no  yes   1739  0

But it seems not every viewer can work with this. Basically I just found chrome browser viewer showing the correct font, but maybe it has JetBrains somwhere with it and is not using the embedded version.
But this gives me the indication that “JetBrainsMono“ without spaces is somehow working as non embedded font.

Unfortunately I can not use rpl_font.py, since there are many hidden text items, the document is not readable then.

So can I embed a font for existing text at all?
At lest I can not embed a font and use an existing name.

Any Idea here?

2

It seems this is similar topic:

I will give it a try

3

@arne123 Nice find - hope the tips in there work - please let us know how it goes!

4

So, I just learned, that white-spaces in PDF names are not allowed.
But “JetBrains Mono” contains white-spaces.
Maybe this should be handled when loading fonts initially (seems the legacy tool knows something here).

So I manually copied the xref and removed white-spaces:


 for key in doc.xref_get_keys(xref_src):
     item = doc.xref_get_key(xref_src, key)
     if key == "BaseFont":
         item = (item[0], item[1].replace(" ", ""))

     doc.xref_set_key(53, key, item[1])
     doc.xref_set_key(50, key, item[1])

So pdffonts looks like this now:

name                                 type              encoding         emb sub uni object ID
------------------------------------ ----------------- ---------------- --- --- --- ---------
JetBrainsMonoRegular                 CID TrueType      Identity-H       yes no  yes     50  0
Courier-Bold                         Type 1            WinAnsi          no  no  no      52  0
JetBrainsMonoRegular                 CID TrueType      Identity-H       yes no  yes     53  0
fstroke3                             TrueType          WinAnsi          no  no  no      55  0
JetBrains Mono Regular               CID TrueType      Identity-H       yes no  yes   1741  0

But all text is gone now in the document, so not yet successful.
I will try an other font next to exclude any name issue.

5

I just had a look again into this, an finally I got it working.

Intention was embed a font into an existing pdf, which wasn’t embedded before,
but which is actually the same font.

The final trick was to set set_simple=True
page.insert_font(fontbuffer=font.buffer, fontname=fontname, set_simple=True)

example:

import pymupdf

doc = pymupdf.open("in.pdf")
page = doc[0]  # First page

fontname = "JBMONO"
fontfile =  "JetBrainsMono-Regular.ttf"

font = pymupdf.Font(fontfile=fontfile)
page.insert_font(fontbuffer=font.buffer, fontname=fontname, set_simple=True)

fts = page.get_fonts(full=True)

xref_dst = []

for ft in fts:
    if ft[1] == 'n/a' and ft[3].startswith("JetBrainsMono"):
         xref_dst.append(ft[0])
         print(f"{ft} < Found dst font: xref_dst = {ft[0]}")
     elif ft[3] == "JetBrainsMono-Regular":
         xref_src = ft[0]
         print(f"{ft} < Found source font: xref_src = {xref_src}")
     else:
         print(ft)

for xref in xref_dst:
    print(f"Updating font reference for xref {xref} to point to xref {xref_src}")
    for key in doc.xref_get_keys(xref_src):
        item = doc.xref_get_key(xref_src, key)
        if key == "BaseFont":
            item = (item[0], item[1].replace(" ", "")) #not sure if needed
        doc.xref_set_key(xref, key, item[1])


print("\nUpdated font references:")

for ft in fts:
    print(ft)

# Save the output
doc.save("out.pdf", garbage=4, deflate=True, use_objstms=1)
1 Like
6

Thanks for the update - not sure why set_simple should make the difference - @HaraldLieder do you have any technical explanation here to enlighten us?

7

You cannot bring existing text in a PDF to use a different font by just replacing its font by another font copying over the new font xref over the old xref.
The reason are multiple (1) almost certainly different fonts have different glyph addresses (numbers), (2) different fonts have different glyph metrics (width, height, ascender, descender, …).

This is a path that almost never can be successful.
… Except: you are lucky enough to use exactly the same font again. In which case a justified question would be: “Why are you doing this at all?”
If setting “simple” to true was successful, then because the original creation of the text did the same [it causes using only the Unicodes up to number 255).