Remove ttconv and implement Type-42 embedding using fontTools #20866
Conversation
447fd1b to
a826065
Compare
8c44828 to
2535508
Compare
de16cf0 to
daee4e5
Compare
016713a to
7cea2d4
Compare
| if fontfile.endswith('ttc'): | ||
| # fix this once we support loading more fonts from a collection | ||
| # https://github.com/matplotlib/matplotlib/issues/3135#issuecomment-571085541 | ||
| options.font_number = 0 |
There was a problem hiding this comment.
I think we test a ttc in CI already, so it could be used to test this line?
| The Type-42 formatted font | ||
| """ | ||
| version, breakpoints = _version_and_breakpoints(font.get('loca'), fontdata) | ||
| post, name = font['post'], font['name'] |
There was a problem hiding this comment.
Are these tables guaranteed to exist?
There was a problem hiding this comment.
Both post and name are required: https://docs.fileformat.com/font/ttf/
| tuple | ||
| ((v1, v2), breakpoints) where v1 is the major version number, | ||
| v2 is the minor version number and breakpoints is a sorted list | ||
| of offsets into fontdata; if loca is not available, just the table | ||
| boundaries |
There was a problem hiding this comment.
For multiple return, you should list the entries separately, like Parameters.
|
Wow, I missed that this came in. 💯 on dropping an extension module! |
|
Dropping to draft until the comments and rebase are addressed. Anyone should feel free to move back to active... |
|
Ping @jkseppan? |
| Returns | ||
| ------- | ||
| fontTools.ttLib.ttFont.TTFont |
There was a problem hiding this comment.
Long term, it could make sense to link to the documentation? (Although fontTools is not in the intersphinx_mapping yet.)
|
We want to try and do mpl 3.7 feature freeze around Jan 1 so I think it is unlikely we will have the bandwidth to get this rebased and reviewed by then, pushing to 3.8. |
7cea2d4 to
a1b41fe
Compare
|
Do we have someone who can pick this up? |
7f2b2a2 to
1393702
Compare
|
FYI, I think I'm okay with this as it stands, but as I did some followup cleanup work, I'm not sure if I should approve it myself. |
oscargus
left a comment
There was a problem hiding this comment.
I am still open to the possibility to add fontTools to the intersphinx-stuff and provide a proper link, but I am also open to do that myself in another PR.
|
Btw, do we (i.e. someone with a Mac(?)) know if #20866 (comment) still holds? |
Maybe @greglucas or @ksunden can have a look at it? |
|
Using the example from the linked issue: # this needs the patch above
import matplotlib
from pathlib import Path
matplotlib.use('ps')
from matplotlib import pyplot as plt
# the following font is in CFF format on my Mac
font = Path('/System/Library/Fonts/AppleSDGothicNeo.ttc')
fig = plt.figure()
plt.axis((0,10,0,10))
plt.text(3,3,'Glib jocks quiz nymph to vex dwarf.', font=font)
matplotlib.rc('ps', fonttype=42)
plt.savefig('fmt42.ps')
matplotlib.rc('ps', fonttype=3)
plt.savefig('fmt3.ps')The type42 is completely blank for me (no axes either). The type3 shows up as expected. I don't get an error though 🤷 |
|
That example fails for me on main. Trying to open the and with this branch I get |
|
Okay, looks like the answer is that this has not fixed #20866 (comment), however that is broken on main anyway, so I'd lean towards not blocking for merging this if we are otherwise satisfied. |
|
The issue also appears to be font specific, dejavu being embedded as type42 works OK. |
Split _backend_pdf_ps.get_glyphs_subset into two functions: get_glyphs_subset returns the font object (which needs to be closed after using) and font_as_file serializes this object into a BytesIO file-like object.
Move the test in lib/matplotlib/tests/test_ttconv.py to lib/matplotlib/tests/test_backend_pdf.py. Actually it has not been a test of ttconv for a while, since type-3 conversion has not been using ttconv.
Fonttools cannot subset these and drops them; avoid the warning.
Also don't end up in an infinite loop if there is a larger gap between breakpoints than we would like.
FT2Font throws RuntimeError for some fonts
Not the subsetted one, since in some cases fontTools.subset does not produce a good name table.
De-wrap a few lines now that we allow longer lines.
1393702 to
45a1654
Compare
This was missed in matplotlib#20866
This was missed in matplotlib#20866
PR Summary
ttconv is now only being used for outputting fonts in Type-42 format in PostScript files, and with fontTools this turns out to be quite doable in pure Python.
PR Checklist
pytestpasses).flake8on changed files to check).flake8-docstringsand runflake8 --docstring-convention=all).doc/users/next_whats_new/(follow instructions in README.rst there).doc/api/next_api_changes/(follow instructions in README.rst there).