8000 Merge pull request #20715 from jkseppan/type1-improved-parsing · matplotlib/matplotlib@0bb36e8 · GitHub
[go: up one dir, main page]
Skip to content

Commit 0bb36e8

Browse files
jklymakjklymak
authored
Merge pull request #20715 from jkseppan/type1-improved-parsing
Improve Type-1 font parsing
2 parents 1a3766c + e98bb83 commit 0bb36e8

File tree

5 files changed

+763
-181
lines changed

5 files changed

+763
-181
lines changed

‎LICENSE/LICENSE_COURIERTEN

Lines changed: 18 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,18 @@
1+
The Courier10PitchBT-Bold.pfb file is a Type-1 version of
2+
Courier 10 Pitch BT Bold by Bitstream, obtained from
3+
<https://ctan.org/tex-archive/fonts/courierten>. It is included
4+
here as test data only, but the following license applies.
5+
6+
7+
(c) Copyright 1989-1992, Bitstream Inc., Cambridge, MA.
8+
9+
You are hereby granted permission under all Bitstream propriety rights
10+
to use, copy, modify, sublicense, sell, and redistribute the 4 Bitstream
11+
Charter (r) Type 1 outline fonts and the 4 Courier Type 1 outline fonts
12+
for any purpose and without restriction; provided, that this notice is
13+
left intact on all copies of such fonts and that Bitstream's trademark
14+
is acknowledged as shown below on all unmodified copies of the 4 Charter
15+
Type 1 fonts.
16+
17+
BITSTREAM CHARTER is a registered trademark of Bitstream Inc.
18+

‎doc/api/next_api_changes/behavior/20715-JKS.rst

Lines changed: 8 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,8 @@
1+
``Type1Font`` objects include more properties
2+
---------------------------------------------
3+
4+
The `.type1font.Type1Font.prop` dictionary now includes more keys, such
5+
as ``CharStrings`` and ``Subrs``. The value of the ``Encoding`` key is
6+
now a dictionary mapping codes to glyph names. The
7+
`.type1font.Type1Font.transform` method now correctly removes
8+
``UniqueID`` properties from the font.

‎lib/matplotlib/tests/Courier10PitchBT-Bold.pfb

37.2 KB
Binary file not shown.

‎lib/matplotlib/tests/test_type1font.py

Lines changed: 91 additions & 10 deletions
EDBE
Original file line numberDiff line numberDiff line change
@@ -1,6 +1,7 @@
11
import matplotlib.type1font as t1f
22
import os.path
33
import difflib
4+
import pytest
45

56

67
def test_Type1Font():
@@ -13,25 +14,49 @@ def test_Type1Font():
1314
assert font.parts[0] == rawdata[0x0006:0x10c5]
1415
assert font.parts[1] == rawdata[0x10cb:0x897f]
1516
assert font.parts[2] == rawdata[0x8985:0x8ba6]
16-
assert font.parts[1:] == slanted.parts[1:]
17-
assert font.parts[1:] == condensed.parts[1:]
1817
assert font.decrypted.startswith(b'dup\n/Private 18 dict dup begin')
1918
assert font.decrypted.endswith(b'mark currentfile closefile\n')
19+
assert slanted.decrypted.startswith(b'dup\n/Private 18 dict dup begin')
20+
assert slanted.decrypted.endswith(b'mark currentfile closefile\n')
21+
assert b'UniqueID 5000793' in font.parts[0]
22+
assert b'UniqueID 5000793' in font.decrypted
23+
assert font._pos['UniqueID'] == [(797, 818), (4483, 4504)]
24+
25+
len0 = len(font.parts[0])
26+
for key in font._pos.keys():
27+
for pos0, pos1 in font._pos[key]:
28+
if pos0 < len0:
29+
data = font.parts[0][pos0:pos1]
30+
else:
31+
data = font.decrypted[pos0-len0:pos1-len0]
32+
assert data.startswith(f'/{key}'.encode('ascii'))
33+
assert {'FontType', 'FontMatrix', 'PaintType', 'ItalicAngle', 'RD'
34+
} < set(font._pos.keys())
35+
36+
assert b'UniqueID 5000793' not in slanted.parts[0]
37+
assert b'UniqueID 5000793' not in slanted.decrypted
38+
assert 'UniqueID' not in slanted._pos
39+
assert font.prop['Weight'] == 'Medium'
40+
assert not font.prop['isFixedPitch']
41+
assert font.prop['ItalicAngle'] == 0
42+
assert slanted.prop['ItalicAngle'] == -45
43+
assert font.prop['Encoding'][5] == 'Pi'
44+
assert isinstance(font.prop['CharStrings']['Pi'], bytes)
45+
assert font._abbr['ND'] == 'ND'
2046

2147
differ = difflib.Differ()
2248
diff = list(differ.compare(
2349
font.parts[0].decode('latin-1').splitlines(),
2450
slanted.parts[0].decode('latin-1').splitlines()))
2551
for line in (
2652
# Removes UniqueID
27-
'- FontDirectory/CMR10 known{/CMR10 findfont dup/UniqueID known{dup',
28-
'+ FontDirectory/CMR10 known{/CMR10 findfont dup',
53+
'- /UniqueID 5000793 def',
2954
# Changes the font name
3055
'- /FontName /CMR10 def',
31-
'+ /FontName /CMR10_Slant_1000 def',
56+
'+ /FontName/CMR10_Slant_1000 def',
3257
# Alters FontMatrix
3358
'- /FontMatrix [0.001 0 0 0.001 0 0 ]readonly def',
34-
'+ /FontMatrix [0.001 0 0.001 0.001 0 0]readonly def',
59+
'+ /FontMatrix [0.001 0 0.001 0.001 0 0] readonly def',
3560
# Alters ItalicAngle
3661
'- /ItalicAngle 0 def',
3762
'+ /ItalicAngle -45.0 def'):
@@ -42,17 +67,73 @@ def test_Type1Font():
4267
condensed.parts[0].decode('latin-1').splitlines()))
4368
for line in (
4469
# Removes UniqueID
45-
'- FontDirectory/CMR10 known{/CMR10 findfont dup/UniqueID known{dup',
46-
'+ FontDirectory/CMR10 known{/CMR10 findfont dup',
70+
'- /UniqueID 5000793 def',
4771
# Changes the font name
4872
'- /FontName /CMR10 def',
49-
'+ /FontName /CMR10_Extend_500 def',
73+
'+ /FontName/CMR10_Extend_500 def',
5074
# Alters FontMatrix
5175
'- /FontMatrix [0.001 0 0 0.001 0 0 ]readonly def',
52-
'+ /FontMatrix [0.0005 0 0 0.001 0 0]readonly def'):
76+
'+ /FontMatrix [0.0005 0 0 0.001 0 0] readonly def'):
5377
assert line in diff, 'diff to condensed font must contain %s' % line
5478

5579

80+
def test_Type1Font_2():
81+
filename = os.path.join(os.path.dirname(__file__),
82+
'Courier10PitchBT-Bold.pfb')
83+
font = t1f.Type1Font(filename)
84+
assert font.prop['Weight'] == 'Bold'
85+
assert font.prop['isFixedPitch']
86+
assert font.prop['Encoding'][65] == 'A' # the font uses StandardEncoding
87+
(pos0, pos1), = font._pos['Encoding']
88+
assert font.parts[0][pos0:pos1] == b'/Encoding StandardEncoding'
89+
assert font._abbr['ND'] == '|-'
90+
91+
92+
def test_tokenize():
93+
data = (b'1234/abc false -9.81 Foo <<[0 1 2]<0 1ef a\t>>>\n'
94+
b'(string with(nested\t\\) par)ens\\\\)')
95+
# 1 2 x 2 xx1
96+
# 1 and 2 are matching parens, x means escaped character
97+
n, w, num, kw, d = 'name', 'whitespace', 'number', 'keyword', 'delimiter'
98+
b, s = 'boolean', 'string'
99+
correct = [
100+
(num, 1234), (n, 'abc'), (w, ' '), (b, False), (w, ' '), (num, -9.81),
101+
(w, ' '), (kw, 'Foo'), (w, ' '), (d, '<<'), (d, '['), (num, 0),
102+
(w, ' '), (num, 1), (w, ' '), (num, 2), (d, ']'), (s, b'\x01\xef\xa0'),
103+
(d, '>>'), (w, '\n'), (s, 'string with(nested\t) par)ens\\')
104+
]
105+
correct_no_ws = [x for x in correct if x[0] != w]
106+
107+
def convert(tokens):
108+
return [(t.kind, t.value()) for t in tokens]
109+
110+
assert convert(t1f._tokenize(data, False)) == correct
111+
assert convert(t1f._tokenize(data, True)) == correct_no_ws
112+
113+
def bin_after(n):
114+
tokens = t1f._tokenize(data, True)
115+
result = []
116+
for _ in range(n):
117+
result.append(next(tokens))
118+
result.append(tokens.send(10))
119+
return convert(result)
120+
121+
for n in range(1, len(correct_no_ws)):
122+
result = bin_after(n)
123+
assert result[:-1] == correct_no_ws[:n]
124+
assert result[-1][0] == 'binary'
125+
assert isinstance(result[-1][1], bytes)
126+
127+
128+
def test_tokenize_errors():
129+
with pytest.raises(ValueError):
130+
list(t1f._tokenize(b'1234 (this (string) is unterminated\\)', True))
131+
with pytest.raises(ValueError):
132+
list(t1f._tokenize(b'/Foo<01234', True))
133+
with pytest.raises(ValueError):
134+
list(t1f._tokenize(b'/Foo<01234abcg>/Bar', True))
135+
136+
56137
def test_overprecision():
57138
# We used to output too many digits in FontMatrix entries and
58139
# ItalicAngle, which could make Type-1 parsers unhappy.

0 commit comments

Comments
 (0)
0