Conversation
|
I'm not sure about the spacing of the result. The rightmost side of the left angle appears to be half as close as the leftmost side of the right angle. For the first three examples, it's about 1 vs 2 pixels, while about 2 vs 4 on the rightmost largest one. If you take your test and call it with only @image_comparison(baseline_images=['math_angle_brackets.png']*2)
def test_math_angle_brackets():
tests = [
r'$\left< x \right>$',
r'$\left< ? \right>$',
r'$\left< \frac{1}{2}\right>$',
r'$\left< \frac{\sum_{0}^{1}}{2}\right>$',
r'$\left<\frac{\sum_{0}^{1}}{\sum_{0}^{1}}\right>$'
]
for left, right in [('<', '>'), (r'\langle', r'\rangle')]:
fig = plt.figure(figsize=(5, 1))
for idx, text in enumerate(tests):
fig.text((idx + 0.3)/len(tests), 0.5,
text.replace('<', left).replace('>', right),
math_fontfamily='cm') |
|
I reverted the changes and tested the code again. The misalignment appears due to the missing sizing of the angle brackets, but reordering the sizing for the glyphs following this diff, generates the following image, which seems to be centered(?) with --- a/lib/matplotlib/_mathtext.py
+++ b/lib/matplotlib/_mathtext.py
@@ -419,10 +419,10 @@ class BakomaFonts(TruetypeFonts):
('ex', '\xbb'), ('ex', '\x26')],
r'\rceil': [('ex', '\xa8'), ('ex', '\x6d'),
('ex', '\xbc'), ('ex', '\x27')],
- r'\langle': [('ex', '\xad'), ('ex', '\x44'),
- ('ex', '\xbf')],
- r'\rangle': [('ex', '\xae'), ('ex', '\x45'),
- ('ex', '\xc0')],
+ r'\langle': [('ex', '\xad'), ('ex', '\x2a'),
+ ('ex', '\xbf'), ('ex', '\x44')],
+ r'\rangle': [('ex', '\xae'), ('ex', '\x2b'),
+ ('ex', '\xc0'), ('ex', '\x45')],
r'\__sqrt__': [('ex', '\x70'), ('ex', '\x71'),
('ex', '\x72'), ('ex', '\x73')],
r'\backslash': [('ex', '\xb2'), ('ex', '\x2f'),
@@ -433,6 +433,8 @@ class BakomaFonts(TruetypeFonts):
('ex', '\x64')],
r'\widetilde': [('rm', '\x7e'), ('ex', '\x65'), ('ex', '\x66'),
('ex', '\x67')],
+ r'<': [('ex', '\x44')],
+ r'>': [('ex', '\x45')]
}Image for CM font: Image for Custom font: |
|
Sorry it's taken a while to get back to this PR. I don't believe it is the correct approach as the glyph order is correct at the moment. The problem is really that the character code gets mapped twice. I've opened #29936 to correct this, which should replace this PR. |
3 participants
PR summary
Fixes #18740
PR checklist