Repository navigation
[ENH]: Use \genfrac in display style? #23008
Description
Problem
Is it possible to render fractions in display style using \genfrac?
The style argument of the _genfrac function in _mathtext.py is received as an Hlist. As a result, fractions will never use display style (since style and self._MathStyle.DISPLAYSTYLE will always compare unequal).
import matplotlib.pyplot as plt
fig = plt.figure()
fig.text(.2, .5, r'$\genfrac{}{}{0}{0}{1}{2}$')
fig.text(.3, .5, r'$\genfrac{}{}{1}{1}{1}{2}$')
plt.show()
Proposed solution
Change
p.genfrac <<= r"\genfrac" - (
"{" + Optional(p.ambi_delim | p.left_delim)("ldelim") + "}"
+ "{" + Optional(p.ambi_delim | p.right_delim)("rdelim") + "}"
+ "{" + p.float_literal("rulesize") + "}"
+ p.simple_group("style") # <- change this line
+ p.required_group("num")
+ p.required_group("den")
| Error("Expected "
r"\genfrac{ldelim}{rdelim}{rulesize}{style}{num}{den}"))
to
p.genfrac <<= r"\genfrac" - (
"{" + Optional(p.ambi_delim | p.left_delim)("ldelim") + "}"
+ "{" + Optional(p.ambi_delim | p.right_delim)("rdelim") + "}"
+ "{" + p.float_literal("rulesize") + "}"
+ "{" + p.int_literal("style") + "}" # <- changed this line
+ p.required_group("num")
+ p.required_group("den")
| Error("Expected "
r"\genfrac{ldelim}{rdelim}{rulesize}{style}{num}{den}"))
with p.int_literal defined using a regex (just like how p.float_literal is defined), following which, the _genfrac function should obtain the enumeration variable from the value. We can hard-code the enumerations
display_style = 0
text_style = 2
script_style = 4
script_script_style = 6
since those are the values specified in TeX: The Program.
I could try to create a PR for this, but my knowledge of language parsing is very limited.
Activity
anntzer commentedon May 8, 2022
Seems reasonable to me, but the numeric values I see in the documentation for genfrac (e.g. in "User's guide for the amsmath package (version 2.1)) are 0/1/2/3, not 0/2/4/6?
(Also, if using scriptstyle or scriptscriptstyle, I guess the
_genfracdefinition needs toshrink()the nodes once or twice more.)oscargus commentedon May 8, 2022
How should the line height behave as consequence of this? #22171
tfpf commentedon May 8, 2022
That's fair; using 0, 1, 2 and 3 sounds good. (Worth noting that those were the numbers originally used in #8151.)
Correct me if I'm wrong, but will that really affect anything? That is for
ruleargument, while this for thestyleargument. (A rebase may be required once that PR is merged, but … other than that?)oscargus commentedon May 8, 2022
Ah, I thought this way about "automatic" rescaling of the text. My bad.