If there's an exception while parsing (because it's something katex doesn't support), the output doesn't get cleared. Typically when this happens, we'll fall back to MathJax and render that in a different element, leaving the previous KaTex-formatted math plus the new MathJax-formatted math next to each other in the page, which is never what you'd want.
Test Plan: Using the fallback mechanism in khan-exercises utils/tex.js, render something katex can deal with, like "3", then using the same element, reprocess with something katex can't deal with, like "\approx 3". Previously the result would be "3 \approx 3" (with the first 3 being left over from the original katex rendering. With this change, only the "\approx 3" remains on the page.
Auditors: alex, alpert
Summary:
Right now, when the size gets bigger, this still doesn't work, so there's a
check to prevent that. However, functions that go smaller (like `\small`,
`\tiny`, etc) do work. Also, we can't seem to use the sizing functions inside
of fractions (so something like `\dfrac{\small\frac{x}{y}}{z}` doesn't work).
However, the most prominent use case is `\small` as the outer-most object, so
this is still helpful. This commit has the parsing and stuff to handle all of
it, but it'll throw an error if you try to do something that doesn't work. (For
the record, "doesn't work" means "looks bad", not "throws an unexpected
error").
Test Plan:
Make sure things like `\small x` work, and things like `\Huge x` and
`\frac{\small x}{y}` don't.
Reviewers: alpert
Reviewed By: alpert
Differential Revision: http://phabricator.khanacademy.org/D3619
Summary:
\nleq and \ngeq are AMS characters, so add support for parsing and
showing those. Also, add font metrics from the ams_regular font.
Test Plan:
View `\ngeq \geq`, make sure that they look the same but with a
slash through `\ngeq`. View `a \ngeq b`, make sure there's the right spacing
for a rel.
Reviewers: alpert
Reviewed By: alpert
Differential Revision: http://phabricator.khanacademy.org/D3547
Summary:
Fixes T1325.
This has the advantage of not having to do the mapping manually. Also it means that characters like \neq are included in the mapping, so this resolves T1323 as well.
I opted to group the metrics by character instead of by metric type because I think it'll gzip better and it makes getCharacterMetrics simpler anyway.
Test Plan: The test page looks identical to the pixel to my eye.
Reviewers: emily
Reviewed By: emily
Maniphest Tasks: T1323, T1325
Differential Revision: http://phabricator.khanacademy.org/D3535
Summary: Add margin to the right of characters that have italic correction.
Test Plan:
Look at `f'` and `C'` before and after. Note that they both now look
good at small font sizes (like 17pt).
Reviewers: alpert
Reviewed By: alpert
Differential Revision: http://phabricator.khanacademy.org/D3534
Summary:
This diff does a couple different things:
- There is now a metrics/ folder, which contains the property files describing
the metrics if the fonts, as well as a script for reading and printing the
metrics in javascript.
- Fractions and superscripts/subscripts are now rendered in slightly different
ways now (notably, no use of inline-table). This allows for much more precise
positioning of the superscripts, subscripts, numerators, and denominators,
while still having an appropriate baseline. Also, there is no longer a
sup/sub/supsub distinction, there are only supsubs with null sup/sub.
- Using the new font metrics and by implementing the formulas found in The TeX
Book, Appendix G, the heights and depths of all of the sub-expressions in a
formula are now calculated. These are currently used to:
- Correctly position superscripts, subscripts, numerators, and denominators
- Adjust the height and depth of the overall expression so it takes up the
appropriate space
- Because we have to add attributes (height and depth) to every attribute, I
have changed the way DOM nodes are assembled. Now, instead of assembling the
DOM elements inline (which is a problem because we need to track
height/depth, and we shouldn't (and can't in IE 8) attach raw attributes to DOM
nodes), we assemble a pseudo-DOM structure with the extra information, and
then actually assemble it at the very end.
The main page also now has an updated expression to show off and test the new
and improved parsing.
Test Plan:
View the main page, make sure that the expression renders. Make sure
that the tests pass. Make sure that expressions have the correct calculated
height (this is most easily tested by viewing them on the main page and making
sure that the top of the expression lines up with the bottom of the input box).
Reviewers: alpert
Reviewed By: alpert
Differential Revision: http://phabricator.khanacademy.org/D3442
Summary:
Add a command to print out a TeX-style logo for KaTeX. I think the
code is okay, but suggestions for the design of the logo are most welcome (I'll
post screenshots soon).
Test Plan: Make sure \KaTeX renders, and the tests still succeed.
Reviewers: alpert
Reviewed By: alpert
Differential Revision: http://phabricator.khanacademy.org/D3443
Summary:
Pull node making into a separate module, make an "options" param for
the make_ functions, and pull the different types of groups into separate
functions.
Test Plan: Open the homepage, make sure everything still works.
Reviewers: alpert
Reviewed By: alpert
Differential Revision: http://phabricator.khanacademy.org/D3368
Summary: Make the makefile copy files into the correct places in all
three repos. (note, the khan-exercises location might change since we
don't want to have two copies of things, but it's where it is in my
setup right now)
Auditors: alpert
Summary: This makes it more annoying to use locally, because it goes
away when you type something bad, but is much more useful in use in
khan-exercises and perseus.
Auditors: alpert
Summary:
Add some things in to make this more useful to khan-exercises.
Notably, make KaTeX.process create and clean up its own .katex node, and rename
.mathmathmath -> .katex.
Test Plan: Make test, make sure the main page still renders stuff
Reviewers: alpert
Reviewed By: alpert
Differential Revision: http://phabricator.khanacademy.org/D3061
Summary:
Add the ability to parse lone `^x` and `_y`, etc. This basically just
involves more checking of cases in the increasingly ugly `parseAtom` function.
Also, now we manually check for the cases of double superscripts and
subscripts.
Test Plan: Make sure the tests pass. Make sure things like `^x` and `_y` parse.
Reviewers: emily
Reviewed By: emily
Differential Revision: http://phabricator.khanacademy.org/D3095
Test Plan: `\orange{x}\blue{-4}`, `\orange{x-}\blue{-4}` look the same after desaturation as `x-4` and `x--4`. In addition, `\orange{x}\blue{{-4}}` looks appropriately atrocious (i.e., has no space whatsoever).
Reviewers: emily
Reviewed By: emily
Differential Revision: http://phabricator.khanacademy.org/D3050
Summary:
Remove all uses of underscore. For the things we actually need, make a
"util" file and put them in there.
Test Plan:
Make sure the test still succeed, and that the main page still
works.
Reviewers: alpert
Reviewed By: alpert
Differential Revision: http://phabricator.khanacademy.org/D3043
Summary: Allow there to be letters immediately after \ , \;, \,, \:.
Test Plan: Run the tests
Reviewers: alpert
Reviewed By: alpert
Differential Revision: http://phabricator.khanacademy.org/D3031
