Subscript not recognized

[retorquere]

This one may be tricky:

@Article{Isobe_1996,
  Title                    = {$T_\mathrm{c}$},
}

It's safe enough to ignore the \mathrm, but the _ applies to its argument, which in this case is \mathrm{c}, so I'd have hoped to see T<sub>c</sub>.

Also, much to my surprise when I tried, $T$ looks visually much more like <i>t</i> than it does like T. I tried $Tt$ and I can't tell them apart.

[retorquere]

Ha, wow. $T\mathrm{c}$ comes out looking like <i>t</i>c.

[retorquere]

Don't know if we would want to take it all the way there though.

[johanneswilm]

We will not be able to fully recreate math mode. That user simply needs to put a small t outside of math mode.

As for the first problem. Currently we ask it to ignore the underscore if there is a command following it.

What we could do is replace "\mathrm" and "\text" with "". That should put the underscore before the brackets, and that would make everything within the brackets get the underscore. Would that work?

[retorquere]

mathmode becomes interesting should mathml ever make it into csl. Other than that it would be rather insane to try.

Yeah, removing those should do the job if they can be removed cleanly; this is a scanning parser without backtracking, right? so if the input at the cursor is \mathrm($|\s|[^a-zA-Z]) or somesuch it should be removable cleanly?

[johanneswilm]

It would just remove the command, not the argument. So $_\mathrm{hello}$ becomes $_{hello}$ which then parses cleanly. See the commit. Does that look like what you would expect?

[retorquere]

Yep, looks great. And for fun, {{Cu$_2$O(1\,1\,1)-Cu$CU_\mathrm{CUS}$}} doesn't make the CU outside the mathrm look like lowercase.

[johanneswilm]

I noticed that there were some "nocase" marks added with this change. Is that correct?

[retorquere]

They do no harm here, and are probably semantically closer to the mark than not having them, because math mode (which is where you will find \mathrm) doesn't apply case protection... but

Title = {Some text and $\mathrm{SOME TEXT WITH SPACES}$}

renders to

“Some text andSOMETEXTWITHSPACES.”

Nice. Although honestly, LaTeX interpretation is hard enough -- math mode interpretation, other than where used to produce subscript/superscript and another few niceties, is probably not a sane scope.

[johanneswilm]

Ok, I think we are done with this issue then.