Latexbranch #464
Latexbranch #464
Conversation
Codecov Report
@@ Coverage Diff @@
## master #464 +/- ##
==========================================
- Coverage 96.47% 93.51% -2.97%
==========================================
Files 35 35
Lines 1617 1680 +63
==========================================
+ Hits 1560 1571 +11
- Misses 57 109 +52
Continue to review full report at Codecov.
|
There was a problem hiding this comment.
Thanks for opening the PR. Looks really interesting. Regarding the font sizes. Do you know why yours are bigger than before?
I just had a look through the code so not a full review yet but maybe something of help. Will try a full review today or tomorrow.
I think you don't have a description of how to run it with the new way yet, right? Do you mind to write down a step by step instruction in the PR for people to just try it out including how to install it. We can later put that in the docs then.
src/latex.jl
Outdated
| end | ||
| LaTeXSVG[text] = svg | ||
| end | ||
| return svg | ||
| end | ||
|
|
||
| function tex2svg(text::LaTeXString; output_dir = "./.TeXfolder") |
There was a problem hiding this comment.
I'm not sure this output dir works for windows systems. I think there is a Julia function to create a temporary folder.
There was a problem hiding this comment.
yes mktemp and mktempdir , will be a neater way to do this ;
will change that .
src/latex.jl
Outdated
| generates svg from TeX; | ||
| default folder placed is ./.TeXfolder, | ||
| pdf's and aux files are deleted , tex and log files and other files remain | ||
| add LaTeX packages in `packages=[]` |
There was a problem hiding this comment.
I don't understand the comment. Is there a reason why they don't get deleted?
Also what do you mean with packages =[] here?
There was a problem hiding this comment.
there's no reason to keep the pdf's around after we're done converting to svg;
i just kept it around incase people want to inspect the log and tex files if they do run into errors
There was a problem hiding this comment.
packages=[] was an earlier way i was passing the packages to be added in the usepackage line of the .tex file ; ive changed it in a later commit to a module level variable Javis.LaTeXusepackages which is a vector holding all the package names.
will remove the add Latex package in ... line
| end | ||
| LaTeXSVG[text] = svg | ||
| end | ||
| return svg | ||
| end | ||
|
|
There was a problem hiding this comment.
Make sure to have doctrings here and not inside the function.
There was a problem hiding this comment.
ah ! Pythonic habits die hard ; will fix that !
src/svg2luxor.jl
Outdated
| defs = Dict{String,Any}() | ||
| for def in collect(child_elements(def_element)) | ||
| defs[attribute(def, "id")] = def | ||
| try |
There was a problem hiding this comment.
For each try catch I would like to know why it can fail and maybe there is an option to avoid the try catch then.
There was a problem hiding this comment.
This try statement checks if we have any def tags , if they exist we add the defs to defs dictionary
if there are no def tags ( like in the case of dvisvgm which draws directly and doesnt predefine
paths to use) get_elements_by_tagname(xroot, "defs") will return an empty iterable and cause a bounds error .
There was a problem hiding this comment.
Yeah! i probably shouldn't have discarded all errors , should throw(e) if its not a BoundsError;
but a try free way would be better , the right way would be to check the lengths and proceed
all_def_tags = get_elements_by_tagname(xroot,"defs")
if length(all_def_tags) !=0
defelements = all_def_tags[1]
...
...
endThere was a problem hiding this comment.
Yes some kind of !isempty would be nicer than a try.
|
yes , i haven't documented how to use dvisvgm over the default , I'll add that. As for the font sizes , beats me ! as for the scaling factor i think iv'e got it right . At the risk of sounding like a bad worker who blames their tools from latex-dvisvgm... |
|
digging a little more deeper into the fonts it seems to me the latex-dvisvgm way is alright However the O has a relative size as defined by the font, one can get this info out of a program called
We see the O is pretty close to 36.25 units, the dimensioning is a bit offset and it might be exactly 36.25 if shift the dimensioning down a little to exactly where the top of O starts , anyway its pretty close to 36.25 , whereas the mathjax font will definitely be way over 36.25 units. This is my reason to conclude that something is the matter with mathjax fonts and not latex and dvisvgm . ill try to get in the changes from the other comments done by tomorrow or day-after , maybe it can be reviewed over the weekend ? or perhaps next week if you're not free this weekend |
|
its not really mathjax's fault , it Javis scaling mathjax fonts the wrong way , thats what |
|
Thanks for checking that out |
src/svg2luxor.jl
Outdated
| @@ -91,6 +102,7 @@ function draw_obj(::Val{:path}, o, defs) | |||
| for pi in 1:length(data_parts) | |||
src/svg2luxor.jl
Outdated
| elseif command == 'Z' | ||
| closepath() | ||
| else | ||
| @warn "Couldn't parse the svg command: $command" | ||
| end | ||
| end | ||
|
|
||
| if stroke_width != nothing |
There was a problem hiding this comment.
Comparing with nothing should use triple equal sign or in this case !==.
src/svg2luxor.jl
Outdated
| defs = Dict{String,Any}() | ||
| for def in collect(child_elements(def_element)) | ||
| defs[attribute(def, "id")] = def | ||
| try |
There was a problem hiding this comment.
Yes some kind of !isempty would be nicer than a try.
|
Regarding the font size I'm super happy you figured it out way more than I did 😉 |
|
@Wikunia , added an example gist in the PR :) |
|
With the gist you mentioned I'm getting the following warning: with this file: https://gist.github.com/Wikunia/75b74eb6806a9fac10b52fc676c0198c not sure whether this is expected. |
|
Thats a small bug ; line 208 in latex.jl (in latexbranch) should be Also the tests fail now because ive changed scaling to 425/1000; lemme know if i |
…rning of failed latexpdf generation , even even latex returned 0
PR Checklist
primarily this PR adds support for using dvisvgm which is distributed with TeXLive to render latex . This is an alternative to the existing mathjax solution. Mathjax still exists as the default setting for now.
All tests dont pass because the images are not exact 1-1 pixel replicas .. but they are close enough
edit1: Former are from dvisvgm and latter are from mathjax tex2svg
Linux: To get this working you will need
latex(symlinked to pdftex binary) anddvisvgmin your path, pdftex, thelatexsymlink, and dvisvgm are made available by installing TeXLive (most distributions package TeXLive in their repos)To try out an example program
https://gist.github.com/ArbitRandomUser/da8d439544945466c1df566df561d5f4
Windows & MacOS: somebody's gotta test this for me , i dont have access to either OS's at the moment.
(by default \begin{equation} adds numbering in latex)
If you are contributing to
Javis.jl, please make sure you are able to check off each item on this list:CHANGELOG.mdwith whatever changes/features I added with this PR?Project.toml+ set an upper bound of the dependency (if applicable)?testdirectory (if applicable)?Link to relevant issue(s)
Closes #
How did you address these issues with this PR? What methods did you use?