This is a Bison grammar to convert TeX math into MathML. It can be used as a standalone program or as a static library. It is written in standard C99. It's been tested on OS X, Ubuntu, and Windows.
The goal of this project is to implement as much of AMS-TeX math as possible.
This code is a fork of the itex2MML project, originally written by Jacques Distler and Paul Gartside. It has vastly more support for AMS-TeX math than the original, with a much larger test suite.
It's true, MathML is not well supported in browsers. (Kudos to the Firefox team for sticking with it!)
However, you can consider MathML as an intermediate format onto greater things, like conversion to SVG via mathematical.
Please refer to SUPPORTED.md for more information on what this lib can do. Right now it has a ~93% compatibility with everything MathJax can do.
The most obvious gap in this library is the inability to define new commands (via \mathop, \def, \mathchoice, etc.). Everything else in standard TeX math should be fine.
By default, inline equations are demarcated by $...$. Display equations are demarcated by $$...$$. You can change these to \(..\) and \[..\] respectively; see below for configuration.
You cannot nest equations; for example, $$...\text{foo $...$ bar}...$$ is not allowed.
Since $ signs appear quite often in plaintext, a basic heuristic is applied to calculate whether $ should be interpreted as math or not:
Basic heuristic is: opening $ can't be followed by whitespace, closing $ can't be followed by a digit or preceded by whitespace." -- @jgm, http://bit.ly/1HVtemx */
Of course, if you type \$, no math is applied as the $ is considered skipped.
Both a static and dynamic library are created as part of the cmake build process. Include one of them during the compilation of your program.
The library exposes the following methods:
-
char * mtex2MML_parse(const char * str, unsigned long strlen, const int options): Converts a single TeX equation instrto MathML. Returns just the MathML equation, as a string. -
char * mtex2MML_global_parse(const char * str, unsigned long strlen, const int options, int global_start): The same asmtex2MML_parse, but allows you to provide a starting integer for equation numbering. Returns just the MathML equation, as a string. -
int mtex2MML_filter(const char * str, unsigned long strlen, const int options): Given a string with a mix of TeX math and non-math elements, this returns a single string containing just the converted math elements. Equations are automatically numbered. Returns astatusindicating success (0) or failure. You must access the resulting string withmtex2MML_output. -
int mtex2MML_text_filter(const char * str, unsigned long strlen, const int options): Given a string with a mix of TeX math and non-math elements, this converts all the math and leaves the rest of the string unmodified. Equations are automatically numbered. Returns astatusindicating success (0) or failure. You must access the resulting string withmtex2MML_output. HTML within a math equation are normalized (eg.<becomes<). -
int mtex2MML_strict_filter(const char * str, unsigned long strlen, const int options): Given a string with a mix of TeX math and non-math elements, this converts all the math and leaves the rest of the string unmodified. Equations are automatically numbered. Returns astatusindicating success (0) or failure. You must access the resulting string withmtex2MML_output. HTML tags are removed completely.
The options argument determines which types of delimiters are to be used:
MTEX2MML_DELIMITER_DEFAULTis the default, using$..$for inline and$$..$$for display.MTEX2MML_DELIMITER_DOLLARuses single dollar signs:$..$for inline.MTEX2MML_DELIMITER_DOUBLEuses double dollar signs:$$..$$for display.MTEX2MML_DELIMITER_PARENSuses parenthesis:\(..\)for inline.MTEX2MML_DELIMITER_BRACKETSuses brackets:\[..\]for brackets.
The tests/basic.c and the tests/delimiters.c suites provides a demonstrate of how these methods can be used.
Like any good Unix program, the mtex2MML binary operates on pipes. That is to say, it works like this:
echo '\sin y' | mtex2MML --inline > math.txt
Use mtex2MML -h to get documentation on the options.
A 0 status indicates a success, while anything else indicates a failure, as per the Bison documentation. If a token cannot be parsed, or if the tokenization is nested too deep, the library will bail with a non-zero status code. An error message is also printed to STDERR.
The tests/maliciousness.c suite has a demonstration on how to perform error handling.
To build mtex2MML, you need:
To fetch dependencies, build, and run the library, call:
script/bootstrap
cd build
cmake ..
make
With Chocolatey, you can run:
choco install cmake winflexbison python3
Then, with Powershell, simply run build.ps1.
Assuming you've built the programs, to run the test suite:
cd build
ctest -V
mtex2MML has a test suite that mimics the one found in MathJax, with a few exceptions:
- Files marked as
.xtexhave features that are not implemented, but probably should be in the future. - Files marked as
.no_texhave features that probably won't be implemented.
During the test run, the suite will list the percentage of features that still need coverage.
To install the mtex2MML binary, you can run:
cd build
make install
Found a bug or want to request a feature? Great! Check out CONTRIBUTING.md for more information!
As with the original version(s), this is Open Source software, released under a triple license: GPL, MPL, and LGPL.