[issue4811] invalid reST markup in several documents

[Gabriel Genellina]

Gabriel Genellina <gagsl-py2 at yahoo.com.ar> added the comment:

This patch includes some (sort of) checker for suspicious constructs 
that resembles markup that has leaked into the final output.

It's a new Builder for Sphinx, and works with the docutils nodes, not 
the source files directly.

"doc-Makefile.diff" updates both make.bat and the Makefile, adding a 
new target "suspicious".

The "ignored.csv" file contains the rules to discard false positives.

"builder.diff" updates builder.py in Sphinx, adding a new builder 
implemented in suspicious.py; the source code has some comments 
explaining how it's used.

Execute `make suspicious`, a "suspicious.csv" file is created in build/
suspicious containing one line per issue found.

Currently, the only relevant message I get is about the :makevar: role 
usage in c-api/init.rst -- looks like docutils doesn't properly 
processes it.

There is something I don't like in the code -- suspicious.py is 
generic, lives inside the Sphinx directory, but knows not to process 
the "documenting" directory inside Python [there are too much false 
positives there -- most of the examples contain actual reST markup]. 
But I don't know enough of Sphinx to make it better :(

_______________________________________
Python tracker <report at bugs.python.org>
<http://bugs.python.org/issue4811>
_______________________________________