Incorrect parsing of reST hyperlink targets

[jgm]

From googlecode Issue #6
Reported by project member fiddlosopher, Feb 11, 2007

RST allows "internal" hyperlink targets that refer to the next element.
They can also be "chained." Pandoc's parser doesn't handle either case
correctly. See
http://docutils.sourceforge.net/docs/ref/rst/restructuredtext.html#hyperlink-targets

What steps will reproduce the problem?

1. Run pandoc on the following input:

`hello`_

.. _hello:

paragraph

Expected output: "hello" should link to the paragraph "paragraph", which
should get id="hello".

1. Run pandoc on the following input:

`hello`_ and `goodbye`_

.. _hello:
.. _goodbye: foo.bar.com

Expected output: Both "hello" and "goodbye" should be links to foo.bar.com.

[jgm]

Implementing this would require the ability to put attributes (at least an id) on arbitrary block elements.

[mpickering]

I think this could now be implemented without too much effort.

[jgm]

To fix the second part of this, we'll need to rewrite the RST reader using the F monad, instead of the current two-phase parse. Otherwise we can't parse

`hello`_

.. _hello:

paragraph

properly, since this requires (a) adding a div around paragraph, which we can't do in the initial parsing phase, and (b) adding something to the table of stateKeys, which we can only do in the initial parsing phase.

[jgm]

Scrap that, I figured out a way to do it.

[jgm]

Feels good to close a bug that has been open 10 years!