For as long as I can remember the following comment has been in the
"""This class is scheduled for removal as part of a refactoring effort."""
My question is; what would that refactor entail? Is there any help I
The reason I ask is highlighted in ticket #1793419 . I'm sure the
quick fix I provide there isn't anywhere near the goal of the
refactor. The way I see it. the problem is that while one could
override the patterns within CorePatterns, is is currently very
difficult to override the code that uses those patterns. It is even
more difficult to add your own patterns.
As the comments suggest just above the CorePatterns class, we should
try to use preprocessors, inlinepatterns or postprocessors if
possible. While understandable, when working with block level markup,
preprocessors don't output to the dom, inlinepatterns will get wrapped
in blocklevel elements defined in CorePatterns (block level elements
would get inappropriately wrapped in <p> tags, for example) and, by
the time we get to postprocessors, it's a little late.
The example I give in the ticket is adding header ids as php markdown
extra does . I could use a preproccesor to extract the id, but now
I need to store it in some way that I know which header it goes with.
Then, how would I add that id to the header? A postprocessor would be
ugly here. The best would be to override the code that puts the header
into the dom. But that is buried in the core. If I'm doing that, I
might as well override the header corepattern.
I'm not posting this to complain. I'd just like to see that refactor
move forward if that's still the plan. If I can help, even better. The
problem is, I don't know what the plan is, so its a little difficult
to help. At the same time I realize this is complex stuff.
PS: As a side note, python-markdown eats php markdown extra's header
id syntax. The header is completely stripped from the output. At the
very least, I would have expected it to be wrapped in a <p> tag. At
best I would have expected the header without the id. Initially, I was
surprised no one implemented this extension yet. Now I know why. It's
currently too hard.