New Code!
After an intermission of a few weeks, Discount has been updated to version 1.2.1 with the addition of a completely reworked emphasis parser. Previously,
it did a naive “turn all *’s and _’s into <em>; turn all ** and __ into <strong>” which led to incorrect XML when I interleaved * and **, but it now attempts to pair matching tokens before it starts spitting out emphasis.
The way I did this was to split the existing second pass (discount has a first pass that breaks the input into blocks, and a second pass that does text substitutions on the contents of those blocks) into two passes; The second pass now converts runs of * into emphasis tokens, interleaved with fully-processed other stuff, and the third pass concatenates them together, matching open and close emphasis together as it goes.
for example, the four variants of
**A*B***produce correct XML now:
***A*B**–><p><strong><em>A</em>B</strong></p>
***A**B*–><p><em><strong>A</strong>B</em></p>
**A*B***–><p><strong>A<em>B</em></strong></p>
*A**B***–><p><em>A<strong>B</strong></em></p>which is much better than the status quo ante.
So it doesn’t dump core (as the presence of this weblog post shows,) it fixes a few memory leaks, and it produces better XML on pathological emphasis cases. And that’s good enough to be New Code! to amaze and educate your friends and family.