$ echo "'''News:''' Stuff here" | node tests/parse.js --normalize <p><b><a href="News:'''">News:'''</a> Stuff here</b></p>
The PHP parser produces <p><b>News:</b> Stuff here</p>.
Originally reported by @gpaumier when trying to edit https://office.wikimedia.org/wiki/User:Guillaume/Goals#Retrospective in VE.
| Subject | Repo | Branch | Lines +/- | |
|---|---|---|---|---|
| Double or triple quotes terminate autolinks | mediawiki/services/parsoid | master | +22 -5 |
Also note that this causes round-trip failures, both in wt2wt and html2html:
$ echo "'''News:''' Stuff here" | node tests/parse.js --wt2wt '''News:''' Stuff here''' $ echo "<p><b>News:</b> Stuff here</p>" | node tests/parse.js --html2html --normalize <p><b><a href="News:'''">News:'''</a> Stuff here</b></p>
Hm, this should be falling into the (recently-added) "autolinks must have at least once character after protocol" test. But it's not -- presumably because the quotes are being sucked into the link. Probably a tokenizer precedence issue, gah. I'll take a look.
Change 240915 had a related patch set uploaded (by Cscott):
Double or triple quotes terminate autolinks
Seems fixed by the above,
λ (master) echo "'''News:''' Stuff here" | node bin/parse.js --normalize <p><b>News:</b> Stuff here</p> λ (master) echo "'''News:''' Stuff here" | node bin/parse.js --wt2wt '''News:''' Stuff here λ (master) echo "<p><b>News:</b> Stuff here</p>" | node bin/parse.js --html2html --normalize <p><b>News:</b> Stuff here</p>