This is probably another "bad data in the DB causes exceptions" GIGO task. I've been poking at T387188 recently and saw a few TypeErrors (for example T387188#10581449)

Error

• mwversion: 1.44.0-wmf.19
• reqId: 3eaf175f-92dc-4e14-b0df-8ccf0b39251c
• Find reqId in Logstash

[{reqId}] {exception_url}   TypeError: Return value of MediaWiki\Parser\Sanitizer::armorFrenchSpaces() must be of the type string, null returned

Impact

Notes

return preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );

preg_replace is returning null, most likely due to some sort of invalid input. Probably indicates a corrupt revision in the DB. (But I haven't poked at undecodable revisions in a while so this isn't me)

• T351953: Various old revisions are encoded as Windows-1252 rather than UTF-8, causing "RuntimeException: PCRE failure" when viewing them <-- I think this is the main task for this
• T387607: Some old revisions on Chinese Wikipedia are truncated, resulting in invalid UTF-8 and "RuntimeException: PCRE failure" when viewing them
• T340750: Detect bad UTF-8 value in PST trasnforms
• T31564: Bad UTF-8 in ThreadSignature breaks display and can't be exported
• T321234: Exception: "Malformed UTF-8 characters" in Parser\MagicWordArray (via WikitextContentHandler)

I had a patch in T351953, but @Ladsgroup wrote a patch to add a maintenance script to fix these up which apparently (T351953#10617788) is almost done. So we might actually be close to a proper fix for these sorts of tasks.