Split the monolithic parseTaskText.ts into a parseTaskText/ directory with
separate files for types, prefixes, prefix parsing, priority parsing, repeat
parsing, date parsing, and text cleanup. Moved parseDate.ts from helpers/time/
into the module since it's only consumed by the task text parser. Barrel export
in index.ts maintains backward compatibility — no consumer import changes needed.
https://claude.ai/code/session_01Aeo1ZunQUGKbWx2watMFdW
When the first regex match is a rejected middle-of-text date, continue
searching for subsequent matches instead of returning null. This fixes
cases like "The 9/11 Report due 10/12" where 9/11 is rejected but
10/12 at the end should still be parsed.
Reworked matchDateAtBoundary() to use a single regex pass instead of
two-pass start/end anchoring. Middle-of-text matches are now accepted
when followed by a time expression (at/@ prefix), so inputs like
"meeting 9/11 at 10:00" still parse correctly while "The 9/11 Report"
is rejected.
- Fixed date parser incorrectly extracting date components from within
words
- "Renovation - 2nd Floor Bath" no longer becomes "Reation - Floor Bath"
with a due date of November 2nd
## Changes
- `getMonthFromText` now requires word boundaries, preventing "nov" from
matching inside "Renovation" or "mar" inside "Remark"
- `getDayFromText` now only matches ordinals when followed by
end-of-string, time expressions, or month names, preventing "2nd Floor"
from being parsed as a date
Resolves
https://community.vikunja.io/t/quick-add-magic-unintended-date-parsing/4259
Reviewed-on: https://kolaente.dev/vikunja/vikunja/pulls/2866
Reviewed-by: konrad <k@knt.li>
Co-authored-by: Harry Martland <HarryEMartland@gmail.com>
Co-committed-by: Harry Martland <HarryEMartland@gmail.com>
Claude
kolaente
Finn Weigand
Harry Martland