collectiveaccess / providence

Cataloguing and data/media management application
GNU General Public License v3.0
290 stars 167 forks source link

Search has error on idno with leading zeroes - bug? #1460

Closed MRMarchivist closed 1 year ago

MRMarchivist commented 1 year ago

CA 1.7.17 - unmodified search.conf file

When searching by idno when it has leading zeroes, most searches past 1 decimal place seem to fail (there are a few that don't, for unknown reasons). Test string in this case was 00020.1.1.

00020 and 00020.1 yielded results, while 00020.1.1 did not. Yesterday, (ca_objects.idno:" 00020.1.1") yielded the correct result, and the search log showed the 0s moving around wildly with other formats (screenshot attached). I'll note that (ca_objects.idno:"00020.1.1") (no preceding space) did not yield a result.

Screenshot 2023-07-07 102145

Today, 00020.1.1 yields nothing, and neither does (ca_objects.idno: " 00020.1.1") or any other variation. 00020 and 00020.1 still yield appropriate results. The search log also does not show the same parsing errors as yesterday. Roughly 95% of records with more than 1 decimal are unsearchable. They are recorded properly ca_objects.

Deletion and reimport does not fix it, nor does reindexing.

Edit: I see now that this issue was brought up in Jan 2022 as well, with request for flexibility with leading zeroes. Has there been any progress on it?

MRMarchivist commented 1 year ago

This appears to have corrected itself with an upgrade to the php8 branch. Still unsure if it was regexes or something else.