Faulty label normalizers for monograph_part labels

[Dan Pearl]

When producing a codified version of a part label into label_sortkey, the algorithm produces incorrectly formatted results which compromises the value's usefulness.

The problem results when detected values appear earlier in the string.

For example:
Label label_sortkey
DISC 15.1 | disc00000000000000000150000000001 This is not correct
DISC 15.2 | disc00000000150000000002 This is correct

Evergreen 2.4
Opensrf all
PG all
Linux all

I have a fix which I will be submitting shortly.

[Ben Shum (bshum)]

I wonder if this is related to bug 1150939. Which is the "part" in your example labels above? And what classification scheme is being employed for the volume?

[Dan Pearl (dpearl)]

This is related only in the conceptual sense; the code to normalize volume labels is different from that of the Dewey normalizer. The "part" can refer to an issue of a serial ("2012.10"), a disc of a multi-disk set ("Disk 2"), etc. The classification scheme is not related to this code, AFAIK.

[Dan Pearl (dpearl)]

Here is the fix. I assume at some point these sort of things get pulled out into upgrade scripts.

http://git.evergreen-ils.org/?p=working/Evergreen.git;a=shortlog;h=refs/heads/user/dpearl/sortkey

[Ben Shum (bshum)]

Gotcha. Assigning new fix target.

[Bill Erickson (berick)]

Hi Dan, actually, could you please create an upgrade script for your changes? Something along the lines of Open-ILS/src/sql/Pg/upgrade/XXXX.schema.lpad-number-repair.sql. See other files in the directory for examples. I'm happy to assist as needed.

If you're feeling adventurous, a few pgtap (Open-ILS/src/sql/Pg/t/*) tests for this function would be great, particularly since it's a small, easily tested function. This would also help demonstrate inputs and expected outputs for testers.

[Dan Pearl (dpearl)]

This commit ought to do it. The previous commit (see #3) addresses new installations. This is the upgrade script and includes some pgTAP tests.

http://git.evergreen-ils.org/?p=working/Evergreen.git;a=shortlog;h=refs/heads/user/dpearl/lpad-number-repair

[Galen Charlton (gmc)]

Pushed to master, rel_2_7, and rel_2_6. Thanks, Dan!

I also included a couple follow-ups that (a) repair a copy-and-paste error in the test case and (b) update the normalized labels in biblio.monograph_parts, asset.call_number_prefix, and asset.call_number_suffix upon upgrade.