kobotoolbox / kpi / 23437958906
82%
master: 76%

LAST BUILD BRANCH: dev-1897-metadata-endpoints
DEFAULT BRANCH: master
Ran
Jobs 10
Files 900
Run time 2min
Badge
Embed ▾
README BADGES
x

If you need to use a raster PNG badge, change the '.svg' to '.png' in the link

Markdown

Textile

RDoc

HTML

Rst
coverage: 81.046% (-0.9%) from 81.951%
23437958906

push

github

web-flow 
fix(qual): allow empty answers to single-select questions DEV-1904 (#6850)

### đŸ“Ŗ Summary
Treat no response selected as a legitimate LLM answer to a select one
question.


### 📖 Description
Previously, if the LLM did not return a selection for a select one
question, it was treated as an error. This PR treats those responses as
legitimate.



### 👀 Preview steps

1. ℹī¸ have an account and a project with an audio question and a
transcription
2. Add a select one QA question that cannot be reasonably answered by
the transcript
3. `Generate with AI`
4. Open the network console to view the response
5. 🔴 [on main] the response comes back with `'status': 'failed',
'error': ''LLM returned no answers for a single select`

- Note: because AI is AI, sometimes the error message ends up being `LLM
returned unexpected response NONE,NONE.` Try to tweak the question to
convince the LLM to return `FALSE,FALSE`, as it should. Anecdotally, it
seems like OSS is slightly smarter than Claude about this

6. đŸŸĸ [on PR] the response comes back with `'status': 'complete',
'value':''`



<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->

## Summary by CodeRabbit

## Bug Fixes
* Improved handling of single-select questions when no options are
selected, returning an empty response instead of an error.
* Enhanced validation logic to gracefully handle edge cases in choice
selection, preventing potential failures during response processing.

<!-- end of auto-generated comment: release notes by coderabbit.ai -->

7490 of 11804 branches covered (63.45%)

0 of 2 new or added lines in 1 file covered. (0.0%)

403 existing lines in 18 files now uncovered.

28926 of 35691 relevant lines covered (81.05%)

5.8 hits per line

New Missed Lines in Diff

Lines Coverage ∆ File
2
32.11
 -0.34% kobo/apps/subsequences/actions/automatic_bedrock_qual.py

Uncovered Existing Lines

Lines Coverage ∆ File
1
40.85
 -1.41% kobo/apps/organizations/admin/organization.py
1
92.31
 -7.69% kobo/apps/user_reports/utils/migrations.py
1
85.0
 -5.0% kpi/urls/__init__.py
2
44.65
 -0.38% kobo/apps/openrosa/libs/utils/logger_tools.py
3
61.21
 -1.82% kobo/apps/organizations/models.py
3
60.53
 -7.89% kobo/apps/organizations/utils.py
6
0.0
 -37.5% kobo/apps/stripe/utils/view_utils.py
7
0.0
 -100.0% kobo/apps/stripe/apps.py
8
0.0
 -100.0% kobo/apps/stripe/urls.py
11
0.0
 -44.0% kobo/apps/stripe/signals.py
12
35.0
 -60.0% kobo/apps/stripe/utils/import_management.py
14
31.82
 -31.82% kobo/apps/stripe/utils/limit_enforcement.py
19
0.0
 -65.52% kobo/apps/stripe/admin.py
36
20.21
 -38.3% kobo/apps/stripe/utils/billing_dates.py
60
26.05
 -50.42% kobo/apps/stripe/utils/subscription_limits.py
63
0.0
 -51.64% kobo/apps/stripe/models.py
75
0.0
 -41.9% kobo/apps/stripe/views.py
81
0.0
 -93.1% kobo/apps/stripe/serializers.py
Jobs
ID Job ID Ran Files Coverage
1 23437958906.1 896
53.07
2 23437958906.2 896
54.6
3 23437958906.3 898
54.31
4 23437958906.4 898
55.86
5 23437958906.5 896
53.3
6 23437958906.6 896
60.12
7 23437958906.7 898
66.04
8 23437958906.8 898
61.26
9 23437958906.9 898
54.67
10 23437958906.10 900
68.3
Source Files on build 23437958906