• Home
  • Features
  • Pricing
  • Docs
  • Announcements
  • Sign In

benwbrum / fromthepage / 26340286607
70%

Build:
DEFAULT BRANCH: development
Ran 23 May 2026 06:48PM UTC
Jobs 1
Files 267
Run time 1min
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

23 May 2026 06:28PM UTC coverage: 70.203% (+0.1%) from 70.068%
26340286607

push

github

web-flow
Add field-based AI transcription support (#5539)

* Add field-based AI transcription support

For field-based collections, the Gemini rake task now builds a structured
JSON prompt listing each transcription field by ID, label, and data type
(including select options and spreadsheet column layouts). The response is
validated for correct JSON, required field keys, and select option values
before being stored in a new `transcription_json` column on ai_transcriptions.
Validation failures store the raw response in `source_text` and mark status
as error so the job error path handles them normally.

The AI tab display renders field values using the same label/value HTML
format as the overview tab, with the human transcription (xml_text) shown
alongside AI values when the comparison toggle is used. CER/WER statistics
now work for field-based pages by deriving AI plain text from
transcription_json when source_text is blank.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* Add RSpec tests for field-based AI transcription

Covers FieldBasedPromptBuilder, FieldBasedResponseValidator,
field-based branches in Create and Generate interactors,
and Page#ai_plaintext / #has_ai_plaintext? for JSON-based results.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* Exclude field labels from AI accuracy CER/WER calculations

Add Page#ground_truth_for_comparison and AiTranscription#text_for_comparison
which return only field values (no labels) for field-based collections,
and fall back to the existing plaintext methods for document-based ones.

Update ai_accuracy_statistics to accept an ai_transcription: kwarg so the
controller can pass the specific record being viewed rather than always using
the most recent one.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* Fix ai_accuracy_calculator_spec doubles missing field_based stub

ground_truth_for_comparison now calls collection.field_based, so the two
specs that stub collection with a plain ... (continued)

2520 of 4124 branches covered (61.11%)

Branch coverage included in aggregate %.

166 of 198 new or added lines in 9 files covered. (83.84%)

10257 of 14076 relevant lines covered (72.87%)

164.59 hits per line

Uncovered Changes

Lines Coverage ∆ File
26
33.33
-16.67% app/helpers/display_helper.rb
5
79.66
app/interactors/ai_transcription/lib/field_based_prompt_builder.rb
1
94.74
app/interactors/ai_transcription/lib/field_based_response_validator.rb
Jobs
ID Job ID Ran Files Coverage
1 26340286607.1 23 May 2026 06:48PM UTC 267
70.2
GitHub Action Run
Source Files on build 26340286607
  • Tree
  • List 267
  • Changed 7
  • Source Changed 7
  • Coverage Changed 6
Coverage ∆ File Lines Relevant Covered Missed Hits/Line Branch Hits Branch Misses
  • Back to Repo
  • Github Actions Build #26340286607
  • 8e67095c on github
  • Prev Build on development (#26335448088)
  • Next Build on development (#26340573297)
STATUS · Troubleshooting · Open an Issue · Sales · Support · CAREERS · ENTERPRISE · START FREE · SCHEDULE DEMO
ANNOUNCEMENTS · TWITTER · TOS & SLA · Supported CI Services · What's a CI service? · Automated Testing

© 2026 Coveralls, Inc