bramp / build-along / 20017191425

89%

Build Type

push

github

Committed by bramp

Commit Message

Add TriviaTextClassifier for detecting flavor/story text on instruction pages

- Add TriviaTextClassifier using union-find spatial clustering to detect
  dense clusters of text blocks (trivia/flavor content)
- Filter numeric text (part numbers, element IDs, counts) from consideration
- Include related images and drawings in the trivia text bounding box
- Exclude large background elements (>50% of page) from expansion
- Clamp final bbox to page bounds to avoid boundary violations
- Add TriviaTextConfig with min_text_blocks, min_total_characters, proximity_margin
- Register classifier in classifier.py and PageClassifier
- Add TriviaText element to Page and iter_elements()
- Update golden test file for page 17

Run Details

130 of 139 new or added lines in 7 files covered. (93.53%)

41 existing lines in 4 files now uncovered.

11230 of 12413 relevant lines covered (90.47%)

0.9 hits per line