OCTO-11477

[OlteanuRares]

• Fix DFXPReader producing None entries in CaptionList when _convert_p_tag_to_caption() returns None for

  elements that pass the get_text().strip() filter but yield zero parsed nodes

• Add None filtering in _convert_div_to_caption_list() to prevent AttributeError: 'NoneType' object has no attribute 'start' in downstream operations (merge_concurrent_captions(), SRTWriter._recreate_lang())

Root Cause

_convert_p_tag_to_caption() explicitly returns None when its internal node list is empty after parsing (e.g., content that passes the text-presence check but fails the indentation-stripping regex).
The list comprehension in _convert_div_to_caption_list() did not filter these None values before constructing the CaptionList.

The fix wraps the existing generator in an outer comprehension that filters out None results.

Tests added:

• empty

  elements do not produce None entries in CaptionList

• concurrent captions where one

  is empty — merge_concurrent_captions() does not crash

• SRTWriter does not crash on such input
• directly forces None return from _convert_p_tag_to_caption and verifies filtering

No re-ingestion needed — production only accepts SCC inputs; DFXP reader path is not exercised

[github-actions]

🟡 PR Compliance Review

Risk Level: MEDIUM

• Compliance Issues: 0 (0 critical)
• Regressions: 0

REVIEW REQUIRED - Address issues before merging

Full report available in workflow artifacts

[github-actions]

🟡 PR Compliance Review

Risk Level: MEDIUM

• Compliance Issues: 0 (0 critical)
• Regressions: 0

REVIEW REQUIRED - Address issues before merging

Full report available in workflow artifacts