OCTO-11513

[OlteanuRares]

Summary

• Base style propagation: ::cue { font-style: italic } now wraps bare text nodes in STYLE openers/closers so downstream writers (VTT, DFXP, SCC) emit inline formatting (<i>, <span tts:fontStyle="italic">) without double-wrapping already-styled cues.
• Line overflow warning: Emits CaptionReadWarning when a multiline cue's line:% position would push text beyond the viewport (based on WebVTT's 15-line grid). Non-fatal — does not alter output.
• REGION roundtrip: Reader stores raw region settings on CaptionSet; writer re-emits them as REGION blocks so region:id cue references remain valid in VTT→VTT conversion.
• CaptionSet.regions: New optional regions parameter (defaults to {}) with get_regions()/set_regions() accessors. Keyword-only in practice — no positional arg shift for existing callers.

Safety

• SCC ingestion pipeline unaffected: SCC reader produces no ::cue styles or regions, so new logic never triggers.
• Skylab's WebVTTSegmentWriter creates CaptionSet({lang: CaptionList(...)}) without styles/regions — confirmed no breakage.
• deepcopy in writers correctly preserves the new _regions dict.
• stacklevel=4 on warning verified to point at the caller's .read() line.

Test plan

• 18 new tests covering all four features (base style wrap, no double-wrap, line overflow warning, region roundtrip)
• Full suite passes: 439 passed in 0.49s
• Edge cases verified: nested tags, partial styling, multiline cues with BREAKs, empty nodes, mismatched close tags
• Cross-format: VTT with ::cue { font-style: italic } → DFXP produces <span tts:fontStyle="italic">

[github-actions]

🟡 PR Compliance Review

Risk Level: MEDIUM

• Compliance Issues: 0 (0 critical)
• Regressions: 1

REVIEW REQUIRED - Address issues before merging

Full report available in workflow artifacts