2.2.26

• Implement WebVTT REGION block parsing in the reader. Cues
  referencing a region now get proper origin/extent geometry computed
  via W3C TTML-WebVTT mapping formulas, enabling correct cross-format
  conversion (e.g., VTT with regions → DFXP produces correct
  tts:origin and tts:extent).
• Fix .x/.y typo in geometry.py fit_to_screen() — the Y-axis safety
  guard was checking X twice instead of both axes.
• Fix _remove_noop_off_on_italics typo in SCC
  specialized_collections.
• Add empty-collection guard in _remove_spaces_at_end_of_the_line.
• Replace magic numbers with named _InstructionNode constants.
• Remove dead _remove_styles() method from WebVTT reader.
• Remove redundant if-guard and unused class attributes in WebVTT
  writer.

2.2.25

• Add drop-frame timecode support to SCCWriter via a new
  drop_frame=False parameter. When enabled, timestamps use the
  standard 10-minute block method with semicolon separators.
• Refactor SCCWriter timing pipeline: exact token counting, monotonic
  frame deduplication, and caption splitting at 80 tokens.
• Extract SCC_TOKENS_PER_CAPTION_MAX as a named constant.
• CI: guard coverage-parsing step on test success.

2.2.24

• Fix SCC ingestion error when a doubled italic-off mid-row code
  (9120 9120) appears before punctuation. The punctuation lookahead
  now skips the error-correction duplicate, preventing an unwanted
  space that pushed lines past the 32-character limit.

2.2.23

• bumps nltk from 3.9.1 to 3.9.4.
• Fix SCC writer producing out-of-order timestamps when a short
  caption is immediately followed by a longer one. The buffer
  lead-time calculation could push a caption's start before its
  predecessor; now clamped to preserve chronological order.
• Fix DFXPReader producing None entries in CaptionList when empty
  <p> elements yield zero parsed nodes, causing AttributeError in
  merge_concurrent_captions() and SRTWriter.

2.2.22

• Fix: EDM (Erase Displayed Memory) command now correctly clears
  displayed captions in paint-on and roll-up modes.

2.2.21

• supports multi-row jumps: small jumps (1-3 rows) emit the correct
  number of line breaks, large jumps (4+ rows) trigger repositioning
  instead
• clear positioning state at caption boundaries (EOC, roll-up,
  paint-on, clear commands)
• Code reformatted with Black/isort across the codebase
• Added missing tests for _PositioningTracker,
  Alignment.from_horizontal_and_vertical_align
  • Bump actions/checkout from v2 to v4 in CI workflow
  • Replace unmaintained archive/github-actions-slack with
    slackapi/slack-github-action@v3.0.2 (fixes Node.js and
    set-output deprecation warnings)

2.2.20

Merge pull request #362 from pbs/OCTO-11150-update-apache-and-copyrig…

2.2.19

Remove support for python 3.18 and 3.19.

2.2.18

update changelog

2.2.17

update nltk from 3.8.0 to 3.9.1