[PR #1826] fix(dvb): Multiple fixes for DVB subtitle extraction from Chinese broadcasts (#224) #2583

New Issue

claunia · 2026-01-29T17:22:55Z

claunia commented

2026-01-29 17:22:55 +00:00

Original Pull Request: https://github.com/CCExtractor/ccextractor/pull/1826

State: closed
Merged: Yes

Summary

This PR addresses multiple issues with DVB subtitle extraction from Chinese broadcasts as reported in #224:

Fix PMT parsing crash: Added bounds checks to prevent segfault on malformed PMT data
Fix negative timestamps: Properly initialize min_pts for DVB subtitle streams
Fix OCR crash: Rewrote ignore_alpha_at_edge() to handle edge cases correctly
Improve OCR accuracy: Added image inversion and contrast normalization for DVB subtitles
Fix --ocrlang parameter: Accept Tesseract language names (chi_tra, chi_sim) directly
Fix --dvblang case sensitivity: Added case-insensitive matching for language codes

Changes

File	Description
`src/lib_ccx/ts_tables.c`	Added bounds checks in `parse_PMT()`
`src/lib_ccx/general_loop.c`	Fixed DVB subtitle timing initialization
`src/lib_ccx/ocr.c`	Fixed cropping, added inversion + contrast enhancement
`src/rust/lib_ccxr/src/common/constants.rs`	Added case-insensitive Language enum
`src/rust/lib_ccxr/src/common/options.rs`	Changed ocrlang to String type
`src/rust/src/parser.rs`	Updated ocrlang handling
`src/rust/src/common.rs`	Updated ocrlang C binding
`docs/CHANGES.TXT`	Added changelog entry

Test Results

Tested with the 12GB sample file from issue #224:

✅ All timestamps now positive (0.235s, 2.594s, etc. instead of -95000s)
✅ OCR accuracy improved from ~70% to ~80-90% for Traditional Chinese
✅ No crashes during full file processing (52 seconds)
✅ --ocrlang chi_tra now works correctly

Test Commands

# Extract with OCR
ccextractor input.ts --codec dvbsub --out=srt -o output.srt --ocrlang chi_tra --no-fontcolor

# Extract without OCR (PNG images)
ccextractor input.ts --codec dvbsub --out=spupng -o output.xml --no-spupngocr

Fixes #224

🤖 Generated with Claude Code

**Original Pull Request:** https://github.com/CCExtractor/ccextractor/pull/1826 **State:** closed **Merged:** Yes --- ## Summary This PR addresses multiple issues with DVB subtitle extraction from Chinese broadcasts as reported in #224: - **Fix PMT parsing crash**: Added bounds checks to prevent segfault on malformed PMT data - **Fix negative timestamps**: Properly initialize min_pts for DVB subtitle streams - **Fix OCR crash**: Rewrote `ignore_alpha_at_edge()` to handle edge cases correctly - **Improve OCR accuracy**: Added image inversion and contrast normalization for DVB subtitles - **Fix --ocrlang parameter**: Accept Tesseract language names (chi_tra, chi_sim) directly - **Fix --dvblang case sensitivity**: Added case-insensitive matching for language codes ## Changes | File | Description | |------|-------------| | `src/lib_ccx/ts_tables.c` | Added bounds checks in `parse_PMT()` | | `src/lib_ccx/general_loop.c` | Fixed DVB subtitle timing initialization | | `src/lib_ccx/ocr.c` | Fixed cropping, added inversion + contrast enhancement | | `src/rust/lib_ccxr/src/common/constants.rs` | Added case-insensitive Language enum | | `src/rust/lib_ccxr/src/common/options.rs` | Changed ocrlang to String type | | `src/rust/src/parser.rs` | Updated ocrlang handling | | `src/rust/src/common.rs` | Updated ocrlang C binding | | `docs/CHANGES.TXT` | Added changelog entry | ## Test Results Tested with the 12GB sample file from issue #224: - ✅ All timestamps now positive (0.235s, 2.594s, etc. instead of -95000s) - ✅ OCR accuracy improved from ~70% to ~80-90% for Traditional Chinese - ✅ No crashes during full file processing (52 seconds) - ✅ `--ocrlang chi_tra` now works correctly ## Test Commands ```bash # Extract with OCR ccextractor input.ts --codec dvbsub --out=srt -o output.srt --ocrlang chi_tra --no-fontcolor # Extract without OCR (PNG images) ccextractor input.ts --codec dvbsub --out=spupng -o output.xml --no-spupngocr ``` ## Related Issues Fixes #224 🤖 Generated with [Claude Code](https://claude.com/claude-code)

claunia added the pull-request label 2026-01-29 17:22:55 +00:00

Sign in to join this conversation.

1 Participants

Notifications

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: starred/ccextractor#2583