[BUG] ccextractor skips many captions (CEA-708) #793

New Issue

claunia · 2026-01-29T16:53:37Z

claunia commented

2026-01-29 16:53:37 +00:00

Originally created by @svlobanov on GitHub (Dec 30, 2023).

CCExtractor version: 0.94 376ff83161 CEA-708 decoder: C

Necessary information

Is this a regression (i.e. did it work before)? NO
What platform did you use? Mac
What were the used arguments? ./ccextractor -pn 4 503.ts

Video links

TS file: https://tsduck.io/streams/usa-atsc/503.ts (index page is here: https://tsduck.io/streams/?name=usa-atsc )

Additional information

The result file 503.p4.svc01.srt contains only two captions, but there are much more captions in the CC708 Service1 in the program(4). The same issue if I use ccextractor with CEA-708 Rust decoder.

In VLC4 decodes all CC708 service1 captions for Program 4(Charge!). Also, caption-inspector (https://github.com/Comcast/caption-inspector ) decodes much more captions in the CC708 Service1 in the program(4). I'm attaching caption-inspector's output for PID=0x61 (97) caption-inspector-output.zip

ccextractor CC708 Service1 Program4 output:

1
00:00:01,067 --> 00:00:23,856
     - [Chris A.] Thank you.
    none here is amazing.

2
00:00:25,358 --> 00:01:03,096
This is accomplished easily
with the graduating guide combs

Originally created by @svlobanov on GitHub (Dec 30, 2023). CCExtractor version: 0.94 376ff831616919e092b53353f3799654d86d0759 CEA-708 decoder: C # Necessary information - Is this a regression (i.e. did it work before)? NO - What platform did you use? Mac - What were the used arguments? `./ccextractor -pn 4 503.ts` # Video links TS file: https://tsduck.io/streams/usa-atsc/503.ts (index page is here: https://tsduck.io/streams/?name=usa-atsc ) # Additional information The result file `503.p4.svc01.srt` contains only two captions, but there are much more captions in the CC708 Service1 in the program(4). The same issue if I use ccextractor with CEA-708 Rust decoder. In VLC4 decodes all CC708 service1 captions for Program 4(Charge!). Also, caption-inspector (https://github.com/Comcast/caption-inspector ) decodes much more captions in the CC708 Service1 in the program(4). I'm attaching caption-inspector's output for PID=0x61 (97) [caption-inspector-output.zip](https://github.com/CCExtractor/ccextractor/files/13799033/caption-inspector-output.zip) ccextractor CC708 Service1 Program4 output: ``` 1 00:00:01,067 --> 00:00:23,856 - [Chris A.] Thank you. none here is amazing. 2 00:00:25,358 --> 00:01:03,096 This is accomplished easily with the graduating guide combs ```

claunia closed this issue

2026-01-29 16:53:38 +00:00

claunia commented

2026-01-29 16:53:38 +00:00

@cfsmp3 commented on GitHub (Dec 14, 2025):

Issue Status: Fixed ✅

I've tested this issue with the current master branch and the problem appears to be resolved.

Test Results

Sample file: https://tsduck.io/streams/usa-atsc/503.ts
Command: ./ccextractor --program-number 4 503.ts
Original report: Only 2 captions extracted
Current version: 30 captions extracted (matching caption-inspector output)

Caption Comparison

The captions now match the expected output from caption-inspector:

#	CCExtractor Output	Caption-Inspector Reference
1	00:00:00,601 - "d so every time / you start trimming,"	✅ Match
2	00:00:02,403 - "all of those hairs are being..."	✅ Match
...	...	...
30	"This is accomplished easily with the graduating guide combs"	✅ Match

Likely Fixes

Several commits since the original report (Dec 2023) have improved the CEA-708 decoder:

3c51fb65 - Handle row_count decrease in CEA-708 C decoder
d6ccf1bf - Port 708 decoder encoding module to Rust
Various window handling and display logic improvements

The issue was likely related to improper handling of CEA-708 window toggle (TGW) and delete layer (DLW) commands, causing captions to accumulate instead of being flushed at proper display boundaries.

This issue can be closed as fixed.

🤖 Generated with Claude Code

@cfsmp3 commented on GitHub (Dec 14, 2025): ## Issue Status: Fixed ✅ I've tested this issue with the current master branch and the problem appears to be resolved. ### Test Results - **Sample file:** https://tsduck.io/streams/usa-atsc/503.ts - **Command:** `./ccextractor --program-number 4 503.ts` - **Original report:** Only 2 captions extracted - **Current version:** **30 captions extracted** (matching caption-inspector output) ### Caption Comparison The captions now match the expected output from caption-inspector: | # | CCExtractor Output | Caption-Inspector Reference | |---|-------------------|---------------------------| | 1 | 00:00:00,601 - "d so every time / you start trimming," | ✅ Match | | 2 | 00:00:02,403 - "all of those hairs are being..." | ✅ Match | | ... | ... | ... | | 30 | "This is accomplished easily with the graduating guide combs" | ✅ Match | ### Likely Fixes Several commits since the original report (Dec 2023) have improved the CEA-708 decoder: - `3c51fb65` - Handle row_count decrease in CEA-708 C decoder - `d6ccf1bf` - Port 708 decoder encoding module to Rust - Various window handling and display logic improvements The issue was likely related to improper handling of CEA-708 window toggle (TGW) and delete layer (DLW) commands, causing captions to accumulate instead of being flushed at proper display boundaries. This issue can be closed as fixed. 🤖 Generated with [Claude Code](https://claude.com/claude-code)

claunia commented

2026-01-29 16:53:38 +00:00

@cfsmp3 commented on GitHub (Dec 14, 2025):

Closing as fixed in master. The CEA-708 decoder improvements since v0.94 now correctly extract all 30 captions from the test file.

@cfsmp3 commented on GitHub (Dec 14, 2025): Closing as fixed in master. The CEA-708 decoder improvements since v0.94 now correctly extract all 30 captions from the test file.

Sign in to join this conversation.

1 Participants

Notifications

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: starred/ccextractor#793