[x15sr71]

In raising this pull request, I confirm the following (please check boxes):

Reason for this PR:

• This PR adds new functionality.
• This PR fixes a bug that I have personally experienced or that a real user has reported and for which a sample exists.
• This PR is porting code from C to Rust.

Sanity check:

• I have read and understood the contributors guide.
• I have checked that another pull request for this purpose does not exist.
• If the PR adds new functionality, I've added it to the changelog. If it's just a bug fix, I have NOT added it to the changelog.
• I am NOT adding new C code unless it's to fix an existing, reproducible bug.

Repro instructions:

RT78, on sample 72 (dcada74…e1ce):

ccextractor <sample>.mpg --autoprogram --out=ttxt --latin1 --datets

Three captions that the RT78 reference shows for several seconds come out on master as a split second. Before #2292 they were zero-length (the third only in its first half).

Caption	Reference	Master
"Morsomt, Dennis, å skremme vettet av folk?"	3.70 s	40 ms
"Forretningspartneren min vil ha avtalen i havn innen en uke. Greit?"	5.62 s	40 ms
"Har dere problemer, får dere hjelp av meg personlig."	4.38 s	split: "Har dere problemer," lasts 20 ms, then "får dere hjelp av meg personlig." alone

---

Cause

This sample carries three pages with the subtitle flag set, all in parallel transmission mode: 333 (the page RT78 extracts), 777, and 444 (headers only).

#725 added a de_ctr counter to telxcc.c (1b3598b, narrowed in 1513b7c). While our page has received rows, a subtitle-flagged header from another magazine is no longer ignored. It reaches process_page(), which ends our page at that header's timestamp and starts a new buffer.

At all three captions, a magazine-7 header (7FF, 7FF, 777) arrives 20–40 ms after page 333's header: after its rows for the first two, and between them for the third, which is why it splits. Page 333's next header comes 3.74, 5.66 and 4.42 s later, with the erase bit set and no rows.

ETS 300 706 §7.2.1, also quoted in the comment above these checks, says a page "is terminated by and excludes the next page header packet having the same magazine address in parallel transmission mode". So a magazine-7 header doesn't end page 333. FFmpeg's libzvbi decoder agrees: it shows each of the three as one page, with no change at the magazine-7 headers.

Issue #699 (last lines missing from .bin output) was fixed by #725's other change, which writes the final page at end of input (rcwt_loop(), telxcc_close()). In #725's review, alexbrt noted that "the fix for the multiple lines shouldn't affect timing at all".

Builds of #725 and its parent (the first two match the platform's output for those commits):

Build	RT78 output	The three captions
4438e6c (#725's parent)	84f651b… (test 676)	3.70 / 5.62 / 4.38 s
fdd5b6b (#725 merged)	1928f66… (test 677)	0 ms, 0 ms, split
fdd5b6b, only telxcc.c reverted	84f651b…	3.70 / 5.62 / 4.38 s

Fix

Remove de_ctr and the three conditions that use it, so headers of other pages are ignored again, as before #725. The end-of-input flush and the multi-page logic from #665 are unchanged.

Testing

Master (d7eb132e) vs master + this commit, built with linux/build -hardsubx (Ubuntu 24.04, aarch64):

• RT78: the three captions last 3.74 / 5.66 / 4.42 s, as in the stream. The output now matches the reference apart from Fix 40ms gap in rolling teletext subtitles #2292's +40 ms on every end time and one extra space that master already has.
• Every other teletext test is byte-identical: RT18, RT19, RT64–77, RT79–83 and RT234–236. A trace build shows the removed check firing only at RT78's three captions, and never on any other sample.
• Few lines missing from output of .bin file [sample attached] #699 is still fixed: on sample 67, extracting from the .bin gives the same 14 cues as the original file (12 before Issue 699 solved #725).

Notes

• Multi-page mode (--tpage 333 --tpage 777, not covered by any test) also changes: one caption split into three fragments becomes one, and two captions go from 40/60 ms to 2.06/3.74 s.

[x15sr71]

The Linux build and format failures aren't from this change: ubuntu-latest moved to Ubuntu 26.04, which has no libgpac-dev. They should be fixed by #2428.