feat(cc608): choose the timecc608 caption mode, defaulting to paint-on

[tobbee]

Replaces #326, which fixed only the pop-on flip placement. With go-608 v0.9.0 generating all three CTA-608 caption modes per unit, timecc608 can name the mode instead, and the default becomes the one whose segments are independent.

The value is now <channel>-<lang>[-<mode>[-<modifier>]]:

URL	mode	caption appears	self-contained segments
timecc608_CC1-eng	paint-on (default)	types on, two characters per frame	yes
timecc608_CC1-eng-roll3	roll-up (roll2, roll3)	types onto a scrolling window	yes
timecc608_CC1-eng-pop	pop-on	whole, exactly on its second	no
timecc608_CC1-eng-pop-sc	pop-on	whole, ~0.5 s late	yes

Paint-on is the default. Each cue clears the screen and writes the caption straight onto it, so no cue's data leaves its own segment: every segment decodes standalone and a client that starts, seeks or joins anywhere is correct from the first cue boundary it sees. The typing doubles as a liveness tell.

Roll-up types each line onto the base row of a window that scrolls the earlier lines up, reset at the start of every segment. Limited to 2 or 3 rows because the caption's own lines put its base row at row 3 — a 4-row window would need a row 0.

Pop-on keeps both placements. -pop puts each EOC on its cue's first frame with the build draining over the frames before it, so the caption is displayed over exactly the interval its text names; the cost is that a cue's build lives in the preceding segment, and a receiver joining mid-stream gets a leading EOC without it — blank on a fresh decoder, one cue of a stale caption on one that keeps 608 state, correct from the next cue on. A sender cannot paper over that: an ENM ahead of the EOC would erase the build about to be flipped. -pop-sc trades the frame accuracy back for a segment that stands alone.

Also in here

• go-608 v0.7.0 → v0.9.0. The scheduling is go-608's per-unit builders throughout; the local copy of that scheduler is gone. Their generate.Unit carries a unit's number, start time and frame count as independent facts, so a segment whose number does not match its media time (startnr_, a non-zero availabilityStartTime) still gets the right clock, and the next unit's start comes from the fragment's media duration in ticks rather than frames × frame duration, which a fractional frame rate would drift on.
• A cue is never shorter than a second — v0.9.0's NumCues divides a segment down into whole periods, so a 1.92 s segment gets one 1.92 s cue instead of two of 0.96 s. Segments of 2 s and 2.002 s are unaffected.
• The language must be a three-letter ISO 639-2 code. It is used in exactly one place, the SCTE 214-1 accessibility descriptor value (value="CC1=eng"), whose scheme is defined in those terms — not DASH's RFC 5646 @lang, which this option never sets. en, en-US and zh-Hans were previously passed straight into a non-conformant descriptor. A single-token language also makes the fields positional, so a mistyped mode or modifier is reported instead of being read as part of the language.

Verification

The tests decode the injected cc_data back in presentation order — the order a conformant receiver reassembles it in — and assert the caption text and the frame it lands on: across two consecutive segments for the cross-segment pop-on case, for AVC and HEVC, at 30, 29.97 and 24 fps, through the low-latency chunked path, and per frame for paint-on's clear-and-type and roll-up's scroll history. Test configurations come from CreateCC608Config, so each test names its mode exactly as a request does.

go test ./... and golangci-lint run are clean. Not yet watched in a real player.

🤖 Generated with Claude Code