Judge a short line on a shorter run, and catch a bare music note

Printing the first line of all 65 pods in the corpus found the one failure that
actually matters -- a break opening on the programme -- in about eight of them.
Three were identical: "I'm Jack Resider.", which is the Darknet Diaries host
saying his own name, misheard. Rhysider is not a word whisper knows.

Four words out of four means one misheard word anywhere in a short line hides
the whole line from the master, so it is never given back. Three of "i m jack
resider" still finds "i m jack" where the publisher wrote "I'm Jack Rhysider".
Short lines are now judged on three; long ones are unchanged, where a full run
is affordable and a looser one would match by accident.

The annotation pattern also missed a bare "♪", which is a music note without
brackets round it.

Measured across the eighteen episodes holding a master:

    before   99% - 106%, mean 101%
    after    92% - 104%, mean 100%

The three that fell furthest are the three Darknet Diaries episodes, and they
fell because the host's sentence is no longer inside the break. Falling below
the length gap is the right answer there: the gap counts advertising, and his
name was never advertising.

Eight suspect pod openings are down to four, and all four are now either a
one-word line too short to judge at all ("Yay!", "- Pessimism.") or a line that
is mostly advertisement with a second of handoff in front of it.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018AwnALZL6bD4VBtATDVv3B

1 files changed, 11 insertions, 1 deletions

diff --git a/server/internal/transcript/trim.go b/server/internal/transcript/trim.go
index f2ded4c..a853800 100644
--- a/server/internal/transcript/trim.go
+++ b/server/internal/transcript/trim.go
@@ -40,6 +40,9 @@ const (
 	// errs towards smaller candidates, which is the safe direction: a skip into the programme
 	// is worse than a missed advertisement.
 	trimRunWords = 4
+	// shortLineWords is where a line is short enough that requiring a full-length run would
+	// mean requiring nearly all of it to be heard correctly.
+	shortLineWords = 5
 	// trimMinWords is the shortest line this can judge. A two-word line gives nothing to look
 	// for, so it is kept rather than guessed at, which is what leaves the music markers alone.
 	trimMinWords = 3
@@ -92,7 +95,7 @@ func trimmed(found []Candidate, master, served []Cue) []Candidate {
 // never anchor -- which leaves them sitting at the front of a break, widening it. Unlike a
 // title card, which this deliberately leaves alone, an annotation says what it is by its shape,
 // so it can be told from an advertisement without guessing.
-var annotation = regexp.MustCompile(`^\s*[\[(][^\])]{0,40}[\])]\s*$`)
+var annotation = regexp.MustCompile(`^\s*(?:[\[(][^\])]{0,40}[\])]|[\x{266a}\x{266b}\x{2669}\x{266c}\s]+)\s*$`)
 
 // inMaster reports whether a line's words appear in the master.
 // trimmable reports whether a line at the edge of a break belongs to the break.
@@ -105,7 +108,14 @@ func inMaster(cue Cue, runs map[string]bool) bool {
 	if len(words) < trimMinWords {
 		return false
 	}
+	// A short line is judged on a shorter run. Four words out of four means one misheard word
+	// anywhere in it hides the whole line -- which is how "I'm Jack Rhysider" came back as
+	// "I'm Jack Resider" and opened three Darknet Diaries breaks on the host's own sentence.
+	// Three of "i m jack resider" still finds "i m jack" in the master.
 	size := trimRunWords
+	if len(words) <= shortLineWords {
+		size = trimMinWords
+	}
 	if len(words) < size {
 		size = len(words)
 	}