Add continuity report with per-junction gaps and copy-pasteable runs

When a clip boundary's audio doesn't match, print a full table of every
junction (capture-time gap, confidence, verdict) plus the input clips
grouped into continuous runs, each formatted as a ready-to-paste clip
list -- so a discontinuous batch can be immediately re-run on just the
subset that's actually one sequence, without manually diffing timestamps.

concat_live_clips.py

@@ -32,6 +32,7 @@ import re
 import subprocess
 import sys
 import tempfile
+from datetime import datetime
 from pathlib import Path
 
 try:
@@ -104,6 +105,61 @@ def best_overlap(a_tail, b_head, sr=SR, kmin_s=0.05, coarse_step_s=0.002, refine
     return best_k / sr, best_score, rms
 
 
+def fetch_creation_time(clip):
+    """Best-effort capture timestamp for a clip, preferring the
+    local-time Apple tag (matches what Photos/Finder show) and falling
+    back to the UTC container tag. Returns None if neither is present or
+    parseable."""
+    out = subprocess.run(
+        ["ffprobe", "-v", "error", "-show_entries",
+         "format_tags=com.apple.quicktime.creationdate,creation_time",
+         "-of", "default=noprint_wrappers=1:nokey=0", str(clip)],
+        check=True, capture_output=True, text=True,
+    ).stdout
+    tags = dict(line.split("=", 1) for line in out.strip().splitlines() if "=" in line)
+    raw = tags.get("TAG:com.apple.quicktime.creationdate") or tags.get("TAG:creation_time")
+    if not raw:
+        return None
+    raw = raw.strip().replace("Z", "+0000")
+    for fmt in ("%Y-%m-%dT%H:%M:%S%z", "%Y-%m-%dT%H:%M:%S.%f%z"):
+        try:
+            return datetime.strptime(raw, fmt)
+        except ValueError:
+            continue
+    return None
+
+
+def format_gap(gap):
+    return f"{gap:.0f}s" if gap is not None else "unknown"
+
+
+def print_continuity_report(clips, results):
+    """results: list of (gap_seconds_or_None, score, rms, verdict) per
+    junction, one entry per consecutive clip pair."""
+    print("\nContinuity report:", file=sys.stderr)
+    header = f"  {'junction':<40} {'gap':>8} {'confidence':>11} {'verdict':>10}"
+    print(header, file=sys.stderr)
+    for i, (gap, score, rms, verdict) in enumerate(results):
+        junction = f"{clips[i].name} -> {clips[i + 1].name}"
+        label = {"match": "match", "quiet": "inconclusive", "no match": "NO MATCH"}[verdict]
+        print(f"  {junction:<40} {format_gap(gap):>8} {score:>11.3f} {label:>10}", file=sys.stderr)
+
+    runs = [[clips[0]]]
+    for i, (_, _, _, verdict) in enumerate(results):
+        if verdict == "no match":
+            runs.append([clips[i + 1]])
+        else:
+            runs[-1].append(clips[i + 1])
+
+    print("\nContinuous runs (copy-paste to re-run on just that subset):", file=sys.stderr)
+    for n, run in enumerate(runs, 1):
+        if len(run) >= 2:
+            print(f"  Run {n} ({len(run)} clips): {' '.join(c.name for c in run)}", file=sys.stderr)
+        else:
+            print(f"  Run {n} (1 clip): {run[0].name} -- nothing to concatenate on its own", file=sys.stderr)
+    print(file=sys.stderr)
+
+
 def detect_overlaps(clips, confidence_threshold, min_signal_rms, allow_discontinuous):
     """Measure the audio overlap at each clip boundary. Junctions with a
     confident match are trimmed accordingly. Junctions with a poor match
@@ -120,33 +176,47 @@ def detect_overlaps(clips, confidence_threshold, min_signal_rms, allow_discontin
             raws.append(raw)
         audios = [load_audio(r) for r in raws]
 
+    creation_times = [fetch_creation_time(c) for c in clips]
+
     overlaps = []
-    discontinuities = []
+    results = []
+    has_discontinuity = False
     for i in range(len(audios) - 1):
         overlap_s, score, rms = best_overlap(audios[i], audios[i + 1])
         pair = f"{clips[i].name} -> {clips[i + 1].name}"
+        gap = None
+        if creation_times[i] is not None and creation_times[i + 1] is not None:
+            gap = (creation_times[i + 1] - creation_times[i]).total_seconds()
+
         if score >= confidence_threshold:
+            verdict = "match"
             print(f"  {clips[i + 1].name}: trimming {overlap_s:.3f}s of duplicated "
                   f"start (confidence {score:.3f})", file=sys.stderr)
         elif rms < min_signal_rms:
+            verdict = "quiet"
             print(f"warning: {pair} is too quiet (rms={rms:.0f}) to verify overlap "
                   f"via audio; assuming continuous with 0 overlap", file=sys.stderr)
             overlap_s = 0.0
         else:
+            verdict = "no match"
             print(f"warning: {pair} audio does not match (confidence {score:.3f}, "
-                  f"rms={rms:.0f}) -- these clips may not be a continuous sequence",
-                  file=sys.stderr)
-            discontinuities.append(pair)
+                  f"rms={rms:.0f}, gap {format_gap(gap)}) -- these clips may not be "
+                  f"a continuous sequence", file=sys.stderr)
+            has_discontinuity = True
             overlap_s = 0.0
-        overlaps.append(overlap_s)
 
-    if discontinuities and not allow_discontinuous:
-        sys.exit(
-            "error: audio doesn't match at " + ", ".join(discontinuities) + " -- "
-            "these clips don't look like a continuous sequence (wrong order, "
-            "missing clip, or unrelated files?). Re-check the input, or pass "
-            "--allow-discontinuous to concatenate them anyway."
-        )
+        overlaps.append(overlap_s)
+        results.append((gap, score, rms, verdict))
+
+    if has_discontinuity:
+        print_continuity_report(clips, results)
+        if not allow_discontinuous:
+            sys.exit(
+                "error: these clips don't look like a single continuous sequence "
+                "(wrong order, a missing clip, or unrelated files?). See the "
+                "continuity report above -- re-run on one of the listed runs, or "
+                "pass --allow-discontinuous to concatenate everything anyway."
+            )
 
     return overlaps