doocus update and meeting list export

ctrl/batch.sh

@@ -7,14 +7,19 @@
 # the mirrored output subfolder.
 #
 # Usage:
-#   ctrl/batch.sh -i <input-dir> -o <output-dir> [-e "mkv mp4 ..."] [-n] \
-#                 [-- <process_meeting.py flags>]
+#   ctrl/batch.sh -i <input-dir> -o <output-dir> [-e "mkv mp4 ..."] \
+#                 [-l <list-file>] [-n] [-- <process_meeting.py flags>]
 #
 # Examples:
 #   # Everything under a mounted share, default extraction flags forwarded
 #   ctrl/batch.sh -i "/mnt/win/trainings" -o output/batch \
 #       -- --embed-images --scene-detection --diarize
 #
+#   # Only the meetings doocus found (relative paths), re-rooted at a new mount,
+#   # output into the doocus folder for later packaging:
+#   ctrl/batch.sh -i "/mnt/drive" -o docs-output -l docs-output/meetings.txt \
+#       -- --embed-images --scene-detection --diarize
+#
 #   # Dry run: just show which videos map to which output folders
 #   ctrl/batch.sh -i "/mnt/win/trainings" -o output/batch -n
 #
@@ -42,6 +47,7 @@ INPUT=""
 OUTPUT=""
 EXTS="mkv mp4 mov avi m4v webm wmv ogg mp3 wav m4a opus flac aac"
 DRY=false
+LIST=""          # optional newline list of RELATIVE paths (re-rooted at INPUT)
 FORWARD=()
 
 while [[ $# -gt 0 ]]; do
@@ -49,6 +55,7 @@ while [[ $# -gt 0 ]]; do
     -i|--input)   INPUT="$2"; shift 2 ;;
     -o|--output)  OUTPUT="$2"; shift 2 ;;
     -e|--ext)     EXTS="$2"; shift 2 ;;
+    -l|--list)    LIST="$2"; shift 2 ;;
     -n|--dry-run) DRY=true; shift ;;
     -h|--help)    usage; exit 0 ;;
     --)           shift; FORWARD=("$@"); break ;;
@@ -59,6 +66,7 @@ done
 [ -n "$INPUT" ]  || { echo "ERROR: -i/--input is required" >&2; exit 1; }
 [ -n "$OUTPUT" ] || { echo "ERROR: -o/--output is required" >&2; exit 1; }
 [ -d "$INPUT" ]  || { echo "ERROR: input dir not found: $INPUT" >&2; exit 1; }
+[ -z "$LIST" ] || [ -f "$LIST" ] || { echo "ERROR: list file not found: $LIST" >&2; exit 1; }
 
 # Absolute paths so relative-path math is stable regardless of where we cd'd.
 INPUT="$(realpath "$INPUT")"
@@ -74,16 +82,35 @@ unset 'find_expr[${#find_expr[@]}-1]'   # drop the trailing -o
 
 echo "Input : $INPUT"
 echo "Output: $OUTPUT"
-echo "Exts  : $EXTS"
+[ -n "$LIST" ] && echo "List  : $LIST (relative paths, re-rooted at input)" || echo "Exts  : $EXTS"
 [ "$DRY" = true ] && echo "(dry run — nothing will be processed)"
 echo
 
+# Producer: either the given list of relative paths (re-rooted at INPUT) or a
+# recursive find by extension. Both emit null-delimited absolute paths.
+producer() {
+  if [ -n "$LIST" ]; then
+    while IFS= read -r rel || [ -n "$rel" ]; do
+      [ -z "$rel" ] && continue
+      printf '%s\0' "$INPUT/$rel"
+    done < "$LIST"
+  else
+    find "$INPUT" -type f \( "${find_expr[@]}" \) -print0 | sort -z
+  fi
+}
+
 total=0 ok=0 fail=0
 # Process substitution (not a pipe) so counters survive into the summary.
 while IFS= read -r -d '' video; do
   total=$((total + 1))
 
   rel="${video#"$INPUT"/}"          # path relative to the input root
+
+  if [ ! -f "$video" ]; then        # list may name a path missing under this root
+    echo "[$total] $rel  !! not found under input (skipped)" >&2
+    fail=$((fail + 1))
+    continue
+  fi
   reldir="$(dirname "$rel")"
   if [ "$reldir" = "." ]; then
     outdir="$OUTPUT"                 # video sat directly in the input root
@@ -110,7 +137,7 @@ while IFS= read -r -d '' video; do
     fail=$((fail + 1))
   fi
   echo
-done < <(find "$INPUT" -type f \( "${find_expr[@]}" \) -print0 | sort -z)
+done < <(producer)
 
 echo "----------------------------------------"
 if [ "$DRY" = true ]; then