#!/usr/bin/env bash
# videocmp_select.sh
#
# Modes:
# 1) Pair mode: compare A and B
#    ./videocmp_select.sh A.mp4 B.mp4 [options]
#
# 2) Directory scan mode: group files by the **first token before the first underscore** in the basename
#    ./videocmp_select.sh --scan-dir [DIR] [--recursive] [options]
#    (DIR defaults to "." if omitted)
#
# Examples of grouping by first token:
#   Becky_California.mp4, Becky_California.1.mp4     → group key "Becky"
#   Bianca_Golden.mp4, Bianca_Golden_Cut.mp4         → group key "Bianca"
#   Emylia.mp4, Emylia_Argan.mp4, Emylia_Wish.mp4    → group key "Emylia"
#
# Pipeline:
#   1) Validate files (ffprobe fields + ffmpeg deep decode)
#   2) Confirm same movie via snapshot SSIM @ --snapshot-time (default 12s)
#   3) Optional: run external compare impl (simple/advanced) for logging
#   4) Pick preferred: prefer --prefer-height (default 720), then longer duration, then larger file
#   5) Act on loser: --action print|move|delete (with --dry-run)
#
# Common options:
#   --snapshot-time SEC        (default: 12)
#   --snapshot-scale WxH       (default: 320:-1)
#   --snapshot-ssim THRESH     (default: 0.97)
#   --impl simple|advanced     (default: simple)   # logs only
#   --impl-simple PATH         (default: ./compare_simple.sh)
#   --impl-advanced PATH       (default: ./compare_advanced.sh)
#   --impl-optional            (default) warn if impl missing
#   --impl-required            error if chosen impl missing
#   --prefer-height N          (default: 720)
#   --duration-eps SEC         (default: 0.0)
#   --action print|move|delete (default: print)
#   --trash-dir PATH           (default: $HOME/.video_trash)
#   --dry-run
#   --verbose
#
# Directory-scan options:
#   --scan-dir [DIR]           Enable directory mode (DIR optional; default ".")
#   --recursive, -r            Recurse into subfolders
#   (Note: delimiter-based grouping is deprecated; this script now groups by first underscore token.)
#
# Exit codes:
#   0 success | 1 differ/broken | 2 usage | 3 missing dependency
set -u

# -------- defaults --------
SNAP_T="12"
SNAP_SCALE="320:-1"
SNAP_SSIM="0.97"

IMPL="simple"
IMPL_SIMPLE="./compare_simple.sh"
IMPL_ADV="./compare_advanced.sh"
IMPL_REQUIRED=0

PREF_HEIGHT=720
DURATION_EPS=0.0

ACTION="print"
TRASH_DIR="${HOME}/.video_trash"
DRY=0
VERBOSE=0

SCAN_DIR=""
RECURSIVE=0

# -------- helpers --------
need() { command -v "$1" >/dev/null 2>&1 || { echo "Missing dependency: $1" >&2; exit 3; }; }
v() { [[ $VERBOSE -eq 1 ]] && echo "[LOG]" "$@" >&2; }
die() { echo "[ERR]" "$@" >&2; exit 1; }

need ffmpeg; need ffprobe; need awk; need grep; need stat; need sed; need tr; need find

# -------- array-based option parser --------
ARGS=("$@")
REM_ARR=()
i=0
while (( i < ${#ARGS[@]} )); do
  arg="${ARGS[i]}"
  case "$arg" in
    --snapshot-time)    SNAP_T="${ARGS[i+1]:-}"; i=$((i+2));;
    --snapshot-scale)   SNAP_SCALE="${ARGS[i+1]:-}"; i=$((i+2));;
    --snapshot-ssim)    SNAP_SSIM="${ARGS[i+1]:-}"; i=$((i+2));;
    --impl)             IMPL="${ARGS[i+1]:-}"; i=$((i+2));;
    --impl-simple)      IMPL_SIMPLE="${ARGS[i+1]:-}"; i=$((i+2));;
    --impl-advanced)    IMPL_ADV="${ARGS[i+1]:-}"; i=$((i+2));;
    --impl-optional)    IMPL_REQUIRED=0; i=$((i+1));;
    --impl-required)    IMPL_REQUIRED=1; i=$((i+1));;
    --prefer-height)    PREF_HEIGHT="${ARGS[i+1]:-}"; i=$((i+2));;
    --duration-eps)     DURATION_EPS="${ARGS[i+1]:-}"; i=$((i+2));;
    --action)           ACTION="${ARGS[i+1]:-}"; i=$((i+2));;
    --trash-dir)        TRASH_DIR="${ARGS[i+1]:-}"; i=$((i+2));;
    --dry-run)          DRY=1; i=$((i+1));;
    --verbose)          VERBOSE=1; i=$((i+1));;
    --scan-dir)
      # optional arg: use next token unless it looks like another option
      next="${ARGS[i+1]:-}"
      if [[ -n "$next" && "$next" != --* ]]; then
        SCAN_DIR="$next"; i=$((i+2))
      else
        SCAN_DIR="."; i=$((i+1))
      fi
      ;;
    --recursive|-r)     RECURSIVE=1; i=$((i+1));;
    --delimiter|--delim) # backward-compat: ignore but warn once
      echo "[WARN] --delimiter is deprecated; grouping now uses first underscore token." >&2
      i=$((i+ (arg=="--delimiter"?2:1) )) ;;
    --help|-h)
      grep -E '^# ' "$0" | sed 's/^# //'
      exit 0 ;;
    *)
      # leave positional for pair mode
      REM_ARR+=("$arg"); i=$((i+1));;
  esac
done

v "[DBG] Options: SCAN_DIR='${SCAN_DIR:-}' RECURSIVE=$RECURSIVE PREF_HEIGHT=$PREF_HEIGHT SNAP_T=$SNAP_T"

# -------- core functions --------
probe_meta() { # file -> "w h dur codec size"
  local f="$1" size w h dur vcodec
  size=$(stat -c%s "$f" 2>/dev/null || stat -f%z "$f")
  w=$(ffprobe -v error -select_streams v:0 -show_entries stream=width  -of default=nw=1:nk=1 "$f" || echo 0)
  h=$(ffprobe -v error -select_streams v:0 -show_entries stream=height -of default=nw=1:nk=1 "$f" || echo 0)
  dur=$(ffprobe -v error -show_entries format=duration -of default=nw=1:nk=1 "$f" || echo 0)
  vcodec=$(ffprobe -v error -select_streams v:0 -show_entries stream=codec_name -of default=nw=1:nk=1 "$f" || echo "?")
  echo "$w $h $dur $vcodec $size"
}

check_ok() { # file -> 0 ok / 1 bad
  local f="$1" w h dur
  w=$(ffprobe -v error -select_streams v:0 -show_entries stream=width  -of default=nw=1:nk=1 "$f" 2>/dev/null || echo "")
  h=$(ffprobe -v error -select_streams v:0 -show_entries stream=height -of default=nw=1:nk=1 "$f" 2>/dev/null || echo "")
  dur=$(ffprobe -v error -show_entries format=duration -of default=nw=1:nk=1 "$f" 2>/dev/null || echo "")
  if [[ -z "$w" || -z "$h" || "$w" = "N/A" || "$h" = "N/A" || "$w" -eq 0 || "$h" -eq 0 ]]; then
    echo "BROKEN(ffprobe: no valid video stream): $f" >&2; return 1; fi
  if [[ -n "$dur" && "$dur" != "N/A" ]]; then
    awk -v d="$dur" 'BEGIN{exit !(d+0>0)}' || { echo "BROKEN(ffprobe: nonpositive duration): $f" >&2; return 1; }
  fi
  if ! ffmpeg -v error -xerror -i "$f" -f null - -nostats >/dev/null 2>&1; then
    echo "BROKEN(ffmpeg decode): $f" >&2; return 1; fi
  return 0
}

snapshot_compare_ssim() { # f1 f2 time scale -> "ssim" (empty if fail)
  local f1="$1" f2="$2" t="$3" sc="$4"
  local tmpd s1 s2 log ssim
  tmpd="$(mktemp -d)"; s1="$tmpd/1.png"; s2="$tmpd/2.png"; log="$tmpd/cmp.log"
  ffmpeg -hide_banner -v error -y -ss "$t" -i "$f1" -frames:v 1 -vf "scale=$sc,format=yuv420p" "$s1" || true
  ffmpeg -hide_banner -v error -y -ss "$t" -i "$f2" -frames:v 1 -vf "scale=$sc,format=yuv420p" "$s2" || true
  if [[ ! -s "$s1" || ! -s "$s2" ]]; then rm -rf "$tmpd"; echo ""; return 0; fi
  ffmpeg -hide_banner -v info -i "$s1" -i "$s2" -lavfi "ssim" -f null - > /dev/null 2> "$log" || true
  ssim="$(grep -Eo 'All:[0-9]+(\.[0-9]+)?' "$log" | head -n1 | cut -d: -f2)"
  [[ -z "$ssim" ]] && ssim="$(grep -Eo 'SSIM [^ ]* All:[0-9]+(\.[0-9]+)?' "$log" | awk -F'All:' '{print $2}' | head -n1)"
  [[ -z "$ssim" ]] && ssim="$(grep -Eo 'SSIM Y:[0-9]+(\.[0-9]+)?' "$log" | head -n1 | cut -d: -f2)"
  rm -rf "$tmpd"; echo "$ssim"
}

run_impl() { # impl, A, B
  local which="$1" f1="$2" f2="$3" path=""
  [[ "$which" == "advanced" ]] && path="$IMPL_ADV" || path="$IMPL_SIMPLE"
  if [[ ! -x "$path" ]]; then
    if [[ $IMPL_REQUIRED -eq 1 ]]; then
      die "Requested --impl=$which but script not found/executable at $path"
    else
      echo "[impl:$which] not found ($path) — skipping" >&2; return 0
    fi
  fi
  echo "[impl:$which] $path \"$f1\" \"$f2\"" >&2
  "$path" "$f1" "$f2" 2>&1 | sed -n '1,12p' >&2
}

score_file() { # file -> "tier720 dur size"
  local f="$1"
  read -r W H DUR VC SIZE <<<"$(probe_meta "$f")"
  local tier=1; [[ "$H" -eq "$PREF_HEIGHT" ]] && tier=0
  echo "$tier $DUR $SIZE"
}

pick_winner() { # A B -> "KEEP|DROP|why"
  local a="$1" b="$2"
  read -r aTier aDur aSize <<<"$(score_file "$a")"
  read -r bTier bDur bSize <<<"$(score_file "$b")"
  v "Quality scores: A[tier=$aTier dur=$aDur size=$aSize] B[tier=$bTier dur=$bDur size=$bSize]"
  if (( aTier < bTier )); then echo "$a|$b|prefer ${PREF_HEIGHT}p (A)"; return 0; fi
  if (( bTier < aTier )); then echo "$b|$a|prefer ${PREF_HEIGHT}p (B)"; return 0; fi
  awk -v A="$aDur" -v B="$bDur" -v eps="$DURATION_EPS" 'BEGIN{
    if ((A-B) > eps) print "A"; else if ((B-A) > eps) print "B"; else print "TIE";
  }' | {
    read who
    if [[ "$who" == "A" ]]; then echo "$a|$b|longer duration (A)"; return 0; fi
    if [[ "$who" == "B" ]]; then echo "$b|$a|longer duration (B)"; return 0; fi
    if (( aSize > bSize )); then echo "$a|$b|larger file size (A)"; else
    if (( bSize > aSize )); then echo "$b|$a|larger file size (B)"; else
      echo "$a|$b|tie-break (keep A)"; fi; fi
  }
}

act_on_loser() { # loser keep
  local loser="$1" keeper="$2"
  case "$ACTION" in
    print)
      echo "[ACTION] Keep:  $keeper"
      echo "[ACTION] Drop:  $loser"
      ;;
    move)
      mkdir -p -- "$TRASH_DIR"
      if [[ $DRY -eq 1 ]]; then
        echo "[ACTION] dry-run: mv \"$loser\" \"$TRASH_DIR/\""
      else
        mv -- "$loser" "$TRASH_DIR/" && echo "[ACTION] moved to trash: $loser -> $TRASH_DIR/"
      fi
      echo "[ACTION] kept: $keeper"
      ;;
    delete)
      if [[ $DRY -eq 1 ]]; then
        echo "[ACTION] dry-run: rm \"$loser\""
      else
        rm -- "$loser" && echo "[ACTION] deleted: $loser"
      fi
      echo "[ACTION] kept: $keeper"
      ;;
    *) echo "[WARN] Unknown --action='$ACTION' → printing only."; echo "[ACTION] Keep: $keeper ; Drop: $loser" ;;
  esac
}

same_movie_or_skip() { # A B -> 0 if same (SSIM>=thr), else 1
  local a="$1" b="$2" ssim
  echo "== Snapshot compare @${SNAP_T}s: ==" >&2
  ssim="$(snapshot_compare_ssim "$a" "$b" "$SNAP_T" "$SNAP_SCALE")"
  if [[ -z "$ssim" ]]; then
    echo "[WARN] Could not compute SSIM for: $a  vs  $b" >&2; return 1; fi
  printf "[INFO] SSIM(All) %s vs %s → %s\n" "$(basename -- "$a")" "$(basename -- "$b")" "$ssim" >&2
  awk -v s="$ssim" -v thr="$SNAP_SSIM" 'BEGIN{exit !(s+0 >= thr+0)}'
}

# ----- directory scan helpers/drivers -----
scan_and_collect() {
  if [[ $RECURSIVE -eq 1 ]]; then
    mapfile -t FILES < <(find "$SCAN_DIR" -type f \( -iname '*.mp4' -o -iname '*.m4v' -o -iname '*.mov' -o -iname '*.mkv' \))
  else
    mapfile -t FILES < <(find "$SCAN_DIR" -maxdepth 1 -type f \( -iname '*.mp4' -o -iname '*.m4v' -o -iname '*.mov' -o -iname '*.mkv' \))
  fi
}

# New grouping rule: key = first token of basename (without extension) before the first underscore.
# If there is no underscore, the whole basename (without extension) is the key.
first_underscore_key() {
  local base extless key
  base="$(basename -- "$1")"
  extless="${base%.*}"
  if [[ "$extless" == *_* ]]; then
    key="${extless%%_*}"
  else
    key="$extless"
  fi
  echo "$key"
}

process_group() { # files...
  local files=("$@") n=${#files[@]}
  if (( n < 2 )); then v "Group <2 files → skip"; return 0; fi
  echo; echo "=== Group (${n} files): first-underscore key '$(first_underscore_key "${files[0]}")' ==="
  echo "Files:"; for f in "${files[@]}"; do echo "  - $f"; done

  local best="${files[0]}"
  if ! check_ok "$best"; then echo "[WARN] Skipping broken file: $best"; return 0; fi

  for ((i=1;i<n;i++)); do
    local cand="${files[i]}"
    if ! check_ok "$cand"; then echo "[WARN] Skipping broken file: $cand"; continue; fi
    echo; echo "--- Compare: $(basename -- "$best")  vs  $(basename -- "$cand") ---"
    if ! same_movie_or_skip "$best" "$cand"; then
      echo "[SKIP] Snapshot says NOT same movie; leaving both in place."; continue; fi
    run_impl "$IMPL" "$best" "$cand"
    read -r keep drop why <<<"$(pick_winner "$best" "$cand" | tr '|' ' ')"
    echo "[DECISION] Keep: $keep"; echo "[DECISION] Drop: $drop"; echo "[REASON]   $why"
    act_on_loser "$drop" "$keep"
    best="$keep"
  done

  echo; echo "=== Group result: kept $(basename -- "$best") ==="
}

dir_mode() {
  [[ -n "${SCAN_DIR:-}" ]] || SCAN_DIR="."
  [[ -d "$SCAN_DIR" ]] || die "Not a directory: $SCAN_DIR"
  echo ">> Directory scan mode on: $SCAN_DIR   (recursive=$RECURSIVE, grouping=first-underscore)"
  scan_and_collect
  if [[ ${#FILES[@]} -eq 0 ]]; then echo "No video files found."; exit 0; fi

  declare -A groups
  local f base key
  for f in "${FILES[@]}"; do
    base="$(basename -- "$f")"
    key="$(first_underscore_key "$base")"
    groups["$key"]+=$'\n'"$f"
  done

  for k in "${!groups[@]}"; do
    IFS=$'\n' read -r -d '' -a grpfiles < <(printf "%s" "${groups[$k]}" | sed '/^$/d' | sort -u && printf '\0')
    process_group "${grpfiles[@]}"
  done

  echo; echo ">> Directory scan complete."
}

# ----- pair comparison driver -----
compare_pair() { # A B
  local A="$1" B="$2"
  echo "== Step 1: Validating files =="; okA=0; okB=0
  check_ok "$A" && okA=1; check_ok "$B" && okB=1
  if (( okA==0 || okB==0 )); then echo "[FAIL] One or both files broken. A_ok=$okA B_ok=$okB" >&2; exit 1; fi
  echo "[OK] Both files decoded cleanly."

  echo; echo "== Step 2: Snapshot compare ==";
  if ! same_movie_or_skip "$A" "$B"; then
    echo "[FAIL] Files are not the same movie (SSIM < ${SNAP_SSIM})." >&2; exit 1
  fi
  echo "[OK] Same movie."

  echo; echo "== Step 3: External compare ($IMPL) =="; run_impl "$IMPL" "$A" "$B"

  echo; echo "== Step 4: Quality selection (prefer ${PREF_HEIGHT}p) ==";
  read -r keep drop why <<<"$(pick_winner "$A" "$B" | tr '|' ' ')"
  echo "[DECISION] Keep: $keep"; echo "[DECISION] Drop: $drop"; echo "[REASON]   $why"

  echo; echo "== Step 5: Action =="; act_on_loser "$drop" "$keep"

  echo; echo "== Summary =="; echo "Kept:   $keep"; echo "Dropped: $drop"
  [[ "$ACTION" == "move" ]] && echo "(Moved loser to: $TRASH_DIR)"
  [[ "$ACTION" == "delete" ]] && echo "(Loser was deleted)"
  [[ $DRY -eq 1 ]] && echo "(Dry-run only; no changes made)"
}

# -------- dispatch --------
if [[ -n "${SCAN_DIR:-}" ]]; then
  dir_mode
  exit 0
fi

# Pair mode
if [[ ${#REM_ARR[@]} -lt 2 ]]; then
  echo "Usage (pair): $0 A.mp4 B.mp4 [options]" >&2
  echo "   or (scan): $0 --scan-dir [DIR] [--recursive] [options]" >&2
  exit 2
fi
A="${REM_ARR[0]}"; B="${REM_ARR[1]}"
[[ -f "$A" ]] || die "File not found: $A"
[[ -f "$B" ]] || die "File not found: $B"
compare_pair "$A" "$B"