#!/usr/bin/env bash
#
# gen-index.sh — keep context/README.md's documentation index in sync with the repo.
#
#   context/gen-index.sh            # CHECK: report drift, exit 1 if the index is stale
#   context/gen-index.sh --write    # rewrite the AUTOGEN block in context/README.md
#   context/gen-index.sh --help
#
# Only the region between the AUTOGEN markers in context/README.md is touched;
# the hand-written header/scope note above it is left alone. Descriptions that
# already exist in the index are PRESERVED (matched by path); files that are new
# to the tree get a stub description pulled from their first heading/lead line,
# ready for you to polish. Vendored/third-party trees are excluded, matching the
# index's stated scope.
#
# Run it from anywhere; it locates the repo via this script's own path.
set -euo pipefail

ROOT="$(cd "$(dirname "${BASH_SOURCE[0]}")/.." && pwd)"
cd "$ROOT"
INDEX="context/README.md"
BEGIN='<!-- BEGIN AUTOGEN (context/gen-index.sh) -->'
END='<!-- END AUTOGEN -->'

MODE="check"
case "${1:-}" in
  --write) MODE="write" ;;
  ""|--check) MODE="check" ;;
  -h|--help) sed -n '3,17p' "$0" | sed 's/^# \{0,1\}//'; exit 0 ;;
  *) echo "unknown option: $1 (try --help)" >&2; exit 2 ;;
esac

# --- ordered sections: a path lands in the FIRST section whose regex matches ---
TITLES=(
  "Orientation"
  "Emulator — plan, phases & operations"
  "Emulator — subsystems"
  "Emulator — deployment & ports"
  "Emulator — devices & support files"
  "Emulator — VelociRender firmware decompilation"
  "Emulator — render bridge"
  "dpl3-revive — DPL3 renderer port & format specs"
  "restoration — scene re-renders"
  "restoration — RIO firmware & hardware"
  "restoration — source410 reconstruction"
)
REGEX=(
  '^(README|HISTORY)\.md$'
  '^emulator/(PLAN|PHASE[0-9].*|LAUNCH|ENV-VARS)\.md$'
  '^emulator/(VDB-NOTES|GAUGE-NOTES|SOUND-NOTES|NET-NOTES|RIO-NOTES|CAMERA-REVIEW-NOTES|RENDER-HARNESS|RENDERER-COLLAB)\.md$'
  '^emulator/(DEPLOYMENT-PLAN|XP-PORT-PLAN|EGRESS-HOLD-FIX-PLAN)\.md$'
  '^emulator/(vpx-device|pod-launch|net-boot|deploy|roms|rio-firmware)/'
  '^emulator/firmware-decomp/'
  '^emulator/render-bridge/'
  '^dpl3-revive/'
  '^restoration/README\.md$'
  '^restoration/rio-'
  '^restoration/source410/'
)

# --- the doc set (same filter as the index scope) ---
list_docs() {
  git ls-files '*.md' '*.MD' \
    | grep -viE '/(vs|libpng|libpdcurses|sdl2|sdl2net|freetype)/' \
    | grep -viE '^emulator/src/(NOTES|\.github)/' \
    | grep -vxF "$INDEX" \
    | sort
}

# --- preserved descriptions: parse the current AUTOGEN block, path -> desc ---
declare -A DESC
if [ -f "$INDEX" ]; then
  in_block=0
  while IFS= read -r line; do
    [ "$line" = "$BEGIN" ] && { in_block=1; continue; }
    [ "$line" = "$END" ] && { in_block=0; continue; }
    [ "$in_block" = 1 ] || continue
    case "$line" in
      "- ["*"](../"*")"*)
        rest="${line#*](../}"; path="${rest%%)*}"; after="${rest#*)}"
        after="${after#"${after%%[![:space:]]*}"}"   # ltrim
        after="${after#—}"; after="${after#-}"        # drop leading em/hyphen dash
        after="${after#"${after%%[![:space:]]*}"}"    # ltrim again
        [ -n "$after" ] && DESC["$path"]="$after"
        ;;
    esac
  done < "$INDEX"
fi

# --- stub description for a file new to the index ---
extract_desc() {
  local f="$1" raw
  raw="$(awk 'NR>1 && $0 !~ /^#/ && $0 !~ /^---/ && $0 !~ /^>/ && $0 !~ /^[[:space:]]*$/ {print; exit}' "$f")"
  raw="$(printf '%s' "$raw" | sed -E 's/[`*]//g; s/[[:space:]]+/ /g; s/^ +//; s/ +$//')"
  [ "${#raw}" -gt 140 ] && raw="${raw:0:137}..."
  [ -n "$raw" ] && printf '%s' "$raw" || printf '(no description — add one)'
}

# --- build the block body on stdout ---
gen_body() {
  local docs; docs="$(list_docs)"
  local -a remaining; mapfile -t remaining <<< "$docs"

  emit_section() {
    local title="$1" re="$2"; shift 2
    local -a kept=() matched=()
    local p
    for p in "${remaining[@]}"; do
      [ -z "$p" ] && continue
      if [[ "$p" =~ $re ]]; then matched+=("$p"); else kept+=("$p"); fi
    done
    remaining=("${kept[@]}")
    [ "${#matched[@]}" -eq 0 ] && return 0
    printf '\n## %s\n\n' "$title"
    for p in "${matched[@]}"; do
      local d="${DESC[$p]:-}"; [ -z "$d" ] && d="$(extract_desc "$p")"
      printf -- '- [%s](../%s) — %s\n' "$p" "$p" "$d"
    done
  }

  local i
  for i in "${!TITLES[@]}"; do
    emit_section "${TITLES[$i]}" "${REGEX[$i]}"
  done

  # anything unmatched still shows up — never silently dropped
  local -a leftover=()
  local p
  for p in "${remaining[@]}"; do [ -n "$p" ] && leftover+=("$p"); done
  if [ "${#leftover[@]}" -gt 0 ]; then
    printf '\n## Uncategorized\n\n'
    printf '%s\n' '> New locations the section rules do not cover yet — assign them in gen-index.sh.'
    printf '\n'
    for p in "${leftover[@]}"; do
      local d="${DESC[$p]:-}"; [ -z "$d" ] && d="$(extract_desc "$p")"
      printf -- '- [%s](../%s) — %s\n' "$p" "$p" "$d"
    done
  fi
}

# --- current block body (between the markers) on stdout ---
current_body() {
  awk -v b="$BEGIN" -v e="$END" '
    $0==b {inb=1; next} $0==e {inb=0} inb==1 {print}
  ' "$INDEX"
}

grep -qF "$BEGIN" "$INDEX" && grep -qF "$END" "$INDEX" || {
  echo "error: AUTOGEN markers not found in $INDEX" >&2; exit 2; }

NEW="$(gen_body)"

if [ "$MODE" = "check" ]; then
  if diff <(current_body) <(printf '%s\n' "$NEW") >/dev/null; then
    echo "context index is current ($(list_docs | grep -c . ) docs)."
    exit 0
  fi
  echo "context index is STALE — run: context/gen-index.sh --write" >&2
  echo "--- diff (current -> generated) ---" >&2
  diff <(current_body) <(printf '%s\n' "$NEW") >&2 || true
  exit 1
fi

# MODE=write: splice NEW between the markers, leave everything else intact
tmp="$(mktemp)"
awk -v b="$BEGIN" -v e="$END" -v body="$NEW" '
  $0==b {print; print body; skip=1; next}
  $0==e {skip=0; print; next}
  skip!=1 {print}
' "$INDEX" > "$tmp"
mv "$tmp" "$INDEX"
echo "wrote $INDEX ($(list_docs | grep -c . ) docs)."