#!/usr/bin/env bash
#
# Refuse tracked text that names a path under status/explorations/.
#
# status/explorations/ is a local lab notebook git ignores
# (status/MAINTAINING.md, "Where things live"), so a tracked file naming a
# path under it points at something another clone does not have. This check
# reads what git would commit (the index plus untracked files the ignore rules
# do not exclude, binaries skipped) for the notebook's directory shape,
# explorations/<topic>-<yyyy-mm>, and leaves out what MAINTAINING.md freezes
# (archived changes, status/archive, status/log), design/ (synced from
# upstream, never edited here) and the notebook itself. The <topic>-<yyyy-mm>
# template MAINTAINING.md and the skills use has no digits, so a description
# of the convention is not a hit.
#
# It lives here and not in internal/lint because it is repository hygiene,
# not part of the shipped program; `make lint` runs it after the program's
# own lint. Exits 1 with one file:line per hit, 0 on a clean tree.
#
# Usage: ./scripts/notebook-citations.sh [repo-root]
set -euo pipefail
cd "${1:-$(dirname "$0")/..}"

pattern='explorations/[a-z0-9]+(-[a-z0-9]+)*-20[0-9]{2}-[0-9]{2}'
hits=$(git grep -n -I -E --untracked "$pattern" -- . \
  ':!design' ':!openspec/changes/archive' ':!status/archive' \
  ':!status/explorations' ':!status/log' ':!test/screens' || true)

if [[ -z "$hits" ]]; then
  echo "notebook-citations: ok" >&2
  exit 0
fi

count=0
while IFS= read -r line; do
  file=${line%%:*}
  rest=${line#*:}
  lineno=${rest%%:*}
  text=${rest#*:}
  cite=$(grep -oE "$pattern" <<<"$text" | head -1)
  printf '%s:%s: [notebook-citation] cites the exploration notebook "%s", which a clone does not have; state the fact or cite its tracked home (status/MAINTAINING.md, "Where things live")\n' \
    "$file" "$lineno" "$cite" >&2
  count=$((count + 1))
done <<<"$hits"
echo >&2
echo "notebook-citations: $count hit(s)" >&2
exit 1