#!/usr/bin/env bash
#
# Regenerates doc/adr/toc.md from the ADR files themselves.
#
# The index is generated, never hand-edited: it drifted from the documents it indexes precisely because
# every ADR merge asked a human to retype it. Run with --check to verify it is current without writing.
#
#   scripts/adr/generate-toc.sh            # rewrite doc/adr/toc.md
#   scripts/adr/generate-toc.sh --check    # exit 1 if doc/adr/toc.md is out of date
#
set -euo pipefail

repo_root="$(cd "$(dirname "${BASH_SOURCE[0]}")/../.." && pwd)"
adr_dir="$repo_root/doc/adr"
toc_path="$adr_dir/toc.md"

check_only=false
if [[ "${1:-}" == "--check" ]]; then
  check_only=true
elif [[ $# -gt 0 ]]; then
  echo "usage: $(basename "$0") [--check]" >&2
  exit 2
fi

# The heading is the source of truth for the title. A legacy "NN. " prefix is stripped so the identifier
# is rendered once, from the filename, rather than depending on whether an author remembered to type it.
title_of() {
  local heading
  heading="$(grep -m 1 '^# ' "$1" || true)"
  [[ -n "$heading" ]] || return 1
  heading="${heading#\# }"
  sed -E 's/^[0-9]+\.[[:space:]]+//' <<<"$heading"
}

entries=()
add_entry() {
  local file="$1" prefix="$2" title
  if ! title="$(title_of "$file")"; then
    echo "error: $(basename "$file") has no '# ' heading to take a title from." >&2
    exit 1
  fi
  entries+=("$(printf '* [%s. %s](%s)' "$prefix" "$title" "$(basename "$file")")")
}

numbered=()
dated=()
while IFS= read -r file; do
  case "$(basename "$file")" in
    toc.md) continue ;;
    # Dated first: a date also opens with four digits, so testing NNNN- first would swallow it.
    [0-9][0-9][0-9][0-9]-[0-9][0-9]-[0-9][0-9]-*.md) dated+=("$file") ;;
    [0-9][0-9][0-9][0-9]-*.md) numbered+=("$file") ;;
    *)
      echo "error: $(basename "$file") matches neither the NNNN- nor the YYYY-MM-DD- naming convention." >&2
      exit 1
      ;;
  esac
done < <(LC_ALL=C find "$adr_dir" -maxdepth 1 -name '*.md' | LC_ALL=C sort)

# Numbered records first, then dated ones: every dated record postdates every numbered one, so one flat
# list stays chronological across the change of convention.
for file in ${numbered[@]+"${numbered[@]}"}; do
  identifier="$(basename "$file")"
  identifier="${identifier%%-*}"
  # 10# so a zero-padded identifier is never read as octal.
  add_entry "$file" "$((10#$identifier))"
done

for file in ${dated[@]+"${dated[@]}"}; do
  identifier="$(basename "$file")"
  add_entry "$file" "${identifier:0:10}"
done

generated="$(
  echo '# Architecture Decision Records'
  echo
  echo '<!-- Generated by scripts/adr/generate-toc.sh. Do not edit by hand. -->'
  echo
  printf '%s\n' ${entries[@]+"${entries[@]}"}
)"

if [[ "$check_only" == true ]]; then
  if ! diff -u "$toc_path" <(printf '%s\n' "$generated"); then
    echo "doc/adr/toc.md is out of date. Run scripts/adr/generate-toc.sh and commit the result." >&2
    exit 1
  fi
  echo "doc/adr/toc.md is up to date."
else
  printf '%s\n' "$generated" > "$toc_path"
  echo "Wrote doc/adr/toc.md."
fi