commit-msg

#!/usr/bin/env sh
# Conventional Commits + scope-enum validator (POSIX sh, no node/pnpm).
#
# Husky v9's runner invokes this hook with `sh -e`, so this script must be
# POSIX-compatible (no bash arrays, no [[ ... ]], no =~, no process
# substitution, no here-strings).
#
# Supports multi-scope syntax, e.g.:
#   feat(coordinator, postman, contracts): add multi-component change

set -eu

MSG_FILE="$1"

# Subject = first non-empty, non-comment line of the commit message.
SUBJECT=$(grep -v '^#' "${MSG_FILE}" | sed '/^$/d' | head -n 1)

# Let merge / fixup / squash / revert auto-commits through unchecked.
case "${SUBJECT}" in
  "Merge "*|"fixup! "*|"squash! "*|"amend! "*|'Revert "'*) exit 0 ;;
esac

# Conventional Commits default type set (from @commitlint/config-conventional).
ALLOWED_TYPES='feat|fix|docs|style|refactor|perf|test|chore|ci|revert|build'

# Allowed scopes
ALLOWED_SCOPES=\
'coordinator prover prover-ray verifier-ray postman tx-exclusion-api '\
'linea-besu contracts sdk-core sdk-ethers sdk-viem '\
'tracer sequencer state-recovery jvm-libs blob-libs '\
'e2e ci docker deps misc maru'

# Match overall structure: <type>(<scope-group>)!?: <subject>
HEADER_RE="^(${ALLOWED_TYPES})\\(([^)]+)\\)!?: .+"

if ! printf '%s' "${SUBJECT}" | grep -qE "${HEADER_RE}"; then
  cat >&2 <<EOF
✗ Commit message does not match the Conventional Commits format.

  Expected:  <type>(<scope>[, <scope>...]): <short description>
  Got:       ${SUBJECT}

  Allowed types:  $(printf '%s' "${ALLOWED_TYPES}" | tr '|' ',' | sed 's/,/, /g')
  Allowed scopes: $(printf '%s' "${ALLOWED_SCOPES}" | sed 's/ /, /g')
EOF
  exit 1
fi

# Extract scope group (capture group 2) using sed.
SCOPES_RAW=$(printf '%s' "${SUBJECT}" | sed -E "s/${HEADER_RE}.*/\\2/")

# Reject malformed scope groups before splitting. Caught patterns:
#   "( )"                    — only whitespace
#   "(,coordinator)"         — leading comma
#   "(coordinator,)"         — trailing comma
#   "(coordinator,,postman)" — empty slot between commas
# (The for-loop below word-splits on whitespace so it would silently swallow
# these otherwise.)
if printf '%s' "${SCOPES_RAW}" \
     | grep -qE '(^[[:space:]]*$|^[[:space:]]*,|,[[:space:]]*$|,[[:space:]]*,)'; then
  cat >&2 <<EOF
✗ Malformed scope group: "(${SCOPES_RAW})"
  Scope name must be present and each comma must be followed by a non-empty scope name.
EOF
  exit 1
fi

# Split on commas, trim each, validate against the enum and detect duplicates.
# `for x in $(...)` word-splits on whitespace; tr collapses commas to spaces
# so each scope ends up as its own field.
INVALID=""
SEEN=""
DUPLICATES=""
for scope in $(printf '%s' "${SCOPES_RAW}" | tr ',' ' '); do
  # Membership check against the enum.
  case " ${ALLOWED_SCOPES} " in
    *" ${scope} "*) ;;
    *) INVALID="${INVALID} ${scope}" ;;
  esac

  # Duplicate detection — record the first occurrence in SEEN, append to
  # DUPLICATES on the second occurrence (and skip on subsequent ones so the
  # message lists each duplicate only once).
  case " ${SEEN} " in
    *" ${scope} "*)
      case " ${DUPLICATES} " in
        *" ${scope} "*) ;;
        *) DUPLICATES="${DUPLICATES} ${scope}" ;;
      esac
      ;;
    *) SEEN="${SEEN} ${scope}" ;;
  esac
done

if [ -n "${INVALID}" ]; then
  cat >&2 <<EOF
✗ Scope(s) not in the allowed list:${INVALID}

  Allowed scopes: $(printf '%s' "${ALLOWED_SCOPES}" | sed 's/ /, /g')
EOF
  exit 1
fi

if [ -n "${DUPLICATES}" ]; then
  cat >&2 <<EOF
✗ Duplicate scope(s) in scope group:${DUPLICATES}
  Each scope must appear at most once inside the parentheses.
EOF
  exit 1
fi

exit 0