Research-Stack/4-Infrastructure/packaging/abiword-local-proofread/abiword-local-proofread

#!/usr/bin/env bash
set -u

usage() {
  cat <<'USAGE'
Usage: abiword-local-proofread [--no-harper] [--no-vale] [--keep-tmp] FILE...

Local proofreading bridge for AbiWord, LibreOffice, and word-processor files.

The wrapper:
  1. Converts word-processor documents to plain text.
     - .abw prefers AbiWord.
     - .odt/.docx/.doc/.rtf and other office files prefer LibreOffice.
  2. Runs harper-cli for local grammar checks.
  3. Runs vale for local prose/style checks.

If a .vale.ini is found in the current directory or a parent directory, it is
used. Otherwise, the bundled conservative Vale config is used.
USAGE
}

find_vale_config() {
  local dir="${PWD}"
  while [[ "${dir}" != "/" ]]; do
    if [[ -f "${dir}/.vale.ini" ]]; then
      printf '%s\n' "${dir}/.vale.ini"
      return 0
    fi
    dir="$(dirname "${dir}")"
  done

  if [[ -f "/usr/share/abiword-local-proofread/vale.ini" ]]; then
    printf '%s\n' "/usr/share/abiword-local-proofread/vale.ini"
    return 0
  fi

  return 1
}

needs_conversion() {
  case "${1,,}" in
    *.txt|*.text|*.md|*.markdown|*.rst|*.adoc|*.html|*.htm)
      return 1
      ;;
    *)
      return 0
      ;;
  esac
}

prefers_abiword() {
  case "${1,,}" in
    *.abw|*.zabw)
      return 0
      ;;
    *)
      return 1
      ;;
  esac
}

convert_with_abiword() {
  local input="$1"
  local output="$2"
  abiword --to=txt --to-name "${output}" "${input}" >/dev/null 2>&1
}

convert_with_libreoffice() {
  local input="$1"
  local tmpdir="$2"
  local base stem output
  base="$(basename "${input}")"
  stem="${base%.*}"
  output="${tmpdir}/${stem}.txt"

  rm -f "${output}"
  if ! libreoffice --headless --nologo --nofirststartwizard \
      --convert-to txt:Text --outdir "${tmpdir}" "${input}" >/dev/null 2>&1; then
    return 1
  fi
  if [[ ! -f "${output}" ]]; then
    return 1
  fi
  printf '%s\n' "${output}"
}

convert_document() {
  local input="$1"
  local tmpdir="$2"
  local base output converted
  base="$(basename "${input}")"
  output="${tmpdir}/${base}.txt"

  if prefers_abiword "${input}"; then
    if convert_with_abiword "${input}" "${output}"; then
      printf '%s\n' "${output}"
      return 0
    fi
    if converted="$(convert_with_libreoffice "${input}" "${tmpdir}")"; then
      printf '%s\n' "${converted}"
      return 0
    fi
    return 1
  fi

  if converted="$(convert_with_libreoffice "${input}" "${tmpdir}")"; then
    printf '%s\n' "${converted}"
    return 0
  fi
  if convert_with_abiword "${input}" "${output}"; then
    printf '%s\n' "${output}"
    return 0
  fi
  return 1
}

run_one() {
  local file="$1"
  local tmpdir="$2"
  local text_file="$file"
  local status=0

  if [[ ! -f "${file}" ]]; then
    printf 'abiword-local-proofread: not a file: %s\n' "${file}" >&2
    return 2
  fi

  if needs_conversion "${file}"; then
    if ! text_file="$(convert_document "${file}" "${tmpdir}")"; then
      printf 'abiword-local-proofread: could not export %s to text with LibreOffice or AbiWord\n' "${file}" >&2
      return 3
    fi
  fi

  printf '\n==> %s\n' "${file}"

  if [[ "${RUN_HARPER}" == "1" ]]; then
    printf '\n-- Harper --\n'
    if ! harper-cli lint --no-color --format compact "${text_file}"; then
      status=1
    fi
  fi

  if [[ "${RUN_VALE}" == "1" ]]; then
    printf '\n-- Vale --\n'
    local vale_config
    if vale_config="$(find_vale_config)"; then
      if ! vale --config="${vale_config}" --no-wrap "${text_file}"; then
        status=1
      fi
    else
      printf 'abiword-local-proofread: no Vale config found; skipped\n' >&2
    fi
  fi

  return "${status}"
}

RUN_HARPER=1
RUN_VALE=1
KEEP_TMP=0
files=()

while [[ $# -gt 0 ]]; do
  case "$1" in
    --help|-h)
      usage
      exit 0
      ;;
    --no-harper)
      RUN_HARPER=0
      ;;
    --no-vale)
      RUN_VALE=0
      ;;
    --keep-tmp)
      KEEP_TMP=1
      ;;
    --)
      shift
      while [[ $# -gt 0 ]]; do
        files+=("$1")
        shift
      done
      break
      ;;
    -*)
      printf 'abiword-local-proofread: unknown option: %s\n' "$1" >&2
      usage >&2
      exit 2
      ;;
    *)
      files+=("$1")
      ;;
  esac
  shift
done

if [[ ${#files[@]} -eq 0 ]]; then
  usage >&2
  exit 2
fi

tmpdir="$(mktemp -d)"
if [[ "${KEEP_TMP}" != "1" ]]; then
  trap 'rm -rf "${tmpdir}"' EXIT
else
  printf 'abiword-local-proofread: keeping temp dir %s\n' "${tmpdir}" >&2
fi

overall=0
for file in "${files[@]}"; do
  if ! run_one "${file}" "${tmpdir}"; then
    overall=1
  fi
done

exit "${overall}"