pdf-text-diff

#!/bin/bash
# Prints a text diff of two PDF files.
# USAGE: pdf-text-diff [-layout|-raw] [DIFF_FLAGS] <file.pdf> <file.pdf>

set -o errexit
set -o nounset

main() {
  BASENAME=$(basename "$0")
  PDF_FLAGS=()
  DIFF_FLAGS=()
  FILES=()
  while (( $# > 0 )); do
    case "$1" in
      -layout) PDF_FLAGS+=( -layout ) ;;
      -raw) PDF_FLAGS+=( -raw ) ;;
      -*) DIFF_FLAGS+=( "$1" ) ;;
      *) FILES+=( "$1" )
    esac
    shift
  done
  if (( ${#FILES[@]} != 2 )); then
    die "USAGE: $BASENAME [-layout|-raw] [DIFF_FLAGS] <file.pdf> <file.pdf>"
  fi
  set -- "${FILES[@]}"

  tempdir=$(mktemp -d "$BASENAME.XXXXXXXX")
  trap "rm -rf '$tempdir'" exit
  mkdir "${tempdir}/a"
  mkdir "${tempdir}/b"

  A="$tempdir/a/$(basename "$1")"
  B="$tempdir/b/$(basename "$2")"

  pdftotext "${PDF_FLAGS[@]}" "$1" "$A"
  pdftotext "${PDF_FLAGS[@]}" "$2" "$B"

  diff-prog "$A" "$B"
}

die() {
  echo "$*" >&2
  exit 1
}

pdf-to-text() {
  pdftotext "${PDF_FLAGS[@]}" "$1"
}

diff-prog() {
  if which git >/dev/null 2>/dev/null; then
    git diff --no-index --no-prefix "${DIFF_FLAGS[@]}" "$@"
  else
    diff "${DIFF_FLAGS[@]}" "$@"
  fi
}

main "$@"