pvw

#!/usr/bin/env bash
# preview a pdf
# depends on: pandoc, wget, fzf, poppler (for pdfunite)
# Copyright (c) 2021 Julia

version="1.3.2"

preview() {
  # make a temporary file, view it and remove it once viewed
  local outputFile
  [[ -s "$1" ]] || printError "'$1' does not exist."

  if [[ "${1: -3}" == pdf ]]; then
    $pdfViewer "$1" &
    return
  fi

  # set the output file based on if it'll be saved or not
  if [[ "${preserve}" ]]; then
    outputFile="${1%.*}.pdf"
  else
    outputFile="$(mktemp --suffix='.pdf')"
  fi
  trap 'rm -f ${outputFile}' EXIT

  # convert and open the file
  pandoc --quiet "$1" -o "${outputFile}" 2>/dev/null &&
    $pdfViewer "${outputFile}" &

  if [[ ! "$preserve" ]]; then
    rm -f -- "${outputFile}"
  fi
}

detectUrl() {
  # detect if input is a url, return a filename
  regex="https?:\/\/(www\.)?[-a-zA-Z0-9@:%._\+~#=]{2,256}\.[a-z]{2,4}\b([-a-zA-Z0-9@:%_\+.~#?&//=]*)"
  if grep -Ei "$regex" <<<"$1" >/dev/null; then
    getFile "$1"
  else
    echo "$1"
  fi

}

getFile() {
  # download a file
  local downloadedFile
  if [[ "$preserve" ]]; then
    # trim the url name to the specific file, and replace any extension with html
    # this exists to remove html anchors from the filename
    local tmp="${1##*/}"
    downloadedFile="${tmp%.*}.html"
    unset tmp
  else
    downloadedFile="$(mktemp --suffix='.html')"
  fi

  wget -q "$1" -O "${downloadedFile}"
  trap 'rm -f ${downloadedFile}' EXIT
  echo "${downloadedFile}"
}

merge() {
  # merge pdfs based on the order they're given as args
  # TODO: (jam) use pdfunite
  :
}

main() {
  checkDeps

  # check for arguments and run preview() on every valid file given
  parseArgs "$@"

  # TODO: (jam) option to combine files into one pdf (pandoc -s)
  # declare arrays for args and opts, sort provided arguments accordingly
  declare -a args opts
  for arg in "$@"; do
    # remove pdfViewer from args
    if [[ "${pdfViewer}" && "${arg}" == "${pdfViewer}" ]]; then
      continue
    fi
    if [[ $arg == [-]* ]]; then
      opts+=("$arg")
    else
      args+=("$arg")
    fi
  done

  if [[ ! "${pdfViewer}" ]]; then
    pdfViewer="xdg-open"
  fi

  # if edit flag is present, edit input files
  if [[ "$edit" ]]; then
    # TODO: (jam) implement detectUrl here
    for f in "${args[@]}"; do
      "$EDITOR" -- "$f"
    done
  fi

  # if the user provided files to operate on, iterate through them
  if (("${#args[@]}" > 0)); then
    for f in "${args[@]}"; do
      preview "$(detectUrl "$f")" || printError "Cannot access file '${f}'"
      shift
    done
    # otherwise, select files interactively
  else
    local preview_file
    preview_file="$(find "." -type f | fzf +m)" || exit 0
    preview "${preview_file}"
  fi
}

checkDeps() {
  # check dependencies
  for com in fzf pandoc wget; do
    hash "$com" 2>/dev/null || printError "$com needs to be installed to run pvw."
  done
}

printError() {
  # print errors to stderr
  tput setaf 1
  printf -- '[ERR] %s\n' "$1" >&2
  tput sgr0
  exit 1
}

usage() {
  # usage
  cat <<EOF
Usage: preview [<args>] [file|url]

Arguments are:
    -e, --edit     Edit input files before generating pdfs
    -h, --help     Display this help
    -k, --keep     Save the generated pdf
    -r, --readme   Preview the README of the current directory
    -v, --viewer   Choose the pdf viewer to use (default: xdg-open)
    --version      Display version information

Note:
    If providing a URL, be sure to prefix it with 'http[s]'.
EOF
  exit 0
}

parseArgs() {
  # parse arguments
  TEMP="$(getopt -o 'ehkrv:' -l 'edit,help,keep,readme,version,viewer:' -n "pvw" -- "$@")"
  eval set -- "$TEMP"
  unset TEMP

  while true; do
    case "$1" in
    '-e' | '--edit')
      edit=true
      shift
      continue
      ;;
    '-h' | '--help')
      usage
      ;;
    '-k' | '--keep')
      # TODO: (jam) change the location the pdf is stored to (opt arg?)
      preserve=true
      shift
      continue
      ;;
    '-r' | '--readme')
      preview "./README.md"
      exit 0
      ;;
    '--version')
      printf -- 'pvw v%s\n' "$version"
      exit 0
      ;;
    '-v' | '--viewer')
      pdfViewer="$2"
      shift 2
      continue
      ;;
    '--')
      shift
      break
      ;;
    *)
      printError "$2 is not a recognized argument"
      ;;
    esac
  done
}

main "$@"
exit 0