Bash script to list and/or download files from The Art Institute of Chicago website

artic.sh

#!/usr/bin/env bash

DownloadFlag=false
DownloadFromFileFlag=false
ListFlag=false

ArtDir="./ArtIC"
ArtDirPD="$ArtDir/PD"
ArtDirNPD="$ArtDir/NPD"
ListName=""
FileListName=""
Iterate=""

while getopts ":fdl" opt; do
  case ${opt} in
    f ) # DOWNLOAD FROM FILE
      DownloadFromFileFlag=true
      ListName="download_from_file__list"
      FileListName="download_from_file__file_list"
      Iterate=$(cat "./codes.txt")
      ;;
    d ) # DOWNLOAD
      DownloadFlag=true
      ListName="download__list"
      FileListName="download__file_list"
      Iterate=$(seq 0 249999)
      ;;
    l ) # LIST
      ListFlag=true
      ListName="full_list"
      FileListName=""
      Iterate=$(seq 0 249999)
      ;;
    \? )
      echo "Unsupported option."
      echo "Usage: [-f -d -l]"
      exit 0
      ;;
  esac
done

if [[ $DownloadFlag = true && $DownloadFromFileFlag = true ]] || \
   [[ $DownloadFlag = true && $ListFlag = true ]] || \
   [[ $DownloadFromFileFlag = true && $ListFlag = true ]]
then
  echo "You have to specify only one option."
  exit 0
fi

if [[ $DownloadFlag = false && $DownloadFromFileFlag = false && $ListFlag = false ]]
then
  echo "You have to specify one option."
  exit 0
fi

mkdir $ArtDir
if [ $ListFlag = false ]
then
  mkdir $ArtDirPD
  mkdir $ArtDirNPD
fi

for i in $Iterate;
do
    IMGLINK="https://www.artic.edu/artworks/${i}/"
    SRC=$(curl -fsL "$IMGLINK")
    if [ "$SRC" == "" ]; then
        echo "$i"
    else
        TITLE=$(echo "$SRC" | pup 'dd[itemprop="name"] span text{}')
        AUTHOR=$(echo "$SRC" | pup 'dd[itemprop="creator"] a text{}')

        TITLE=${TITLE//—/--}
        AUTHOR=${AUTHOR//$'\n'/, }
        AUTHOR=${AUTHOR//—/--}

        TITLE=$(echo "$TITLE" | recode HTML..LATIN1)
        AUTHOR=$(echo "$AUTHOR" | recode HTML..LATIN1)

        # Save current IFS (SOURCE: https://stackoverflow.com/questions/24628076/bash-convert-n-delimited-strings-into-array)
        SAVEIFS=$IFS
        # Change IFS to new line.
        IFS=$'\n'

        DOWNLINK=($(echo "$SRC" | pup 'button attr{data-gallery-img-download-url}'))
        PUBDOM=($(echo "$SRC"| pup 'button attr{data-gallery-img-credit}'))

        # Restore IFS
        IFS=$SAVEIFS

        echo "No.: $i; Author: $AUTHOR; Title: $TITLE; Copyright: $PUBDOM"
        echo "No.: $i; Author: $AUTHOR; Title: $TITLE; Copyright: $PUBDOM" >> "$ArtDir/$ListName.txt"

        if [ $ListFlag = false ]
        then
          NUMDOWN=${#DOWNLINK[@]}

          TITLE=${TITLE//\//-}
          TITLE=${TITLE//[<>:\\|?*]/_}

          AUTHOR=${AUTHOR//\//-}
          AUTHOR=${AUTHOR//[<>:\\|?*]/_}

          FILENAME="Author_ ${AUTHOR:0:50}; Title_ ${TITLE:0:50}; No._ ${i}; Copyright_ $PUBDOM"

          if [ "$DOWNLINK" != "" ]; then
              if [ "$NUMDOWN" == "1" ]; then
                  echo "File: $FILENAME.jpg; URL: $DOWNLINK" >> "$ArtDir/$FileListName.txt"

                  if [ "$PUBDOM" == "CC0 Public Domain Designation" ]; then
                      wget -q -O "$ArtDirPD/$FILENAME.jpg" "$DOWNLINK"
                  else
                      wget -q -O "$ArtDirNPD/$FILENAME.jpg" "$DOWNLINK"
                  fi
                  sleep 0.25s
              else
                  for ((j=0;j<NUMDOWN;j++));
                  do
                      echo "File: $FILENAME-${j}.jpg; URL: ${DOWNLINK[j]}" >> "$ArtDir/$FileListName.txt"

                      if [ "${PUBDOM[j]}" == "CC0 Public Domain Designation" ]; then
                          wget -q -O "$ArtDirPD/$FILENAME-${j}.jpg" "${DOWNLINK[j]}"
                      else
                          wget -q -O "$ArtDirNPD/$FILENAME-${j}.jpg" "${DOWNLINK[j]}"
                      fi
                      sleep 0.25s
                  done
              fi
          fi
        fi
    fi
done

codes.txt

4
9
11