Substack data downloads don't include the post images, only the HTML. This script reads all the html files in the current folder, and downloads the image from each url in the srcset into the ./images folder

substack_image_getter.sh

#!/bin/bash
# Requires wget, which doesnt come as standard on a mac. Can be installed via homebrew.
# Written with help from ChatGPT-o3-mini-high

# Trap SIGINT (Ctrl+C) so the whole script exits when interrupted.
trap "exit" INT

mkdir -p images

for file in *.html; do
  base="${file%.html}"
  image_index=1

  # Instead of piping into the while loop (which spawns a subshell),
  # we use process substitution so that the loop runs in the main shell.
  while IFS= read -r srcset; do
    while IFS= read -r src; do
      # Trim any extra whitespace.
      candidate=$(echo "$src" | sed 's/^[[:space:]]*//; s/[[:space:]]*$//')
      # Extract the descriptor
      descriptor="${src##* }"
      # Remove the descriptor to obtain the full URL.
      url=$(echo "$candidate" | sed -E 's/ [0-9]+[wx]$//')
      
      # Remove any query string for proper extension extraction.
      url_clean=${url%%\?*}
      extension="${url_clean##*.}"

      outfile="images/${base}_${image_index}_${descriptor}.${extension}"

      echo "Downloading $descriptor image from $base"
      echo "URL:     $url"
      echo "Outfile: $outfile"
      echo "SRC:     $src"
      echo " ";
      wget -q -O "$outfile" "$url"
      echo "wget done";
      echo " ";
    done < <(echo "$srcset" | sed 's/, /\n/g')
    image_index=$((image_index + 1))
  done < <(grep -oE 'srcset="[^"]+"' "$file" | sed 's/srcset="//; s/"$//')
done