#!/usr/bin/env bash
set -euo pipefail

jq_script=$(cat <<-'EOF'
    [.pages[] | .images[] | .object] as $imageIDs |
    .objects as $objects | 
    [$imageIDs[] | { key: ., value: $objects[.] }] as $images | 
    [$images[] | .value["/SMask"] | values] as $softMasks |
    $images[] |
        .key, 
         if .value["/Filter"] == "/DCTDecode" or .value["/Filter"] == ["/DCTDecode"] then 
            "jpg" 
         else 
            "png" 
         end,
        .value["/Width"],
        .value["/Height"],
        .value["/BitsPerComponent"],
        .value["/SMask"] // "None",
        (if (.value["/ColorSpace"] | type) == "string" then 
            $objects[.value["/ColorSpace"]] // .value["/ColorSpace"]
        else 
            .value["/ColorSpace"] 
        end) as $colorSpace | if ($colorSpace | type) == "array" then
            $colorSpace[] 
        else
            $colorSpace
        end
EOF
)

pdf="$1"
output="$2"

pdfname="$(basename "$pdf")"

echo "PDF: $pdf" >&2
i=1
while read -r imageID && \
      read -r format && \
      read -r width && \
      read -r height && \
      read -r bits && \
      read -r softMask && \
      read -r colorSpace; do
    if [[ "$colorSpace" == '/ICCBased' ]]; then
        read -r colorSpaceID
    fi

    echo "$imageID: $format ${width}x${height} $colorSpace" >&2

    name="$(printf -- '%s/%s-%06d.%s' "$output" "$pdfname" "$i" "$format")"

    if [[ "$format" == "jpg" ]]; then
        qpdf --show-object="$imageID" --raw-stream-data "$pdf" >"$name"
    else
        if [[ "$colorSpace" == '/DeviceGray' ]]; then
            magickFormat='gray'
        else
            magickFormat='RGB'
        fi
        qpdf --show-object="$imageID" --filtered-stream-data "$pdf" | convert -size "${width}x${height}" -depth "$bits" "$magickFormat":- "$name"
        if [[ "$colorSpace" == '/ICCBased' ]]; then
            exiftool -overwrite_original "-icc_profile<="<(qpdf --show-object="$colorSpaceID" --filtered-stream-data "$pdf") "$name"
        fi
        if [[ "$softMask" != "None" ]]; then
            convert "$name" <(qpdf --show-object="$softMask" --filtered-stream-data "$pdf" | convert -size "${width}x${height}" -depth "$bits" gray:- PNG:-) -alpha off -compose copy-opacity -composite "$name.tmp.png" && mv "$name.tmp.png" "$name"
        fi
    fi
    i=$((i+1))
done < <(qpdf --json "$pdf" | jq --raw-output -c "$jq_script")