do-convert-single-file

#!/bin/bash

set -e

ensure_slimerjs_profile_template() {
  # Create a profile "template" on first run. We'll copy that profile directory
  # for every run, which is faster than forcing Firefox to build a fresh one.
  if [ ! -d /tmp/slimerjs-profile-template ]; then
    mkdir /tmp/slimerjs-profile-template
    echo 'slimer.exit()' > /tmp/noop.js
    slimerjs \
      --profile /tmp/slimerjs-profile-template \
      --headless \
      /tmp/noop.js
  fi
}

reset_slimerjs_profile() {
  # Sets /tmp/slimerjs-profile to be a fresh Firefox profile directory.
  ensure_slimerjs_profile_template
  rm -rf /tmp/slimerjs-profile
  cp -a /tmp/slimerjs-profile-template /tmp/slimerjs-profile
}

convert_html() {
  # Output 0.blob and 0.txt

  reset_slimerjs_profile

  mv input.blob input.html

  # Output 0.blob (PDF) and 0.txt
  slimerjs \
    --profile /tmp/slimerjs-profile \
    --headless \
    /app/convert.js
}

convert_rtf() {
  Ted --saveTo input.blob 0.txt

  # --saveTo uses document paper size; --printToFile uses letter paper size.
  # We want --saveTo
  Ted --saveTo input.blob 0.ps
  gs \
    -sDEVICE=pdfwrite \
    -sOutputFile=0.blob \
    -dSAFER \
    -dBATCH \
    -dNOPAUSE \
    -dNOPROMPT \
    -dQUIET \
    -dCompatibilityLevel=1.4 \
    -q \
    -sDocumentUUID=00000000-0000-0000-0000-000000000000 \
    -sInstanceUUID=00000000-0000-0000-0000-000000000000 \
    -c "100000000 setvmthreshold" \
    -f0.ps
}

# Put HTML input in "input.html", converting if needed
CONTENT_TYPE=$(echo "$1" | jq -r .contentType)
case "$CONTENT_TYPE" in
  application/rtf)
    convert_rtf
    ;;
  *)
    convert_html
    ;;
esac

# SlimerJS renders screenshots, but they're of the _page_, not the _PDF_. We'll
# generate a screenshot of the PDF.
#
# Ditto for RTF conversion
pdftocairo \
    -jpeg \
    -singlefile \
    -scale-to 700 \
    0.blob \
    0-thumbnail

# Output JSON
echo "$1" | jq '{ filename: .filename, languageCode: .languageCode, wantOcr: false, wantSplitByPage: .wantSplitByPage, contentType: "application/pdf", metadata: .metadata }' > 0.json