Stirling-PDF/scripts/init.sh

#!/bin/bash

# Copy the original tesseract-ocr files to the volume directory without overwriting existing files
echo "Copying original files without overwriting existing files"
mkdir -p /usr/share/tessdata
cp -rn /usr/share/tessdata-original/* /usr/share/tessdata

if [ -d /usr/share/tesseract-ocr/4.00/tessdata ]; then
        cp -r /usr/share/tesseract-ocr/4.00/tessdata/* /usr/share/tessdata || true;
fi

if [ -d /usr/share/tesseract-ocr/5/tessdata ]; then
        cp -r /usr/share/tesseract-ocr/5/tessdata/* /usr/share/tessdata || true;
fi

# Check if TESSERACT_LANGS environment variable is set and is not empty
if [[ -n "$TESSERACT_LANGS" ]]; then
  # Convert comma-separated values to a space-separated list
  SPACE_SEPARATED_LANGS=$(echo $TESSERACT_LANGS | tr ',' ' ')
  pattern='^[a-zA-Z]{2,4}(_[a-zA-Z]{2,4})?$'
  # Install each language pack
  for LANG in $SPACE_SEPARATED_LANGS; do
     if [[ $LANG =~ $pattern ]]; then
      apk add --no-cache "tesseract-ocr-data-$LANG"
     else
      echo "Skipping invalid language code"
     fi
  done
fi

# Ensure temp directory exists with correct permissions before running main init
mkdir -p /tmp/stirling-pdf || true
chown -R stirlingpdfuser:stirlingpdfgroup /tmp/stirling-pdf || true
chmod -R 755 /tmp/stirling-pdf || true

/scripts/init-without-ocr.sh "$@"