Islandora Hash extraction

fedora_checksums.sh

#!/bin/bash

# chmod +x fedora_checksums.sh
# ./fedora_checksums.sh

# This script is used to extract Fedora checksums for files referenced in Drupal nodes.
# It must be run from within the Drupal container.
# The script connects to Fedora via http://fcrepo:8080 and uses Drush
# to query the Drupal database for file URIs.
# The script creates a fedora_checksums.json file with node IDs and their
# corresponding Fedora file checksums.

# Help function
show_help() {
  echo "Usage: $0 [OPTIONS]"
  echo ""
  echo "Extract Fedora checksums for files referenced in Drupal nodes."
  echo ""
  echo "OPTIONS:"
  echo "  -h, --help    Show this help message"
  echo ""
  echo "IMPORTANT: This script must be run from within the Drupal container."
  echo "The script connects to Fedora via http://fcrepo:8080 and uses Drush"
  echo "to query the Drupal database for file URIs."
  echo ""
  echo "Output: Creates fedora_checksums.json with node IDs and their"
  echo "        corresponding Fedora file checksums."
  exit 0
}

# Parse command line arguments
while [[ $# -gt 0 ]]; do
  case $1 in
    -h|--help)
      show_help
      ;;
    *)
      echo "Unknown option: $1"
      echo "Use --help for usage information."
      exit 1
      ;;
  esac
  shift
done

echo "🔍 Checking if running in a container..."
if ! curl -s http://fcrepo:8080 > /dev/null; then
  echo "❌ Could not resolve http://fcrepo:8080. Are you running in a container?"
  exit 1
fi

echo "🔍 Running Fedora checksum extraction script..."
echo "📦 Using Drupal SQL connection via Drush"
echo "📡 Running SQL query to fetch node IDs and file URIs..."

SQL_QUERY="
  SELECT n.nid, fm.uri
  FROM node_field_data n
  JOIN media__field_media_of mo ON mo.field_media_of_target_id = n.nid
  JOIN media_field_data mfd ON mfd.mid = mo.entity_id
  JOIN media__field_media_file mff ON mfd.mid = mff.entity_id
  JOIN file_managed fm ON mff.field_media_file_target_id = fm.fid
  WHERE fm.uri LIKE 'fedora://%'
    AND fm.uri NOT LIKE '%Extracted Text%'
    AND fm.uri NOT LIKE '%FITS File%';
"

echo "🐚 Executing query..."
if ! NODE_ROWS=$(drush sql:query "$SQL_QUERY"); then
  echo "❌ SQL query failed. Exiting."
  exit 1
else
  echo "✅ SQL query successful."
  NODE_COUNT=$(echo "$NODE_ROWS" | wc -l)
  echo "🔍 Number of results: $NODE_COUNT"
fi

echo "📊 Processing rows..."
declare -A RESULTS
JSON_OUTPUT="{"

while IFS=$'\t' read -r NID URI; do
  [[ "$NID" =~ ^[0-9]+$ ]] || continue  # Skip header row if it exists

  if [[ "$URI" == fedora://* ]]; then
    FEDORA_PATH="${URI#fedora://}"
    FEDORA_URL="http://fcrepo:8080/fcrepo/rest/${FEDORA_PATH}"

    echo "🌐 Fetching checksum for $FEDORA_URL"
    DIGEST_HEADER=$(curl -sI -H "Want-Digest: sha-256" "$FEDORA_URL" | grep -i ^Digest)

    if [[ -z "$DIGEST_HEADER" ]]; then
      echo "⚠️  No Digest returned for $FEDORA_URL"
      CHECKSUM="MISSING"
    else
      CHECKSUM=$(echo "$DIGEST_HEADER" | sed -n 's/^Digest: sha-256=\(.*\)/\1/p')
    fi

  else
    echo "❓ Unknown URI scheme: $URI"
    continue
  fi

  # Append to JSON output
  JSON_OUTPUT+="\"$NID\": {
    \"fedora_uri\": \"${URI}\",
    \"checksum\": \"${CHECKSUM}\"
  },"

done <<< "$NODE_ROWS"

# Remove trailing comma and close JSON
JSON_OUTPUT="${JSON_OUTPUT%,}"
JSON_OUTPUT+="}"

echo "$JSON_OUTPUT" > fedora_checksums.json
echo "✅ Checksums saved to fedora_checksums.json"