PDF Parsy

Extract text and images from PDFs through and endpoint that runs as a lambda function built with the serverless framework

Setup application

docker-compose build
docker-compose run --rm app pip install -r requirements_dev.txt
docker-compose run --rm -e S3_SAMPLE_OBJECT_KEY=my_s3_bucket -e S3_SAMPLE_IMAGES_FOLDER=my_s3_folder app python initial_setup.py

Invoke local

docker-compose run --rm -e S3_BUCKET=my_bucket -e S3_ACCESS_KEY_ID=my_access_key_id -e S3_SECRET_ACCESS_KEY=my_aws_secret_access_key app sls invoke local -f pdf_to_text -p fixtures/pdf_input.json

AWS credentials should have permissions to:

• Read from the provided S3 bucket that stores the PDF to be analyzed

• Write the extracted images from the PDF on the folder set by S3_SAMPLE_IMAGES_FOLDER on the previous step

• Run the function:

docker-compose run --rm app sls invoke local -f pdf_to_text -p fixtures/pdf_input.json

Deploy

docker-compose run --rm -e AWS_ACCESS_KEY_ID=my_access_key_id -e AWS_SECRET_ACCESS_KEY=my_aws_secret_access_key app sls deploy --verbose

• Set the following env vars on the AWS page for your newly created lamba functions:
• S3_BUCKET
• S3_ACCESS_KEY_ID
• S3_SECRET_ACCESS_KEY

Endpoints

• GET /pdf_to_text?s3_pdf_key=key_to_object_on_s3
  • Extracts text from PDF and returns in a json response
• POST /pdf_images?s3_pdf_key=key_to_object_to_s3&s3_images_folder=s3_folder_to_store_images
  • Extracts images from PDF and uploads to the specified s3_images_folder on the bucket configured in the application