This Python web application uses Flask to handle PDF files, extract their text content, and restructure the output based on a template JSON. It leverages various libraries including PyPDFLoader for reading PDFs, Flask for the web server, and Google Cloud's AI and machine learning services.
We recommend that you setup the API within a python venv to ensure you only have the modules needed and there are not circular imports.
Along with that, use a python version between 3.10.3 - 3.11 to ensure the modules work as expected.
Currently, the requirements.txt only works for MacOS
- requirements.txt
- [requirements_windows.txt] In progress
To install these dependencies, use pip install -r requirements.txt
Along with those dependencies, you will need to set up a GCP with Vertex AI service turned on.
Following that, you must authenticate your platform for gcloud CLI. gcloud CLI steps
Create a .env file in the same directory as the api. The .env should include GOOGLE_API_KEY, GOOGLE_LOCATION_ID, and GOOGLE_PROJECT_ID.
- Input: Multipart form-data containing a PDF file.
- Output: A JSON response with the processed data or an error message.
- File Handling: The uploaded PDF file is saved to a temporary location.
- PDF Processing: pdf_to_text function extracts text from each page of the PDF using PyPDFLoader.
- Data Structuring: structureOutput function restructures the extracted text based on a predefined template JSON loaded at application startup.
- Error Handling: Proper error responses are generated if any part of the process fails.
- Cleanup: Temporary files are deleted after processing.
The application can be started by running the module directly, which will also load the JSON template data and run the Flask app in debug mode.