docs and logging

document_intelligence/README.md

@@ -1 +1,99 @@
+# NougatEquation Task
+
+This project implements a FastAPI application to process PDF files, extract images, and recognize LaTeX equations from the images using Hugging Face's VisionEncoderDecoderModel. The application is designed to handle PDF uploads, extract images from the pages, and run them through a model to extract LaTeX mathematical notations.
+
+## Features
+
+- Accepts PDF files via a FastAPI endpoint.
+- Extracts images from the PDF.
+- Recognizes LaTeX equations in images.
+- Processes image batches to ensure efficient resource usage.
+- Returns the extracted LaTeX equations in a JSON response.
+
+## Requirements
+
+Before running the project, ensure you have the following dependencies installed:
+
+- Python 3.8+
+- `torch` (PyTorch)
+- `transformers` (Hugging Face Transformers)
+- `Pillow` (Python Imaging Library for image processing)
+- `PyMuPDF` (for PDF processing)
+- `fastapi` (web framework for the API)
+- `uvicorn` (ASGI server)
+
+### Installation
+
+1. **Clone the repository**:
+
+    ```bash
+    git clone https://github.com/yourusername/nougat-equation-task.git
+    cd nougat-equation-task
+    ```
+
+2. **Create a virtual environment**:
+
+    ```bash
+    python3 -m venv nougat_env
+    source nougat_env/bin/activate
+    ```
+
+3. **Install the dependencies**:
+
+    First, create a `requirements.txt` file:
+
+    ```txt
+    torch
+    transformers
+    Pillow
+    PyMuPDF
+    fastapi
+    uvicorn
+    ```
+
+    Then install the dependencies using `pip`:
+
+    ```bash
+    pip install -r requirements.txt
+    ```
+
+### Usage
+
+1. **Start the FastAPI server**:
+
+    You can run the FastAPI server using `uvicorn`:
+
+    ```bash
+    uvicorn main:app --reload --host 0.0.0.0 --port 8000
+    ```
+
+    This will start the server at `http://0.0.0.0:8000`.
+
+2. **API Endpoint**:
+
+    - **POST /process_pdf**: Accepts a PDF file and extracts LaTeX equations from the images in the file.
+
+    Example usage with `curl`:
+
+    ```bash
+    curl -X 'POST' \
+      'http://127.0.0.1:8000/process_pdf' \
+      -H 'accept: application/json' \
+      -H 'Content-Type: multipart/form-data' \
+      -F '[email protected]'
+    ```
+
+    The response will be a JSON object with the extracted LaTeX equations.
+
+### Example
+
+Suppose you upload a PDF that contains images of mathematical equations. The API will process the PDF, extract the images, and return LaTeX code for the detected equations in JSON format.
+
+Example response:
+
+```json
+{
+  "0": ["\\( x^2 + y^2 = z^2 \\)"],
+  "1": ["\\[ e = mc^2 \\]"]
+}
 

document_intelligence/nougat/run.py

@@ -84,6 +84,7 @@ def __init__(self, images: List[Image.Image], batch_size: int):
         self.index = 0
 
     def next_batch(self) -> Tuple[List[Image.Image], int]:
+        logger.info(f"Processing batch {self.index} to {self.index + self.batch_size}")
         if self.index >= len(self.images):
             return None, self.index