Added sandbox for extracting figure and text out of PDF's and writting them out as JSON, ready to import into lambda-feedback.

Author: ggorman

wizard/README.md

@@ -1,3 +1,36 @@
 This is a folder where we'll generate a 'wizard' to automatically process input documents ready for in2lambda to reacte a JSON (or to directly create a JSON)
 
-We'll work on this branch 'hackathon'
+We'll work on this branch 'hackathon'
+
+
+# README
+
+## Overview
+This Jupyter Notebook (`sandbox.ipynb`) is designed for processing scientific documents, extracting mathematical expressions, and formatting them in Markdown. It leverages Azure OpenAI's LLM capabilities for text transformation.
+
+## Features
+- Loads PDFs and extracts text using `UnstructuredPDFLoader` and `PyMuPDF`.
+- Converts mathematical expressions into properly formatted Markdown.
+- Uses `langchain` and `AzureChatOpenAI` for text processing.
+- Supports structured output parsing using `pydantic`.
+
+## Requirements
+Ensure you have the following installed:
+- Python 3.8+
+- `pip install -r requirements.txt`
+- `langchain`, `langchain_openai`, `pydantic`, `dotenv`, `PyMuPDF`, `PIL`
+
+## Setup
+1. Create a `.env` file in the root directory and add your Azure OpenAI API keys:
+   ```env
+   AZURE_OPENAI_API_KEY=<your-api-key>
+   AZURE_OPENAI_ENDPOINT=<your-endpoint>
+   ```
+4. Open `sandbox.ipynb` and execute the cells to process your documents.
+
+## Notes
+- Ensure your API key and endpoint are correct, as they are required for LLM functionality.
+- The notebook is designed for scientific documents, but can be extended to other text formats.
+
+
+

wizard/requirements.txt

@@ -0,0 +1,13 @@
+langchain
+pandas
+langchain-openai
+pypdf
+chromadb
+langchain_community
+unstructured
+PyMuPDF
+Pillow
+pdfminer.six
+unstructured[pdf]
+ipywidgets
+jupyter