Apppublic

mohanm18/CET_Rigged_Parser

sourceHugging Faceupdated 3mo agoView on Hugging Face

1likes

App README

MHT-CET College Hunt

A read-only Streamlit app for exploring official MHT-CET Engineering cutoff data.

Architecture

PDF → Docling JSON → Parser Script (Offline) → master_cutoffs.parquet (Supabase Storage) → Streamlit App

This application never parses PDFs or JSON at runtime. It downloads the pre-built Parquet dataset from Supabase Storage, caches it using @st.cache_data, and serves search, filtering, visualization, and prediction functionality.

Setup

bash

python -m venv venv

Windows:

bash

venv\Scripts\activate

Linux / macOS:

bash

source venv/bin/activate

Install dependencies:

bash

pip install -r requirements.txt

Required Environment Variables

Configure the following secrets before running the application:

SUPABASE_URL
SUPABASE_KEY
SUPABASE_BUCKET
PARQUET_FILE

Run

bash

streamlit run app.py

Project Layout

app.py                     # Application entry point

components/
    advanced_search.py
    charts.py
    explorer.py
    filters.py
    predictor.py

utils/
    analytics.py
    data_loader.py
    exports.py
    helpers.py

.streamlit/
    secrets.toml           # Local secrets (not committed)

requirements.txt
README.md

Features

—🎯 College Predictor
—🏛️ College Explorer
—🌿 Branch Explorer
—🔎 Advanced Search
—📊 Interactive Plotly Charts
—📥 CSV & Excel Export
—⚡ Cached Parquet Loading
—☁️ Dataset served from Supabase Storage
—🚀 Optimized for Streamlit deployment

Notes

—The dataset is generated offline using the parser.
—Streamlit only downloads the latest Parquet file from Supabase Storage.
—The dataset is cached after the first load for fast subsequent access.
—Institute type is derived automatically from the status column.
—All filters are optional and combined using logical AND.
—The Predictor tab filters colleges based on the entered percentile.