Deploy: OCR support for PDF/Images/DOCX

.dockerignore

@@ -0,0 +1,10 @@
+node_modules
+dist
+.git
+.gitignore
+.env
+README.md
+Dockerfile
+.dockerignore
+.lovable
+supabase

.env

@@ -0,0 +1,3 @@
+VITE_SUPABASE_PROJECT_ID="momxyrzmibxccqkmqbay"
+VITE_SUPABASE_PUBLISHABLE_KEY="eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJpc3MiOiJzdXBhYmFzZSIsInJlZiI6Im1vbXh5cnptaWJ4Y2Nxa21xYmF5Iiwicm9sZSI6ImFub24iLCJpYXQiOjE3NzY2MTcwNjksImV4cCI6MjA5MjE5MzA2OX0.en2svoWyapwx4olnlOdlVGVy9_2zC4z_05hh-zL7elI"
+VITE_SUPABASE_URL="https://momxyrzmibxccqkmqbay.supabase.co"

.lovable/plan.md

@@ -0,0 +1,32 @@
+
+# Revised Plan — Local Whisper Note
+
+## The constraint
+You want `openai-whisper` (the local Python package) instead of the OpenAI Whisper API for audio/video transcription.
+
+## Why this can't run inside Lovable's stack
+Lovable apps run as **React + Vite (browser)** + **Supabase Edge Functions (Deno runtime)**. There is no Python process, no GPU, and Edge Functions have a ~150s timeout and small memory limits — local Whisper (PyTorch model, multi-GB, CPU/GPU heavy) cannot run there.
+
+## Two viable paths
+
+**Option A — Self-hosted Whisper microservice (true local Whisper)**
+- You run a tiny FastAPI service somewhere you control (your laptop, a VPS, RunPod, Modal, Fly.io GPU, etc.) that exposes `POST /transcribe`.
+- That service does `pip install -U openai-whisper` + `whisper.load_model("base"|"small"|"medium")` and returns `{ text }`.
+- Our Supabase Edge Function `ingest` uploads the audio file to that endpoint via its public URL + a shared secret, then stores `raw_text`.
+- You give us: `WHISPER_ENDPOINT_URL` + `WHISPER_AUTH_TOKEN` (stored as Lovable Cloud secrets).
+- Pros: real local Whisper, zero per-minute cost. Cons: you maintain the box.
+
+**Option B — Hosted Whisper-compatible API (no API-key vendor lock)**
+- Use a provider that runs the same `openai-whisper` model server-side (Replicate, Groq Whisper-large-v3, Deepgram Nova, Fal.ai). One key, no infra.
+- Same code path as A, just a different URL.
+
+## Recommended
+Go with **Option A**. I'll wire the Edge Function to call a configurable `WHISPER_ENDPOINT_URL`, and provide you a ready-to-run `whisper_server.py` (FastAPI + openai-whisper) plus a `requirements.txt` and run instructions, so you can `pip install -U openai-whisper && python whisper_server.py` on any machine and it just works. If you'd rather start with a hosted provider for now, swap the URL later — no code change.
+
+## Plan changes vs. previous
+- Remove "OpenAI Whisper API key" from required secrets.
+- Add `WHISPER_ENDPOINT_URL` + `WHISPER_AUTH_TOKEN` secrets.
+- Slice 5 (Audio sources) deliverables now include: the standalone `whisper_server.py` script + README, plus the Edge Function client that POSTs the audio bytes and handles long-running transcription via the existing `jobs` table + Realtime progress.
+- YouTube path unchanged (still needs a transcript API since `yt-dlp` also can't run in Edge Functions). If you'd prefer, the same self-hosted box can run `yt-dlp` + Whisper and we route YouTube through it too — tell me and I'll fold that in.
+
+Everything else in the previously approved plan (dark UI, Notes/Flashcards/Quiz/Podcast/Chat tabs, pgvector RAG, jobs + Realtime, hash de-dup, Lovable AI Gateway for Gemini) stays the same.

Dockerfile

@@ -0,0 +1,23 @@
+# Build stage
+FROM node:20-alpine AS build
+WORKDIR /app
+COPY package*.json ./
+RUN npm install
+COPY . .
+# These will be provided by Hugging Face Secrets during build
+# Vite automatically picks up VITE_ prefixed variables from the environment
+RUN npm run build
+
+# Production stage
+FROM nginx:stable-alpine
+COPY --from=build /app/dist /usr/share/nginx/html
+RUN echo 'server { \
+    listen 7860; \
+    location / { \
+        root /usr/share/nginx/html; \
+        index index.html; \
+        try_files $uri $uri/ /index.html; \
+    } \
+}' > /etc/nginx/conf.d/default.conf
+EXPOSE 7860
+CMD ["nginx", "-g", "daemon off;"]

README.md

@@ -0,0 +1,14 @@
+---
+title: SOURCE.IO
+emoji: 🚀
+colorFrom: indigo
+colorTo: blue
+sdk: docker
+app_port: 7860
+---
+
+# SOURCE.AI
+
+SOURCE TO YOUR STUDIES
+
+This project is a React/Vite application deployed on Hugging Face Spaces.

components.json

@@ -0,0 +1,20 @@
+{
+  "$schema": "https://ui.shadcn.com/schema.json",
+  "style": "default",
+  "rsc": false,
+  "tsx": true,
+  "tailwind": {
+    "config": "tailwind.config.ts",
+    "css": "src/index.css",
+    "baseColor": "slate",
+    "cssVariables": true,
+    "prefix": ""
+  },
+  "aliases": {
+    "components": "@/components",
+    "utils": "@/lib/utils",
+    "ui": "@/components/ui",
+    "lib": "@/lib",
+    "hooks": "@/hooks"
+  }
+}