Improvements

.env.example

@@ -0,0 +1,10 @@
+API_BASE_URL=http://localhost:8000                  # Temporary btw
+OLLAMA_BASE_URL=http://127.0.0.1:11434              # Ollama's default
+PRECIS_ALLOWED_ORIGINS=http://localhost:5173        # Just front end, might make it more/less strict but I just need something consistent rn
+PRECIS_API_KEY=replace-with-a-long-random-secret    # Once the API is actually up, this'll be needed
+PRECIS_DEFAULT_MODEL=phi4-mini:latest
+PRECIS_AVAILABLE_MODELS=phi4-mini:latest,qwen:4b    # Only here so both front and backend have it
+MAX_SUMMARY_TOKENS=120
+TEMPERATURE=0.2                                     # Random choice right now, will probably tweak
+PRECIS_MAX_UPLOAD_BYTES=10485760
+PRECIS_MAX_TRANSCRIPT_CHARS=250000

.github/workflows/hf.yml

@@ -1,18 +1,15 @@
-# Sync HuggingFace
-
-name: Sync
+name: Sync to Hugging Face hub
 on: [push]
+
 jobs:
-  sync:
+  sync-to-hub:
     runs-on: ubuntu-latest
-
     steps:
       - uses: actions/checkout@v3
-      - uses: actions/setup-python@v4
-        with: { python-version: '3.9' }
-
-      - run: pip install huggingface_hub
-
-      - env: { HF_TOKEN: '${{ secrets.HF_TOKEN }}' }
-        run: python -c "import os; from huggingface_hub import HfApi; HfApi().upload_folder(repo_id='compendious/precis', folder_path='.', repo_type='space', token=os.environ['HF_TOKEN'], ignore_patterns=['.git*'])"
-  
+        with:
+          fetch-depth: 0
+          lfs: true
+      - name: Push to hub
+        env:
+          HF_TOKEN: ${{ secrets.HF_TOKEN }}
+        run: git push --force https://compendious:${HF_TOKEN}@huggingface.co/spaces/compendious/precis main

.gitignore

@@ -2,7 +2,10 @@
 # Python Remainders
 **cache**
 *.ipynb
-.venv
+*.venv
+.env
+.env.*
+!.env.example
 
 # Front end
 node_modules

README.md

@@ -2,15 +2,28 @@
 
 A system for compressing long-form content into clear, structured summaries. Précis is designed for videos, articles, and papers. Paste a YouTube link, drop in an article, or upload a text file. Précis will pulls the key facts into a single sentence using a local LLM via [Ollama](https://ollama.com).
 
-## Stack
+## Features
 
-| Layer    | Tech |
-|----------|------|
-| Frontend | React 19 + Vite |
-| Backend  | FastAPI (Python) |
-| LLM      | Ollama (phi4-mini, qwen-4b) |
+- **YouTube summarization**: paste a URL, transcript is fetched automatically via `youtube-transcript-api`
+- **Article / transcript**: paste any text directly
+- **File upload**: drag-and-drop `.txt` files
+- **Streaming**: summaries stream token-by-token from Ollama via NDJSON
+- **Model switching**: choose between available Ollama models from the UI
+
+## API Endpoints
 
-## Setup
+| Method  |       Path              |     Description       |
+|---------|-------------------------|-----------------------|
+| `GET`   | `/health`               | Health check          |
+| `GET`   | `/status`               | Ollama statuses, etc. |
+| `GET`   | `/models`               | List available models |
+| `POST`  | `/summarize/transcript` | Raw text summary      |
+| `POST`  | `/summarize/youtube`    | YouTube video by URL  |
+| `POST`  | `/summarize/file`       | `.txt` file summary   |
+
+All `/summarize/*` endpoints accept an optional `model` field to override the default.
+
+## Local Setup
 
 ### Prerequisites
 
@@ -43,26 +56,79 @@ npm run dev
 
 Runs on `http://localhost:5173`.
 
-## Features
+<!-- ## Data -->
 
-- **YouTube summarization**: paste a URL, transcript is fetched automatically via `youtube-transcript-api`
-- **Article / transcript**: paste any text directly
-- **File upload**: drag-and-drop `.txt` files
-- **Streaming**: summaries stream token-by-token from Ollama via NDJSON
-- **Model switching**: choose between available Ollama models from the UI
+<!-- Later, for fine-tuning data details -->
 
-## API Endpoints
+<!-- Interview Dataset -->
+<!-- 
 
-| Method | Path | Description |
-|--------|------|-------------|
-| `GET`  | `/health` | Health check |
-| `GET`  | `/status` | Service status, available models, Ollama reachability |
-| `GET`  | `/models` | List available models |
-| `POST` | `/summarize/transcript` | Summarize raw text (NDJSON stream) |
-| `POST` | `/summarize/youtube` | Summarize a YouTube video by URL (NDJSON stream) |
-| `POST` | `/summarize/file` | Summarize an uploaded `.txt` file (NDJSON stream) |
+@article{zhu2021mediasum,
+  title={MediaSum: A Large-scale Media Interview Dataset for Dialogue Summarization},
+  author={Zhu, Chenguang and Liu, Yang and Mei, Jie and Zeng, Michael},
+  journal={arXiv preprint arXiv:2103.06410},
+  year={2021}
+} 
 
-All `/summarize/*` endpoints accept an optional `model` field to override the default.
+-->
+
+<!--------------------------------------------------------------------------------------------------->
+
+<!-- 
+
+@inproceedings{chen-etal-2021-dialogsum,
+    title = "{D}ialog{S}um: {A} Real-Life Scenario Dialogue Summarization Dataset",
+    author = "Chen, Yulong  and
+      Liu, Yang  and
+      Chen, Liang  and
+      Zhang, Yue",
+    booktitle = "Findings of the Association for Computational Linguistics: ACL-IJCNLP 2021",
+    month = aug,
+    year = "2021",
+    address = "Online",
+    publisher = "Association for Computational Linguistics",
+    url = "https://aclanthology.org/2021.findings-acl.449",
+    doi = "10.18653/v1/2021.findings-acl.449",
+    pages = "5062--5074",
+}
+
+-->
+
+<!------------------------------------------------------------------------------------------------->
+
+<!-- "Single question followed by an answer" dataset -->
+
+<!--
+
+@article{wang2022squality,
+  title        = {SQuALITY: Building a Long-Document Summarization Dataset the Hard Way},
+  author       = {Wang, Alex and Pang, Richard Yuanzhe and Chen, Angelica and Phang, Jason and Bowman, Samuel R.},
+  journal      = {arXiv preprint arXiv:2205.11465},
+  year         = {2022},
+  archivePrefix = {arXiv},
+  eprint       = {2205.11465},
+  primaryClass = {cs.CL},
+  doi          = {10.48550/arXiv.2205.11465},
+  url          = {https://doi.org/10.48550/arXiv.2205.11465}
+}
+
+-->
+
+<!------------------------------------------------------------------------------------------------->
+
+<!-- High Quality Query-Answer (concise) examples -->
+
+<!--
+
+@inproceedings{nguyen2016msmarco,
+  title     = {MS MARCO: A Human Generated MAchine Reading COmprehension Dataset},
+  author    = {Nguyen, Tri and Rosenberg, Mir and Song, Xia and Gao, Jianfeng and Tiwary, Saurabh and Majumder, Rangan and Deng, Li},
+  booktitle = {Proceedings of the Workshop on Cognitive Computation: Integrating Neural and Symbolic Approaches 2016},
+  year      = {2016},
+  publisher = {CEUR-WS.org}
+}
+
+-->
 
 ## License
 

backend/app.py

@@ -2,11 +2,17 @@ import asyncio
 from typing import Optional
 
 import httpx
-from fastapi import FastAPI, HTTPException, UploadFile, File
+from fastapi import FastAPI, HTTPException, UploadFile, File, Header, Request
 from fastapi.middleware.cors import CORSMiddleware
-from fastapi.responses import StreamingResponse
 
-from config import OLLAMA_BASE_URL, DEFAULT_MODEL, AVAILABLE_MODELS
+from config import (
+    OLLAMA_BASE_URL,
+    DEFAULT_MODEL,
+    AVAILABLE_MODELS,
+    ALLOWED_ORIGINS,
+    API_KEY,
+    MAX_UPLOAD_BYTES,
+)
 from schemas import TranscriptRequest, YouTubeRequest
 from ollama import stream_summary
 from youtube import extract_video_id, fetch_transcript
@@ -19,13 +25,23 @@ app = FastAPI(
 
 app.add_middleware(
     CORSMiddleware,
-    allow_origins=["*"],
-    allow_credentials=True,
-    allow_methods=["*"],
-    allow_headers=["*"],
+    allow_origins=ALLOWED_ORIGINS,
+    allow_credentials=False,
+    allow_methods=["POST", "GET", "OPTIONS"],
+    allow_headers=["Content-Type", "X-API-Key"],
 )
 
 
+def verify_api_key(x_api_key: Optional[str] = Header(default=None, alias="X-API-Key")):
+    if not API_KEY:
+        raise HTTPException(
+            status_code=500,
+            detail="Server misconfigured: PRECIS_API_KEY must be set.",
+        )
+    if x_api_key != API_KEY:
+        raise HTTPException(status_code=401, detail="Invalid API key.")
+
+
 @app.get("/health")
 async def health():
     return {"status": "healthy", "service": "precis"}
@@ -56,25 +72,51 @@ async def list_models():
 
 
 @app.post("/summarize/transcript")
-async def summarize_transcript(request: TranscriptRequest):
+async def summarize_transcript(
+    request: TranscriptRequest,
+    x_api_key: Optional[str] = Header(default=None, alias="X-API-Key"),
+):
+    verify_api_key(x_api_key)
     if not request.text.strip():
         raise HTTPException(status_code=400, detail="Text must not be empty.")
     return stream_summary(request.text, title=request.title, model=request.model)
 
 
 @app.post("/summarize/youtube")
-async def summarize_youtube(request: YouTubeRequest):
+async def summarize_youtube(
+    request: YouTubeRequest,
+    x_api_key: Optional[str] = Header(default=None, alias="X-API-Key"),
+):
+    verify_api_key(x_api_key)
     video_id = extract_video_id(request.url)
     text = await asyncio.to_thread(fetch_transcript, video_id)
     return stream_summary(text, model=request.model)
 
 
 @app.post("/summarize/file")
-async def summarize_file(file: UploadFile = File(...), model: Optional[str] = None):
+async def summarize_file(
+    req: Request,
+    file: UploadFile = File(...),
+    model: Optional[str] = None,
+    x_api_key: Optional[str] = Header(default=None, alias="X-API-Key"),
+):
+    verify_api_key(x_api_key)
+    content_length = req.headers.get("content-length")
+    if content_length and int(content_length) > MAX_UPLOAD_BYTES:
+        raise HTTPException(status_code=413, detail="Uploaded file is too large.")
+
     if not file.filename.endswith(".txt"):
         raise HTTPException(status_code=400, detail="Only .txt files are supported.")
+
     content = await file.read()
-    text = content.decode("utf-8")
+    if len(content) > MAX_UPLOAD_BYTES:
+        raise HTTPException(status_code=413, detail="Uploaded file is too large.")
+
+    try:
+        text = content.decode("utf-8")
+    except UnicodeDecodeError:
+        raise HTTPException(status_code=400, detail="File must be valid UTF-8 text.")
+
     if not text.strip():
         raise HTTPException(status_code=400, detail="Uploaded file is empty.")
     return stream_summary(text, title=file.filename, model=model)

backend/config.py

@@ -1,5 +1,41 @@
-OLLAMA_BASE_URL = "http://127.0.0.1:11434"
-DEFAULT_MODEL = "phi4-mini:latest"
-AVAILABLE_MODELS = ["phi4-mini:latest", "qwen:4b"]
-MAX_SUMMARY_TOKENS = 120
-TEMPERATURE = 0.2
+import os
+from pathlib import Path
+
+from dotenv import load_dotenv
+
+
+ROOT_ENV_PATH = Path(__file__).resolve().parents[1] / ".env"
+load_dotenv(ROOT_ENV_PATH)
+
+
+def _csv_env(name: str, default: list[str]) -> list[str]:
+	raw = os.getenv(name, "")
+	if not raw.strip():
+		return default
+	values = [value.strip() for value in raw.split(",") if value.strip()]
+	return values or default
+
+
+def _required_env(name: str) -> str:
+	value = os.getenv(name, "").strip()
+	if not value:
+		raise RuntimeError(f"Missing required environment variable: {name}")
+	return value
+
+
+OLLAMA_BASE_URL = _required_env("PRECIS_OLLAMA_BASE_URL")
+DEFAULT_MODEL = _required_env("PRECIS_DEFAULT_MODEL")
+AVAILABLE_MODELS = _csv_env("PRECIS_AVAILABLE_MODELS", [DEFAULT_MODEL])
+if DEFAULT_MODEL not in AVAILABLE_MODELS:
+	AVAILABLE_MODELS = [DEFAULT_MODEL, *AVAILABLE_MODELS]
+
+ALLOWED_ORIGINS = _csv_env("PRECIS_ALLOWED_ORIGINS", [])
+if not ALLOWED_ORIGINS:
+	raise RuntimeError("Missing required environment variable: PRECIS_ALLOWED_ORIGINS")
+
+API_KEY = _required_env("PRECIS_API_KEY")
+
+MAX_SUMMARY_TOKENS = int(os.getenv("PRECIS_MAX_SUMMARY_TOKENS", "120"))
+TEMPERATURE = float(os.getenv("PRECIS_TEMPERATURE", "0.2"))
+MAX_UPLOAD_BYTES = int(os.getenv("PRECIS_MAX_UPLOAD_BYTES", "10485760"))
+MAX_TRANSCRIPT_CHARS = int(os.getenv("PRECIS_MAX_TRANSCRIPT_CHARS", "120000"))

backend/schemas.py

@@ -1,13 +1,15 @@
 from typing import Optional
-from pydantic import BaseModel
+from pydantic import BaseModel, Field
+
+from config import MAX_TRANSCRIPT_CHARS
 
 
 class YouTubeRequest(BaseModel):
-    url: str
+    url: str = Field(min_length=10, max_length=2048)
     model: Optional[str] = None
 
 
 class TranscriptRequest(BaseModel):
-    text: str
+    text: str = Field(min_length=1, max_length=MAX_TRANSCRIPT_CHARS)
     title: Optional[str] = None
     model: Optional[str] = None

frontend/src/App.jsx

@@ -2,17 +2,15 @@ import { useState, useRef } from 'react'
 import InlineResult from './components/InlineResult'
 import { useStreaming } from './hooks/useStreaming'
 import logoSvg from './assets/logo.svg'
+import { API_BASE, AVAILABLE_MODELS, DEFAULT_MODEL } from './config'
 import './App.css'
 
-const API_BASE = 'http://localhost:8000'
-const MODELS = ['phi4-mini:latest', 'qwen:4b']
-
 function App() {
   const [activeTab, setActiveTab] = useState('youtube')
   const [youtubeUrl, setYoutubeUrl] = useState('')
   const [transcript, setTranscript] = useState('')
   const [selectedFile, setSelectedFile] = useState(null)
-  const [selectedModel, setSelectedModel] = useState(MODELS[0])
+  const [selectedModel, setSelectedModel] = useState(DEFAULT_MODEL)
   const fileInputRef = useRef(null)
 
   const { loading, response, error, streamingText, submit } = useStreaming()
@@ -54,7 +52,7 @@ function App() {
             onChange={(e) => setSelectedModel(e.target.value)}
             disabled={loading}
           >
-            {MODELS.map((m) => <option key={m} value={m}>{m}</option>)}
+            {AVAILABLE_MODELS.map((m) => <option key={m} value={m}>{m}</option>)}
           </select>
           <a href={`${API_BASE}/docs`} target="_blank" rel="noopener noreferrer" className="btn" style={{ textDecoration: 'none' }}>
             API Docs

frontend/src/config.js

@@ -0,0 +1,25 @@
+const parseCsv = (raw, fallback = []) => {
+  if (!raw || !raw.trim()) return fallback
+  return raw.split(',').map((part) => part.trim()).filter(Boolean)
+}
+
+const requiredEnv = (name) => {
+  const value = import.meta.env[name]
+  if (!value || !String(value).trim()) {
+    throw new Error(`Missing required environment variable: ${name}`)
+  }
+  return String(value).trim()
+}
+
+export const API_BASE = requiredEnv('PRECIS_API_BASE_URL')
+export const API_KEY = requiredEnv('PRECIS_API_KEY')
+
+export const DEFAULT_MODEL = requiredEnv('PRECIS_DEFAULT_MODEL')
+export const AVAILABLE_MODELS = parseCsv(
+  import.meta.env.PRECIS_AVAILABLE_MODELS,
+  [DEFAULT_MODEL],
+)
+
+export const authHeaders = (headers = {}) => (
+  API_KEY ? { ...headers, 'X-API-Key': API_KEY } : headers
+)

frontend/src/hooks/useStreaming.js

@@ -1,6 +1,5 @@
 import { useState, useRef } from 'react'
-
-const API_BASE = 'http://localhost:8000'
+import { API_BASE, authHeaders } from '../config'
 
 export function useStreaming() {
     const [loading, setLoading] = useState(false)
@@ -47,9 +46,10 @@ export function useStreaming() {
         }
 
         if (json) {
-            fetchOpts.headers = { 'Content-Type': 'application/json' }
+            fetchOpts.headers = authHeaders({ 'Content-Type': 'application/json' })
             fetchOpts.body = JSON.stringify(json)
         } else if (formData) {
+            fetchOpts.headers = authHeaders()
             fetchOpts.body = formData
         }
 

frontend/vite.config.js

@@ -3,5 +3,7 @@ import react from '@vitejs/plugin-react'
 
 // https://vite.dev/config/
 export default defineConfig({
+  envDir: '..',
+  envPrefix: ['VITE_', 'PRECIS_'],
   plugins: [react()],
 })

requirements.txt

@@ -13,3 +13,5 @@ uvicorn
 httpx                   # async HTTP client for Ollama calls
 python-multipart        # required by FastAPI for file uploads
 youtube-transcript-api  # YouTube transcript fetching
+python-dotenv           # .env loading for backend/frontend config
+