Add Dockerfile for Hugging Face Space deployment, update README with deployment instructions and API usage, and create requirements.txt for dependencies

Dockerfile

@@ -0,0 +1,31 @@
+# Use PyTorch base image with CUDA support
+FROM pytorch/pytorch:2.9.1-cuda12.6-cudnn9-runtime
+
+# Set up a new user named "user" with user ID 1000
+RUN useradd -m -u 1000 user
+
+# Switch to the "user" user
+USER user
+
+# Set home to the user's home directory
+ENV HOME=/home/user \
+    PATH=/home/user/.local/bin:$PATH
+
+# Set the working directory to the user's home directory
+WORKDIR $HOME/app
+
+# Try and run pip command after setting the user with `USER user` to avoid permission issues with Python
+RUN pip install --no-cache-dir --upgrade pip
+
+# Copy requirements.txt and install dependencies
+COPY --chown=user requirements.txt $HOME/app/
+RUN pip install --no-cache-dir -r requirements.txt
+
+# Copy the current directory contents into the container at $HOME/app setting the owner to the user
+COPY --chown=user . $HOME/app
+
+# Expose port 7860 (Hugging Face Space default port)
+EXPOSE 7860
+
+# Set the startup command
+CMD ["uvicorn", "marker.scripts.server:app", "--host", "0.0.0.0", "--port", "7860"]

README.md

@@ -1,3 +1,12 @@
+---
+title: Marker PDF Converter
+emoji: 📄
+colorFrom: blue
+colorTo: purple
+sdk: docker
+app_port: 7860
+---
+
 # Marker
 
 Marker converts documents to markdown, JSON, chunks, and HTML quickly and accurately.
@@ -433,6 +442,76 @@ requests.post("http://localhost:8001/marker", data=json.dumps(post_data)).json()
 
 Note that this is not a very robust API, and is only intended for small-scale use.  If you want to use this server, but want a more robust conversion option, you can use the hosted [Datalab API](https://www.datalab.to/plans).
 
+## Deploying to Hugging Face Space
+
+This project can be deployed to Hugging Face Space as a Docker Space. The API includes token authentication for security.
+
+### Setting up Secrets
+
+1. Go to your Space Settings on Hugging Face
+2. Navigate to the "Variables and secrets" section
+3. Create a new secret named `API_TOKEN`
+4. Set the secret value to your desired authentication token
+
+The token will be automatically injected as an environment variable at runtime.
+
+### Using the API
+
+Once deployed, the API requires authentication via Bearer token in the request header:
+
+```python
+import requests
+
+# Your token (set in Space secrets)
+token = "your-token-here"
+
+# Example: Convert PDF via file upload
+url = "https://your-space.hf.space/marker/upload"
+headers = {
+    "Authorization": f"Bearer {token}"
+}
+files = {
+    "file": ("document.pdf", open("document.pdf", "rb"), "application/pdf")
+}
+data = {
+    "output_format": "markdown",
+    "force_ocr": False
+}
+
+response = requests.post(url, headers=headers, files=files, data=data)
+result = response.json()
+print(result["output"])
+```
+
+Or using the JSON endpoint:
+
+```python
+import requests
+import json
+
+url = "https://your-space.hf.space/marker"
+headers = {
+    "Authorization": f"Bearer {token}",
+    "Content-Type": "application/json"
+}
+data = {
+    "filepath": "/path/to/file.pdf",
+    "output_format": "markdown"
+}
+
+response = requests.post(url, headers=headers, data=json.dumps(data))
+result = response.json()
+```
+
+### API Endpoints
+
+- `GET /` - API information page (no authentication required)
+- `GET /docs` - Interactive API documentation (no authentication required)
+- `POST /marker` - Convert PDF using file path (requires authentication)
+- `POST /marker/upload` - Convert PDF via file upload (requires authentication)
+
+All POST endpoints require the `Authorization: Bearer <token>` header with a valid token.
+
 # Troubleshooting
 
 There are some settings that you may find useful if things aren't working the way you expect:

marker/scripts/server.py

@@ -5,6 +5,8 @@ import os
 
 from pydantic import BaseModel, Field
 from starlette.responses import HTMLResponse
+from fastapi import FastAPI, Form, File, UploadFile, Depends, HTTPException, status
+from fastapi.security import HTTPBearer, HTTPAuthorizationCredentials
 
 from marker.config.parser import ConfigParser
 from marker.output import text_from_rendered
@@ -14,7 +16,6 @@ from contextlib import asynccontextmanager
 from typing import Optional, Annotated
 import io
 
-from fastapi import FastAPI, Form, File, UploadFile
 from marker.converters.pdf import PdfConverter
 from marker.models import create_model_dict
 from marker.settings import settings
@@ -38,6 +39,30 @@ async def lifespan(app: FastAPI):
 
 app = FastAPI(lifespan=lifespan)
 
+# Token verification
+security = HTTPBearer()
+
+def verify_token(credentials: HTTPAuthorizationCredentials = Depends(security)):
+    """
+    Verify the token from Authorization header.
+    Token is read from environment variable API_TOKEN.
+    """
+    token = os.environ.get("API_TOKEN")
+    
+    if not token:
+        # If no token is set in environment, allow access (for local development)
+        # In production, you should raise an exception here
+        return True
+    
+    if credentials.credentials != token:
+        raise HTTPException(
+            status_code=status.HTTP_401_UNAUTHORIZED,
+            detail="Invalid authentication token",
+            headers={"WWW-Authenticate": "Bearer"},
+        )
+    
+    return True
+
 
 @app.get("/")
 async def root():
@@ -128,7 +153,7 @@ async def _convert_pdf(params: CommonParams):
 
 
 @app.post("/marker")
-async def convert_pdf(params: CommonParams):
+async def convert_pdf(params: CommonParams, token_verified: bool = Depends(verify_token)):
     return await _convert_pdf(params)
 
 
@@ -141,6 +166,7 @@ async def convert_pdf_upload(
     file: UploadFile = File(
         ..., description="The PDF file to convert.", media_type="application/pdf"
     ),
+    token_verified: bool = Depends(verify_token),
 ):
     upload_path = os.path.join(UPLOAD_DIRECTORY, file.filename)
     with open(upload_path, "wb+") as upload_file:

requirements.txt

@@ -0,0 +1,36 @@
+# Core dependencies for marker
+# Note: PyTorch 2.9.1 is already included in the base image (pytorch/pytorch:2.9.1-cuda12.6-cudnn9-runtime)
+# So we don't need to install torch separately
+
+# FastAPI and server dependencies
+fastapi>=0.115.4
+uvicorn>=0.32.0
+python-multipart>=0.0.16
+
+# Marker core dependencies
+Pillow>=10.1.0
+pydantic>=2.4.2
+pydantic-settings>=2.0.3
+transformers>=4.45.2
+python-dotenv>=1.0.0
+tqdm>=4.66.1
+ftfy>=6.1.1
+rapidfuzz>=3.8.1
+surya-ocr>=0.17.0
+regex>=2024.4.28
+pdftext~=0.6.3
+markdownify>=1.1.0
+click>=8.2.0
+markdown2>=2.5.2
+filetype>=1.2.0
+google-genai>=1.0.0
+anthropic>=0.46.0
+scikit-learn>=1.6.1
+openai>=1.65.2
+
+# Optional dependencies for full document support
+mammoth>=1.9.0
+openpyxl>=3.1.5
+python-pptx>=1.0.2
+ebooklib>=0.18
+weasyprint>=63.1