|
| 1 | +from fastapi import FastAPI, HTTPException, Response, UploadFile, Form, File |
| 2 | +from pydantic import BaseModel |
| 3 | +import aiohttp |
| 4 | +from typing import Optional, Dict, Any |
| 5 | +import os |
| 6 | +from fastapi.staticfiles import StaticFiles |
| 7 | +from llama_parse import LlamaParse |
| 8 | + |
| 9 | + |
| 10 | +app = FastAPI() |
| 11 | +app.mount("/", StaticFiles(directory="app/static", html=True), name="static") |
| 12 | + |
| 13 | +ALLOWED_PREFIXES = os.getenv('ALLOWED_PREFIXES', '').split(',') |
| 14 | + |
| 15 | +class ProxyRequest(BaseModel): |
| 16 | + method: str |
| 17 | + url: str |
| 18 | + headers: Optional[Dict[str, str]] = None |
| 19 | + body: Optional[Any] = None |
| 20 | + |
| 21 | +@app.post('/cors/proxy') |
| 22 | +async def proxy(request: ProxyRequest): |
| 23 | + if not any(request.url.startswith(prefix) for prefix in ALLOWED_PREFIXES): |
| 24 | + raise HTTPException(status_code=403, detail='URL not allowed') |
| 25 | + |
| 26 | + kwargs = { |
| 27 | + 'method': request.method, |
| 28 | + 'url': request.url, |
| 29 | + 'headers': request.headers or {} |
| 30 | + } |
| 31 | + |
| 32 | + if request.body is not None: |
| 33 | + if isinstance(request.body, (dict, list)): |
| 34 | + kwargs['json'] = request.body |
| 35 | + else: |
| 36 | + kwargs['data'] = request.body |
| 37 | + |
| 38 | + async with aiohttp.ClientSession() as session: |
| 39 | + try: |
| 40 | + async with session.request(**kwargs) as response: |
| 41 | + content = await response.read() |
| 42 | + |
| 43 | + return Response(content=content, status_code=response.status) |
| 44 | + except Exception as e: |
| 45 | + raise HTTPException(status_code=500, detail=str(e)) |
| 46 | + |
| 47 | +@app.post('/doc-parse/parse') |
| 48 | +async def parse_document( |
| 49 | + file: UploadFile = File(...), |
| 50 | + language: Optional[str] = Form(default='en'), |
| 51 | + target_pages: Optional[str] = Form(default=None) |
| 52 | +): |
| 53 | + parser = LlamaParse( |
| 54 | + result_type='markdown', |
| 55 | + language=language, |
| 56 | + target_pages=target_pages |
| 57 | + ) |
| 58 | + |
| 59 | + file_content = await file.read() |
| 60 | + |
| 61 | + try: |
| 62 | + documents = await parser.aload_data( |
| 63 | + file_content, |
| 64 | + {'file_name': file.filename} |
| 65 | + ) |
| 66 | + |
| 67 | + return { |
| 68 | + 'success': True, |
| 69 | + 'content': [{'text': doc.text, 'meta': doc.metadata} for doc in documents] |
| 70 | + } |
| 71 | + |
| 72 | + except Exception as e: |
| 73 | + return { |
| 74 | + 'success': False, |
| 75 | + 'error': str(e) |
| 76 | + } |
| 77 | + |
| 78 | + finally: |
| 79 | + await file.close() |
0 commit comments