Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -6,19 +6,25 @@ import os
|
|
| 6 |
from uuid import uuid4
|
| 7 |
from docling.document_converter import DocumentConverter
|
| 8 |
from threading import Lock
|
|
|
|
|
|
|
| 9 |
|
| 10 |
app = FastAPI()
|
| 11 |
|
| 12 |
-
#
|
| 13 |
app.add_middleware(
|
| 14 |
CORSMiddleware,
|
| 15 |
-
allow_origins=["*"],
|
| 16 |
allow_credentials=True,
|
| 17 |
-
allow_methods=["*"],
|
| 18 |
-
allow_headers=["*"],
|
| 19 |
)
|
| 20 |
|
| 21 |
-
#
|
|
|
|
|
|
|
|
|
|
|
|
|
| 22 |
class PdfConverterSingleton:
|
| 23 |
_instance = None
|
| 24 |
_lock = Lock()
|
|
@@ -37,14 +43,18 @@ class PdfConverterSingleton:
|
|
| 37 |
|
| 38 |
def get_text(self, pdf_path: str) -> str:
|
| 39 |
result = self.converter.convert(pdf_path)
|
| 40 |
-
|
| 41 |
return result.document.export_to_markdown()
|
| 42 |
|
| 43 |
-
#
|
| 44 |
-
def
|
| 45 |
return PdfConverterSingleton().get_text(pdf_path)
|
| 46 |
|
| 47 |
-
#
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 48 |
@app.post("/extract-pdf-text")
|
| 49 |
async def extract_pdf_text(file: UploadFile = File(...)):
|
| 50 |
if file.content_type != "application/pdf":
|
|
@@ -55,7 +65,7 @@ async def extract_pdf_text(file: UploadFile = File(...)):
|
|
| 55 |
with open(temp_filename, "wb") as buffer:
|
| 56 |
shutil.copyfileobj(file.file, buffer)
|
| 57 |
|
| 58 |
-
text =
|
| 59 |
return JSONResponse(content={"markdown_text": text})
|
| 60 |
|
| 61 |
except Exception as e:
|
|
|
|
| 6 |
from uuid import uuid4
|
| 7 |
from docling.document_converter import DocumentConverter
|
| 8 |
from threading import Lock
|
| 9 |
+
from concurrent.futures import ThreadPoolExecutor
|
| 10 |
+
import asyncio
|
| 11 |
|
| 12 |
app = FastAPI()
|
| 13 |
|
| 14 |
+
# CORS for all
|
| 15 |
app.add_middleware(
|
| 16 |
CORSMiddleware,
|
| 17 |
+
allow_origins=["*"],
|
| 18 |
allow_credentials=True,
|
| 19 |
+
allow_methods=["*"],
|
| 20 |
+
allow_headers=["*"],
|
| 21 |
)
|
| 22 |
|
| 23 |
+
# Thread pool sized according to vCPU
|
| 24 |
+
MAX_WORKERS = os.cpu_count() or 2 # fallback to 2 if os.cpu_count() returns None
|
| 25 |
+
thread_pool = ThreadPoolExecutor(max_workers=MAX_WORKERS)
|
| 26 |
+
|
| 27 |
+
# Singleton class for DocumentConverter
|
| 28 |
class PdfConverterSingleton:
|
| 29 |
_instance = None
|
| 30 |
_lock = Lock()
|
|
|
|
| 43 |
|
| 44 |
def get_text(self, pdf_path: str) -> str:
|
| 45 |
result = self.converter.convert(pdf_path)
|
|
|
|
| 46 |
return result.document.export_to_markdown()
|
| 47 |
|
| 48 |
+
# Run sync function in threadpool
|
| 49 |
+
def sync_extract_text(pdf_path: str) -> str:
|
| 50 |
return PdfConverterSingleton().get_text(pdf_path)
|
| 51 |
|
| 52 |
+
# Async wrapper for thread pool
|
| 53 |
+
async def async_extract_text(pdf_path: str) -> str:
|
| 54 |
+
loop = asyncio.get_event_loop()
|
| 55 |
+
return await loop.run_in_executor(thread_pool, sync_extract_text, pdf_path)
|
| 56 |
+
|
| 57 |
+
# Main endpoint
|
| 58 |
@app.post("/extract-pdf-text")
|
| 59 |
async def extract_pdf_text(file: UploadFile = File(...)):
|
| 60 |
if file.content_type != "application/pdf":
|
|
|
|
| 65 |
with open(temp_filename, "wb") as buffer:
|
| 66 |
shutil.copyfileobj(file.file, buffer)
|
| 67 |
|
| 68 |
+
text = await async_extract_text(temp_filename)
|
| 69 |
return JSONResponse(content={"markdown_text": text})
|
| 70 |
|
| 71 |
except Exception as e:
|