Spaces:
Sleeping
Sleeping
Commit
·
3a171a3
1
Parent(s):
a187f48
update app.py
Browse files
app.py
CHANGED
|
@@ -2,7 +2,6 @@ import os
|
|
| 2 |
import tempfile
|
| 3 |
from fastapi import FastAPI, UploadFile, File, HTTPException, Header
|
| 4 |
from fastapi.middleware.cors import CORSMiddleware
|
| 5 |
-
from fastapi.responses import FileResponse, JSONResponse
|
| 6 |
from typhoon_ocr import ocr_document
|
| 7 |
from PyPDF2 import PdfReader
|
| 8 |
from docx import Document
|
|
@@ -69,10 +68,10 @@ async def ocr_endpoint(
|
|
| 69 |
x_api_key: str | None = Header(None)
|
| 70 |
):
|
| 71 |
if API_KEY and x_api_key != API_KEY:
|
| 72 |
-
|
| 73 |
|
| 74 |
if page_num < 0:
|
| 75 |
-
|
| 76 |
|
| 77 |
suffix = os.path.splitext(file.filename)[-1].lower()
|
| 78 |
content = await file.read()
|
|
@@ -83,33 +82,15 @@ async def ocr_endpoint(
|
|
| 83 |
try:
|
| 84 |
md, md_path, docx_path = process_file_with_typhoon(tmp_path, task_type, page_num)
|
| 85 |
except Exception as e:
|
| 86 |
-
|
| 87 |
|
| 88 |
-
# ส่งข้อมูล Markdown + ลิงก์ดาวน์โหลดไฟล์
|
| 89 |
return {
|
| 90 |
"markdown": md,
|
| 91 |
-
"
|
| 92 |
-
"
|
| 93 |
"task_type": task_type,
|
| 94 |
}
|
| 95 |
|
| 96 |
-
@app.get("/download/md/{filename}")
|
| 97 |
-
async def download_md(filename: str):
|
| 98 |
-
tmp_dir = tempfile.gettempdir()
|
| 99 |
-
file_path = os.path.join(tmp_dir, filename)
|
| 100 |
-
if not os.path.exists(file_path):
|
| 101 |
-
raise HTTPException(status_code=404, detail="ไฟล์ .md ไม่เจอ")
|
| 102 |
-
return FileResponse(file_path, media_type="text/markdown", filename="ocr_extraction_peace.md")
|
| 103 |
-
|
| 104 |
-
@app.get("/download/docx/{filename}")
|
| 105 |
-
async def download_docx(filename: str):
|
| 106 |
-
tmp_dir = tempfile.gettempdir()
|
| 107 |
-
file_path = os.path.join(tmp_dir, filename)
|
| 108 |
-
if not os.path.exists(file_path):
|
| 109 |
-
raise HTTPException(status_code=404, detail="ไฟล์ .docx ไม่เจอ")
|
| 110 |
-
return FileResponse(file_path, media_type="application/vnd.openxmlformats-officedocument.wordprocessingml.document", filename="ocr_extraction_peace.docx")
|
| 111 |
-
|
| 112 |
-
|
| 113 |
def gradio_handler(file, task_type, page_num):
|
| 114 |
if file is None:
|
| 115 |
return "❌ อัปโหลดไฟล์ก่อน....", None, None
|
|
|
|
| 2 |
import tempfile
|
| 3 |
from fastapi import FastAPI, UploadFile, File, HTTPException, Header
|
| 4 |
from fastapi.middleware.cors import CORSMiddleware
|
|
|
|
| 5 |
from typhoon_ocr import ocr_document
|
| 6 |
from PyPDF2 import PdfReader
|
| 7 |
from docx import Document
|
|
|
|
| 68 |
x_api_key: str | None = Header(None)
|
| 69 |
):
|
| 70 |
if API_KEY and x_api_key != API_KEY:
|
| 71 |
+
raise HTTPException(status_code=401, detail="API key ผิดพ่อง")
|
| 72 |
|
| 73 |
if page_num < 0:
|
| 74 |
+
raise HTTPException(status_code=400, detail="เลขหน้าต้องเป็น 0 หรือบวกเท่านั้น")
|
| 75 |
|
| 76 |
suffix = os.path.splitext(file.filename)[-1].lower()
|
| 77 |
content = await file.read()
|
|
|
|
| 82 |
try:
|
| 83 |
md, md_path, docx_path = process_file_with_typhoon(tmp_path, task_type, page_num)
|
| 84 |
except Exception as e:
|
| 85 |
+
raise HTTPException(status_code=400, detail=str(e))
|
| 86 |
|
|
|
|
| 87 |
return {
|
| 88 |
"markdown": md,
|
| 89 |
+
"md_file": md_path,
|
| 90 |
+
"docx_file": docx_path,
|
| 91 |
"task_type": task_type,
|
| 92 |
}
|
| 93 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 94 |
def gradio_handler(file, task_type, page_num):
|
| 95 |
if file is None:
|
| 96 |
return "❌ อัปโหลดไฟล์ก่อน....", None, None
|