Spaces:
Sleeping
Sleeping
Upload folder using huggingface_hub
Browse files
app.py
CHANGED
|
@@ -599,49 +599,49 @@ async def extract_bill_data(payload: BillRequest):
|
|
| 599 |
doc_url = payload.document
|
| 600 |
file_bytes = None
|
| 601 |
|
| 602 |
-
# ---------------------------
|
| 603 |
-
# 1. LOCAL FILE SUPPORT
|
| 604 |
-
# ---------------------------
|
| 605 |
-
if doc_url.startswith("file://"):
|
| 606 |
-
|
| 607 |
-
|
| 608 |
-
|
| 609 |
-
|
| 610 |
-
|
| 611 |
-
|
| 612 |
-
|
| 613 |
-
|
| 614 |
-
|
| 615 |
-
|
| 616 |
-
|
| 617 |
-
|
| 618 |
-
# ---------------------------
|
| 619 |
-
# 2. REMOTE URL DOWNLOAD
|
| 620 |
-
# ---------------------------
|
| 621 |
-
else:
|
| 622 |
-
|
| 623 |
-
|
| 624 |
-
|
| 625 |
-
|
| 626 |
-
|
| 627 |
-
|
| 628 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 629 |
return {
|
| 630 |
"is_success": False,
|
| 631 |
-
"error":
|
| 632 |
"data": {"pagewise_line_items": [], "total_item_count": 0},
|
| 633 |
"token_usage": {"total_tokens": 0, "input_tokens": 0, "output_tokens": 0}
|
| 634 |
}
|
| 635 |
|
| 636 |
-
# Safety check
|
| 637 |
-
if not file_bytes:
|
| 638 |
-
return {
|
| 639 |
-
"is_success": False,
|
| 640 |
-
"error": "No file bytes found.",
|
| 641 |
-
"data": {"pagewise_line_items": [], "total_item_count": 0},
|
| 642 |
-
"token_usage": {"total_tokens": 0, "input_tokens": 0, "output_tokens": 0}
|
| 643 |
-
}
|
| 644 |
-
|
| 645 |
images = []
|
| 646 |
clean_url = doc_url.split("?", 1)[0].lower()
|
| 647 |
try:
|
|
|
|
| 599 |
doc_url = payload.document
|
| 600 |
file_bytes = None
|
| 601 |
|
| 602 |
+
# ---------------------------
|
| 603 |
+
# 1. LOCAL FILE SUPPORT
|
| 604 |
+
# ---------------------------
|
| 605 |
+
if doc_url.startswith("file://"):
|
| 606 |
+
local_path = doc_url.replace("file://", "")
|
| 607 |
+
try:
|
| 608 |
+
with open(local_path, "rb") as f:
|
| 609 |
+
file_bytes = f.read()
|
| 610 |
+
except Exception as e:
|
| 611 |
+
return {
|
| 612 |
+
"is_success": False,
|
| 613 |
+
"error": f"Local file read error: {e}",
|
| 614 |
+
"data": {"pagewise_line_items": [], "total_item_count": 0},
|
| 615 |
+
"token_usage": {"total_tokens": 0, "input_tokens": 0, "output_tokens": 0}
|
| 616 |
+
}
|
| 617 |
+
|
| 618 |
+
# ---------------------------
|
| 619 |
+
# 2. REMOTE URL DOWNLOAD
|
| 620 |
+
# ---------------------------
|
| 621 |
+
else:
|
| 622 |
+
try:
|
| 623 |
+
headers = {"User-Agent": "Mozilla/5.0"}
|
| 624 |
+
resp = requests.get(doc_url, headers=headers, timeout=30)
|
| 625 |
+
if resp.status_code != 200:
|
| 626 |
+
raise RuntimeError(f"Download failed status={resp.status_code}")
|
| 627 |
+
file_bytes = resp.content
|
| 628 |
+
except Exception as e:
|
| 629 |
+
return {
|
| 630 |
+
"is_success": False,
|
| 631 |
+
"error": f"HTTP error: {e}",
|
| 632 |
+
"data": {"pagewise_line_items": [], "total_item_count": 0},
|
| 633 |
+
"token_usage": {"total_tokens": 0, "input_tokens": 0, "output_tokens": 0}
|
| 634 |
+
}
|
| 635 |
+
|
| 636 |
+
# Final safety check
|
| 637 |
+
if not file_bytes:
|
| 638 |
return {
|
| 639 |
"is_success": False,
|
| 640 |
+
"error": "No file bytes found.",
|
| 641 |
"data": {"pagewise_line_items": [], "total_item_count": 0},
|
| 642 |
"token_usage": {"total_tokens": 0, "input_tokens": 0, "output_tokens": 0}
|
| 643 |
}
|
| 644 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 645 |
images = []
|
| 646 |
clean_url = doc_url.split("?", 1)[0].lower()
|
| 647 |
try:
|