Upload folder using huggingface_hub

app.py

@@ -28,17 +28,34 @@ app = FastAPI(title="Bajaj Datathon - Bill Extractor")
 
 class BillRequest(BaseModel):
     document: str
-# ---------------- FASTAPI APP ----------------
-
-app = FastAPI()
-
+# ---------------- Helpers: number normalization & detection ----------------
+NUM_RE = re.compile(r"[-+]?\d{1,3}(?:[,0-9]*)(?:\.\d+)?")  # matches numbers with commas, decimals
+TOTAL_KEYWORDS = re.compile(r"(grand\s*total|net\s*payable|total\s*amount|amount\s*payable|bill\s*amount|final\s*amount|balance\s*due|sub\s*total|subtotal)", re.I)
+
+def normalize_num_str(s: str) -> Optional[float]:
+    if s is None:
+        return None
+    s = s.strip()
+    # remove currency symbols, whitespace
+    s = re.sub(r"[^\d\-\+\,\.()]", "", s)
+    # remove parentheses meaning negative
+    negative = False
+    if s.startswith("(") and s.endswith(")"):
+        negative = True
+        s = s[1:-1]
+    s = s.replace(",", "")
+    if s == "":
+        return None
+    try:
+        val = float(s)
+        return -val if negative else val
+    except Exception:
+        return None
 
-class BillRequest(BaseModel):
-    """
-    Request body model.
-    Expects a public URL to a bill document (image/PDF).
-    """
-    document: str
+def looks_like_amount_token(token: str) -> bool:
+    if not token or token.strip() == "":
+        return False
+    return NUM_RE.search(token) is not None
 
 
 # ---------------- FALLBACK REGEX EXTRACTOR ----------------