Update src/streamlit_app.py

src/streamlit_app.py

@@ -196,11 +196,17 @@ def clean_float(x) -> float:
     except ValueError:
         return 0.0
 
-def normalize_date(date_str) -> str:
+def normalize_date(date_str, currency=None) -> str:
     """
     Normalize various date formats:
     - Full dates (day-month-year) → dd-MMM-yyyy (e.g., 01-Jan-2025)
     - Month-year only → MMM-yyyy (e.g., Aug-2025)
+    
+    Currency-aware parsing:
+    - If currency is USD and date is numeric format (11/09/2025, 11-09-2025), 
+      treat as MM/DD/YYYY
+    - For text formats (06-Nov-2025, December 6, 2025), parse normally
+    
     Returns empty string if date cannot be parsed
     """
     if not date_str or date_str == "":
@@ -210,17 +216,55 @@ def normalize_date(date_str) -> str:
         date_str = date_str.strip()
         if date_str == "":
             return ""
+        
+        # EXTRA CLEANING: Replace various unicode spaces and clean up
+        # Non-breaking space, thin space, etc. → regular space
+        date_str = re.sub(r'[\u00A0\u2000-\u200B\u202F\u205F\u3000]', ' ', date_str)
+        # Remove zero-width characters
+        date_str = re.sub(r'[\u200B-\u200D\uFEFF]', '', date_str)
+        # Normalize multiple spaces to single space
+        date_str = re.sub(r'\s+', ' ', date_str).strip()
+
+    # Clean ordinal suffixes FIRST (1st, 2nd, 3rd, 4th, 06th, etc.)
+    cleaned_date = date_str
+    if isinstance(date_str, str):
+        # Handle ordinals: "06th December 2025" → "06 December 2025"
+        # Also handles: "December 6th, 2025" → "December 6, 2025"
+        cleaned_date = re.sub(r'(\d+)(st|nd|rd|th)\b', r'\1', date_str, flags=re.IGNORECASE)
 
-    # FULL DATE FORMATS (day-month-year) - try these first
+    # Check if date is NUMERIC format (contains only digits and separators)
+    # Pattern: XX/XX/XXXX, XX-XX-XXXX, XX.XX.XXXX (with 2 or 4 digit year)
+    is_numeric_format = bool(re.match(r'^\d{1,2}[/\-\.]\d{1,2}[/\-\.]\d{2,4}$', cleaned_date))
+    
+    # US FORMAT PRIORITY: If currency is USD and date is numeric, try MM/DD/YYYY first
+    if currency and currency.upper() == 'USD' and is_numeric_format:
+        us_formats = [
+            "%m/%d/%Y",           # 01/15/2025
+            "%m-%d-%Y",           # 01-15-2025
+            "%m.%d.%Y",           # 01.15.2025
+            "%m/%d/%y",           # 01/15/25
+            "%m-%d-%y",           # 01-15-25
+            "%m.%d.%y",           # 01.15.25
+        ]
+        for fmt in us_formats:
+            try:
+                parsed_date = datetime.strptime(cleaned_date, fmt)
+                return parsed_date.strftime("%d-%b-%Y")
+            except (ValueError, TypeError):
+                continue
+
+    # FULL DATE FORMATS (day-month-year) - standard parsing
     full_date_formats = [
-        # ISO formats (4-digit year)
+        # ISO formats (4-digit year) - these are unambiguous
         "%Y-%m-%d",           # 2025-01-15
         "%Y/%m/%d",           # 2025/01/15
         "%Y.%m.%d",           # 2025.01.15
         "%Y %m %d",           # 2025 01 15
         "%Y%m%d",             # 20250115 (compact)
         
-        # European formats with full month names (4-digit year)
+        # European formats with full month names (4-digit year) - UNAMBIGUOUS
+        "%d %B, %Y",          # 15 December, 2025 (with comma)
+        "%d %b, %Y",          # 15 Dec, 2025 (with comma)
         "%d %B %Y",           # 15 January 2025
         "%d %b %Y",           # 15 Jan 2025
         "%d-%B-%Y",           # 15-January-2025
@@ -230,13 +274,15 @@ def normalize_date(date_str) -> str:
         "%d/%B/%Y",           # 15/January/2025
         "%d/%b/%Y",           # 15/Jan/2025
         
-        # US formats with full month names (4-digit year)
+        # US formats with full month names (4-digit year) - UNAMBIGUOUS
         "%B %d, %Y",          # January 15, 2025
         "%b %d, %Y",          # Jan 15, 2025
         "%B %d %Y",           # January 15 2025
         "%b %d %Y",           # Jan 15 2025
         "%B-%d-%Y",           # January-15-2025
         "%b-%d-%Y",           # Jan-15-2025
+        "%B %d,%Y",           # January 15,2025 (no space after comma)
+        "%b %d,%Y",           # Jan 15,2025
         
         # European formats - Day first (4-digit year)
         "%d-%m-%Y",           # 15-01-2025
@@ -244,7 +290,7 @@ def normalize_date(date_str) -> str:
         "%d.%m.%Y",           # 15.01.2025
         "%d %m %Y",           # 15 01 2025
         
-        # US formats - Month first (4-digit year)
+        # US formats - Month first (4-digit year) - only if not USD or not numeric
         "%m-%d-%Y",           # 01-15-2025
         "%m/%d/%Y",           # 01/15/2025
         "%m.%d.%Y",           # 01.15.2025
@@ -273,7 +319,9 @@ def normalize_date(date_str) -> str:
         "%d%m%y",             # 150125
         "%m%d%y",             # 011525
         
-        # European formats with abbreviated month (2-digit year)
+        # European formats with abbreviated month (2-digit year) - UNAMBIGUOUS
+        "%d %B, %y",          # 15 December, 25 (with comma)
+        "%d %b, %y",          # 15 Dec, 25 (with comma)
         "%d-%b-%y",           # 15-Jan-25
         "%d/%b/%y",           # 15/Jan/25
         "%d.%b.%y",           # 15.Jan.25
@@ -281,7 +329,7 @@ def normalize_date(date_str) -> str:
         "%d-%B-%y",           # 15-January-25
         "%d/%B/%y",           # 15/January/25
         
-        # US formats with abbreviated month (2-digit year)
+        # US formats with abbreviated month (2-digit year) - UNAMBIGUOUS
         "%b %d, %y",          # Jan 15, 25
         "%b %d %y",           # Jan 15 25
         "%B %d, %y",          # January 15, 25
@@ -295,25 +343,14 @@ def normalize_date(date_str) -> str:
         "%Y%d%m",             # 20251501
     ]
 
-    # Try full date formats first → output as dd-MMM-yyyy
+    # Try full date formats → output as dd-MMM-yyyy
     for fmt in full_date_formats:
         try:
-            parsed_date = datetime.strptime(str(date_str), fmt)
+            parsed_date = datetime.strptime(cleaned_date, fmt)
             return parsed_date.strftime("%d-%b-%Y")
         except (ValueError, TypeError):
             continue
 
-    # Try with ordinal suffixes removed (1st, 2nd, 3rd, etc.)
-    if isinstance(date_str, str):
-        cleaned = re.sub(r'(\d+)(st|nd|rd|th)\b', r'\1', date_str, flags=re.IGNORECASE)
-        if cleaned != date_str:
-            for fmt in full_date_formats:
-                try:
-                    parsed_date = datetime.strptime(cleaned, fmt)
-                    return parsed_date.strftime("%d-%b-%Y")
-                except (ValueError, TypeError):
-                    continue
-
     # MONTH-YEAR ONLY FORMATS - output as MMM-yyyy
     month_year_formats = [
         # Full month name with year
@@ -354,7 +391,7 @@ def normalize_date(date_str) -> str:
     # Try month-year formats → output as MMM-yyyy (no day)
     for fmt in month_year_formats:
         try:
-            parsed_date = datetime.strptime(str(date_str), fmt)
+            parsed_date = datetime.strptime(cleaned_date, fmt)
             return parsed_date.strftime("%b-%Y")  # Aug-2025 format
         except (ValueError, TypeError):
             continue
@@ -362,10 +399,10 @@ def normalize_date(date_str) -> str:
     # If no format matched, return empty string
     return ""
 
-def parse_date_to_object(date_str):
+def parse_date_to_object(date_str, currency=None):
     """
     Parse a date string to a datetime.date object for date_input widget
-    Handles: ISO, US, EU, Asian, two-digit years, and 50+ worldwide date formats
+    Currency-aware: If USD and numeric format, treat as MM/DD/YYYY
     Returns None if date cannot be parsed
     """
     if not date_str or date_str == "":
@@ -375,147 +412,89 @@ def parse_date_to_object(date_str):
         date_str = date_str.strip()
         if date_str == "":
             return None
+        
+        # EXTRA CLEANING: Replace various unicode spaces and clean up
+        date_str = re.sub(r'[\u00A0\u2000-\u200B\u202F\u205F\u3000]', ' ', date_str)
+        date_str = re.sub(r'[\u200B-\u200D\uFEFF]', '', date_str)
+        date_str = re.sub(r'\s+', ' ', date_str).strip()
+
+    # Clean ordinal suffixes FIRST (1st, 2nd, 3rd, 4th, 06th, etc.)
+    cleaned_date = str(date_str)
+    if isinstance(date_str, str):
+        cleaned_date = re.sub(r'(\d+)(st|nd|rd|th)\b', r'\1', date_str, flags=re.IGNORECASE)
 
-    # Comprehensive list of date formats to try (same as normalize_date)
+    # Check if date is NUMERIC format (contains only digits and separators)
+    is_numeric_format = bool(re.match(r'^\d{1,2}[/\-\.]\d{1,2}[/\-\.]\d{2,4}$', cleaned_date))
+    
+    # US FORMAT PRIORITY: If currency is USD and date is numeric, try MM/DD/YYYY first
+    if currency and currency.upper() == 'USD' and is_numeric_format:
+        us_formats = [
+            "%m/%d/%Y",           # 01/15/2025
+            "%m-%d-%Y",           # 01-15-2025
+            "%m.%d.%Y",           # 01.15.2025
+            "%m/%d/%y",           # 01/15/25
+            "%m-%d-%y",           # 01-15-25
+            "%m.%d.%y",           # 01.15.25
+        ]
+        for fmt in us_formats:
+            try:
+                parsed_date = datetime.strptime(cleaned_date, fmt)
+                return parsed_date.date()
+            except (ValueError, TypeError):
+                continue
+
+    # Standard formats
     formats = [
         # ISO formats (4-digit year)
-        "%Y-%m-%d",           # 2025-01-15
-        "%Y/%m/%d",           # 2025/01/15
-        "%Y.%m.%d",           # 2025.01.15
-        "%Y %m %d",           # 2025 01 15
-        "%Y%m%d",             # 20250115 (compact)
+        "%Y-%m-%d", "%Y/%m/%d", "%Y.%m.%d", "%Y %m %d", "%Y%m%d",
         
-        # European formats with full month names (4-digit year)
-        "%d %B %Y",           # 15 January 2025
-        "%d %b %Y",           # 15 Jan 2025
-        "%d-%B-%Y",           # 15-January-2025
-        "%d-%b-%Y",           # 15-Jan-2025
-        "%d.%B.%Y",           # 15.January.2025
-        "%d.%b.%Y",           # 15.Jan.2025
-        "%d/%B/%Y",           # 15/January/2025
-        "%d/%b/%Y",           # 15/Jan/2025
+        # Text month formats with comma - MUST BE FIRST for "06 December, 2025"
+        "%d %B, %Y", "%d %b, %Y",          # 06 December, 2025 / 06 Dec, 2025
         
-        # US formats with full month names (4-digit year)
-        "%B %d, %Y",          # January 15, 2025
-        "%b %d, %Y",          # Jan 15, 2025
-        "%B %d %Y",           # January 15 2025
-        "%b %d %Y",           # Jan 15 2025
-        "%B-%d-%Y",           # January-15-2025
-        "%b-%d-%Y",           # Jan-15-2025
+        # Text month formats - UNAMBIGUOUS
+        "%d %B %Y", "%d %b %Y", "%d-%B-%Y", "%d-%b-%Y",
+        "%d.%B.%Y", "%d.%b.%Y", "%d/%B/%Y", "%d/%b/%Y",
+        "%B %d, %Y", "%b %d, %Y", "%B %d %Y", "%b %d %Y",
+        "%B-%d-%Y", "%b-%d-%Y", "%B %d,%Y", "%b %d,%Y",
         
-        # European formats - Day first (4-digit year)
-        "%d-%m-%Y",           # 15-01-2025
-        "%d/%m/%Y",           # 15/01/2025
-        "%d.%m.%Y",           # 15.01.2025
-        "%d %m %Y",           # 15 01 2025
+        # European formats - Day first
+        "%d-%m-%Y", "%d/%m/%Y", "%d.%m.%Y", "%d %m %Y",
+        "%d-%m-%y", "%d/%m/%y", "%d.%m.%y", "%d %m %y",
         
-        # US formats - Month first (4-digit year)
-        "%m-%d-%Y",           # 01-15-2025
-        "%m/%d/%Y",           # 01/15/2025
-        "%m.%d.%Y",           # 01.15.2025
-        "%m %d %Y",           # 01 15 2025
-        
-        # European formats with 2-digit year - Day first
-        "%d-%m-%y",           # 15-01-25
-        "%d/%m/%y",           # 15/01/25 or 25/09/25 ← FIXES YOUR ISSUE!
-        "%d.%m.%y",           # 15.01.25
-        "%d %m %y",           # 15 01 25
-        
-        # US formats with 2-digit year - Month first
-        "%m-%d-%y",           # 01-15-25
-        "%m/%d/%y",           # 01/15/25
-        "%m.%d.%y",           # 01.15.25
-        "%m %d %y",           # 01 15 25
+        # US formats - Month first
+        "%m-%d-%Y", "%m/%d/%Y", "%m.%d.%Y", "%m %d %Y",
+        "%m-%d-%y", "%m/%d/%y", "%m.%d.%y", "%m %d %y",
         
         # ISO with 2-digit year
-        "%y-%m-%d",           # 25-01-15
-        "%y/%m/%d",           # 25/01/15
-        "%y.%m.%d",           # 25.01.15
-        "%y %m %d",           # 25 01 15
-        
-        # Compact formats with 2-digit year
-        "%y%m%d",             # 250115
-        "%d%m%y",             # 150125
-        "%m%d%y",             # 011525
-        
-        # European formats with abbreviated month (2-digit year)
-        "%d-%b-%y",           # 15-Jan-25
-        "%d/%b/%y",           # 15/Jan/25
-        "%d.%b.%y",           # 15.Jan.25
-        "%d %b %y",           # 15 Jan 25
-        "%d-%B-%y",           # 15-January-25
-        "%d/%B/%y",           # 15/January/25
-        
-        # US formats with abbreviated month (2-digit year)
-        "%b %d, %y",          # Jan 15, 25
-        "%b %d %y",           # Jan 15 25
-        "%B %d, %y",          # January 15, 25
-        "%B %d %y",           # January 15 25
-        "%b-%d-%y",           # Jan-15-25
-        "%B-%d-%y",           # January-15-25
+        "%y-%m-%d", "%y/%m/%d", "%y.%m.%d", "%y %m %d",
         
-        # Compact 8-digit formats
-        "%d%m%Y",             # 15012025
-        "%m%d%Y",             # 01152025
-        "%Y%d%m",             # 20251501
+        # Compact formats
+        "%y%m%d", "%d%m%y", "%m%d%y", "%d%m%Y", "%m%d%Y", "%Y%d%m",
         
-        # ========== MONTH-YEAR ONLY FORMATS (defaults to 1st of month) ==========
-        # Full month name with year
-        "%B %Y",              # August 2025
-        "%b %Y",              # Aug 2025
-        "%B, %Y",             # August, 2025
-        "%b, %Y",             # Aug, 2025
-        "%B-%Y",              # August-2025
-        "%b-%Y",              # Aug-2025
-        "%B/%Y",              # August/2025
-        "%b/%Y",              # Aug/2025
+        # Text month with 2-digit year (with comma)
+        "%d %B, %y", "%d %b, %y",          # 06 December, 25 / 06 Dec, 25
+        "%d-%b-%y", "%d/%b/%y", "%d.%b.%y", "%d %b %y",
+        "%d-%B-%y", "%d/%B/%y",
+        "%b %d, %y", "%b %d %y", "%B %d, %y", "%B %d %y",
+        "%b-%d-%y", "%B-%d-%y",
         
-        # Numeric month-year (4-digit year)
-        "%m/%Y",              # 08/2025
-        "%m-%Y",              # 08-2025
-        "%m.%Y",              # 08.2025
-        "%m %Y",              # 08 2025
-        "%Y-%m",              # 2025-08
-        "%Y/%m",              # 2025/08
-        "%Y.%m",              # 2025.08
-        "%Y %m",              # 2025 08
-        
-        # Numeric month-year (2-digit year)
-        "%m/%y",              # 08/25
-        "%m-%y",              # 08-25
-        "%m.%y",              # 08.25
-        "%m %y",              # 08 25
-        "%y-%m",              # 25-08
-        "%y/%m",              # 25/08
-        
-        # Full month name with 2-digit year
-        "%B %y",              # August 25
-        "%b %y",              # Aug 25
-        "%B-%y",              # August-25
-        "%b-%y",              # Aug-25
+        # Month-year only
+        "%B %Y", "%b %Y", "%B, %Y", "%b, %Y",
+        "%B-%Y", "%b-%Y", "%B/%Y", "%b/%Y",
+        "%m/%Y", "%m-%Y", "%m.%Y", "%m %Y",
+        "%Y-%m", "%Y/%m", "%Y.%m", "%Y %m",
+        "%m/%y", "%m-%y", "%m.%y", "%m %y",
+        "%y-%m", "%y/%m",
+        "%B %y", "%b %y", "%B-%y", "%b-%y",
     ]
 
-    # Try parsing with each format
     for fmt in formats:
         try:
-            parsed_date = datetime.strptime(str(date_str), fmt)
+            parsed_date = datetime.strptime(cleaned_date, fmt)
             return parsed_date.date()
         except (ValueError, TypeError):
             continue
 
-    # If still not parsed, try removing ordinal suffixes
-    if isinstance(date_str, str):
-        import re
-        cleaned = re.sub(r'(\d+)(st|nd|rd|th)\b', r'\1', date_str, flags=re.IGNORECASE)
-        
-        if cleaned != date_str:
-            for fmt in formats:
-                try:
-                    parsed_date = datetime.strptime(cleaned, fmt)
-                    return parsed_date.date()
-                except (ValueError, TypeError):
-                    continue
-
     return None
 
 
@@ -762,11 +741,14 @@ def parse_vllm_json(raw_json_text):
         header = data.get("header", {})
         summary = data.get("summary", {})
         items = data.get("items", [])
+        
+        # Get currency first for date parsing (USD uses MM/DD/YYYY for numeric dates)
+        currency = summary.get("currency", "")
 
         result = {
             "Invoice Number": header.get("invoice_no", ""),
-            "Invoice Date": normalize_date(header.get("invoice_date", "")),
-            "Due Date": normalize_date(header.get("due_date", "")),
+            "Invoice Date": normalize_date(header.get("invoice_date", ""), currency),
+            "Due Date": normalize_date(header.get("due_date", ""), currency),
             "Sender Name": header.get("sender_name", ""),
             "Sender Address": header.get("sender_addr", ""),
             "Sender": {
@@ -792,7 +774,7 @@ def parse_vllm_json(raw_json_text):
             "Tax Percentage": clean_amount(summary.get("tax_rate", "0")),
             "Total Tax": clean_amount(summary.get("tax_amount", "0")),
             "Total Amount": clean_amount(summary.get("total_amount", "0")),
-            "Currency": summary.get("currency", ""),
+            "Currency": currency,
             "Itemized Data": []
         }
 
@@ -1193,9 +1175,12 @@ def map_prediction_to_ui(pred):
                         return s
         return None
 
+    # Get currency first for date parsing (USD uses MM/DD/YYYY for numeric dates)
+    currency = (pick_first("currency") or "").strip()
+    
     ui["Invoice Number"] = pick_first("invoice_no", "invoice_number", "invoiceid", "invoice id") or ""
-    ui["Invoice Date"] = normalize_date(pick_first("invoice_date", "date", "invoice date") or "")
-    ui["Due Date"] = normalize_date(pick_first("due_date", "due_date", "due") or "")
+    ui["Invoice Date"] = normalize_date(pick_first("invoice_date", "date", "invoice date") or "", currency)
+    ui["Due Date"] = normalize_date(pick_first("due_date", "due_date", "due") or "", currency)
     ui["Sender Name"] = pick_first("sender_name", "sender") or ""
     ui["Sender Address"] = pick_first("sender_addr", "sender_address", "sender addr") or ""
     ui["Recipient Name"] = pick_first("rcpt_name", "recipient_name", "recipient", "rcpt") or ""
@@ -1215,7 +1200,7 @@ def map_prediction_to_ui(pred):
     ui["Tax Percentage"] = clean_number(pick_first("tax_rate", "tax_percentage", "tax pct", "tax percentage") or 0.0)
     ui["Total Tax"] = clean_number(pick_first("tax_amount", "tax", "total_tax") or 0.0)
     ui["Total Amount"] = clean_number(pick_first("total_amount", "grand_total", "total", "amount") or 0.0)
-    ui["Currency"] = (pick_first("currency") or "").strip()
+    ui["Currency"] = currency
 
     items_rows = []
 
@@ -1535,18 +1520,21 @@ elif len(st.session_state.batch_results) > 0:
 
     # --------- Initialize widget state - ONLY IF NOT EXISTS (avoid overwriting user edits) ----------
     bank = form_data.get("Bank Details", {}) if isinstance(form_data.get("Bank Details", {}), dict) else {}
+    
+    # Get currency for date parsing (USD uses MM/DD/YYYY for numeric dates)
+    form_currency = form_data.get('Currency', '')
 
     # Only initialize if key doesn't exist - this preserves user edits between reruns
     if f"Invoice Number_{selected_hash}" not in st.session_state:
         st.session_state[f"Invoice Number_{selected_hash}"] = form_data.get('Invoice Number', '')
 
-    # Parse dates to date objects for date_input widgets
+    # Parse dates to date objects for date_input widgets (pass currency for US date handling)
     if f"Invoice Date_{selected_hash}" not in st.session_state:
-        invoice_date_obj = parse_date_to_object(form_data.get('Invoice Date', ''))
+        invoice_date_obj = parse_date_to_object(form_data.get('Invoice Date', ''), form_currency)
         st.session_state[f"Invoice Date_{selected_hash}"] = invoice_date_obj
     
     if f"Due Date_{selected_hash}" not in st.session_state:
-        due_date_obj = parse_date_to_object(form_data.get('Due Date', ''))
+        due_date_obj = parse_date_to_object(form_data.get('Due Date', ''), form_currency)
         st.session_state[f"Due Date_{selected_hash}"] = due_date_obj
 
     if f"Currency_{selected_hash}" not in st.session_state: