fix OCR

ocr_parser.py

@@ -161,15 +161,19 @@ def parse_revolut_format(text: str) -> Dict[str, float]:
     """
     Parse Revolut-specific format.
 
-    Revolut format (typically 2-3 lines per stock):
-    Line 1: @ Company Name 3420,14$
-    Line 2: 8,31 MU - 411,50'$ 4123,26%
-    (Sometimes line 3 if company name is long)
+    Revolut format (typically 2 lines per stock):
+    Line 1: [icon] Company Name [portfolio_value]$
+    Line 2: [shares] TICKER[separator] [price_per_share]$ [change%]
+
+    Examples:
+    Line 1: "@ Micron Technology 3 212,85 $"
+    Line 2: "8,31 MU» 386,56 $ 4 109,73%"
 
     Handles variations:
     - Spaces in numbers: "3 256,40"
-    - Different separators: "-", ":", "*"
+    - Different separators after ticker: "-", ":", "*", "»", "«"
     - Numbers without decimals: "172312"
+    - Negative values in change column
 
     Args:
         text: Extracted text from OCR
@@ -185,9 +189,26 @@ def parse_revolut_format(text: str) -> Dict[str, float]:
     while i < len(lines):
         current_line = lines[i].strip()
 
+        # Skip empty lines
+        if not current_line:
+            i += 1
+            continue
+
+        # Check if this is a TICKER line (not a value line)
+        # Ticker lines start with: [shares] [TICKER][separator]
+        # Example: "8,31 MU» 386,56 $" or "52,03 AMKR: 51$" or "0,94LLY -1080"
+        is_ticker_line = re.match(r'^[\d,]+[.,]?\d*\s*[A-Z]{2,5}[\s\-–:*«»]', current_line)
+
+        if is_ticker_line:
+            # This is a ticker line, skip it (it's already been processed as lookahead)
+            i += 1
+            continue
+
         # Look for portfolio value line (contains amount with $, €, £)
-        # Handle spaces in numbers: "3 256,40" or "172312" or "3420,14"
-        value_match = re.search(r'([\d\s,]+(?:[.,]\d{1,2})?)\s*[\$€£]', current_line)
+        # IMPORTANT: Match dollar amounts that are NOT preceded by a negative sign
+        # Avoid matching negative change values like "-1080,46$"
+        # Allow optional colon/apostrophe before currency: "3 120,52: $" or "240,92'$"
+        value_match = re.search(r'(?<![\-–])([\d\s,]+(?:[.,]\d{1,2})?)[:\']?\s*[\$€£]', current_line)
 
         if value_match:
             portfolio_value_str = value_match.group(1)
@@ -208,7 +229,9 @@ def parse_revolut_format(text: str) -> Dict[str, float]:
 
             try:
                 amount = float(clean_value)
-                if amount < 1:  # Skip invalid amounts
+                # Filter out very small amounts (likely percentages, share counts, or other data)
+                # Portfolio positions are typically > 50 (even small positions)
+                if amount < 50:
                     i += 1
                     continue
             except ValueError:
@@ -223,19 +246,22 @@ def parse_revolut_format(text: str) -> Dict[str, float]:
 
                 check_line = lines[i + lookahead].strip()
 
-                # Match ticker patterns (more flexible):
-                # "8,31 MU - 411,50" or "52,03 AMKR: 51$" or "GOOGL* 335,15"
-                # Ticker can be followed by: -, :, *, space, or «
-                ticker_match = re.search(r'[\d,]+[.,]?\d*\s+([A-Z]{2,5})[\s\-–:*«]', check_line)
+                # Match ticker patterns: [shares] [TICKER][separator]
+                # Examples: "8,31 MU -" or "52,03 AMKR:" or "5,06 GOOGL*" or "5,06 TSM «"
+                # Also handles OCR errors with missing space: "0,94LLY"
+                # Ticker can be followed by: -, :, *, », «, space, or end of significant text
+                ticker_match = re.search(r'[\d,]+[.,]?\d*\s*([A-Z]{2,5})[\s\-–:*«»]', check_line)
 
                 if ticker_match:
                     ticker = ticker_match.group(1)
 
                     # Validate ticker (not a word fragment or common false positive)
-                    if len(ticker) >= 2 and ticker not in ['AM', 'PM', 'USD', 'EUR', 'GBP', 'JPY']:
-                        portfolio[ticker] = amount
+                    if len(ticker) >= 2 and ticker not in ['AM', 'PM', 'USD', 'EUR', 'GBP', 'JPY', 'CHF']:
+                        # Only add if not already present (avoid duplicates)
+                        if ticker not in portfolio:
+                            portfolio[ticker] = amount
                         ticker_found = True
-                        i += lookahead + 1  # Skip processed lines
+                        i += lookahead + 1  # Skip to line after ticker line
                         break
 
             if not ticker_found: