Price currency and amount fixes

changedetectionio/processors/restock_diff/__init__.py

@@ -6,8 +6,76 @@ import re
 
 class Restock(dict):
 
-    def parse_currency(self, raw_value: str) -> Union[float, None]:
-        # Clean and standardize the value (ie 1,400.00 should be 1400.00), even better would be store the whole thing as an integer.
+    def _normalize_currency_code(self, currency: str) -> str:
+        """
+        Normalize currency symbol or code to ISO 4217 code for consistency.
+        Uses iso4217parse for accurate conversion.
+        """
+        if not currency:
+            return currency
+
+        # If already a 3-letter code, likely already normalized
+        if len(currency) == 3 and currency.isupper():
+            return currency
+
+        try:
+            import iso4217parse
+
+            # Parse the currency - returns list of possible matches
+            currencies = iso4217parse.parse(currency)
+
+            if currencies:
+                # For ambiguous symbols, prefer common currencies
+                if currency == '$':
+                    # Prefer USD for $ symbol
+                    usd = [c for c in currencies if c.alpha3 == 'USD']
+                    if usd:
+                        return 'USD'
+                elif currency == '£':
+                    # Prefer GBP for £ symbol
+                    gbp = [c for c in currencies if c.alpha3 == 'GBP']
+                    if gbp:
+                        return 'GBP'
+                elif currency == '¥':
+                    # Prefer JPY for ¥ symbol
+                    jpy = [c for c in currencies if c.alpha3 == 'JPY']
+                    if jpy:
+                        return 'JPY'
+
+                # Return first match for unambiguous symbols
+                return currencies[0].alpha3
+        except Exception:
+            pass
+
+        # Fallback: return as-is if can't normalize
+        return currency
+
+    def parse_currency(self, raw_value: str) -> Union[dict, None]:
+        """
+        Parse price and currency from text, handling messy formats with extra text.
+        Returns dict with 'price' and 'currency' keys (ISO 4217 code), or None if parsing fails.
+        """
+        try:
+            from price_parser import Price
+            # price-parser handles:
+            # - Extra text before/after ("Beginning at", "tax incl.")
+            # - Various number formats (1 099,00 or 1,099.00)
+            # - Currency symbols and codes
+            price_obj = Price.fromstring(raw_value)
+
+            if price_obj.amount is not None:
+                result = {'price': float(price_obj.amount)}
+                if price_obj.currency:
+                    # Normalize currency symbol to ISO 4217 code for consistency with metadata
+                    normalized_currency = self._normalize_currency_code(price_obj.currency)
+                    result['currency'] = normalized_currency
+                return result
+
+        except Exception as e:
+            from loguru import logger
+            logger.trace(f"price-parser failed on '{raw_value}': {e}, falling back to manual parsing")
+
+        # Fallback to existing manual parsing logic
         standardized_value = raw_value
 
         if ',' in standardized_value and '.' in standardized_value:
@@ -24,7 +92,7 @@ class Restock(dict):
 
         if standardized_value:
             # Convert to float
-            return float(parse_decimal(standardized_value, locale='en'))
+            return {'price': float(parse_decimal(standardized_value, locale='en'))}
 
         return None
 
@@ -51,7 +119,15 @@ class Restock(dict):
         # Custom logic to handle setting price and original_price
         if key == 'price' or key == 'original_price':
             if isinstance(value, str):
-                value = self.parse_currency(raw_value=value)
+                parsed = self.parse_currency(raw_value=value)
+                if parsed:
+                    # Set the price value
+                    value = parsed.get('price')
+                    # Also set currency if found and not already set
+                    if parsed.get('currency') and not self.get('currency'):
+                        super().__setitem__('currency', parsed.get('currency'))
+                else:
+                    value = None
 
         super().__setitem__(key, value)
 

changedetectionio/processors/restock_diff/processor.py

@@ -180,18 +180,24 @@ def get_price_data_availability(html_content, price_change_custom_include_filter
 
     if filtered_content.strip():
         # Convert HTML to text
-        price_text = html_tools.html_to_text(
-            html_content=filtered_content,
-            render_anchor_tag_content=False,
-            is_rss=False
-        ).strip()
+        import re
+        price_text = re.sub(
+            r'[\r\n\t]+', ' ',
+            html_tools.html_to_text(
+                html_content=filtered_content,
+                render_anchor_tag_content=False,
+                is_rss=False
+            ).strip()
+        )
 
         # Parse the price from text
         try:
-            parsed_price = value.parse_currency(price_text)
-            if parsed_price is not None:
-                value['price'] = parsed_price
-                logger.debug(f"Extracted price from custom selector: {parsed_price} (from text: '{price_text}')")
+            parsed_result = value.parse_currency(price_text)
+            if parsed_result:
+                value['price'] = parsed_result.get('price')
+                if parsed_result.get('currency'):
+                    value['currency'] = parsed_result.get('currency')
+                logger.debug(f"Extracted price from custom selector: {parsed_result.get('price')} {parsed_result.get('currency', '')} (from text: '{price_text}')")
         except Exception as e:
             logger.warning(f"Failed to parse price from '{price_text}': {e}")
 

requirements.txt

@@ -105,6 +105,8 @@ extruct
 
 # For cleaning up unknown currency formats
 babel
+# For normalizing currency symbols to ISO 4217 codes
+iso4217parse
 
 levenshtein