Prioritize numbers next to currencies

Author: Gallaecio

price_parser/parser.py

@@ -1,7 +1,7 @@
 # -*- coding: utf-8 -*-
 import re
 import string
-from typing import Callable, Optional, Pattern, List, Tuple
+from typing import Callable, Match, Optional, Pattern, List, Tuple
 from decimal import Decimal, InvalidOperation
 
 import attr
@@ -36,11 +36,17 @@ def fromstring(cls, price: Optional[str],
         ``price`` string, it could be **preferred** over a value extracted
         from ``currency_hint`` string.
         """
-        amount_text = extract_price_text(price) if price is not None else None
+        currency_match, source = _extract_currency_symbol(price, currency_hint)
+        if price is not None:
+            _currency_match = currency_match if source == price else None
+            amount_text = extract_price_text(price, _currency_match)
+        else:
+            amount_text = None
         amount_num = parse_number(amount_text) if amount_text is not None else None
-        currency = extract_currency_symbol(price, currency_hint)
-        if currency is not None:
-            currency = currency.strip()
+        if currency_match is not None:
+            currency = currency_match.group(0).strip()
+        else:
+            currency = None
         return Price(
             amount=amount_num,
             currency=currency,
@@ -120,11 +126,11 @@ def or_regex(symbols: List[str]) -> Pattern:
 _search_unsafe_currency = or_regex(OTHER_CURRENCY_SYMBOLS).search
 
 
-def extract_currency_symbol(price: Optional[str],
-                            currency_hint: Optional[str]) -> Optional[str]:
+def _extract_currency_symbol(price: Optional[str], currency_hint: Optional[str]) -> Tuple[Optional[Match], Optional[str]]:
     """
-    Guess currency symbol from extracted price and currency strings.
-    Return an empty string if symbol is not found.
+    Guess the currency symbol from extracted price and currency strings.
+    Return a (`match object`_, source_string) tuple with the symbol found and
+    the string where it was found, or (None, None) if no symbol is found.
     """
     methods: List[Tuple[Callable, Optional[str]]] = [
         (_search_safe_currency, price),
@@ -142,17 +148,32 @@ def extract_currency_symbol(price: Optional[str],
     for meth, attr in methods:
         m = meth(attr) if attr else None
         if m:
-            return m.group(0)
+            return m, attr
+
+    return None, None
 
+
+def extract_currency_symbol(price: Optional[str],
+                            currency_hint: Optional[str]) -> Optional[str]:
+    """
+    Guess currency symbol from extracted price and currency strings.
+    Return the symbol as found as a string, or None if no symbol is found.
+    """
+    match, _ = _extract_currency_symbol(price, currency_hint)
+    if match:
+        return match.group(0)
     return None
 
 
-def extract_price_text(price: str) -> Optional[str]:
+def extract_price_text(price: str, currency_match: Optional[Match] = None) -> Optional[str]:
     """
     Extract text of a price from a string which contains price and
-    maybe some other text. If multiple price-looking substrings are present,
-    the first is returned (FIXME: it is better to return a number
-    which is near a currency symbol).
+    maybe some other text.
+
+    If a match object of the currency within the `price` string is provided,
+    amounts before or after the matched currency substring are prioritized.
+    Otherwise, if multiple price-looking substrings are present, the first is
+    returned.
 
     >>> extract_price_text("price: $12.99")
     '12.99'
@@ -189,16 +210,39 @@ def extract_price_text(price: str) -> Optional[str]:
         """, price, re.VERBOSE)
         if m:
             return m.group(0).replace(' ', '')
+
+    def number_from_match(m):
+        return m.group(1).strip(',.').strip()
+
+    if currency_match is not None:
+
+        m = re.search(r"""
+            (\d[\d\s.,]*)  # number, probably with thousand separators
+            \s*$           # only match right before the currency symbol
+            """, price[:currency_match.start(0)], re.VERBOSE)
+        if m:
+            return number_from_match(m)
+
+        m = re.search(r"""
+        ^\s*           # only match right after the currency symbol
+        (\d[\d\s.,]*)  # number, probably with thousand separators
+        \s*            # skip whitespace
+        (?:[^%\d]|$)   # capture next symbol - it shouldn't be %
+        """, price[currency_match.end(0):], re.VERBOSE)
+        if m:
+            return number_from_match(m)
+
     m = re.search(r"""
         (\d[\d\s.,]*)  # number, probably with thousand separators
         \s*            # skip whitespace
         (?:[^%\d]|$)   # capture next symbol - it shouldn't be %
         """, price, re.VERBOSE)
-
     if m:
-        return m.group(1).strip(',.').strip()
+        return number_from_match(m)
+
     if 'free' in price.lower():
         return '0'
+
     return None
 
 

tests/test_price_parsing.py

@@ -17,6 +17,7 @@
 import pytest
 
 from price_parser import Price
+from price_parser.parser import extract_currency_symbol
 
 
 class Example(Price):
@@ -625,8 +626,6 @@ def __eq__(self, other):
             'Р', '30', 30),
     Example('€', '€ 139.00',
             '€', '139.00', 139),
-    Example('There are 163 products.', 'From 26 to 50 €',
-            '€', '26', 26),
     Example('Pris NOK 1 999,00', '139,00',
             'NOK', '139,00', 139),
     Example('/sqft', '1.52',
@@ -1909,15 +1908,55 @@ def __eq__(self, other):
             'CHF', '19.90', 19.90),
     Example('', '530,42 Zł',
             'Zł', '530,42', 530.42),
+
+    # Prefer values next to currency symbols
+    Example('3 Ausgaben für nur 14,85 EUR', '3 Ausgaben für nur 14,85 EUR',
+            'EUR', '14,85', 14.85),
+    Example(None, '2 items at 24,00€',
+            '€', '24,00', 24.00),
+    Example(None, '2 items at 24,00 €',
+            '€', '24,00', 24.00),
+    Example(None, '2 items at €24,00',
+            '€', '24,00', 24.00),
+    Example(None, '2 items at € 24,00',
+            '€', '24,00', 24.00),
+    Example(None, '2 items at 24,00€ or 30,00€',
+            '€', '24,00', 24.00),
+    Example(None, '2 items at 24,00€ or 30,00 €',
+            '€', '24,00', 24.00),
+    Example(None, '2 items at 24,00€ or €30,00',
+            '€', '24,00', 24.00),
+    Example(None, '2 items at 24,00€ or € 30,00',
+            '€', '24,00', 24.00),
+    Example(None, '2 items at 24,00 € or 30,00€',
+            '€', '24,00', 24.00),
+    Example(None, '2 items at 24,00 € or 30,00 €',
+            '€', '24,00', 24.00),
+    Example(None, '2 items at 24,00 € or €30,00',
+            '€', '24,00', 24.00),
+    Example(None, '2 items at 24,00 € or € 30,00',
+            '€', '24,00', 24.00),
+    Example(None, '2 items at €24,00 or 30,00€',
+            '€', '24,00', 24.00),
+    Example(None, '2 items at €24,00 or 30,00 €',
+            '€', '24,00', 24.00),
+    Example(None, '2 items at €24,00 or €30,00',
+            '€', '24,00', 24.00),
+    Example(None, '2 items at €24,00 or € 30,00',
+            '€', '24,00', 24.00),
+    Example(None, '2 items at € 24,00 or 30,00€',
+            '€', '24,00', 24.00),
+    Example(None, '2 items at € 24,00 or 30,00 €',
+            '€', '24,00', 24.00),
+    Example(None, '2 items at € 24,00 or €30,00',
+            '€', '24,00', 24.00),
+    Example(None, '2 items at € 24,00 or € 30,00',
+            '€', '24,00', 24.00),
 ]
 
 
 PRICE_PARSING_EXAMPLES_XFAIL = [
     # amount is picked as a price
-    Example('3 Ausgaben für nur 14,85 EUR', '3 Ausgaben für nur 14,85 EUR',
-            'EUR', '14,85', 14.85),
-    Example(None, 'Buy Now - 2 Litre Was $120.00 Now $60.00',
-            '$', '60.00', 60),
     Example('Цена: уточняйте (мин. заказ: 1 )', 'Цена: уточняйте (мин. заказ: 1 )',
             None, None, None),
     Example(None, '50 - $2.00 100 - $2.75 400 - $4.50 1,000 - $9.00 2,000 - $17.00 3,000 - $24.00 10,000 - $75.00',
@@ -1931,6 +1970,14 @@ def __eq__(self, other):
     Example('Cuneo', '61.858 L',  # Romanian New Leu
             'L', '61.858', 61858),
 
+    # no handling of price ranges
+    Example('There are 163 products.', 'From 26 to 50 €',
+            '€', '26', 26),
+
+    # no handling of old-vs-new prices
+    Example(None, 'Buy Now - 2 Litre Was $120.00 Now $60.00',
+            '$', '60.00', 60),
+
     # "р" / "руб" is detected as currency
     Example('>', 'См. цену в прайсе',
             None, None, None),
@@ -1977,6 +2024,30 @@ def test_parsing(example: Example):
     assert parsed == example
 
 
+@pytest.mark.parametrize(
+    "input_string,symbol",
+    (
+        # no currency
+        ('', None),
+        ('1', None),
+
+        # fictional currency
+        ('10 eddies', None),
+
+        # cyrrency code
+        ('5 CNY', 'CNY'),
+
+        # cyrrency name
+        ('5 euros', 'euro'),
+
+        # currency symbol
+        ('$4', '$'),
+    )
+)
+def test_extract_currency_symbol(input_string, symbol):
+    assert extract_currency_symbol(input_string, None) == symbol
+
+
 @pytest.mark.parametrize(
     "amount,amount_float",
     (