by_symbol_match : improve heuristic

iso4217parse/__init__.py

@@ -199,14 +199,17 @@ def by_symbol_match(value, country_code=None):
         List[Currency]: Currency objects found in `value`; filter by country_code.
     """
     res = None
-    for s, group in _symbols():
-        if s.lower() in value.lower():
+    for symbol, group in _symbols():
+        symbol_pattern = symbol
+        symbol_pattern = symbol_pattern.replace("$", r"\$")  # $ means endline
+        symbol_pattern = symbol_pattern.replace(".", r"\.")  # . means anything
+        if re.search(rf"(^|\b|\d|\s){symbol_pattern}([^A-Z]|$)", value, re.I):
             if group == 'symbol':
-                res = by_symbol(s, country_code)
+                res = by_symbol(symbol, country_code)
             if group == 'alpha3':
-                res = [by_alpha3(s)]
+                res = [by_alpha3(symbol)]
             if group == 'name':
-                res = [_data()['name'][s]]
+                res = [_data()['name'][symbol]]
             if res and country_code is not None:
                 res = [
                     currency

tests/test_by_symbol.py

@@ -55,4 +55,32 @@ def test_by_symbol_match_filters_country_code(
 def test_parse_by_symbol_value_is_case_insensitive(text, expected_alpha3):
     res = iso4217parse.by_symbol_match(text)
     assert len(res) == 1
-    assert res[0].alpha3 == expected
+    assert res[0].alpha3 == expected_alpha3
+
+
+@pytest.mark.parametrize(
+    'text, ambiguous_alpha3, wanted_alpha3',
+    (
+        # ambiguous words + symbol
+        ('cost $100', 'WST', 'USD'),  # st$ => WST
+        # only letters found in words
+        ('durée 100', 'SZL', None),  # e => SWL / no currency
+        ('maximum 100', 'MRO', None),  # um => MRO / no currency
+        ('flowers', 'NPR', None),  # Re => NPR / no currency
+        ('flowers', 'LKR', None),  # Re => LKR / no currency
+        ('flowers', 'PKR', None),  # Re => PRK / no currency
+        ('amount : 100 currency : Ks', 'NPR', 'MMK'),  # Re => NPR / Ks is MMK
+        ('yes: 100l', 'SOS', 'ALL'),  # s gives SOS / l is ALL or LSL
+        # alpha 3 codes found in words
+        ('course', 'COU', None),  # COU => COU
+        ('finance', 'ANG', None),  # ANG => ANG
+    ),
+)
+def test_parse_by_symbol_value_disambiguation(text, ambiguous_alpha3, wanted_alpha3):
+    assert iso4217parse.by_alpha3(ambiguous_alpha3) not in (iso4217parse.by_symbol_match(text) or [])
+
+    if wanted_alpha3:
+        assert iso4217parse.by_alpha3(wanted_alpha3) in iso4217parse.by_symbol_match(text)
+    else:
+        assert iso4217parse.by_symbol_match(text) is None
+