Improved parsing performance

fin_defs/parse_price.py

@@ -2,7 +2,7 @@ import logging
 import re
 from decimal import Decimal
 
-from .data import DKK, USD, Asset, AssetAmount, FiatCurrency
+from .data import DKK, USD, Asset, AssetAmount, FiatCurrency, CURRENCY_CODES, CURRENCY_SYMBOLS
 
 logger = logging.getLogger(__name__)
 
@@ -10,6 +10,54 @@ RE_PRICE_RAW = r'\b(?:dkk|sek|usd|nok|eur)?\s*([1-9][\d.]*[\d](?:,\d+)?)\s*(?:,-
 
 RE_PRICE = re.compile(RE_PRICE_RAW, flags=re.IGNORECASE)
 
+RE_PRODUCT_PRICE_DK = r'-?\d{1,3}(?:\.?\d{3})*(?:,\d\d\.?)?'
+RE_PRODUCT_PRICE_EN = r'-?\d{1,3}(?:,?\d{3})*(?:\.\d\d)?'
+RE_PRODUCT_PRICE_AMOUNT = r'(' + RE_PRODUCT_PRICE_DK + '|' + RE_PRODUCT_PRICE_EN + ')'
+
+
+def parse_amount(price: str) -> Decimal:
+    if re.fullmatch(RE_PRODUCT_PRICE_DK, price):
+        price = price.replace('.', '').replace(',', '.')
+    else:
+        price = price.replace(',', '')
+    return Decimal(price)
+
+
+RE_CURRENCY_CODES = '(?P<code>' + '|'.join(re.escape(c) for c in CURRENCY_CODES) + ')'
+
+RE_CURRENCY_SYMBOLS = (
+    r'(?P<sym>[' + ''.join(re.escape(c) for c in CURRENCY_SYMBOLS.values()) + '])'
+)
+
+RE_SYM_AMOUNT_CODE = re.compile(RE_CURRENCY_SYMBOLS
+        + r'\s*'
+        + RE_PRODUCT_PRICE_AMOUNT
+        + r'(?:\s+'
+        + RE_CURRENCY_CODES
+        + r')?', flags=re.IGNORECASE)
+
+RE_AMOUNT_SYM_CODE = re.compile(
+        RE_PRODUCT_PRICE_AMOUNT
+        + r'\s*'
+        + RE_CURRENCY_SYMBOLS
+        + r'(?:\s+'
+        + RE_CURRENCY_CODES
+        + ')?'
+        , flags=re.IGNORECASE)
+
+RE_AMOUNT_CODE = re.compile(
+        RE_PRODUCT_PRICE_AMOUNT + r'\s+' + RE_CURRENCY_CODES
+        , flags=re.IGNORECASE)
+
+RE_KR_AMOUNT = re.compile(
+        r'kr\.?\s*(' + RE_PRODUCT_PRICE_AMOUNT + ')',
+        flags=re.IGNORECASE)
+
+RE_AMOUNT_KR= re.compile(
+        '(' + RE_PRODUCT_PRICE_AMOUNT + r')\s*kr\.?',
+        flags=re.IGNORECASE)
+
+
 def parse_price(text: str, default_currency: Asset) -> AssetAmount | None:
     """
     Attempts to parse price from the given text.
@@ -18,49 +66,24 @@ def parse_price(text: str, default_currency: Asset) -> AssetAmount | None:
     """
     if isinstance(text, AssetAmount):
         return text
-    text = str(text)
+    text = str(text).lower().strip()
 
-    if m := re.match(r'^Kr\s*([\d.]+(?:,\d+))?$', text):
-        return AssetAmount(
-            DKK,
-            Decimal(m.group(1).replace('.', '').replace(',', '.')),
-        )
-    if m := re.match(r'^(\d+)\s*DKK$', text):
-        return AssetAmount(DKK, Decimal(m.group(1)))
+    if text == 'free':
+        return AssetAmount(default_currency, Decimal(0))
 
-    if m := re.match(r'^\$\s*([0-9.]+)(\s+USD)?$', text):
-        return AssetAmount(USD, Decimal(m.group(1)))
+    code, sym, amount_text = None, None, None
 
-    if text.lower().strip() == 'free':
-        return AssetAmount(default_currency, Decimal(0.0))
-
-    text = str(text).strip().lower().removesuffix('.')
-    if m := RE_PRICE.fullmatch(text):
-        currency = default_currency
-        price_tag = m.group(1).replace('.', '').replace(',', '.')  # TODO
-        if text.endswith('dkk') or text.startswith('dkk'):
-            currency = FiatCurrency('DKK')
-        elif text.endswith('sek') or text.startswith('sek'):
-            currency = FiatCurrency('SEK')
-        elif text.endswith('nok') or text.startswith('nok'):
-            currency = FiatCurrency('NOK')
-        elif text.endswith('usd') or text.startswith('usd'):
-            currency = FiatCurrency('USD')
-
-        return AssetAmount(currency, Decimal(price_tag))
-
-    logger.warning('Unknown price format: %s', text)
-    return None
-
-
-def parse_usd_price(s: str | int) -> AssetAmount:
-    assert s is not None
-    if isinstance(s, str) or isinstance(s, int):
-        text = str(s)
+    if m := RE_SYM_AMOUNT_CODE.fullmatch(text):
+        code, sym, amount_text = m.group('code'), m.group('sym'), m.group(2)
+    elif m := RE_AMOUNT_SYM_CODE.fullmatch(text):
+        code, sym, amount_text = m.group('code'), m.group('sym'), m.group(1)
+    elif m := RE_AMOUNT_CODE.fullmatch(text):
+        code, amount_text = m.group('code'), m.group(1)
+    elif m := (RE_KR_AMOUNT.fullmatch(text) or RE_AMOUNT_KR.fullmatch(text)):
+        code, amount_text = 'DKK', m.group(1)
     else:
-        text = s.text_content()
-    text = text.strip().replace(',', '').removeprefix('$')
-    if text in {'-', ''}:
-        return AssetAmount(USD, Decimal(0))  # TODO
-    dollar_amount = Decimal(text)
-    return AssetAmount(USD, dollar_amount)
+        return None
+
+    currency = CURRENCY_CODES[code.upper()] if code else FiatCurrency.from_currency_symbol(sym)
+    assert currency is not None
+    return AssetAmount(currency, parse_amount(amount_text))

test/test_parse_price.py

@@ -28,9 +28,36 @@ PRICES_UNPARSABLE = [
 
 @pytest.mark.parametrize(('price_string', 'parsed_amount'), PRICES_PARSABLE)
 def test_parse_price(price_string: str, parsed_amount: AssetAmount):
-    result = parse_price(price_string, parsed_amount.asset)
+    result = parse_price(price_string, FiatCurrency.JPY)
     assert result == parsed_amount
 
 @pytest.mark.parametrize('price_string', PRICES_UNPARSABLE)
 def test_parse_unparsable(price_string: str):
     assert parse_price(price_string, USD) is None
+
+
+def parse_asset_amount(text: str) -> AssetAmount:
+    return parse_price(text, FiatCurrency.JPY)
+
+def test_parse_asset_amount_dkk():
+    assert parse_asset_amount('1338 DKK').amount == 1338
+    assert parse_asset_amount('1338,00 DKK').amount == 1338
+    assert parse_asset_amount('13,38 DKK').amount == Decimal('13.38')
+    assert parse_asset_amount('13.38 DKK').amount == Decimal('13.38')
+    assert parse_asset_amount('1338,00. DKK').amount == 1338
+    assert parse_asset_amount('99,00 kr.').amount == 99
+
+    assert parse_asset_amount('kr 825.00').amount == 825
+    assert parse_asset_amount('kr 825.00').asset == DKK
+    assert parse_asset_amount('kr 825,00').amount == 825
+    assert parse_asset_amount('kr 825,00').asset == DKK
+    assert parse_asset_amount('kr. 825.00').amount == 825
+    assert parse_asset_amount('kr. 825.00').asset == DKK
+
+def test_parse_asset_amount_usd():
+    assert parse_asset_amount('$99').amount == 99
+    assert parse_asset_amount('$99').asset == USD
+
+    assert parse_asset_amount('99$ USD').amount == 99
+    assert parse_asset_amount('99$ USD').asset == USD
+