def get_money_annotations(text: str, float_digits=4) \
-> Generator[MoneyAnnotation, None, None]:
for match in CURRENCY_PTN_RE.finditer(text):
capture = match.capturesdict()
if not (capture['prefix']
or capture['postfix']) and not (capture['trigger_word']):
continue
prefix = capture['prefix']
postfix = capture['postfix']
amount = list(
get_amounts(capture['amount'][0], float_digits=float_digits))
if len(amount) != 1:
continue
if prefix:
prefix = prefix[0].lower()
currency_type = CURRENCY_SYMBOL_MAP.get(prefix)\
or CURRENCY_PREFIX_MAP.get(prefix)\
or prefix.upper()
elif postfix:
postfix = postfix[0].lower()
currency_type = CURRENCY_TOKEN_MAP.get(postfix) or (
capture['postfix'][0]).upper()
else:
currency_type = None
if not currency_type:
currency_type = DEFAULT_CURRENCY
text = capture['text'][0].strip(
string.punctuation.replace('$', '') + string.whitespace)
ant = MoneyAnnotation(coords=match.span(),
amount=amount[0],
text=text,
currency=currency_type)
yield ant
def get_money(text, return_sources=False, float_digits=4) -> Generator:
for match in CURRENCY_PTN_RE.finditer(text):
capture = match.capturesdict()
if not (capture['prefix'] or capture['postfix']):
continue
prefix = capture['prefix']
postfix = capture['postfix']
amount = list(
get_amounts(capture['amount'][0], float_digits=float_digits))
if len(amount) != 1:
continue
if prefix:
prefix = prefix[0].lower()
currency_type = CURRENCY_SYMBOL_MAP.get(prefix)\
or CURRENCY_PREFIX_MAP.get(prefix)\
or prefix.upper()
else:
postfix = postfix[0].lower()
currency_type = CURRENCY_TOKEN_MAP.get(postfix) or (
capture['postfix'][0]).upper()
item = (amount[0], currency_type)
if return_sources:
item += (capture['text'][0].strip(
string.punctuation.replace('$', '') + string.whitespace), )
yield item
('chinese yuan', 'CNY'),
('dollars', 'USD'),
('dollar', 'USD'),
('euros', 'EUR'),
('euro', 'EUR'),
('pounds', 'GBP'),
('pound', 'GBP'),
('renminbi', 'CNY'),
('yens', 'JPY'),
('yen', 'JPY'),
('yuans', 'CNY'),
('yuan', 'CNY')
])
CURRENCY_ABBR_LIST = set(
list(CURRENCY_SYMBOL_MAP.values()) +
list(CURRENCY_TOKEN_MAP.values()) +
list(CURRENCY_PREFIX_MAP.values())
)
CURRENCY_PREFIXES = set(
list(CURRENCY_PREFIX_MAP.keys()) +
list(CURRENCY_SYMBOL_MAP.values())
)
CURR_NUM_PTN = NUM_PTN.replace('(?<=\\W|^)', '')
CURRENCY_PTN = r"""
(?P<text>
(?P<prefix>{currency_prefixes}|[{currency_symbols}])\s*
(?P<amount>{num_ptn_1})
__version__ = "1.4.0"
__maintainer__ = "LexPredict, LLC"
__email__ = "*****@*****.**"
DEFAULT_CURRENCY = 'USD'
CURRENCY_TOKEN_MAP = OrderedDict([('chinese yuans', 'CNY'),
('chinese yuan', 'CNY'), ('dollars', 'USD'),
('dollar', 'USD'), ('euros', 'EUR'),
('euro', 'EUR'), ('pounds', 'GBP'),
('pound', 'GBP'), ('renminbi', 'CNY'),
('yens', 'JPY'), ('yen', 'JPY'),
('yuans', 'CNY'), ('yuan', 'CNY')])
CURRENCY_ABBR_LIST = set(
list(CURRENCY_SYMBOL_MAP.values()) + list(CURRENCY_TOKEN_MAP.values()) +
list(CURRENCY_PREFIX_MAP.values()))
CURRENCY_PREFIXES = set(
list(CURRENCY_PREFIX_MAP.keys()) + list(CURRENCY_SYMBOL_MAP.values()))
CURR_NUM_PTN = NUM_PTN.replace('(?<=\\W|^)', '')
TRIGGER_WORDS = ['price', 'cost']
CURRENCY_PTN = r"""
(?P<text>
(?P<prefix>{currency_prefixes}|[{currency_symbols}])\s*
(?P<amount>{num_ptn_1})
|
(?P<amount>{num_ptn_2})\s*