nojhan/weboob-devel
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

More robust currency guessing

Browse source 
This allows to parse PayPal balances which includes both symbol and letters, for
instance "$ 42,00 USD".
Also, it does not require adding any new symbols to the regular
expression.
It might break modules, though numerous cases were tested.
This commit is contained in:
Laurent Bachelier 2013-02-04 15:16:03 +01:00
commit 81e3da8bb0
2 changed files with 27 additions and 4 deletions
Download patch file Download diff file Expand all files Collapse all files

3
setup.cfg
View file

@ -1,5 +1,6 @@
[nosetests]
verbosity = 2
detailed-errors = 1
with-doctest = 1
where = weboob
tests = weboob.tools.capabilities.paste,weboob.tools.path
tests = weboob.tools.capabilities.paste,weboob.tools.path,weboob.capabilities.bank

28
weboob/capabilities/bank.py
View file

@ -41,7 +41,7 @@ class TransferError(UserError):
A transfer has failed.
"""
class Currency:
class Currency(object):
CUR_UNKNOWN = 0
CUR_EUR = 1
CUR_CHF = 2
@ -54,10 +54,32 @@ class Currency:
u'USD': CUR_USD,
}
EXTRACTOR = re.compile(r'[\d\s,\.\-]', re.UNICODE)
@classmethod
def get_currency(klass, text):
text = re.sub(u'[^A-Z€]', '', text.upper())
return klass.TXT2CUR.get(text, klass.CUR_UNKNOWN)
u"""
>>> Currency.get_currency(u'42')
0
>>> Currency.get_currency(u'42 €')
1
>>> Currency.get_currency(u'$42')
3
>>> Currency.get_currency(u'42.000,00€')
1
>>> Currency.get_currency(u'$42 USD')
3
>>> Currency.get_currency(u'%42 USD')
3
>>> Currency.get_currency(u'US1D')
0
"""
curtexts = klass.EXTRACTOR.sub(' ', text.upper()).split()
for curtext in curtexts:
cur = klass.TXT2CUR.get(curtext)
if cur is not None:
return cur
return klass.CUR_UNKNOWN
class Recipient(CapBaseObject):