return classes
class PersonClassifier(BaseClassifier):
def is_person(self, symbol):
# TODO : test the precision of this method of determining is_person
infoboxes = get_infoboxes(symbol)
for ibx in infoboxes:
if ibx.wikipedia_class() == 'wikipedia-person' or \
ibx.get('birth-date'):
return True
from wikipediabase.resolvers import PersonResolver
if PersonResolver().birth_date(symbol, 'birth-date'):
return True
return False
def classify(self, symbol, fetcher=None):
if fetcher:
self.fetcher = fetcher
classes = []
if self.is_person(symbol):
classes += ['wikibase-person']
return classes
WIKIBASE_CLASSIFIERS = subclasses(BaseClassifier)
return self.val is None
def val_str(self):
return u'nil'
class LispNumber(_LispLiteral):
def should_parse(self):
return isinstance(self.val, Number)
def val_str(self):
return str(self.val)
WIKIBASE_LISP_TYPES = subclasses(LispType, instantiate=False)
def lispify(obj, typecode=None, infobox_attr=None):
"""
Return a Lisp-like encoded string.
infobox_attr is the name of the infobox attribute where obj came from
"""
if isinstance(obj, LispType):
return obj
for T in WIKIBASE_LISP_TYPES:
t = T(obj, typecode, infobox_attr=infobox_attr)
if t.valid:
def title(self, symbol, fetcher=None):
fetcher = fetcher or WIKIBASE_FETCHER
html = fetcher.html_source(symbol)
match = re.search(TITLE_REGEX, html)
if match:
return match.group(1)
def induce(self, symbol, fetcher=None):
synonyms = []
fetcher = fetcher or WIKIBASE_FETCHER
title = self.title(symbol, fetcher=fetcher)
if title:
sym = string_reduce(symbol)
if title != sym:
synonyms.extend(lexical_synonyms(title))
return synonyms
class LexicalInducer(BaseInducer):
"""
Induce just the lexicals of the symbol
"""
def induce(self, symbol):
return lexical_synonyms(symbol)
WIKIBASE_INDUCERS = subclasses(BaseInducer)
