def find(self, document):
strainer = SoupStrainer(attrs={'class': self.GEO_CLASS})
if not isinstance(document, BeautifulSoup):
elements = BeautifulSoup(document, parseOnlyThese=strainer)
else:
elements = document.findAll(strainer)
for element in elements:
preformatted = element.name == 'pre'
lat_element = element.find(attrs={'class': self.LATITUDE_CLASS})
long_element = element.find(attrs={'class': self.LONGITUDE_CLASS})
latitude = None
longitude = None
if lat_element and long_element:
latitude = self._get_value(lat_element, preformatted)
longitude = self._get_value(long_element, preformatted)
elif self.shorthand:
lat_long = re.split(self.SEP, self._get_value(element), 1)
if len(lat_long) == 2:
latitude, longitude = lat_long
if latitude and longitude:
lat_match = FLOAT_RE.match(unescape(latitude))
long_match = FLOAT_RE.match(unescape(longitude))
if lat_match and long_match:
latitude = float(lat_match.group(1))
longitude = float(long_match.group(1))
text = unescape(self._get_text(element).strip())
name = re.sub('\s+', ' ', text)
yield Location(name, (latitude, longitude))
def find(self, document):
strainer = SoupStrainer(attrs={'class': self.GEO_CLASS})
if not isinstance(document, BeautifulSoup):
elements = BeautifulSoup(document, parseOnlyThese=strainer)
else:
elements = document.findAll(strainer)
for element in elements:
preformatted = element.name == 'pre'
lat_element = element.find(attrs={'class': self.LATITUDE_CLASS})
long_element = element.find(attrs={'class': self.LONGITUDE_CLASS})
latitude = None
longitude = None
if lat_element and long_element:
latitude = self._get_value(lat_element, preformatted)
longitude = self._get_value(long_element, preformatted)
elif self.shorthand:
lat_long = re.split(self.SEP, self._get_value(element), 1)
if len(lat_long) == 2:
latitude, longitude = lat_long
if latitude and longitude:
lat_match = FLOAT_RE.match(unescape(latitude))
long_match = FLOAT_RE.match(unescape(longitude))
if lat_match and long_match:
latitude = float(lat_match.group(1))
longitude = float(long_match.group(1))
text = unescape(self._get_text(element).strip())
name = re.sub('\s+', ' ', text)
yield Location(name, (latitude, longitude))
def find(self, document):
strainer = SoupStrainer('meta', attrs={'name': self.META_NAME})
if not isinstance(document, BeautifulSoup):
elements = BeautifulSoup(document, parseOnlyThese=strainer)
else:
elements = document.findAll(strainer)
for element in elements:
lat_long = element.get('content')
if lat_long or not self.ignore_invalid:
try:
point = Point(unescape(lat_long))
except (TypeError, ValueError):
if not self.ignore_invalid:
raise
else:
yield Location(None, point)
def find(self, document):
strainer = SoupStrainer('meta', attrs={'name': self.META_NAME})
if not isinstance(document, BeautifulSoup):
elements = BeautifulSoup(document, parseOnlyThese=strainer)
else:
elements = document.findAll(strainer)
for element in elements:
lat_long = element.get('content')
if lat_long or not self.ignore_invalid:
try:
point = Point(unescape(lat_long))
except (TypeError, ValueError):
if not self.ignore_invalid:
raise
else:
yield Location(None, point)
def find(self, document):
strainer = SoupStrainer('meta', attrs={'name': self.META_NAME})
if not isinstance(document, BeautifulSoup):
elements = BeautifulSoup(document, parseOnlyThese=strainer)
else:
elements = document.findAll(strainer)
attrs = {}
for element in elements:
meta_name = element['name']
attr_name = re.match(self.META_NAME, meta_name).group(1)
value = element.get('content')
if attr_name in attrs:
location = self._get_location(attrs)
if location is not None:
yield location
attrs.clear()
attrs[attr_name] = value and unescape(value)
location = self._get_location(attrs)
if location is not None:
yield location
def find(self, document):
strainer = SoupStrainer('meta', attrs={'name': self.META_NAME})
if not isinstance(document, BeautifulSoup):
elements = BeautifulSoup(document, parseOnlyThese=strainer)
else:
elements = document.findAll(strainer)
attrs = {}
for element in elements:
meta_name = element['name']
attr_name = re.match(self.META_NAME, meta_name).group(1)
value = element.get('content')
if attr_name in attrs:
location = self._get_location(attrs)
if location is not None:
yield location
attrs.clear()
attrs[attr_name] = value and unescape(value)
location = self._get_location(attrs)
if location is not None:
yield location
