def get_case_name(self):
raw = self.__find_by_class_name('h1', 'detail-title-content')
name = clean.sub('', raw.text) if raw is not None else ""
name = punctuation_cleaner.sub('', name)
return name
def get_host_role(self):
raw = self.__find_by_class_name('span', 'info-span-msg')
host_role = raw.text if raw is not None else ""
host_role = punctuation_cleaner.sub('', host_role)
return host_role
def get_price(self):
key = u'租金'
price = self.infos[key] if self.__is_key(key) else "0"
price = punctuation_cleaner.sub('', price)
price = re.search('\d+', price).group()
return float(price), u'元/月'
def get_case_name(self):
case_no = self.get_case_number()
raw = find_by_css(self.soup, 'div.h1table h1')
case_name = raw[0].text.replace(case_no, '') if raw else ''
case_name = punctuation_cleaner.sub('', case_name)
return case_name
def get_price(self):
raw = find_by_css(self.soup, 'div.price')
price = re.search('\d+', punctuation_cleaner.sub('', raw[0].text)) \
.group() if raw else 0.0
raw = find_by_css(self.soup, 'div.price b')
unit = raw[0].text if raw else u"元/月"
return float(price), unit
def get_host_company(self):
raw = find_by_css(self.soup, 'div.auatarSonBox')
clean_text = punctuation_cleaner.sub('', raw[0].text) if raw else ''
m = re.search(u'公司名：(\W+)分店：(\W+)', clean_text)
company = m.group(1) if m else ""
branch = m.group(2) if m else ""
return ' '.join([company, branch])
def __get_house_infos(self):
infos = {}
rows = find_by_css(self.soup, 'div.object-list li')
for row in rows:
clean_text = punctuation_cleaner.sub('', row.text)
if ':' in clean_text:
attr, value = clean_text.split(':')
infos[attr] = value
return infos
def __get_house_infos(self):
infos = {}
attr_raw = find_by_css(self.soup, '.labelList .one')
value_raw = find_by_css(self.soup, '.labelList .two')
for attr, value in zip(attr_raw, value_raw):
infos[attr.text] = value.text.replace(u'：', '')
rows = find_by_css(self.soup, 'ul.attr li')
for row in rows:
clean_text = punctuation_cleaner.sub('', row.text)
attr, value = clean_text.split(':')
infos[attr] = value
return infos
def get_host_company(self):
host_store = ""
raw = self.__find_by_class_name('div', 'info-detail-show')
text = clean.sub('', raw.text) if raw is not None else ""
text = punctuation_cleaner.sub('', text)
m = re.search(u'公司名：(\W+)分公司：(\W+)', text)
try:
headquarter = m.group(1)
branch = m.group(2)
host_store = " ".join([headquarter, branch])
except Exception as e:
print(e)
return host_store
def get_case_name(self):
raw = find_by_css(self.soup, 'div.top h1')
case_name = raw[0].text if raw else ''
case_name = punctuation_cleaner.sub('', case_name)
return case_name
def get_case_number(self):
raw = find_by_css(self.soup, 'div.h1table span.color-gray')
case_number = punctuation_cleaner.sub('', raw[0].text) if raw else ''
return case_number