def parse_content_html(raw):
pattern = re.compile(r"(?s)<div class=TRS_Editor>(.*?)<\/div>")
m = pattern.search(raw)
content = m.groups()[0]
province = ProvinceData(provinceName, provinceKey)
pattern_confirm = re.compile(r"累计.*?确诊病例.*?>(\d+)<.*?例")
cm = pattern_confirm.search(content)
if cm is not None:
province.Confirmed = int(cm.groups()[0])
pattern_heal = re.compile(r"治愈出院.*?>(\d+)<.*?例")
hm = pattern_heal.search(content)
if hm is not None:
province.Healed = int(hm.groups()[0])
city = {}
pattern_data = re.compile(r"[，、]([\u4E00-\u9FA5]+).*?>(\d+)<.*?例")
for i in pattern_data.finditer(content[content.rfind("确诊病例中"):]):
name = utils.remove_preposition(i.groups()[0])
if name in cities.keys():
id = cities[name]
if id not in city.keys():
d = CityData(name, id)
d.Confirmed = int(i.groups()[1])
city[id] = d
return province, city
def parse_content_html(raw):
pattern = re.compile(r"(?s)<!--content begin -->(.*)<!--content end -->")
m = pattern.search(raw)
content = m.groups()[0]
province = ProvinceData(provinceName, provinceKey)
pattern_confirm = re.compile(r"累计确诊.*?(\d+)例")
cm = pattern_confirm.search(content)
if cm is not None:
province.Confirmed = int(cm.groups()[0])
pattern_heal = re.compile(r"出院(\d+)例")
hm = pattern_heal.search(content)
if hm is not None:
province.Healed = int(hm.groups()[0])
pattern_dead = re.compile(r"死亡(\d+)例")
dm = pattern_dead.search(content)
if dm is not None:
province.Dead = int(dm.groups()[0])
city = {}
pattern_data = re.compile(r"[。，、]([\u4E00-\u9FA5]+)(\d+)例")
for i in pattern_data.finditer(content[content.rfind("累计确诊"):]):
name = utils.remove_preposition(i.groups()[0])
if name in cities.keys():
id = cities[name]
if id not in city.keys():
d = CityData(name, id)
d.Confirmed = int(i.groups()[1])
city[id] = d
return province, city
def parse_content_html(raw):
pattern = re.compile(r"(?s)<div class=\"ze-art\" style=\"width: 100%;\">(.*)<\/div>")
m = pattern.search(raw)
content = m.groups()[0]
province = ProvinceData(provinceName, provinceKey)
pattern_confirm = re.compile(r"累计.*?确诊.*?(\d+)例")
cm = pattern_confirm.search(content)
if cm is not None:
province.Confirmed = int(cm.groups()[0])
pattern_heal = re.compile(r"累计治愈出院(\d+)例")
hm = pattern_heal.search(content)
if hm is not None:
province.Healed = int(hm.groups()[0])
city = {}
pattern_data = re.compile(r"([\u4E00-\u9FA5]+)(\d+)例")
for i in pattern_data.finditer(content[content.rfind("累计报告"):content.rfind("累计治愈")]):
name = utils.remove_preposition(i.groups()[0])
if name in cities.keys():
id = cities[name]
if id not in city.keys():
d = CityData(name, id)
d.Confirmed = int(i.groups()[1])
city[id] = d
return province, city
def parse_content_html(raw):
pattern = re.compile(
r"(?s)<!------------------------- mian开始 ------------------------->(.*)<!--------责任编辑相关---------->"
)
m = pattern.search(raw)
content = m.groups()[0]
province = ProvinceData(provinceName, provinceKey)
pattern_confirm = re.compile(r"累计.*?确诊.*?(\d+)例")
cm = pattern_confirm.search(content)
if cm is not None:
province.Confirmed = int(cm.groups()[0])
pattern_heal = re.compile(r"，出院病例(\d+)例")
for hm in pattern_heal.finditer(content):
province.Healed = int(hm.groups()[0])
city = {}
pattern_data = re.compile(r"[\.\u200b]([\u4E00-\u9FA5]+)(\d+)例")
for i in pattern_data.finditer(
content[content.rfind("累计报告"):content.rfind("重症病例")]):
name = utils.remove_preposition(i.groups()[0])
if '盟' in name[:-1]:
name = name[:name.find('盟') + 1]
if '市' in name[:-1]:
name = name[:name.find('市') + 1]
if name in cities.keys():
id = cities[name]
if id not in city.keys():
d = CityData(name, id)
d.Confirmed = int(i.groups()[1])
city[id] = d
else:
city[id].Confirmed += int(i.groups()[1])
return province, city
def parse_content_html(raw):
pattern = re.compile(r"<p>(.*?累计.*?)<\/p>")
m = pattern.search(raw)
content = m.groups()[0]
province = ProvinceData(provinceName, provinceKey)
pattern_confirm = re.compile(r"确诊病例(\d+)例")
cm = pattern_confirm.search(content)
if cm is not None:
province.Confirmed = int(cm.groups()[0])
pattern_heal = re.compile(r"治愈出院(\d+)例")
hm = pattern_heal.search(content)
if hm is not None:
province.Healed = int(hm.groups()[0])
pattern_dead = re.compile(r"死亡(\d+)例（([\u4E00-\u9FA5]+)）")
dm = pattern_dead.search(content)
dc = {}
if dm is not None:
province.Dead = int(dm.groups()[0])
dc[dm.groups()[1]] = int(dm.groups()[0]) # SPECIAL HANDLE
city = {}
pattern_data = re.compile(r"[（，、。]([\u4E00-\u9FA5]+)(\d+)例")
for i in pattern_data.finditer(content):
name = utils.remove_preposition(i.groups()[0])
if name in alia_cities.keys():
name = alia_cities[name]
if name in cities.keys():
id = cities[name]
if id not in city.keys():
d = CityData(name, id)
d.Confirmed = int(i.groups()[1])
city[id] = d
else:
city[id].Confirmed += int(i.groups()[1])
# SPECIAL HANDLE
for dd in dc.items():
if dd[0] in cities.keys():
id = cities[dd[0]]
if id not in city.keys():
d = CityData(dd[0], id)
d.Confirmed = dd[1]
city[id] = d
else:
city[id].Confirmed += dd[1]
return province, city
def test_remove_preposition(self):
s = '其中广州市'
self.assertEqual(utils.remove_preposition(s), "广州市")