def topFirst(self):
# 获取评论数量前十的景区
ten_site = site_db.select_sql(
['site_name', 'num_comment'],
sqlFilter=f"order by num_comment desc limit 10")
ten_site = sorted(ten_site,
key=lambda x: int(x['num_comment']),
reverse=True)
# 获取评分最高的10个酒店
ten_hotel = hotel_db.select_sql(
['hotel_name', 'comment_num'],
sqlFilter=f"order by comment_num desc limit 10")
ten_hotel = sorted(ten_hotel,
key=lambda x: int(x['comment_num']),
reverse=True)
# 获取游记数量最多的10个地区
ten_site_info = site_db.select_sql(
['site_name', 'num_ginfo'],
sqlFilter=f"order by num_ginfo desc limit 10")
ten_site_info = sorted(ten_site_info,
key=lambda x: int(x['num_ginfo']),
reverse=True)
print(ten_site_info)
return ten_site, ten_hotel, ten_site_info
def picData(self):
ten_site = site_db.select_sql(
['site_name', 'num_comment', 'num_ginfo'],
sqlFilter=f"order by num_comment desc limit 14")
ten_site = sorted(ten_site,
key=lambda x: int(x['num_comment']),
reverse=True)
res = []
for item in ten_site:
temp = {
'name': item['site_name'],
'value1': item['num_comment'],
'value2': item['num_ginfo'],
}
res.append(temp)
return res
def second(self):
# 获取详情页面内容
for url in allSite:
targetUrl = self.detail_info_url + url
print('second', targetUrl)
detailRoot = self.request(targetUrl)
imgs = self.secondImg(detailRoot)
img = json.dumps(imgs, ensure_ascii=False)
name = self.secondTitle(detailRoot)
brief = self.secondBrief(detailRoot)
site = self.secondSite(detailRoot)
site_id = produce_id()
# 获取评论信息
comment_num = self.commentNum(detailRoot)
comment_category = self.commentCategory(detailRoot)
other = {
'comment_num': comment_num,
'comment_category': comment_category
}
other = json.dumps(other, ensure_ascii=False)
# fixme 数据存储
is_exist = site_db.select_sql(
target=['id'], sqlFilter=f"where site_name = '{name}'")[0]
print('is_exist =============== ', is_exist)
# if not is_exist and brief:
# data = [site_id, name, site, brief, img, other]
# target = ['id', 'site_name', 'site', 'brief', 'imgs', 'other']
# site_db.insert_sql(target, data)
#
# for index in range(1, 16):
# comment_id = self.commentContent(detailRoot, index)
# data = [site_id, comment_id]
# target = ['site_id', 'comment_id']
# relationship_db.insert_sql(target, data)
if is_exist:
for index in range(1, 16):
comment_id = self.commentContent(detailRoot, index)
site_id = is_exist.get('id')
data = [site_id, comment_id]
target = ['site_id', 'comment_id']
relationship_db.insert_sql(target, data)
driver.close()
def getContent(id):
site = site_db.select_sql(["*"], sqlFilter=f'where id = "{id}"')
return site
def getManySite():
allSite = site_db.select_sql(["*"], sqlFilter=f'where img != "" limit 16')
return allSite
def getSite():
allSite = site_db.select_sql(
["*"], sqlFilter=f'order by num_comment desc limit 12')
return allSite