def getImages(self, url, title):
try:
r = self.get(url)
html = r.text
soup = BeautifulSoup(html, 'html.parser')
p_images = soup.find_all('p', style='text-align:center')
if len(p_images) > 0:
# 新增专题
now = int(time.time())
topic_id = Mysql.create("INSERT INTO lab_topic (title, create_time, update_time) VALUES ('%s', %s, %s)" % (title, now, now))
images_values = []
for p_image in p_images:
url = p_image.img['src']
p_name = p_image.img['alt']
p_object = re.search( r'id=(\d*)\..*', p_name, re.I)
p_id = int(p_object.group(1)) if p_object else 0
# 获取文件后缀名
etc = os.path.splitext(url)[1]
date = time.strftime('%Y%m%d',time.localtime(time.time()))
old_name = date + str(round(time.time() * 1000)) + p_name
name = hashlib.md5(old_name.encode(encoding='UTF-8')).hexdigest() + etc
downloadPath = os.path.join(self.downloadPath, date)
self.downloadImage(url, downloadPath, name)
save_path = date + '/' + name
images_values.append("('%s', %s, '%s', %s, %s, %s)" % (save_path, topic_id, p_name, p_id, now, now))
create_sql = 'INSERT INTO lab_image (url, topic_id, name, p_id, create_time, update_time) VALUES ' + (','.join(images_values))
Mysql.execute(create_sql)
print('\033[1;32m--------------------已创建：', title, '\033[0m')
except Exception as e:
# TODO: log
print(e)
return
def createNews(self, url):
try:
r = self.get(url)
html = r.text
soup = BeautifulSoup(html, 'html.parser')
# 获取相关文章的链接
linkHtml = soup.find('a', class_='dec_img')
viewUrl = str(linkHtml['href'])
if self.oldUrl == viewUrl:
return -2
image = str(linkHtml.img['src'])
subject = str(linkHtml['title']).replace("'", "''")
introHtml = linkHtml.parent.parent
intro = introHtml.find('p', class_='com_about').get_text().replace("'", "''")
catalogName = introHtml.find('span', class_='bq_ico').get_text()
if catalogName == '美图':
# 美图类型不获取
return -2
category_id = self.getCatalog(catalogName)
content = self.getNewsView(viewUrl)
now = int(time.time())
user_id = 1
author = '网络'
sql = "INSERT INTO news (user_id, author, catalog_id, title, intro, content, cover, ctime, utime) VALUES (%s, '%s', %s, '%s', '%s', '%s', '%s', %s, %s)" % (user_id, author, category_id, subject, intro, content, image, now, now)
newsID = Mysql.create(sql)
return '{ "id": ' + str(newsID) + ', "url": "' + viewUrl + '"}'
except Exception as e:
# TODO: log
# print(e)
return -1