news_get_hello_github.py 5.1 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130
  1. # -*- coding: utf-8 -*-
  2. '''
  3. Hello Github
  4. '''
  5. import os
  6. import sys
  7. sys.path.append(os.path.join(os.getcwd().split('auto')[0], 'auto'))
  8. import threading
  9. import time
  10. from datetime import datetime
  11. import httpx
  12. from tools_mongo_handle import MongoHandle
  13. from tools_logs_handle import LogsHandle
  14. from tools_send_email import SendEmail
  15. class HelloGithub(object):
  16. def __init__(self):
  17. self.logs_handle = LogsHandle()
  18. self.now_day = time.strftime('%Y-%m-%d', time.localtime())
  19. self.base_url = 'https://api.hellogithub.com/v1/?sort_by=last&tid=&page={}'
  20. self.headers = {
  21. 'User-Agent': 'Mozilla/4.0 (compatible; MSIE 8.0; Windows NT 6.1; WOW64; Trident/4.0; SLCC2; Media Center PC 6.0; InfoPath.2; MS-RTC LM 8'
  22. }
  23. self.db = 'NEWS'
  24. self.collection = 'HelloGithub_info'
  25. self.source_url = 'https://hellogithub.com/repository/'
  26. self.send_email_datas = []
  27. def main(self):
  28. self.logs_handle.logs_write('HelloGithub', '开始获取 HelloGithub 数据', 'start', False)
  29. targets = ['last', 'hot']
  30. response_datas = []
  31. for target in targets:
  32. response_data = self.req(target)
  33. response_datas += response_data
  34. if response_datas:
  35. self.save_to_mongo(response_datas)
  36. else:
  37. self.logs_handle.logs_write('HelloGithub', '获取 HelloGithub 数据失败', 'error', False)
  38. self.logs_handle.logs_write('HelloGithub', 'HelloGithub 数据获取完成', 'done', False)
  39. print('获取 HelloGithub 数据 done')
  40. if self.send_email_datas:
  41. self.send_to_email()
  42. else:
  43. print('没有新数据, 不发送邮件')
  44. def req(self, target):
  45. print('开始获取 HelloGithub {} 数据'.format(target))
  46. response_data = []
  47. for i in range(1, 5):
  48. response = httpx.get(url='https://api.hellogithub.com/v1/?sort_by={}&tid=&page={}'.format(target, i),
  49. headers=self.headers)
  50. if response.status_code != 200:
  51. self.logs_handle.logs_write('HelloGithub', '请求失败, 状态码: %s' % response.status_code, 'error',
  52. False)
  53. exit(0)
  54. json_data = response.json()
  55. for d in json_data.setdefault('data'):
  56. response_data.append({
  57. "title": d.setdefault('title', ''),
  58. "context": d.setdefault('summary', '') + ' --- ' + d.setdefault('description'),
  59. "source_url": 'https://hellogithub.com',
  60. 'link': self.source_url + d.setdefault('item_id'),
  61. "article_type": '',
  62. "article_source": target,
  63. "img_url": '',
  64. 'keyword': '',
  65. "posted_date": d.setdefault('updated_at'),
  66. "create_time": int(time.time()),
  67. "create_datetime": datetime.now().strftime('%Y-%m-%d %H:%M:%S')
  68. })
  69. if response_data:
  70. return response_data
  71. else:
  72. self.logs_handle.logs_write('HelloGithub', '获取数据失败', 'error', False)
  73. def save_to_mongo(self, data):
  74. print(f'开始储存 HelloGithub 数据')
  75. for data_to_insert in data:
  76. mongo = MongoHandle(db=self.db, collection=self.collection, del_db=False, del_collection=False, auto_remove=0)
  77. try:
  78. # 检查数据库中是否存在匹配的文档
  79. filter_criteria = {'title': data_to_insert.get('title', '')} # 确保 title 字段有值
  80. count = mongo.collection.count_documents(filter_criteria)
  81. if count == 0:
  82. # 如果没有找到匹配的文档,插入新文档
  83. result = mongo.collection.insert_one(data_to_insert)
  84. # 准备发送邮件的数据
  85. self.send_email_datas.append(data_to_insert)
  86. except TypeError as te:
  87. print('\n%s' % te)
  88. self.logs_handle.logs_write('HelloGithub', '写入数据库报错: %s' % te, 'error', False)
  89. return 0
  90. print(f'处理 HelloGithub 数据完成')
  91. def send_to_email(self):
  92. title = 'HelloGithub - info'
  93. subject = 'HelloGithub - info'
  94. text = '********************************************************\n'
  95. for data in self.send_email_datas:
  96. text += '标题: {}\n'.format(data['title'])
  97. text += '正文: {}\n'.format(data['context'])
  98. text += '文章地址: {}\n'.format(data['source_url'])
  99. text += '文章时间: {}\n'.format(data['posted_date'])
  100. text += '获取时间: {}\n'.format(data['create_datetime'])
  101. text += '********************************************************\n\n'
  102. send_email = SendEmail(subject=subject, title=title, text=text)
  103. send_email.send()
  104. self.logs_handle.logs_write('HelloGithub', f'{title}-发送邮件完成', 'done', False)
  105. if __name__ == "__main__":
  106. H = HelloGithub()
  107. H.main()