hello_github.py 5.7 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143
  1. # -*- coding: utf-8 -*-
  2. '''
  3. Hello Github
  4. '''
  5. import os
  6. import sys
  7. import time
  8. from datetime import datetime
  9. import httpx
  10. sys.path.append(os.path.join(os.path.abspath(__file__).split('AutoInfo')[0] + 'AutoInfo'))
  11. from utils.utils import *
  12. config_json = LoadConfig().load_config()
  13. DEFAULT_RE_PUSH_TIMES = config_json.get('DEFAULT_RE_PUSH_TIMES')
  14. class HelloGithub(object):
  15. def __init__(self):
  16. self.logs_handle = LogsHandle()
  17. self.now_day = time.strftime('%Y-%m-%d', time.localtime())
  18. self.base_url = 'https://api.hellogithub.com/v1/?sort_by=last&tid=&page={}'
  19. self.headers = {
  20. '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'
  21. }
  22. self.db = 'NEWS'
  23. self.collection = 'HelloGithub_info'
  24. self.source_url = 'https://hellogithub.com/repository/'
  25. self.send_email_datas = []
  26. self.send_email_now = 0
  27. def main(self):
  28. self.logs_handle.logs_write('HelloGithub', '开始获取 HelloGithub 数据', 'start', False)
  29. targets = ['featured']
  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_now:
  41. if self.send_email_datas:
  42. self.send_to_email()
  43. else:
  44. print('没有新数据, 不发送邮件')
  45. def req(self, target):
  46. print('开始获取 HelloGithub {} 数据'.format(target))
  47. response_data = []
  48. for i in range(1, 5):
  49. url = 'https://api.hellogithub.com/v1/?sort_by={}&tid=&page={}'.format(target, i)
  50. try:
  51. response = httpx.get(url=url, headers=self.headers)
  52. except Exception as e:
  53. print("请求出错{}, \nurl: {}".format(e, url))
  54. continue
  55. if response.status_code != 200:
  56. print(
  57. '获取 HelloGithub {} 数据, 状态码: {}, 程序退出\n检查目标地址: https://api.hellogithub.com/v1/?sort_by={}&tid=&page={}'.format(
  58. target, response.status_code, target, i))
  59. self.logs_handle.logs_write('HelloGithub', '请求失败, 状态码: %s' % response.status_code, 'error',
  60. False)
  61. exit(0)
  62. json_data = response.json()
  63. for d in json_data.setdefault('data'):
  64. response_data.append({
  65. "title": d.setdefault('title', ''),
  66. "context": '---'.join([d.setdefault('summary', ''), d.setdefault('description', '')]),
  67. "source_url": 'https://hellogithub.com',
  68. 'link': self.source_url + d.setdefault('item_id'),
  69. "article_type": '',
  70. "article_source": target,
  71. "img_url": '',
  72. 'keyword': '',
  73. "posted_date": d.setdefault('updated_at'),
  74. "create_time": int(time.time()),
  75. "create_datetime": datetime.now().strftime('%Y-%m-%d %H:%M:%S'),
  76. "repush_times": DEFAULT_RE_PUSH_TIMES
  77. })
  78. if response_data:
  79. return response_data
  80. else:
  81. self.logs_handle.logs_write('HelloGithub', '获取数据失败', 'error', False)
  82. def save_to_mongo(self, data):
  83. print(f'开始储存 HelloGithub 数据')
  84. for data_to_insert in data:
  85. mongo = MongoHandle(db=self.db, collection=self.collection, del_db=False, del_collection=False,
  86. auto_remove=0)
  87. try:
  88. # 检查数据库中是否存在匹配的文档
  89. filter_criteria = {'title': data_to_insert.get('title', '')} # 确保 title 字段有值
  90. count = mongo.collection.count_documents(filter_criteria)
  91. if count == 0:
  92. # 如果没有找到匹配的文档,插入新文档
  93. result = mongo.collection.insert_one(data_to_insert)
  94. # 准备发送邮件的数据
  95. self.send_email_datas.append(data_to_insert)
  96. except TypeError as te:
  97. print('\n%s' % te)
  98. self.logs_handle.logs_write('HelloGithub', '写入数据库报错: %s' % te, 'error', False)
  99. return 0
  100. print(f'处理 HelloGithub 数据完成', datetime.now().strftime('%Y-%m-%d %H:%M:%S'))
  101. def send_to_email(self):
  102. title = 'HelloGithub - info'
  103. subject = 'HelloGithub - info'
  104. text = '********************************************************\n'
  105. for data in self.send_email_datas:
  106. text += '标题: {}\n'.format(data['title'])
  107. text += '正文: {}\n'.format(data['context'])
  108. text += '文章地址: {}\n'.format(data['source_url'])
  109. text += '文章时间: {}\n'.format(data['posted_date'])
  110. text += '获取时间: {}\n'.format(data['create_datetime'])
  111. text += '********************************************************\n\n'
  112. SendEmail(subject=subject, title=title, text=text).send()
  113. # GotifyNotifier(title=title, message=text, token_name='news').send_message()
  114. self.logs_handle.logs_write('HelloGithub', f'{title}-发送邮件完成', 'done', False)
  115. if __name__ == "__main__":
  116. HelloGithub().main()