import peter

seek/the_paper_com/base.py

@@ -0,0 +1,32 @@
+from DrissionPage.errors import ElementNotFoundError
+
+from database.tinformationsource.model import TInformationSource
+from database.tnews.model import TNews
+from log.log_manager import logger
+from seek.seek_base import SeekBase
+from utils.time_utils import process_time
+
+
+class Base(SeekBase):
+    def __init__(self, information_source: TInformationSource):
+        super().__init__(information_source)
+
+    def get_news(self):
+        news_result = []
+        _news_list = self.session.s_ele('.index_cards__AdZtA').s_eles('.ant-col ant-col-6')
+
+        for _news in _news_list:
+            tnews = TNews()
+            try:
+                tnews.title = _news.s_ele('tag:a').text
+                tnews.url = _news.s_ele('tag:a').link
+                _time = _news.s_ele('.small_text__dR01h').s_eles('tag:span')[1].text
+                tnews.occurrence_date = process_time(_time)
+                tnews.source = self.information_source.title
+                news_result.append(tnews)
+            except ElementNotFoundError as e:
+                logger.error(f"ElementNotFoundError {tnews.title}: {e} - Failed to find element in news item.")
+            except Exception as e:
+                logger.error(f'Unexpected error occurred: {e}')
+
+        return news_result

seek/the_paper_com/content.py

@@ -0,0 +1,50 @@
+import datetime
+
+from DrissionPage.errors import ElementNotFoundError
+
+from database.tinformationsource.model import TInformationSource
+from database.tnews.model import TNews
+from log.log_manager import logger
+from seek.content_base import ContentBase
+
+
+class ThePaperContent(ContentBase):
+    def __init__(self, news: TNews):
+        super().__init__(news)
+
+    def get_content(self):
+        content_ = ''
+        try:
+            content_ = self.session.s_ele('.^index_cententWrap').text
+        except ElementNotFoundError as e:
+            try:
+                # 视频
+                content_ = self.session.s_ele('.^header_desc').text
+            except ElementNotFoundError as e:
+                content_ = 'not found element'
+        return content_
+
+
+def get_content(information_source: TInformationSource) -> list:
+    the_paper_content = ThePaperContent(information_source)
+    content = the_paper_content.get_content()
+    the_paper_content.finish()
+    return content
+
+
+def content_task(news: TNews):
+    logger.info(f'{news.title} news_task start execute at {datetime.datetime.now()}', )
+    ofweek_com_ai = ThePaperContent(news)
+    ofweek_com_ai.do_seek_task()
+    ofweek_com_ai.finish()
+    logger.info(f'{news.title} news_task end execute at {datetime.datetime.now()}')
+
+
+if __name__ == '__main__':
+    logger.info('This module is not for direct call!')
+    news_ = TNews()
+    news_.is_static = True
+    news_.url = 'https://www.thepaper.cn/newsDetail_forward_29745442'
+    content = get_content(news_)
+    logger.info(content)
+    logger.info('Done.')

seek/the_paper_com/international.py

@@ -0,0 +1,38 @@
+import datetime
+
+from database.tinformationsource.model import TInformationSource
+from log.log_manager import logger
+from seek.the_paper_com.base import Base
+
+
+class International(Base):
+    def __init__(self, information_source: TInformationSource):
+        super().__init__(information_source)
+
+
+def get_news(information_source: TInformationSource) -> list:
+    instance = International(information_source)
+    news_list = instance.get_news()
+    instance.finish()
+    return news_list
+
+
+def news_task(information_source: TInformationSource):
+    logger.info(f'{information_source.title} news_task start execute at {datetime.datetime.now()}', )
+    instance = International(information_source)
+    instance.do_seek_task()
+    instance.finish()
+    logger.info(f'{information_source.title} news_task end execute at {datetime.datetime.now()}')
+
+
+if __name__ == '__main__':
+    logger.info('This module is not for direct call!')
+    information_source_ = TInformationSource()
+    information_source_.is_static = True
+    information_source_.url = 'https://www.thepaper.cn/channel_122908'
+    information_source_.title = '国际_澎湃新闻'
+    # news_task(information_source_)
+    news_list_ = get_news(information_source_)
+    for news in news_list_:
+        print(news)
+    logger.info('Done.')

seek/the_paper_com/tech.py

@@ -0,0 +1,38 @@
+import datetime
+
+from database.tinformationsource.model import TInformationSource
+from log.log_manager import logger
+from seek.the_paper_com.base import Base
+
+
+class Tech(Base):
+    def __init__(self, information_source: TInformationSource):
+        super().__init__(information_source)
+
+
+def get_news(information_source: TInformationSource) -> list:
+    instance = Tech(information_source)
+    news_list = instance.get_news()
+    instance.finish()
+    return news_list
+
+
+def news_task(information_source: TInformationSource):
+    logger.info(f'{information_source.title} news_task start execute at {datetime.datetime.now()}', )
+    instance = Tech(information_source)
+    instance.do_seek_task()
+    instance.finish()
+    logger.info(f'{information_source.title} news_task end execute at {datetime.datetime.now()}')
+
+
+if __name__ == '__main__':
+    logger.info('This module is not for direct call!')
+    information_source_ = TInformationSource()
+    information_source_.is_static = True
+    information_source_.url = 'https://www.thepaper.cn/channel_119908'
+    information_source_.title = '科技_澎湃新闻'
+    news_task(information_source_)
+    # news_list_ = get_news(information_source_)
+    # for news in news_list_:
+    #     print(news)
+    logger.info('Done.')