【1901100139】⾃学训练营17群 DAY11

exercises/1901100139/d10/main.py

@@ -30,4 +30,4 @@ def main():
         logging.exception(e)
 
 if __name__=='__main__':
-    main()        
+    main()    

exercises/1901100139/d11/main.py

@@ -0,0 +1,34 @@
+
+import yagmail
+import requests
+import pyquery
+import getpass
+import logging
+from mymodule import stats_word
+
+
+logging.basicConfig(format='file:%(filename)s|line:%(lineno)d|message:%(message)s',level=logging.DEBUG)
+
+def get_article():
+    r=requests.get('https://mp.weixin.qq.com/s/pLmuGoc4bZrMNl7MSoWgiA')
+    document= pyquery.PyQuery(r.text)
+    return document('#js_content').text()
+
+def main():
+    try:
+        article=get_article()
+        result=stats_word.stats_text_cn(article,10)
+        logging.info('%s %s',type(result),str(result))
+        sender=input('请输入邮箱号：')
+        password=getpass.getpass('请输入密码：')
+        recipients=input('请输入收件邮箱')
+        yag=yagmail.SMTP(sender,password,'smtp.qq.com')
+        yag.send(recipients,'【1901100139】⾃学训练营 DAY11 PassionPit',str(result))
+        logging.info('已发送')
+    except Exception as e:
+        logging.exception(e) 
+
+
+
+if __name__=='__main__':
+    main()        

exercises/1901100139/d11/mymodule/stats_word.py

@@ -0,0 +1,22 @@
+from collections import Counter
+def stats_text_en(text,count):
+    elements = text.split()
+    words = []
+    symbols = ',.*-!'
+    for element in elements:
+        for symbol in symbols:
+            element = element.replace(symbol,'')
+        if len(element) and element.isascii():
+            words.append(element)
+    return Counter(words).most_common(count)
+def stats_text_cn(text,count):
+    cn_characters = []
+    for character in text:
+        if '\u4e00' <= character <= '\u9fff':
+            cn_characters.append(character)
+    return Counter(cn_characters).most_common(count)
+def stats_text(text,count):
+    '''
+    合并中英词频的结果
+    ''' 
+    return stats_text_en(text,count) + stats_text_cn(text,count)