Написали класс для сканирования

main.py

@@ -1,63 +1,79 @@
+import requests
+from bs4 import BeautifulSoup
+import re
 import sqlite3
+from datetime import datetime
 
-db_name = 'mydb.db'
+class Spyder:
-
+    def __init__(self):
-class User:
+        self.db_name = 'database.db'
-    def __init__(self,name,email,age):
+        self.conn = sqlite3.connect(self.db_name)
-        self.__id = None
+        self.cursor = self.conn.cursor()
-        self.__name = name
+    def init_db(self):
-        self.__email = email
+        query = '''CREATE TABLE links(
-        self.__age = age
+            id INTEGER PRIMARY KEY AUTOINCREMENT,
-
+            link TEXT NOT NULL,
-    @property
+            last_scan INTEGER DEFAULT 0,
-    def name(self):
+            title TEXT DEFAULT ''
-        return self.__name
+        )'''
-    @name.setter
+        self.cursor.execute(query)
-    def name(self,value):
+    def scan(self,url):
-        self.__name = value
+        if url.endswith('.pdf'):
-        
+            return
-    @property
+        headers = {
-    def email(self):
+            'User-Agent':'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/148.0.0.0 Safari/537.36'
-        return self.__email
+            }
-
+        try:
-    @email.setter
+            response = requests.get(
-    def email(self,value):
+                url,
-        if '@' in value:
+                headers=headers,
-            self.__email = value
+                timeout=5
+            )
+        except:
+            self.update_url(url,'Scan error')
+            return
+        html = response.text
+        soup = BeautifulSoup(html,'html.parser')
+        result = soup.find_all("a", href=True)
+        title = soup.title.text
+        self.update_url(url,title)
+        for el in result:
+            if el['href'].endswith('.pdf'):
+                continue
+            if not el['href'].startswith('http'):
+                match = re.findall(r'https?://[A-Za-z0-9.-]+',url)
+                link = f"{match[0]}{el['href']}"
             else:
-            raise ValueError('Не корректный адрес почты')
+                link = el['href']
-    @property
+            query = f'SELECT id FROM links WHERE link = ?'
-    def age(self):
+            self.cursor.execute(query,(link,))
-        return self.__age
+            if self.cursor.fetchone():
-    @age.setter
+                continue
-    def age(self,value):
+            query = f'INSERT INTO links(link) VALUES(?)'
-        if 0 <= value <= 120:
+            self.cursor.execute(query,(link,))
-            self.__age = value
+            self.conn.commit()
-        else:
+    def update_url(self,url,title):
-            raise ValueError('Неправильно указан возраст')
+        query = "SELECT id FROM links WHERE link = ?"
+        self.cursor.execute(query,(url,))
+        link_id = self.cursor.fetchone()
+        if link_id:
+            query = "UPDATE links SET title=?, last_scan=? WHERE id=?"
+            self.cursor.execute(query,(title,datetime.now().timestamp(),link_id[0]))
+            self.conn.commit()
+    def get_url(self):
+        query = "SELECT link FROM links ORDER BY last_scan LIMIT 1"
+        self.cursor.execute(query)
+        return self.cursor.fetchone()
+    def close(self):
+        self.conn.close()
 
-    def save(self):
+if __name__ == '__main__':
-        conn = sqlite3.connect(db_name)
+    spyder = Spyder()
-        cursor = conn.cursor()
+    #spyder.init_db()
-        property_list = [f"{k}" for k, v in User.__dict__.items() if isinstance(v, property)]
+    #spyder.scan('https://habr.com/ru/news/1046177/')
-        if '_User__id' in self.__dict__.keys():
+    for _ in range(5):
-            update_string = ','.join([f'`{p_name}`= ? ' for p_name in property_list])
+        url = spyder.get_url()
-            update_values = [getattr(self,p) for p in property_list] + [self.__id]
+        print(url[0])
-            query = f"UPDATE {User.__name__} SET {update_string} WHERE id = ?"
+        spyder.scan(url[0])
-            print(query,update_values)
+    spyder.close()
-        else:
-            property_count = ('?,'*len(property_list))[:-1] #','.join(['?' for _ in range(len(property_list))])
-            property_names = ','.join(property_list)
-            property_values = [getattr(self,p) for p in property_list]
-            query = f"INSERT INTO {User.__name__}({property_names}) VALUES({property_count})"
-            print(query,property_values)
-            cursor.execute(query,property_values)
-            conn.commit()
-            self.__id = cursor.lastrowid
-        conn.close()
-#print(User.__dict__)
 
-user1 = User('Igor','igor@example.com',44)
-user1.save()
-user1.age = 55
-user1.save()