added errors page and routes, moved html parser, fixed db.py app.py

Makefile

@@ -17,6 +17,9 @@ build:
 lint:
 	poetry run flake8 page_analyzer
 
+test:
+	poetry run pytest
+
 selfcheck:
 	poetry check
 

page_analyzer/app.py

@@ -1,9 +1,9 @@
 import os
 import requests
 from dotenv import load_dotenv
-from page_analyzer.url_validator import validate
 from urllib.parse import urlparse
-from bs4 import BeautifulSoup
+from page_analyzer.url_validator import validate
+from page_analyzer.html_parser import parse_page
 from flask import (
     Flask,
     render_template,
@@ -28,6 +28,16 @@
 app.config['SECRET_KEY'] = os.getenv('SECRET_KEY')
 
 
+@app.errorhandler(404)
+def page_not_found(e):
+    return render_template('404.html'), 404
+
+
+@app.errorhandler(500)
+def internal_server_error(e):
+    return render_template('500.html'), 500
+
+
 @app.get('/')
 def page_analyzer():
     message = get_flashed_messages(with_categories=True)
@@ -102,15 +112,9 @@ def add_check(id):
 
     status_code = response.status_code
 
-    html_data = BeautifulSoup(response.text, 'html.parser')
-
-    title = html_data.title.string if html_data.title else None
-    h1 = html_data.h1.string if html_data.h1 else None
-    description = html_data.find('meta', {'name': 'description'})
-    if description:
-        description = description.get('content')
+    page_data = parse_page(response.text)
 
-    add_check_to_db(id, status_code, h1, title, description)
+    add_check_to_db(id, status_code, page_data)
 
     flash('Страница успешно проверена', 'success')
 

page_analyzer/db.py

@@ -51,7 +51,7 @@ def get_url_by_id(url_id):
     return url_data
 
 
-def add_check_to_db(url_id, status_code, h1, title, description):
+def add_check_to_db(url_id, status_code, page_data):
     conn = get_connection()
 
     with conn.cursor() as cur:
@@ -63,21 +63,26 @@ def add_check_to_db(url_id, status_code, h1, title, description):
                     "description "
                     ") "
                     "VALUES (%s, %s, %s, %s, %s)",
-                    (url_id, status_code, h1, title, description)
+                    (url_id,
+                     status_code,
+                     page_data['h1'],
+                     page_data['title'],
+                     page_data['description']
+                     )
                     )
         conn.commit()
         conn.close()
 
 
 def get_urls_with_latest_check():
     conn = get_connection()
-    query = "SELECT urls.id, "\
-            "urls.name, "\
-            "COALESCE(url_checks.status_code::text, '') as status_code, "\
-            "COALESCE(MAX(url_checks.created_at)::text, '') as latest_check "\
-            "FROM urls "\
-            "LEFT JOIN url_checks ON urls.id = url_checks.url_id "\
-            "GROUP BY urls.id, url_checks.status_code "\
+    query = "SELECT urls.id, " \
+            "urls.name, " \
+            "COALESCE(url_checks.status_code::text, '') as status_code, " \
+            "COALESCE(MAX(url_checks.created_at)::text, '') as latest_check " \
+            "FROM urls " \
+            "LEFT JOIN url_checks ON urls.id = url_checks.url_id " \
+            "GROUP BY urls.id, url_checks.status_code " \
             "ORDER BY urls.id DESC"
 
     all_urls_with_latest_check = fetch_all(conn, query)
@@ -87,14 +92,14 @@ def get_urls_with_latest_check():
 
 def get_checks_desc(url_id):
     conn = psycopg2.connect(DATABASE_URL)
-    query = "SELECT id, "\
-            "status_code, "\
-            "COALESCE(h1, '') as h1, "\
-            "COALESCE(title, '') as title, "\
-            "COALESCE(description, '') as description, "\
-            "created_at::text "\
-            "FROM url_checks "\
-            "WHERE url_id = %s "\
+    query = "SELECT id, " \
+            "status_code, " \
+            "COALESCE(h1, '') as h1, " \
+            "COALESCE(title, '') as title, " \
+            "COALESCE(description, '') as description, " \
+            "created_at::text " \
+            "FROM url_checks " \
+            "WHERE url_id = %s " \
             "ORDER BY id DESC"
     value = (url_id,)
 

page_analyzer/html_parser.py

@@ -0,0 +1,15 @@
+from bs4 import BeautifulSoup
+
+
+def parse_page(response_text):
+    html_data = BeautifulSoup(response_text, 'html.parser')
+    page_data = {'title': html_data.title.string if html_data.title else None,
+                 'h1': html_data.h1.string if html_data.h1 else None}
+
+    description = html_data.find('meta', {'name': 'description'})
+    if description:
+        description = description.get('content')
+
+    page_data['description'] = description
+
+    return page_data

page_analyzer/templates/404.html

@@ -0,0 +1,7 @@
+{% extends "layout.html" %}
+{% block content %}
+  <h1>Страница не найдена</h1>
+  <p>Проверьте корректность адреса</p>
+  <a href="{ url_for('page_analyzer') }">Вернуться на главную</a>
+{% endblock %}
+

page_analyzer/templates/500.html

@@ -0,0 +1,6 @@
+{% extends "layout.html" %}
+{% block content %}
+  <h1>Ошибка сервера</h1>
+  <p>Извините, попробуте выполнить ваш запрос позже</p>
+  <a href="{ url_for('page_analyzer') }">Вернуться на главную</a>
+{% endblock %}

page_analyzer/url_validator.py

@@ -6,5 +6,3 @@ def validate(url):
         return 'URL превышает 255 символов'
     elif validators.url(url) is not True:
         return 'Некорректный URL'
-    else:
-        return ''