Вынес парсер в отдельный файл

.gitignore

@@ -0,0 +1,2 @@
+__pycache__/
+*.py[cod]

Dockerfile

@@ -11,7 +11,8 @@ RUN pip3 install -r requirements.txt
 COPY crontab /tmp/crontab
 RUN cat /tmp/crontab > /etc/crontabs/root
 
-COPY main.py main.py
+COPY start_telegram.py start_telegram.py
+COPY parser.py parser.py
 
 # run crond as main process of container
-CMD ["crond", "-f", "-l", "2"]
+CMD ["crond", "-f", "-l", "2"]

README.md

@@ -23,6 +23,24 @@ docker run -d --name VodokanalBot \
 docker compose up -d --build
 ```
 
+## Интеграции
+
+Общая логика парсинга, работы с `data/db.json` и поиска новых сообщений вынесена в [`parser.py`](./parser.py).
+
+Для новой интеграции достаточно создать свой адаптер отправки и передать его в `publish_new_posts`:
+
+``` Python
+from parser import create_session, publish_new_posts
+
+session = create_session()
+
+publish_new_posts(
+    send_message=lambda message: print(message),
+    session=session,
+    url="http://www.tgnvoda.ru/avarii.php",
+)
+```
+
 ## Библиотеки
 
 * [Requests](https://requests.readthedocs.io/en/latest/)

crontab

@@ -1,2 +1,2 @@
-@reboot cd /app && python3 /app/main.py 
+@reboot cd /app && python3 /app/start_telegram.py 
-*/10 * * * *  cd /app && python3 /app/main.py 
+*/10 * * * *  cd /app && python3 /app/start_telegram.py 

main.py

@@ -1,101 +0,0 @@
-import requests, json, os
-from bs4 import BeautifulSoup
-from datetime import datetime
-
-# Config
-
-URL = os.environ.get('VODOKANAL_URL', 'http://www.tgnvoda.ru/avarii.php')
-SEND_SILENT = os.environ.get('SEND_SILENT', False)
-TELEGRAM_TOKEN = os.environ.get('TELEGRAM_TOKEN', '')
-TELEGRAM_CHANNEL = os.environ.get('TELEGRAM_CHANNEL', '')
-PROXY_URL = os.environ.get('PROXY_URL', '')
-
-if TELEGRAM_TOKEN == '':
-    print("Telegram token is not set")
-    exit()
-
-if TELEGRAM_CHANNEL == '':
-    print("Telegram channel is not set")
-    exit()
-
-# Configure HTTP client
-
-session = requests.Session()
-
-if PROXY_URL != '':
-    session.proxies.update({
-        'http': PROXY_URL,
-        'https': PROXY_URL,
-    })
-
-# Load database
-
-db = None
-if (os.path.isfile('data/db.json')):
-    with open('data/db.json', 'r', encoding='utf-8') as f:
-        db = json.load(f)
-else:
-    print("Database not loaded")
-
-# Get data
-
-req = session.get(URL)
-
-if (req.status_code != 200):
-    print("Request error: " + str(req.status_code))
-    exit()
-
-soup = BeautifulSoup(req.content, "html.parser")
-
-elements = []
-for tag in soup.find_all('font', size='2', face='VERDANA'):
-    date = tag.select_one('font:nth-of-type(1)').b.text
-    if not(date.split('.')[0] == str(datetime.today().day).zfill(2) and date.split('.')[1] == str(datetime.today().month).zfill(2)):
-        continue
-    elements.append(date + "$" + tag.select_one('font:nth-of-type(2)').text.replace('\n', ''))
-
-if elements == []:
-    print("No posts")
-    exit()
-
-print("The number of posts for this day:", len(elements))
-
-# Send telegram message
-
-def send_message(message):
-    req = session.get(
-        "https://api.telegram.org/bot" + TELEGRAM_TOKEN + "/sendMessage",
-        params={
-            'chat_id': TELEGRAM_CHANNEL,
-            'disable_notification': str(SEND_SILENT),
-            'text': message,
-        },
-    )
-    if (req.status_code != 200):
-        print("Telegram request error: " + str(req.status_code))
-        exit()
-    else:
-        print("Telegram message sent, mess id: " + str(req.json()['result']['message_id']))
-
-# Compare db and elements
-
-if db is not None:
-    diff = set(elements) - set(db)
-    if not diff:
-        print("No new posts")
-        exit()
-
-    for i in diff:
-        send_message(i.split("$",1)[1])
-else:
-    for element in elements:
-        send_message(element.split("$",1)[1])
-
-# Save database
-
-if not os.path.exists("data"):
-   os.makedirs("data")
-
-with open('data/db.json', 'w', encoding='utf-8') as f:
-    json.dump(elements, f, ensure_ascii=False)
-    print("Database updated")

parser.py

@@ -0,0 +1,110 @@
+import json
+import os
+from dataclasses import dataclass
+from datetime import datetime
+
+import requests
+from bs4 import BeautifulSoup
+
+
+DEFAULT_DB_PATH = "data/db.json"
+
+
+@dataclass(frozen=True)
+class Post:
+    date: str
+    text: str
+
+    @property
+    def key(self):
+        return self.date + "$" + self.text
+
+
+def create_session(proxy_url=""):
+    session = requests.Session()
+
+    if proxy_url:
+        session.proxies.update({
+            "http": proxy_url,
+            "https": proxy_url,
+        })
+
+    return session
+
+
+def load_database(path=DEFAULT_DB_PATH):
+    if not os.path.isfile(path):
+        print("Database not loaded")
+        return None
+
+    with open(path, "r", encoding="utf-8") as f:
+        return json.load(f)
+
+
+def save_database(posts, path=DEFAULT_DB_PATH):
+    directory = os.path.dirname(path)
+    if directory and not os.path.exists(directory):
+        os.makedirs(directory)
+
+    with open(path, "w", encoding="utf-8") as f:
+        json.dump([post.key for post in posts], f, ensure_ascii=False)
+        print("Database updated")
+
+
+def fetch_posts(session, url, today=None):
+    req = session.get(url)
+
+    if req.status_code != 200:
+        raise RuntimeError("Request error: " + str(req.status_code))
+
+    return parse_posts(req.content, today=today)
+
+
+def parse_posts(content, today=None):
+    today = today or datetime.today()
+    soup = BeautifulSoup(content, "html.parser")
+
+    posts = []
+    for tag in soup.find_all("font", size="2", face="VERDANA"):
+        date = tag.select_one("font:nth-of-type(1)").b.text
+        if not is_today(date, today):
+            continue
+
+        text = tag.select_one("font:nth-of-type(2)").text.replace("\n", "")
+        posts.append(Post(date=date, text=text))
+
+    return posts
+
+
+def is_today(date, today):
+    day, month = date.split(".")[:2]
+    return day == str(today.day).zfill(2) and month == str(today.month).zfill(2)
+
+
+def get_new_posts(posts, database):
+    if database is None:
+        return posts
+
+    database_keys = set(database)
+    return [post for post in posts if post.key not in database_keys]
+
+
+def publish_new_posts(send_message, session, url, db_path=DEFAULT_DB_PATH):
+    database = load_database(db_path)
+    posts = fetch_posts(session, url)
+
+    if not posts:
+        print("No posts")
+        return
+
+    print("The number of posts for this day:", len(posts))
+
+    new_posts = get_new_posts(posts, database)
+    if not new_posts:
+        print("No new posts")
+        return
+
+    for post in new_posts:
+        send_message(post.text)
+
+    save_database(posts, db_path)

start_telegram.py

@@ -0,0 +1,51 @@
+import os
+
+from parser import create_session, publish_new_posts
+
+
+URL = os.environ.get('VODOKANAL_URL', 'http://www.tgnvoda.ru/avarii.php')
+SEND_SILENT = os.environ.get('SEND_SILENT', False)
+TELEGRAM_TOKEN = os.environ.get('TELEGRAM_TOKEN', '')
+TELEGRAM_CHANNEL = os.environ.get('TELEGRAM_CHANNEL', '')
+PROXY_URL = os.environ.get('PROXY_URL', '')
+
+
+def send_telegram_message(session, message):
+    req = session.get(
+        "https://api.telegram.org/bot" + TELEGRAM_TOKEN + "/sendMessage",
+        params={
+            'chat_id': TELEGRAM_CHANNEL,
+            'disable_notification': str(SEND_SILENT),
+            'text': message,
+        },
+    )
+    if (req.status_code != 200):
+        print("Telegram request error: " + str(req.status_code))
+        exit()
+    else:
+        print("Telegram message sent, mess id: " + str(req.json()['result']['message_id']))
+
+
+def main():
+    if TELEGRAM_TOKEN == '':
+        print("Telegram token is not set")
+        exit()
+
+    if TELEGRAM_CHANNEL == '':
+        print("Telegram channel is not set")
+        exit()
+
+    session = create_session(PROXY_URL)
+    try:
+        publish_new_posts(
+            send_message=lambda message: send_telegram_message(session, message),
+            session=session,
+            url=URL,
+        )
+    except RuntimeError as error:
+        print(error)
+        exit()
+
+
+if __name__ == "__main__":
+    main()