Initial commit: LookingWeb project

.gitignore

@@ -0,0 +1,6 @@
+__pycache__/
+*.pyc
+*.pyo
+.env
+.venv/
+venv/

LookingWebMain.py

@@ -0,0 +1,96 @@
+import time
+
+import requests
+from loguru import logger
+from bs4 import BeautifulSoup
+from dataService.dto.SeedInfo import SeedInfo
+from dataService import MySqlService as mysql
+from ariaDownloaderService import AriaDownloader as ariaDownloader
+
+"""
+获取页面信息
+"""
+def get_page(url):
+    logger.debug("开始获取页面信息：{}", url)
+    try:
+        response = requests.get(url)
+        soup = BeautifulSoup(response.content, 'html.parser')
+        logger.debug("页面信息获取成功。URL: {}", url)
+        return soup
+    except Exception as e:
+        logger.error("获取页面信息失败。URL: {}, 错误: {}", url, str(e))
+        return None
+
+
+"""
+用于获取目标元素上的详情页url
+"""
+def get_target_element_list(page_content):
+    logger.debug("开始提取可下载内容的链接")
+    downloadable_url_list = []
+    try:
+        for element in page_content.find_all("div", {"class": "card-body"}):
+            if "可下" in element.find("div", {"class": "video-list-item-tag-wrapper"}).text:
+                url = element.find("a")["href"]
+                downloadable_url_list.append(url)
+                logger.debug("发现可下载链接: {}", url)
+        logger.debug("链接提取完成。共发现 {} 个链接。", len(downloadable_url_list))
+        return downloadable_url_list
+    except Exception as e:
+        logger.error("提取链接时出现错误: {}", str(e))
+        return []
+
+
+def get_seed_from_element_detail(element_detail):
+    logger.debug("开始解析页面中的种子信息")
+    seedInfoList = []
+    time.sleep(2)
+    try:
+        from_url = element_detail.find("link")["href"]
+        title = element_detail.find("meta", {"name": "description"})["content"]
+        for e in element_detail.find("tbody").find_all("a", class_="magnet-name text-truncate d-block small"):
+            seedInfo = SeedInfo(title, e["href"], "1", url=from_url)
+            seedInfoList.append(seedInfo)
+            logger.debug("解析到种子信息: 标题：{}，链接：{}", title, e["href"])
+        logger.debug("页面种子信息解析完成。共 {} 个种子。", len(seedInfoList))
+        return seedInfoList
+    except Exception as e:
+        logger.error("解析种子信息时出现错误: {}", str(e))
+        return []
+
+
+def main():
+    # 1-18
+    for page in range(5):
+        url = f"https://freejavbt.com/censored/filter?c=212&page={page+1}"
+        logger.info("程序启动。开始处理 URL: {}", url)
+        new_seed = []
+        # 获取页面信息
+        page = get_page(url)
+        if page is not None:
+            # 获取页面中的目标元素所包含的详情页信息(url列表)
+            targetElementList = get_target_element_list(page)
+            for targetElement in targetElementList:
+                element_detail = get_page(targetElement)
+                if element_detail is not None:
+                    seed_list = get_seed_from_element_detail(element_detail)
+                    # 将新发现的种子存储进new_seed列表
+                    for seed in seed_list:
+                        if not mysql.seed_is_exist(seed):
+                            new_seed.append(seed)
+                            logger.debug("新增种子到下载列表: {}", seed.name)
+
+            # 推送new_seed进入aria2进行种子下载
+            for seed in new_seed:
+                logger.info("开始下载种子: {}", seed.name)
+                status = ariaDownloader.download_by_seed(seed)  # 返回值取值1，0，-1
+                seed.haveDownload = status
+                seed.seave_to_dataBase()
+                logger.info("种子下载完成。标题: {}, 状态: {}", seed.name, "成功" if status == "1" else "失败")
+        else:
+            logger.error("无法获取网页内容。URL: {}", url)
+
+        logger.info("程序处理完成。")
+
+if __name__ == "__main__":
+    main()

ariaDownloaderService/AriaDownloader.py

@@ -0,0 +1,30 @@
+import requests
+from loguru import logger
+
+def download_by_seed(seed_info):
+    url = "http://192.168.123.199:6800/jsonrpc"
+    payload = {
+        "jsonrpc": "2.0",
+        "method": "aria2.addUri",
+        "id": "QXJpYU5nXzE3MTczMTQ0NjBfMC41Nzc4ODU5Mjc2NjQ2MzE4",
+        "params": [
+            "token:guoxy5157",
+            [seed_info.seedUrl],
+            {}
+        ]
+    }
+
+    logger.debug("开始向 Aria2 发送请求，种子URL: {}", seed_info.seedUrl)
+
+    try:
+        response = requests.post(url, json=payload)
+
+        if response.status_code == 200:
+            logger.info("请求成功，种子 {} 已添加至下载队列", seed_info.seedUrl)
+            return "1"
+        else:
+            logger.warning("请求返回非200状态码，种子 {} 添加失败。状态码: {}", seed_info.seedUrl, response.status_code)
+            return "-1"  # 请求成功，但失败了
+    except Exception as e:
+        logger.error("请求过程中出现异常，种子 {} 添加失败。错误信息: {}", seed_info.seedUrl, str(e))
+        return "0"  # 请求过程中发生异常

dataService/MySqlService.py

@@ -0,0 +1,122 @@
+import pymysql
+from loguru import logger
+
+class DatabaseConfig:
+    HOST = "192.168.123.199"
+    PORT = 3306
+    USER = "root"
+    PASSWORD = "GUOxy5157"
+    DB = "LookingWeb"
+
+
+class MySQLDatabase:
+    def __init__(self, host, prot, user, password, db):
+        self.port = prot
+        self.host = host
+        self.user = user
+        self.password = password
+        self.db = db
+        self.connection = None
+        self.connect()
+
+    def __enter__(self):
+        self.connect()
+        return self
+
+    def __exit__(self, exc_type, exc_val, exc_tb):
+        self.connection.close()
+
+    @classmethod
+    def create_connection(cls):
+        return cls(DatabaseConfig.HOST, DatabaseConfig.PORT, DatabaseConfig.USER, DatabaseConfig.PASSWORD,
+                   DatabaseConfig.DB)
+
+
+    def connect(self):
+        if not self.connection or not self.connection.open:
+            try:
+                self.connection = pymysql.connect(
+                    host=self.host, port=self.port, user=self.user,
+                    password=self.password, db=self.db, charset='utf8mb4',
+                    cursorclass=pymysql.cursors.DictCursor)
+                logger.info("数据库连接成功：{}@{}", self.user, self.host)
+            except pymysql.MySQLError as e:
+                logger.error("数据库连接失败：{}", str(e))
+
+    def close(self):
+        if self.connection and self.connection.open:
+            self.connection.close()
+
+    def query(self, sql, params=None):
+        self.connect()
+        with self.connection.cursor() as cursor:
+            try:
+                cursor.execute(sql, params or ())
+                result = cursor.fetchall()
+                logger.debug("查询成功：{}", sql)
+                return result
+            except pymysql.MySQLError as e:
+                logger.error("查询失败：{}，错误：{}", sql, str(e))
+                return None
+
+    def execute(self, sql, params=None):
+        self.connect()
+        with self.connection.cursor() as cursor:
+            cursor.execute(sql, params or ())
+            self.connection.commit()
+
+
+def is_connected():
+    logger.debug("正在检查数据库连接...")
+    with MySQLDatabase.create_connection() as mysql:
+        try:
+            mysql.query("SELECT * FROM seeds_info where id = 1")
+            logger.info("数据库连接检查通过")
+            return True
+        except pymysql.MySQLError as e:
+            logger.error("数据库连接检查失败：{}", str(e))
+            return False
+
+
+def seed_is_exist(seed):
+    with MySQLDatabase.create_connection() as mysql:
+        try:
+            sql = "SELECT * FROM seeds_info where seed_url = %s"
+            result = mysql.query(sql, (seed.seedUrl,))
+            if not result:
+                logger.info("种子不存在: {}", seed.seedUrl)
+                return False
+            else:
+                logger.info("种子已存在: {}", seed.seedUrl)
+                return True
+        except pymysql.MySQLError as e:
+            logger.error("检查种子存在失败: {}，错误: {}", seed.seedUrl, str(e))
+            return False
+
+def save_seed(seed_info):
+    with MySQLDatabase.create_connection() as connection:
+        query = """
+        INSERT INTO seeds_info (name, seed_url, have_download, create_time, update_time, url)
+        VALUES (%s, %s, %s, %s, %s, %s)
+        """
+        values = (seed_info.name, seed_info.seedUrl, seed_info.haveDownload, seed_info.createTime, seed_info.updateTime,
+                  seed_info.url)
+        try:
+            connection.execute(query, values)
+            logger.info("种子信息保存成功: {}", seed_info.seedUrl)
+            return True
+        except pymysql.MySQLError as e:
+            logger.error("保存种子信息失败: {}，错误: {}", seed_info.seedUrl, str(e))
+            return False
+
+def delete_seed(seed_info):
+    query = "DELETE FROM seeds_info WHERE seed_url = %s"
+    value = (seed_info.seedUrl,)
+    with MySQLDatabase.create_connection() as connection:
+        try:
+            connection.execute(query, value)
+            logger.info("种子删除成功: {}", seed_info.seedUrl)
+            return True
+        except pymysql.MySQLError as e:
+            logger.error("删除种子失败: {}，错误: {}", seed_info.seedUrl, str(e))
+            return False

dataService/dto/SeedInfo.py

@@ -0,0 +1,22 @@
+import time
+from datetime import datetime
+from dataService import MySqlService as mysqlService
+
+class SeedInfo(object):
+    def __init__(self, name=None, seedUrl=None, haveDownload=None, createTime=None, updateTime=None, url=None):
+        self.name = name
+        self.seedUrl = seedUrl
+        self.haveDownload = haveDownload
+        self.createTime = self.format_time(createTime)
+        self.updateTime = self.format_time(updateTime)
+        self.url = url
+
+    @staticmethod
+    def format_time(timestamp):
+        if timestamp is None:
+            timestamp = time.time()
+        return datetime.fromtimestamp(timestamp).strftime('%Y-%m-%d %H:%M:%S')
+
+    def seave_to_dataBase(self):
+        mysqlService.save_seed(self)
+        return