基于 abp vNext 和 .NET Core 开发博客项目 - 定时任务最佳实战（二）

基于 abp vNext 和 .NET Core 开发博客项目 – 定时任务最佳实战（二）

2020 年 5 月 30 日
筆記
.NET Core, abp vnext, HtmlAgilityPack, xpath

上一篇(//www.cnblogs.com/meowv/p/12971041.html)使用HtmlAgilityPack抓取壁纸数据成功将图片存入数据库，本篇继续来完成一个全网各大平台的热点新闻数据的抓取。

同样的，可以先预览一下我个人博客中的成品：//meowv.com/hot 😝😝😝，和抓取壁纸的套路一样，大同小异。

本次要抓取的源有18个，分别是博客园、V2EX、SegmentFault、掘金、微信热门、豆瓣精选、IT之家、36氪、百度贴吧、百度热搜、微博热搜、知乎热榜、知乎日报、网易新闻、GitHub、抖音热点、抖音视频、抖音正能量。

还是将数据存入数据库，按部就班先将实体类和自定义仓储创建好，实体取名HotNews。贴一下代码：

//HotNews.cs
using System;
using Volo.Abp.Domain.Entities;

namespace Meowv.Blog.Domain.HotNews
{
    public class HotNews : Entity<Guid>
    {
        /// <summary>
        /// 标题
        /// </summary>
        public string Title { get; set; }

        /// <summary>
        /// 链接
        /// </summary>
        public string Url { get; set; }

        /// <summary>
        /// SourceId
        /// </summary>
        public int SourceId { get; set; }

        /// <summary>
        /// 创建时间
        /// </summary>
        public DateTime CreateTime { get; set; }
    }
}

剩下的大家自己完成，最终数据库生成一张空的数据表，meowv_hotnews 。

然后还是将我们各大平台放到一个枚举类HotNewsEnum.cs中。

//HotNewsEnum.cs
using System.ComponentModel;

namespace Meowv.Blog.Domain.Shared.Enum
{
    public enum HotNewsEnum
    {
        [Description("博客园")]
        cnblogs = 1,

        [Description("V2EX")]
        v2ex = 2,

        [Description("SegmentFault")]
        segmentfault = 3,

        [Description("掘金")]
        juejin = 4,

        [Description("微信热门")]
        weixin = 5,

        [Description("豆瓣精选")]
        douban = 6,

        [Description("IT之家")]
        ithome = 7,

        [Description("36氪")]
        kr36 = 8,

        [Description("百度贴吧")]
        tieba = 9,

        [Description("百度热搜")]
        baidu = 10,

        [Description("微博热搜")]
        weibo = 11,

        [Description("知乎热榜")]
        zhihu = 12,

        [Description("知乎日报")]
        zhihudaily = 13,

        [Description("网易新闻")]
        news163 = 14,

        [Description("GitHub")]
        github = 15,

        [Description("抖音热点")]
        douyin_hot = 16,

        [Description("抖音视频")]
        douyin_video = 17,

        [Description("抖音正能量")]
        douyin_positive = 18
    }
}

和上一篇抓取壁纸一样，做一些准备工作。

在.Application.Contracts层添加HotNewsJobItem<T>，在.BackgroundJobs层添加HotNewsJob用来处理爬虫逻辑，用构造函数方式注入仓储IHotNewsRepository。

//HotNewsJobItem.cs
using Meowv.Blog.Domain.Shared.Enum;

namespace Meowv.Blog.Application.Contracts.HotNews
{
    public class HotNewsJobItem<T>
    {
        /// <summary>
        /// <see cref="Result"/>
        /// </summary>
        public T Result { get; set; }

        /// <summary>
        /// 来源
        /// </summary>
        public HotNewsEnum Source { get; set; }
    }
}

//HotNewsJob.CS
using Meowv.Blog.Domain.HotNews.Repositories;
using System;
using System.Net.Http;
using System.Threading.Tasks;

namespace Meowv.Blog.BackgroundJobs.Jobs.HotNews
{
    public class HotNewsJob : IBackgroundJob
    {
        private readonly IHttpClientFactory _httpClient;
        private readonly IHotNewsRepository _hotNewsRepository;

        public HotNewsJob(IHttpClientFactory httpClient,
                          IHotNewsRepository hotNewsRepository)
        {
            _httpClient = httpClient;
            _hotNewsRepository = hotNewsRepository;
        }

        public async Task ExecuteAsync()
        {
            throw new NotImplementedException();
        }
    }
}

接下来明确数据源地址，因为以上数据源有的返回是HTML，有的直接返回JSON数据。为了方便调用，我这里还注入了IHttpClientFactory。

整理好的待抓取数据源列表是这样的。

...
var hotnewsUrls = new List<HotNewsJobItem<string>>
{
    new HotNewsJobItem<string> { Result = "//www.cnblogs.com", Source = HotNewsEnum.cnblogs },
    new HotNewsJobItem<string> { Result = "//www.v2ex.com/?tab=hot", Source = HotNewsEnum.v2ex },
    new HotNewsJobItem<string> { Result = "//segmentfault.com/hottest", Source = HotNewsEnum.segmentfault },
    new HotNewsJobItem<string> { Result = "//web-api.juejin.im/query", Source = HotNewsEnum.juejin },
    new HotNewsJobItem<string> { Result = "//weixin.sogou.com", Source = HotNewsEnum.weixin },
    new HotNewsJobItem<string> { Result = "//www.douban.com/group/explore", Source = HotNewsEnum.douban },
    new HotNewsJobItem<string> { Result = "//www.ithome.com", Source = HotNewsEnum.ithome },
    new HotNewsJobItem<string> { Result = "//36kr.com/newsflashes", Source = HotNewsEnum.kr36 },
    new HotNewsJobItem<string> { Result = "//tieba.baidu.com/hottopic/browse/topicList", Source = HotNewsEnum.tieba },
    new HotNewsJobItem<string> { Result = "//top.baidu.com/buzz?b=341", Source = HotNewsEnum.baidu },
    new HotNewsJobItem<string> { Result = "//s.weibo.com/top/summary/summary", Source = HotNewsEnum.weibo },
    new HotNewsJobItem<string> { Result = "//www.zhihu.com/api/v3/feed/topstory/hot-lists/total?limit=50&desktop=true", Source = HotNewsEnum.zhihu },
    new HotNewsJobItem<string> { Result = "//daily.zhihu.com", Source = HotNewsEnum.zhihudaily },
    new HotNewsJobItem<string> { Result = "//news.163.com/special/0001386F/rank_whole.html", Source = HotNewsEnum.news163 },
    new HotNewsJobItem<string> { Result = "//github.com/trending", Source = HotNewsEnum.github },
    new HotNewsJobItem<string> { Result = "//www.iesdouyin.com/web/api/v2/hotsearch/billboard/word", Source = HotNewsEnum.douyin_hot },
    new HotNewsJobItem<string> { Result = "//www.iesdouyin.com/web/api/v2/hotsearch/billboard/aweme", Source = HotNewsEnum.douyin_video },
    new HotNewsJobItem<string> { Result = "//www.iesdouyin.com/web/api/v2/hotsearch/billboard/aweme/?type=positive", Source = HotNewsEnum.douyin_positive },
};
...

其中有几个比较特殊的，掘金、百度热搜、网易新闻。

掘金需要发送Post请求，返回的是JSON数据，并且需要指定特有的请求头和请求数据，所有使用IHttpClientFactory创建了HttpClient对象。

百度热搜、网易新闻两个老大哥玩套路，网页编码是GB2312的，所以要专门为其指定编码方式，不然取到的数据都是乱码。

...
var web = new HtmlWeb();
var list_task = new List<Task<HotNewsJobItem<object>>>();

hotnewsUrls.ForEach(item =>
{
    var task = Task.Run(async () =>
    {
        var obj = new object();

        if (item.Source == HotNewsEnum.juejin)
        {
            using var client = _httpClient.CreateClient();
            client.DefaultRequestHeaders.Add("User-Agent", "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/83.0.4103.14 Safari/537.36 Edg/83.0.478.13");
            client.DefaultRequestHeaders.Add("X-Agent", "Juejin/Web");
            var data = "{\"extensions\":{\"query\":{ \"id\":\"21207e9ddb1de777adeaca7a2fb38030\"}},\"operationName\":\"\",\"query\":\"\",\"variables\":{ \"first\":20,\"after\":\"\",\"order\":\"THREE_DAYS_HOTTEST\"}}";
            var buffer = data.SerializeUtf8();
            var byteContent = new ByteArrayContent(buffer);
            byteContent.Headers.ContentType = new MediaTypeHeaderValue("application/json");

            var httpResponse = await client.PostAsync(item.Result, byteContent);
            obj = await httpResponse.Content.ReadAsStringAsync();
        }
        else
        {
            Encoding.RegisterProvider(CodePagesEncodingProvider.Instance);
            obj = await web.LoadFromWebAsync(item.Result, (item.Source == HotNewsEnum.baidu || item.Source == HotNewsEnum.news163) ? Encoding.GetEncoding("GB2312") : Encoding.UTF8);
        }

        return new HotNewsJobItem<object>
        {
            Result = obj,
            Source = item.Source
        };
    });
    list_task.Add(task);
});
Task.WaitAll(list_task.ToArray());

循环 hotnewsUrls ，可以看到HotNewsJobItem我们返回的是object类型，因为有JSON又有HtmlDocument对象。所以这里为了能够统一接收，就是用了object。

针对掘金做了单独处理，使用HttpClient发送Post请求，返回JSON字符串数据。

针对百度热搜和网易新闻，使用Encoding.RegisterProvider(CodePagesEncodingProvider.Instance);注册编码提供程序，然后在web.LoadFromWebAsync(...)加载网页数据的时候指定网页编码，我使用了一个三元表达式来处理。

完成上面这一步，就可以循环 list_task，使用XPath语法，或者解析JSON数据，去拿到数据了。

...
var hotNews = new List<HotNews>();
foreach (var list in list_task)
{
    var item = await list;
    var sourceId = (int)item.Source;

    ...

    if (hotNews.Any())
    {
        await _hotNewsRepository.DeleteAsync(x => true);
        await _hotNewsRepository.BulkInsertAsync(hotNews);
    }
}

这个爬虫同样很简单，只要拿到标题和链接即可，所以主要目标是寻找到页面上的a标签列表。这个我觉得也没必要一个个去分析了，直接上代码。

// 博客园
 if (item.Source == HotNewsEnum.cnblogs)
 {
     var nodes = ((HtmlDocument)item.Result).DocumentNode.SelectNodes("//div[@class='post_item_body']/h3/a").ToList();
     nodes.ForEach(x =>
     {
         hotNews.Add(new HotNews
         {
             Title = x.InnerText,
             Url = x.GetAttributeValue("href", ""),
             SourceId = sourceId,
             CreateTime = DateTime.Now
         });
     });
 }

// V2EX
if (item.Source == HotNewsEnum.v2ex)
{
    var nodes = ((HtmlDocument)item.Result).DocumentNode.SelectNodes("//span[@class='item_title']/a").ToList();
    nodes.ForEach(x =>
    {
        hotNews.Add(new HotNews
        {
            Title = x.InnerText,
            Url = $"//www.v2ex.com{x.GetAttributeValue("href", "")}",
            SourceId = sourceId,
            CreateTime = DateTime.Now
        });
    });
}

 // SegmentFault
 if (item.Source == HotNewsEnum.segmentfault)
 {
     var nodes = ((HtmlDocument)item.Result).DocumentNode.SelectNodes("//div[@class='news__item-info clearfix']/a").Where(x => x.InnerText.IsNotNullOrEmpty()).ToList();
     nodes.ForEach(x =>
     {
         hotNews.Add(new HotNews
         {
             Title = x.SelectSingleNode(".//h4").InnerText,
             Url = $"//segmentfault.com{x.GetAttributeValue("href", "")}",
             SourceId = sourceId,
             CreateTime = DateTime.Now
         });
     });
 }

// 掘金
if (item.Source == HotNewsEnum.juejin)
{
    var obj = JObject.Parse((string)item.Result);
    var nodes = obj["data"]["articleFeed"]["items"]["edges"];
    foreach (var node in nodes)
    {
        hotNews.Add(new HotNews
        {
            Title = node["node"]["title"].ToString(),
            Url = node["node"]["originalUrl"].ToString(),
            SourceId = sourceId,
            CreateTime = DateTime.Now
        });
    }
}

// 微信热门
if (item.Source == HotNewsEnum.weixin)
{
    var nodes = ((HtmlDocument)item.Result).DocumentNode.SelectNodes("//ul[@class='news-list']/li/div[@class='txt-box']/h3/a").ToList();
    nodes.ForEach(x =>
    {
        hotNews.Add(new HotNews
        {
            Title = x.InnerText,
            Url = x.GetAttributeValue("href", ""),
            SourceId = sourceId,
            CreateTime = DateTime.Now
        });
    });
}

// 豆瓣精选
if (item.Source == HotNewsEnum.douban)
{
    var nodes = ((HtmlDocument)item.Result).DocumentNode.SelectNodes("//div[@class='channel-item']/div[@class='bd']/h3/a").ToList();
    nodes.ForEach(x =>
    {
        hotNews.Add(new HotNews
        {
            Title = x.InnerText,
            Url = x.GetAttributeValue("href", ""),
            SourceId = sourceId,
            CreateTime = DateTime.Now
        });
    });
}

// IT之家
if (item.Source == HotNewsEnum.ithome)
{
    var nodes = ((HtmlDocument)item.Result).DocumentNode.SelectNodes("//div[@class='lst lst-2 hot-list']/div[1]/ul/li/a").ToList();
    nodes.ForEach(x =>
    {
        hotNews.Add(new HotNews
        {
            Title = x.InnerText,
            Url = x.GetAttributeValue("href", ""),
            SourceId = sourceId,
            CreateTime = DateTime.Now
        });
    });
}

// 36氪
if (item.Source == HotNewsEnum.kr36)
{
    var nodes = ((HtmlDocument)item.Result).DocumentNode.SelectNodes("//div[@class='hotlist-main']/div[@class='hotlist-item-toptwo']/a[2]|//div[@class='hotlist-main']/div[@class='hotlist-item-other clearfloat']/div[@class='hotlist-item-other-info']/a").ToList();
    nodes.ForEach(x =>
    {
        hotNews.Add(new HotNews
        {
            Title = x.InnerText,
            Url = $"//36kr.com{x.GetAttributeValue("href", "")}",
            SourceId = sourceId,
            CreateTime = DateTime.Now
        });
    });
}

// 百度贴吧
if (item.Source == HotNewsEnum.tieba)
{
    var obj = JObject.Parse(((HtmlDocument)item.Result).ParsedText);
    var nodes = obj["data"]["bang_topic"]["topic_list"];
    foreach (var node in nodes)
    {
        hotNews.Add(new HotNews
        {
            Title = node["topic_name"].ToString(),
            Url = node["topic_url"].ToString().Replace("amp;", ""),
            SourceId = sourceId,
            CreateTime = DateTime.Now
        });
    }
}

// 百度热搜
if (item.Source == HotNewsEnum.baidu)
{
    var nodes = ((HtmlDocument)item.Result).DocumentNode.SelectNodes("//table[@class='list-table']//tr/td[@class='keyword']/a[@class='list-title']").ToList();
    nodes.ForEach(x =>
    {
        hotNews.Add(new HotNews
        {
            Title = x.InnerText,
            Url = x.GetAttributeValue("href", ""),
            SourceId = sourceId,
            CreateTime = DateTime.Now
        });
    });
}

// 微博热搜
if (item.Source == HotNewsEnum.weibo)
{
    var nodes = ((HtmlDocument)item.Result).DocumentNode.SelectNodes("//table/tbody/tr/td[2]/a").ToList();
    nodes.ForEach(x =>
    {
        hotNews.Add(new HotNews
        {
            Title = x.InnerText,
            Url = $"//s.weibo.com{x.GetAttributeValue("href", "").Replace("#", "%23")}",
            SourceId = sourceId,
            CreateTime = DateTime.Now
        });
    });
}

// 知乎热榜
if (item.Source == HotNewsEnum.zhihu)
{
    var obj = JObject.Parse(((HtmlDocument)item.Result).ParsedText);
    var nodes = obj["data"];
    foreach (var node in nodes)
    {
        hotNews.Add(new HotNews
        {
            Title = node["target"]["title"].ToString(),
            Url = $"//www.zhihu.com/question/{node["target"]["id"]}",
            SourceId = sourceId,
            CreateTime = DateTime.Now
        });
    }
}

// 知乎日报
if (item.Source == HotNewsEnum.zhihudaily)
{
    var nodes = ((HtmlDocument)item.Result).DocumentNode.SelectNodes("//div[@class='box']/a").ToList();
    nodes.ForEach(x =>
    {
        hotNews.Add(new HotNews
        {
            Title = x.InnerText,
            Url = $"//daily.zhihu.com{x.GetAttributeValue("href", "")}",
            SourceId = sourceId,
            CreateTime = DateTime.Now
        });
    });
}

// 网易新闻
if (item.Source == HotNewsEnum.news163)
{
    var nodes = ((HtmlDocument)item.Result).DocumentNode.SelectNodes("//div[@class='area-half left']/div[@class='tabBox']/div[@class='tabContents active']/table//tr/td[1]/a").ToList();
    nodes.ForEach(x =>
    {
        hotNews.Add(new HotNews
        {
            Title = x.InnerText,
            Url = x.GetAttributeValue("href", ""),
            SourceId = sourceId,
            CreateTime = DateTime.Now
        });
    });
}

// GitHub
if (item.Source == HotNewsEnum.github)
{
    var nodes = ((HtmlDocument)item.Result).DocumentNode.SelectNodes("//article[@class='Box-row']/h1/a").ToList();
    nodes.ForEach(x =>
    {
        hotNews.Add(new HotNews
        {
            Title = x.InnerText.Trim().Replace("\n", "").Replace(" ", ""),
            Url = $"//github.com{x.GetAttributeValue("href", "")}",
            SourceId = sourceId,
            CreateTime = DateTime.Now
        });
    });
}

// 抖音热点
if (item.Source == HotNewsEnum.douyin_hot)
{
    var obj = JObject.Parse(((HtmlDocument)item.Result).ParsedText);
    var nodes = obj["word_list"];
    foreach (var node in nodes)
    {
        hotNews.Add(new HotNews
        {
            Title = node["word"].ToString(),
            Url = $"#{node["hot_value"]}",
            SourceId = sourceId,
            CreateTime = DateTime.Now
        });
    }
}

// 抖音视频 & 抖音正能量
if (item.Source == HotNewsEnum.douyin_video || item.Source == HotNewsEnum.douyin_positive)
{
    var obj = JObject.Parse(((HtmlDocument)item.Result).ParsedText);
    var nodes = obj["aweme_list"];
    foreach (var node in nodes)
    {
        hotNews.Add(new HotNews
        {
            Title = node["aweme_info"]["desc"].ToString(),
            Url = node["aweme_info"]["share_url"].ToString(),
            SourceId = sourceId,
            CreateTime = DateTime.Now
        });
    }
}

将item.Result转换成指定类型，最终拿到数据后，我们先删除所有数据后再批量插入。

然后新建扩展方法UseHotNewsJob()，在模块类中调用。

//MeowvBlogBackgroundJobsExtensions.cs
...
        /// <summary>
        /// 每日热点数据抓取
        /// </summary>
        /// <param name="context"></param>
        public static void UseHotNewsJob(this IServiceProvider service)
        {
            var job = service.GetService<HotNewsJob>();

            RecurringJob.AddOrUpdate("每日热点数据抓取", () => job.ExecuteAsync(), CronType.Hour(1, 2));
        }
...

指定定时任务为每2小时运行一次。

...
        public override void OnApplicationInitialization(ApplicationInitializationContext context)
        {
            ...
            var service = context.ServiceProvider;
            ...
            service.UseHotNewsJob();
        }

编译运行，此时周期性作业就会出现我们的定时任务了。