static void Main(string[] args)
{
var config = new Config
{
Name = "quanmin",
ScanUrls = "https://www.quanmin.tv/game/all",
Fields = new[]
{
new Field
{
Name = "title",
Selectortype = SelectorType.JsonPath,
Selector = "$.title"
},
new Field
{
Name = "username",
Selectortype = SelectorType.JsonPath,
Selector = "$.nick"
},
new Field
{
Name = "online",
Selectortype = SelectorType.JsonPath,
Selector = "$.view",
Type = FieldType.Int,
},
new Field
{
Name = "fanscount",
Selectortype = SelectorType.JsonPath,
Selector = "$.follow",
Type = FieldType.Int,
},
new Field
{
Name = "cate",
Selector = "$.category_name",
Selectortype = SelectorType.JsonPath
}
},
RepeatWhen = RepeatWhenEver.hour,
RepeatAt = new TimeSpan(0, 5, 0),
};
crawler = new CrawlerDotNet.Core.Crawler();
var curPage = 1;
crawler.BeforeCrawl = () =>
{
curPage = 1;
};
crawler.Downloader.AfterDownloadPage = p =>
{
if (p.Request.Type != PageType.ContextUrl)
{
return;
}
//处理页面
var r = new Regex("{\"uid([\\s\\S]*)\"ignore_ad\":true}");
var m = r.Match(p.Html);
p.Html = m.Value;
};
crawler.Processor.OnProcessScanPage = p =>
{
var r = new Regex("total:([0-9]*),");
var m = r.Match(p.Html.Replace(" ", string.Empty));
//得到页码
var page = int.Parse(m.Groups[1].Value);
for (int i = 1; i <= page; i++)
{
crawler.Schduler.AddUrl($"https://www.quanmin.tv/game/all?p={i}", PageType.HelperUrl);
}
p.SkipExtract();
};
crawler.Processor.OnProcessHelperPage = p =>
{
var r = new Regex("\"evtname\":\"([0-9]*)\"");
var ms = r.Matches(p.Html);
foreach (Match m in ms)
{
crawler.Schduler.AddUrl("https://www.quanmin.tv/" + m.Groups[1].Value);
}
p.SkipExtract();
};
crawler.Setup(config);
crawler.Start();
Console.WriteLine("end");
Console.ReadKey();
}
static void Main(string[] args)
{
var config = new Config
{
Name = "longzhu",
ScanUrls = "http://api.plu.cn/tga/streams?max-results=50&start-index=0&sort-by=views&filter=0&game=0",
Fields = new[]
{
new Field
{
Name = "title",
Selectortype = SelectorType.JsonPath,
Selector = "$.channel.status"
},
new Field
{
Name = "username",
Selectortype = SelectorType.JsonPath,
Selector = "$.channel.name"
},
new Field
{
Name = "online",
Selectortype = SelectorType.JsonPath,
Selector = "$.viewers",
Type = FieldType.Int,
},
new Field
{
Name = "fanscount",
Selectortype = SelectorType.JsonPath,
Selector = "$.channel.followers",
Type = FieldType.Int,
},
new Field
{
Name = "cate",
Selector = "$.game[0].name",
Selectortype = SelectorType.JsonPath
}
},
RepeatWhen = RepeatWhenEver.hour,
RepeatAt = new TimeSpan(0, 25, 0),
};
crawler = new Crawler();
crawler.Downloader.AfterDownloadPage = p =>
{
};
crawler.Processor.OnCustomExtract = p =>
{
var j = JObject.Parse(p.Html);
var jr = JArray.FromObject(j["data"]["items"]);
for (int i = 0; i < jr.Count; i++)
{
var exres = new ExtractResults();
var info = jr[i];
foreach (var f in config.Fields)
{
var res = new Result(f.Name, info.SelectToken(f.Selector).ToString());
exres.Add(res);
}
p.Results.Add(exres);
}
};
crawler.Processor.OnProcessScanPage = p =>
{
var totalcount = p.GetJson("$.data.totalItems");
var pagecount = int.Parse(totalcount) / 50 + 1;
for (int i = 1; i <= pagecount; i++)
{
crawler.Schduler.AddUrl($"http://api.plu.cn/tga/streams?max-results=200&start-index={i * 50}&sort-by=views&filter=0&game=0");
}
};
crawler.Setup(config);
crawler.Start();
Console.WriteLine("end");
Console.ReadKey();
}
static void Main(string[] args)
{
var config = new Config
{
Name = "chushou",
ScanUrls = "https://chushou.tv/live/down-v2.htm",
Fields = new[]
{
new Field
{
Name = "title",
Selectortype = SelectorType.JsonPath,
Selector = "$.name"
},
new Field
{
Name = "username",
Selectortype = SelectorType.JsonPath,
Selector = "$.meta.creator"
},
new Field
{
Name = "online",
Selectortype = SelectorType.JsonPath,
Selector = "$.meta.onlineCount",
Type = FieldType.Int,
},
new Field
{
Name = "fanscount",
Selectortype = SelectorType.JsonPath,
Selector = "$.meta.subscriberCount",
Type = FieldType.Int,
},
new Field
{
Name = "cate",
Selector = "$.meta.gameName",
Selectortype = SelectorType.JsonPath
}
},
RepeatWhen = RepeatWhenEver.hour,
RepeatAt = new TimeSpan(0, 35, 0),
};
crawler = new CrawlerDotNet.Core.Crawler();
string lastpoint = "";
crawler.Processor.OnProcessScanPage = p =>
{
var point = p.GetJson("$.data.breakpoint");
crawler.Schduler.AddUrl("https://chushou.tv/live/down-v2.htm?&breakpoint=" + point, point != lastpoint ? PageType.ScanUrl : PageType.ContextUrl);
lastpoint = point;
};
crawler.Processor.OnCustomExtract = p =>
{
var j = JObject.Parse(p.Html);
var jr = JArray.FromObject(j["data"]["items"]);
for (int i = 0; i < jr.Count; i++)
{
var exres = new ExtractResults();
var info = jr[i];
foreach (var f in config.Fields)
{
var res = new Result(f.Name, info.SelectToken(f.Selector).ToString());
exres.Add(res);
}
p.Results.Add(exres);
}
};
crawler.Setup(config);
crawler.Start();
Console.WriteLine("end");
Console.ReadKey();
}
static void douyuSample()
{
//https://www.douyu.com/directory/all
//https://www.douyu.com/directory/all?page=1&isAjax=1
//http://open.douyucdn.cn/api/RoomApi/room/
#region config
var c = new Config
{
Name = "douyu",
ScanUrls = "https://www.douyu.com/directory/all",
ContentUrlRegexes = new Regex("room"),
HelperUrlRegexes = new Regex("page"),
Fields = new[]
{
new Field
{
Name = "title",
Selectortype = SelectorType.JsonPath,
Selector = "$.data.room_name"
}, new Field
{
Name = "username",
Selectortype = SelectorType.JsonPath,
Selector = "$.data.owner_name"
}, new Field
{
Name = "online",
Selectortype = SelectorType.JsonPath,
Selector = "$.data.online",
Type = FieldType.Int,
}, new Field
{
Name = "fanscount",
Selectortype = SelectorType.JsonPath,
Selector = "$.data.fans_num",
Type = FieldType.Int,
}, new Field
{
Name = "cate",
Selector = "$.data.cate_name",
Selectortype = SelectorType.JsonPath
}, new Field
{
Name = "startat",
Selectortype = SelectorType.JsonPath,
Selector = "$.data.start_time",
Type = FieldType.String,
},
},
RepeatWhen = RepeatWhenEver.hour,
RepeatAt = new TimeSpan(0, 20, 0),
};
#endregion
douyu = new CrawlerDotNet.Core.Crawler();
douyu.Setup(c);
douyu.Processor.OnProcessHelperPage = p =>
{
var r = new Regex("href=\"([\\s\\S]*?)\" title=");
var ms = r.Matches(p.Html);
foreach (Match m in ms)
{
douyu.Schduler.AddUrl("http://open.douyucdn.cn/api/RoomApi/room/" + m.Groups[1].Value, p.Request.Deth + 1);
}
p.SkipExtract();
};
douyu.Processor.OnProcessScanPage = p =>
{
var r = new Regex(@"count:(.+),");
var m = r.Match(p.Html);
var count = int.Parse(m.Groups[1].Value.Replace("\"", string.Empty));
#if DEBUG
// count = 0;
#endif
for (int i = 0; i < count; i++)
{
douyu.Schduler.AddUrl($"https://www.douyu.com/directory/all?page={ i + 1}&isAjax=1", PageType.HelperUrl, p.Request.Deth + 1);
}
p.SkipExtract();
};
douyu.Start();
}
static void Main(string[] args)
{
var config = new Config
{
Name = "zhanqi",
ScanUrls = "http://www.zhanqi.tv/api/static/v2.1/live/list/200/1.json",
Fields = new[]
{
new Field
{
Name = "title",
Selectortype = SelectorType.JsonPath,
Selector = "$.title"
},
new Field
{
Name = "username",
Selectortype = SelectorType.JsonPath,
Selector = "$.nickname"
},
new Field
{
Name = "online",
Selectortype = SelectorType.JsonPath,
Selector = "$.online",
Type = FieldType.Int,
},
new Field
{
////*[@id="js-room-anchor-info-area"]/div[2]/div[1]/div/span[1]
Name = "fanscount",
Selectortype = SelectorType.Regex,
Selector = "js-room-follow-num\">([0-9]*)<",
Type = FieldType.Int,
},
new Field
{
Name = "cate",
Selector = "$.newGameName",
Selectortype = SelectorType.JsonPath
},
new Field
{
Name = "childcate",
Selector = "$.gameName",
Selectortype = SelectorType.JsonPath
}
},
RepeatWhen = RepeatWhenEver.hour,
RepeatAt = new TimeSpan(0, 10, 0),
};
crawler = new CrawlerDotNet.Core.Crawler();
var curPage = 1;
crawler.BeforeCrawl = () =>
{
curPage = 1;
};
crawler.Downloader.AfterDownloadPage = p =>
{
//是不是有数据.有数据加入下一个json
var rooms = p.GetJson("$.data.rooms");
if (rooms != "[]")
{
curPage++;
crawler.Schduler.AddUrl($"http://www.zhanqi.tv/api/static/v2.1/live/list/200/{curPage}.json");
}
};
crawler.Processor.OnCustomExtract = p =>
{
var j = JObject.Parse(p.Html);
var jr = JArray.FromObject(j["data"]["rooms"]);
for (int i = 0; i < jr.Count; i++)
{
var exres = new ExtractResults();
var info = jr[i];
foreach (var f in config.Fields)
{
if (f.Name == "fanscount")
{
//请求订阅
var fanspage = crawler.Downloader.DownloaderOnly(new Request(crawler.Schduler)
{
Url =
"https://www.zhanqi.tv" +
info.SelectToken("$.url").ToString()
});
var r = BaseProcessor.DoRegex(fanspage.Html, f);
if (r.Value == "")
{
r.Value = "0";
}
exres.Add(r);
continue;
}
var res = new Result(f.Name, info.SelectToken(f.Selector).ToString());
exres.Add(res);
}
p.Results.Add(exres);
}
};
crawler.Setup(config);
crawler.Start();
Console.WriteLine("end");
Console.ReadKey();
}
