/// <summary>
/// 详情
/// </summary>
public void RunArticleDetail()
{
var rule = new SpliderRule()
{
ContentXPath = "//div[@id='post_detail']",
EachXPath = "",
Url = " https://www.cnblogs.com/fancunwei/p/9581168.html",
RuleFields = new List <RuleField>()
{
new RuleField()
{
DisplayName = "标题", XPath = "*//div[@class='post']//a[@id='cb_post_title_url']", IsFirstInnerText = true
},
new RuleField()
{
DisplayName = "详情", XPath = "*//div[@class='postBody']//div[@class='blogpost-body']", Attribute = "", IsFirstInnerText = false
}
}
};
var splider = new ArticleSplider();
var list = splider.GetByRule(rule);
foreach (var item in list)
{
var msg = string.Empty;
item.Fields.ForEach(M =>
{
Console.WriteLine($"{M.DisplayName}:{M.Value}");
});
Console.WriteLine(msg);
}
}
public AppInfo WandoujiaAppDetailUnitTest()
{
var rule = new SpliderRule()
{
ContentXPath = "//dl[@class='infos-list']",
EachXPath = "",
Url = "",
RuleFields = new List <RuleField>()
{
new RuleField()
{
DisplayName = "Company", XPath = "//span[@class='dev-sites']", Attribute = "", IsFirstInnerText = false
},
//new RuleField(){ DisplayName="Name",XPath="div[@class='app-desc']/h2/a", IsFirstInnerText=true },
//new RuleField(){ DisplayName="URL",XPath="a[@class='detail-check-btn']",Attribute="href", IsFirstInnerText=false },
}
};
var splider = new AppsSplider();
string fileName = @"E:\DevProjects\html-agility-pack\CsharpFanDemo-fork\CsharpFanDemo\Demo.DataSplider\testfiles\detailpage.html";
List <SpliderContent> list = splider.GetByRuleFromFile(rule, fileName);
AppInfo appInfo = new AppInfo();
foreach (var item in list)
{
var msg = string.Empty;
int id = 0;
item.Fields.ForEach(M =>
{
switch (M.DisplayName)
{
case "Company":
appInfo.Company = M.Value;
break;
case "Description":
appInfo.Description = M.Value;
break;
case "URL":
appInfo.URL = M.Value;
break;
default:
break;
}
msg += $"{M.DisplayName}:{M.Value} ";
});
Console.WriteLine(msg);
}
return(appInfo);
}
/// <summary>
/// 天气预报 - 15 days
/// </summary>
public void RunWeather15Days()
{
var rule = new SpliderRule()
{
ContentXPath = "//div[@id='15d']",
EachXPath = "*//li",
Url = "http://www.weather.com.cn/weather15d/101020100.shtml",
RuleFields = new List <RuleField>()
{
new RuleField()
{
DisplayName = "日期", XPath = "span[@class='time']", IsFirstInnerText = true
},
new RuleField()
{
DisplayName = "天气", XPath = "span[@class='wea']", Attribute = "", IsFirstInnerText = false
},
new RuleField()
{
DisplayName = "区间", XPath = "span[@class='tem']", Attribute = "", IsFirstInnerText = false
},
new RuleField()
{
DisplayName = "风向", XPath = "span[@class='wind']", Attribute = "", IsFirstInnerText = false
},
new RuleField()
{
DisplayName = "风力", XPath = "span[@class='wind1']", Attribute = "", IsFirstInnerText = false
},
}
};
var splider = new ArticleSplider();
var list = splider.GetByRule(rule);
foreach (var item in list)
{
var msg = string.Empty;
item.Fields.ForEach(M =>
{
msg += $"{M.DisplayName}:{M.Value} ";
});
Console.WriteLine(msg);
}
}
/// <summary>
/// 天气预报 - 7 days
/// </summary>
public void RunWeather7Days()
{
var rule = new SpliderRule()
{
//ContentXPath = "//div[@id='7d']",
ContentXPath = "//ul[@class='t clearfix']",
EachXPath = "li[@class]",
Url = "http://www.weather.com.cn/weather/101250101.shtml", //长沙7天天气
RuleFields = new List <RuleField>()
{
new RuleField()
{
DisplayName = "日期", XPath = "h1", IsFirstInnerText = true
},
new RuleField()
{
DisplayName = "天气", XPath = "p[@class='wea']", Attribute = "", IsFirstInnerText = true
},
new RuleField()
{
DisplayName = "高", XPath = "p[@class='tem']/span", Attribute = "", IsFirstInnerText = true
},
new RuleField()
{
DisplayName = "低", XPath = "p[@class='tem']/i", Attribute = "", IsFirstInnerText = true
},
//new RuleField(){ DisplayName="风向",XPath="p[@class='win']",Attribute="", IsFirstInnerText=false },
//new RuleField(){ DisplayName="风力",XPath="p[@class='win']",Attribute="", IsFirstInnerText=false },
}
};
var splider = new ArticleSplider();
var list = splider.GetByRule(rule);
foreach (var item in list)
{
var msg = string.Empty;
item.Fields.ForEach(M =>
{
msg += $"{M.DisplayName}:{M.Value} ";
});
Console.WriteLine(msg);
}
}
/// <summary>
///
/// </summary>
public void RunArticleRule()
{
var postitembodyXPath = "div[@class='post_item_body']//";
var postitembodyFootXPath = postitembodyXPath + "div[@class='post_item_foot']//";
var rule = new SpliderRule()
{
ContentXPath = "//div[@id='post_list']",
EachXPath = "div[@class='post_item']",
Url = "https://www.cnblogs.com",
RuleFields = new List <RuleField>()
{
new RuleField()
{
DisplayName = "推荐", XPath = "*//span[@class='diggnum']", IsFirstInnerText = true
},
new RuleField()
{
DisplayName = "标题", XPath = postitembodyXPath + "a[@class='titlelnk']", IsFirstInnerText = true
},
new RuleField()
{
DisplayName = "URL", XPath = postitembodyXPath + "a[@class='titlelnk']", Attribute = "href", IsFirstInnerText = true
},
new RuleField()
{
DisplayName = "简要", XPath = postitembodyXPath + "p[@class='post_item_summary']", IsFirstInnerText = true
},
new RuleField()
{
DisplayName = "作者", XPath = postitembodyFootXPath + "a[@class='lightblue']", IsFirstInnerText = true
},
new RuleField()
{
DisplayName = "作者URL", XPath = postitembodyFootXPath + "a[@class='lightblue']", Attribute = "href", IsFirstInnerText = true
},
new RuleField()
{
DisplayName = "讨论数", XPath = "span[@class='article_comment']", IsFirstInnerText = true, InnerTextRegex = @"[^0-9]+"
},
new RuleField()
{
DisplayName = "阅读数", XPath = postitembodyFootXPath + "span[@class='article_view']", IsFirstInnerText = true, InnerTextRegex = @"[^0-9]+"
},
}
};
var splider = new ArticleSplider();
var list = splider.GetByRule(rule);
foreach (var item in list)
{
var msg = string.Empty;
item.Fields.ForEach(M =>
{
if (M.DisplayName != "简要" && !M.DisplayName.Contains("URL"))
{
msg += $"{M.DisplayName}:{M.Value}";
}
});
Console.WriteLine(msg);
}
}
public List <AppInfo> WandoujiaAppList(string content)
{
var rule = new SpliderRule()
{
//ContentXPath = "//ul[@id='j-tag-list']",//这是从网页上找。
ContentXPath = @"/", //这是从网页上找。
EachXPath = "li[@class='card']",
Url = "", //豌豆荚 某个分类下的应用列表。url应该作为一个参数输入。
RuleFields = new List <RuleField>()
{
new RuleField()
{
DisplayName = "Id", XPath = "a[@class='detail-check-btn']", Attribute = "data-app-id", IsFirstInnerText = false
},
new RuleField()
{
DisplayName = "Name", XPath = "div[@class='app-desc']/h2/a", IsFirstInnerText = true
},
new RuleField()
{
DisplayName = "URL", XPath = "a[@class='detail-check-btn']", Attribute = "href", IsFirstInnerText = false
},
//new RuleField(){ DisplayName="图标",XPath="p[@class='tem']/span",Attribute="", IsFirstInnerText=true },
new RuleField()
{
DisplayName = "InstallCount", XPath = "div[@class='app-desc']/div[@class='meta']/span[@class='install-count']", Attribute = "", IsFirstInnerText = true
},
}
};
var splider = new AppsSplider();
List <SpliderContent> list = splider.GetByRule(rule, content);
//打印出来
List <AppInfo> appList = new List <AppInfo>();
//AppList appList = new AppList();
foreach (var item in list)
{
AppInfo appInfo = new AppInfo();
var msg = string.Empty;
int id = 0;
item.Fields.ForEach(M =>
{
switch (M.DisplayName)
{
case "Id":
appInfo.Id = M.Value;
break;
case "Name":
appInfo.Name = M.Value;
break;
case "URL":
appInfo.URL = M.Value;
break;
default:
break;
}
appList.Add(appInfo);
msg += $"{M.DisplayName}:{M.Value} ";
});
Console.WriteLine(msg);
}
return(appList);
}