/// <summary>
/// 获取工作描述
/// </summary>
/// <param name="url"></param>
/// <returns></returns>
private Dictionary <string, string> getjobdesc(string url)
{
try
{
IHtmlDocument html = new JumonyParser().LoadDocument(url);
Regex jgr = new Regex("(?<jg>\\d+)");
var desc = html.Find(".main-detail .desc").FirstOrDefault().InnerHtml();
var zz = html.Find(".name a").FirstOrDefault().InnerText();
var number = html.Find(".main-top .number").FirstOrDefault().InnerText();
var price = html.Find(".main-top .detail-row .budgets .budget span").FirstOrDefault().InnerText();
var prices = jgr.Matches(price);
Dictionary <string, string> dic = new Dictionary <string, string>();
dic.Add("desc", desc);
dic.Add("zz", zz);
dic.Add("number", number);
if (prices.Count > 0)
{
for (int i = 0; i < prices.Count; i++)
{
dic.Add("price_" + (i + 1), prices[i].Groups["jg"].Value);
}
}
else
{
dic.Add("price_min", "");
dic.Add("price_max", "");
}
return(dic);
}
catch (Exception e)
{
log.Error(e.Message);
throw;
}
}
private async Task <bool> GetAgentsAsync(Uri cityUri)
{
LogHelper.Info(cityUri.ToString());
var pageSource = await HttpClient.GetStringAsync(cityUri);
while (!pageSource.Contains("查企业"))
{
if (pageSource.StartsWith("<script>window.location"))
{
VertifyCode(new Uri(pageSource.Split("'")[1]));
pageSource = await HttpClient.GetStringAsync(cityUri);
}
else if (pageSource.Contains("小查还没找到数据"))
{
return(false);
}
}
var block = JumonyParser.Parse(pageSource).Find(".m_srchList tbody tr td:nth-child(3)");
foreach (var item in block)
{
await VertifyAsync(item.InnerHtml());
}
if (block.Count() < PageSize)
{
return(false);
}
return(true);
}
static void get_items()
{
var msg_prefix = "-分析列表:";
var y = Encoding.Default.GetByteCount(msg_prefix);
Console.Write(msg_prefix);
var charArr = @"-\|/".ToArray();
var resp = browser.Get("https://materials.ulprospector.com/zh/results?pageSize=100");
JumonyParser parser = new JumonyParser();
var document = parser.Parse(resp);
var total = document.FindFirst(".pagination.pull-right").InnerText().Split(' ').Last().ToInt();
for (int i = 1; i < 2; i++)
{
resp = browser.Get("https://materials.ulprospector.com/zh/results?pageNum=" + i + "pageSize=100");
document = parser.Parse(resp);
var eles = document.Find("table.results tbody tr. ");
foreach (var v in eles)
{
var item = new Item();
item.Url = v.FindFirst(".entry a").Attribute("href").Value();
item.Id = v.FindFirst("input").Attribute("value").Value().ToInt();
Items.Add(item);
}
Console.CursorLeft = 0;
Console.Write(charArr[i % charArr.Length]);
Console.CursorLeft = y;
Console.Write((int)(i * 100 / total) + "%");
}
Console.WriteLine();
}
public override IList <ExamItem> Process(string strResponse, int moduleId)
{
var result = new List <ExamItem>();
if (string.IsNullOrEmpty(strResponse))
{
return(result);
}
var document = new JumonyParser().Parse(strResponse);
//所有题目
var dataTable = document.Descendants("body>div>table").ElementAt(1);
var AllTrs = dataTable.Elements("tr");
var htmlExamItems = AllTrs.Where(x => x.InnerHtml().Contains("【"));
foreach (var item in htmlExamItems)
{
var model = BuildEntity(moduleId, item);
if (model == null)
{
continue;
}
result.Add(model);
}
if (htmlExamItems.Count() > result.Count)
{
string msg = string.Format("Html:[{0}]个，解析：[{1}]个。", result.Count, htmlExamItems.Count());
WriteLog(strResponse, msg);
}
return(result);
}
public Class1[] GetHref_(int page)
{
System.Threading.Thread.Sleep(1 * 1000);
string URL = "http://liaoyang.58.com/ershoufang/0";
Class1[] RTc = new Class1[160];
URL = page == 1 ? URL : URL + "/pn" + page.ToString() + "/";
//抓取关键字对应的url
WebClient client = new WebClient();
client.Encoding = Encoding.UTF8;
string html = client.DownloadString(URL);
IHtmlDocument document = new JumonyParser().Parse(html);
GetUrlText_1(RTc, document);
int b = 0;
b = GetCount(RTc, b);
Class1[] rtcc = new Class1[b];
for (int a = 0; a < rtcc.Length; a++)
{
rtcc[a] = RTc[a];
}
return(rtcc);
}
protected void Page_Load( object sender, EventArgs e )
{
var client = new WebClient();
var html = client.DownloadString( "http://www.cnblogs.com/" );
var parser = new JumonyParser();
var document = parser.Parse( html );
var links = document.Find( "a[href]" );
var baseUrl = new Uri( "http://www.cnblogs.com" );
var data = from hyperLink in links
let url = new Uri( baseUrl, hyperLink.Attribute( "href" ).Value() )
orderby url.AbsoluteUri
select new
{
Url = url.AbsoluteUri,
IsLinkingOut = !url.Host.EndsWith( "cnblogs.com" ),
Target = hyperLink.Attribute( "target" ).Value() ?? "_self"
};
DataList.DataSource = data;
DataBind();
}
public void Test1()
{
var context = new ControllerContext(HttpContext.Request.RequestContext, new TestController());
var result = ViewEngines.Engines.FindView(context, "~/ActionUrlTest/Test1.html", null);
Assert.NotNull(result.View, "找不到视图");
IHtmlDocument document;
using (var writer = new StringWriter())
{
result.View.Render(new ViewContext(context, result.View, new ViewDataDictionary(), new TempDataDictionary(), writer), writer);
document = new JumonyParser().Parse(writer.ToString());
}
var link = document.FindFirst("a");
Assert.NotNull(link);
Assert.AreEqual(link.Attribute("href").Value(), "/TestController/TestAction?arg=args");
}
protected void Page_Load(object sender, EventArgs e)
{
var client = new WebClient();
var html = client.DownloadString("http://www.cnblogs.com/");
var parser = new JumonyParser();
var document = parser.Parse(html);
var links = document.Find("a[href]");
var baseUrl = new Uri("http://www.cnblogs.com");
var data = from hyperLink in links
let url = new Uri(baseUrl, hyperLink.Attribute("href").Value())
orderby url.AbsoluteUri
select new
{
Url = url.AbsoluteUri,
IsLinkingOut = !url.Host.EndsWith("cnblogs.com"),
Target = hyperLink.Attribute("target").Value() ?? "_self"
};
DataList.DataSource = data;
DataBind();
}
static void Main(string[] args)
{
var id = Guid.NewGuid();
var path = Path.Combine(tempDirectory, id.ToString());
Directory.CreateDirectory(path);
SmtpClient smtp = new SmtpClient();
smtp.EnableSsl = false;
smtp.DeliveryMethod = SmtpDeliveryMethod.SpecifiedPickupDirectory;
smtp.PickupDirectoryLocation = path;
var parser = new JumonyParser();
var document = parser.LoadDocument("http://blog.sina.com.cn/s/blog_4701280b010183ny.html");
MailMessage message = CreateMail(document);
smtp.Send(message);
var directory = new DirectoryInfo(path);
var file = directory.GetFiles().Single();
file.MoveTo(Path.Combine(tempDirectory, id.ToString() + ".mht"));
directory.Delete(true);
}
public sys_job GetJobInfo(string url)
{
try
{
IHtmlDocument html = new JumonyParser().LoadDocument(url, Encoding.UTF8);
int pos = html.InnerHtml().IndexOf("product-info-summary");
if (pos < 0)
{
return(new sys_job());
}
string jobtitle = html.Find(".product-info-summary .row h4").FirstOrDefault().InnerText();
string author = html.Find(".product-info-summary .row small").FirstOrDefault().InnerText().Replace("发布者：", "");
string price = html.Find(".product-info-summary .row .p-desc").FirstOrDefault().InnerText().Replace(" 预算： ", "");
string rq = html.Find("#p-other ul li:first-child").FirstOrDefault().InnerText();
string xqh = html.Find("#p-other ul li:nth-child(3)").SingleOrDefault().InnerText();
string describe = html.Find("#wrap").SingleOrDefault().InnerHtml();
string t = @"<script\b[^<]*(?:(?!<\/script>)<[^<]*)*<\/script>";
describe = Regex.Replace(describe, t, "");
describe = Regex.Replace(describe, "<.*?>", "").Replace("\t", "").Replace("\r", "").Replace("\n", "");
return(new sys_job
{
title = jobtitle,
author = author,
desc = describe,
rq = rq,
tag = xqh,
price_min = price
});
}
catch (Exception e)
{
log.Error(url + "----" + e.Message);
return(new sys_job());
}
}
/// <summary>
/// 核心加载
/// </summary>
/// <param name="context"></param>
/// <param name="url"></param>
/// <param name="htmlContent">HTML内容</param>
/// <param name="result">目标结果</param>
/// <returns></returns>
protected override bool LoadCore(HttpContext <string> context, string url, string htmlContent, ResourceSearchInfo result)
{
var parser = new JumonyParser();
var doc = parser.Parse(htmlContent, new Uri(ReferUrlPage));
var node = doc.Find("#archiveResult tr").Skip(1);
foreach (var row in node)
{
var title = row.FindFirstOrDefault("td.name")?.InnerText();
//var size = row.FindFirstOrDefault("td.size")?.InnerText();
var date = row.FindFirstOrDefault("td.date")?.InnerText()?.ToDateTimeNullable();
var has = Regex.Match(row.FindFirstOrDefault("td.action a:nth-child(1)").Attribute("href").AttributeValue, @"/([a-z\d]{40})", RegexOptions.IgnoreCase).GetGroupValue(1);
var item = CreateResourceInfo(has, title);
//item.DownloadSize = size;
item.UpdateTime = date;
result.Add(item);
}
var pager = doc.FindFirstOrDefault("div.pagination");
result.HasPrevious = pager?.FindFirstOrDefault("*:first-child")?.Name == "a";
result.HasMore = pager?.FindFirstOrDefault("*:last-child")?.Name == "a";
return(base.LoadCore(context, url, htmlContent, result));
}
public List <string> GetPageUrl(string url)
{
try
{
url = rooturl;
List <string> list = new List <string>();
IHtmlDocument html = new JumonyParser().LoadDocument(url, Encoding.UTF8);
string entityCount = html.Find("#entityCount").SingleOrDefault().Attribute("value").Value();
string maxEntityPerPage = html.Find("#maxEntityPerPage").SingleOrDefault().Attribute("value").Value();
string maxPagePerRow = html.Find("#maxPagePerRow").SingleOrDefault().Attribute("value").Value();
string pageCount = html.Find("#pageCount").SingleOrDefault().Attribute("value").Value();
string currentPage = html.Find("#currentPage").SingleOrDefault().Attribute("value").Value();
string currentPageRow = html.Find("#currentPageRow").SingleOrDefault().Attribute("value").Value();
string pageRowCount = html.Find("#pageRowCount").SingleOrDefault().Attribute("value").Value();
Int32 count = Convert.ToInt32(pageCount);
Int32 current = Convert.ToInt32(currentPage);
string query = string.Empty;
for (Int32 i = current; i <= count; i++)
{
query = "entityCount=" + entityCount + "&maxEntityPerPage=" + maxEntityPerPage +
"&maxPagePerRow=" + maxPagePerRow + "&pageCount=" + pageCount +
"¤tPage=" + i + "¤tPageRow=" + currentPageRow +
"&pageRowCount=" + pageRowCount +
"&cBudget=0-1000000000&budgetTo=&statusBy=&categoryBy=&typeBy=&typeName=&orderByClause=a.c_postDate+desc";
string pageurl = url + "?" + query;
list.Add(pageurl);
}
return(list);
}
catch (Exception e)
{
log.Error(e.Message);
return(new List <string>());
}
}
public void GetPageUrlToRedis(string url)
{
try
{
IHtmlDocument html = new JumonyParser().LoadDocument(url);
var pagelist = html.Find(".co_content8 .x a");
var last_index = pagelist.Count() - 3;
var end_index = pagelist.Count() - 1;
var last_url = pagelist.ToList()[last_index].Attribute("href").Value();
var end_txt = pagelist.ToList()[end_index].InnerText();
this.isendpage = end_txt.IndexOf("末页") >= 0 ? false : true;
looplast_index = isendpage ? pagelist.Count() : pagelist.Count() - 1;
var last_full_url = page_baseurl + last_url;
int i = 0;
i = isfirstpage ? 0 : 2;
for (; i < looplast_index; i++)
{
IHtmlElement item = pagelist.ToList()[i];
string pageurl = item.Attribute("href").Value();
string pagefullurl = page_baseurl + pageurl;
db.SortedSetAdd("filmpageurl", pagefullurl, (double)index++);
if (i == last_index && !isendpage)
{
isfirstpage = false;
GetPageUrl(pagefullurl);
}
}
}
catch (Exception e)
{
log.Error(e.Message);
throw;
}
}
public override IList <ExamItem> Process(string strResponse, int moduleId)
{
var result = new List <ExamItem>();
if (string.IsNullOrEmpty(strResponse))
{
return(result);
}
var document = new JumonyParser().Parse(strResponse);
//所有题目
var htmlExamItems = document.Descendants(@"div.st");
foreach (var item in htmlExamItems)
{
var model = BuildEntity(moduleId, item);
if (model == null)
{
continue;
}
result.Add(model);
}
if (htmlExamItems.Count() > result.Count)
{
string msg = string.Format("Html:[{0}]个，解析：[{1}]个。", result.Count, htmlExamItems.Count());
WriteLog(strResponse, msg);
}
return(result);
}
public override List <SubjectModule> Process(string strResponse, int moduleId)
{
var list = new List <SubjectModule>();
var document = new JumonyParser().Parse(strResponse);
var trs = document.Descendants("tbody>tr");
foreach (IHtmlElement tr in trs)
{
string title = tr.FindFirst("td").InnerText();
string href = tr.FindLast("td a").Attribute("href").Value();
list.Add(new SubjectModule()
{
Id = GetId(moduleId, title),
Handler = "SweetFly.Job.Handler.NormalHandler,SweetFly.Job",
HtmlDataSource = new HtmlDataSource()
{
Encoding = "GB2312",
Uri = @"http://learning.cmr.com.cn/student/acourse/HomeworkCenter/" + href
}
});
}
Console.WriteLine("{0} - {1}", trs.Count(), list.Count);
return(list);
}
public void SetStyleTest()
{
var element = new JumonyParser().Parse("<div></div>").Elements().First();
element.Style("display", "none");
Assert.AreEqual(element.Attribute("style").Value(), "display:none", ".Style( name, value ) 测试不通过");
element.Style().SetValue("color", "red");
Assert.AreEqual(element.Attribute("style").Value(), "display:none;color:red", ".Style().SetValue( name, value ) 测试不通过");
element.Style().SetValue("display", "block");
Assert.AreEqual(element.Attribute("style").Value(), "display:block;color:red", ".Style().SetValue( name, value ) 测试不通过");
element.Style().SetValue("display", null);
Assert.AreEqual(element.Attribute("style").Value(), "color:red", ".Style().SetValue( name, null ) 测试不通过");
element.Style().Clear();
Assert.AreEqual(element.Attribute("style").Value(), "", ".Style().Clear() 测试不通过");
element.Style().SetValue("padding", "10px");
Assert.AreEqual(element.Style().GetValue("padding-left"), "10px", "shorthand 展开测试不通过");
element.Style().SetValue("padding-left", "0px");
Assert.AreEqual(element.Style().GetValue("padding-left"), "0px", "shorthand 展开测试不通过");
Assert.AreEqual(element.Style().GetValue("padding-top"), "10px", "shorthand 展开测试不通过");
element.Style().SetValue("margin", "5px");
Assert.AreEqual(element.Style().GetValue("margin-left"), "5px", "margin shorthand 展开测试不通过");
}
static void GetDataByJumony()
{
Console.WriteLine("开始跑数据");
var db = DB.GetInstance();
for (var i = 1; i < 51; i++)
{
try
{
var path = "http://cn.coovee.com/company/s1.html?p=" + i.ToString();
IHtmlDocument source = new JumonyParser().LoadDocument(path, System.Text.Encoding.GetEncoding("utf-8"));
var company = source.Find(".company-l-item");
Console.WriteLine(company.Count());
foreach (var item in company)
{
try
{
var a = item.Find(".dt h4 a").FirstOrDefault();
Console.WriteLine("公司名=" + a.InnerText());
var span = item.Find(".dt p span").FirstOrDefault();
Console.WriteLine("联系人=" + span.InnerText());
var zhuying = item.Find(".dd p").FirstOrDefault();
Console.WriteLine("主营=" + zhuying.InnerText().Replace("主营：", ""));
var url = item.Find(".dd ul li").Last().Find("a").FirstOrDefault().Attribute("href").Value();
Console.WriteLine(url);
System.Threading.Thread.Sleep(1000);
source = new JumonyParser().LoadDocument(url, System.Text.Encoding.GetEncoding("utf-8"));
var tel = source.Find(".company-contact-info p").ToList <IHtmlElement>()[1].InnerText();
Console.WriteLine("电话=" + tel.Replace("联系电话：", ""));
var adress = source.Find(".company-contact-info p").ToList <IHtmlElement>()[3].InnerText();
Console.WriteLine("地址=" + adress.Replace("公司地址：", "").Replace(" ", ""));
COMPANY comp = new COMPANY()
{
ID = System.Guid.NewGuid().ToString(),
COMPANY_NAME = a.InnerText(),
ADDRESS = adress.Replace("公司地址：", "").Replace(" ", ""),
LINK_MAN = span.InnerText(),
SALE_PRODUCT = zhuying.InnerText().Replace("主营：", ""),
TEL = tel.Replace("联系电话：", "")
};
db.Insertable <COMPANY>(comp).ExecuteCommand();
System.Threading.Thread.Sleep(2000);
}
catch
{ }
}
System.Threading.Thread.Sleep(2000);
}
catch
{ }
}
}
private void BaiXingNewHTMLhreper()
{
string url = "http://liaoyang.baixing.com/qiufang/";
string ThisHtml = BXGetHTMLstr(url);
using (var ctx = new oaEntities())
{
DateTime DBtime = Convert.ToDateTime(DateTime.Now.Year.ToString() + "-" + DateTime.Now.Month.ToString() + "-" + DateTime.Now.Day.ToString());
IHtmlDocument document = new JumonyParser().Parse(ThisHtml);
IEnumerable<IHtmlElement> result = document.Find(".media-body-title");
List<Class1> Ncss = new List<Class1>();
foreach (var item in result)
{
Class1 Class1 = new Class1();
item.FindFirst("a");
Class1.href = item.Exists("a") ? item.FindFirst("a").Attribute("href").Value() : string.Empty;
Class1.TextName = MainWindow.GetN_value(item, "a");
//开始读取子连接
#region 读取子连接
string ThisZ = BXGetHTMLstr(Class1.href);
IHtmlDocument document_1 = new JumonyParser().Parse(ThisZ);
IEnumerable<IHtmlElement> restime = document_1.Find("div>.viewad-topMeta");
foreach (var tm in restime)
{
}
#endregion
Ncss.Add(Class1);
}
}
//media - body - title
}
void webbrowser2_LoadCompleted(object sender, NavigationEventArgs e)
{
mshtml.HTMLDocument mhtml = (mshtml.HTMLDocument)Webbrowser2.Document;
Webbrowser2.Navigate("http://#");
string html = mhtml.body.innerHTML;
IHtmlDocument document_1 = new JumonyParser().Parse(html);
}
/// <summary>
/// 获取每一次请求的Ids
/// </summary>
/// <param name="address"></param>
/// <returns></returns>
public Hashtable GetList(string address = "")
{
var listAddress = new List <string>();
// 获取返回信息
var result = JsonConvert.DeserializeObject <DuoWan.DwResult>(GetUrlString(address));
var document = new JumonyParser().Parse(result.html);
var cells = document.Find("li>a");
// 开始遍历
foreach (var li in cells)
{
var detailUrl = li.Attribute("href").Value();
// 获取id
listAddress.Add((from each in detailUrl where each.ToString().ToInt(-1) > 0 select each).Join(""));
}
var o = new Hashtable()
{
{ "more", result.more },
{ "ids", listAddress }
};
return(o);
}
public static void getContent(string id)
{
HttpClient httpClient = new HttpClient();
try
{
httpClient
.PostAsync("http://hd.huachuan.gov.cn/aspx/gkml_view.aspx?id=" + id, null)
.ContinueWith((postTask) =>
{
HttpResponseMessage response = postTask.Result;
response.Content.ReadAsStringAsync().ContinueWith((readTask) =>
{
try
{
IHtmlDocument source = new JumonyParser().Parse(readTask.Result);
string content = source.FindSingle(".zwnr").InnerHtml();
string sql = string.Format("update t_spider_zwgk t set t.content='{0}' where t.id={1}", content, id);
int count = DbHelperMySQL.ExecuteSql(sql);
}
catch (Exception e)
{
Debug.WriteLine("----->【" + id + "】内容存储异常<-----：" + e);
}
});
});
}
catch (Exception e)
{
Debug.WriteLine("----->【" + id + "】内容存储异常<-----：" + e);
}
}
public static void getByPage(int page)
{
string url = "";
if (page == 0)
{
url = "http://www.huachuan.gov.cn/zwgk/xxgksyzl/fgfgg/index.html";
}
else
{
string p = page.ToString();
if (page < 10)
{
p = "0" + p;
}
url = "http://www.huachuan.gov.cn/system/more/zwgk/xxgksyzl/fgfgg/index/page_" + p + ".html";
}
string pageStr = Util.getHtmlStr(url, Encoding.Default);
IHtmlDocument source = new JumonyParser().Parse(pageStr);
var items = source.Find(".listmain ul li");
foreach (var item in items)
{
string id = "0";
string path = item.FindFirst("div").FindSingle("a").Attribute("href").Value();
string title = item.FindFirst("div").FindSingle("a").InnerText();
string time = item.FindFirst("div").NextElement().InnerText();
if (path.StartsWith("http"))
{
id = new Random().Next(1000000, 9999999).ToString();
string sql = string.Format("select count(*) from t_spider_bslc t where t.title='{0}'", title);
int count = Convert.ToInt32(DbHelperMySQL.GetSingle(sql));
if (count == 0)
{
sql = string.Format("insert into t_spider_bslc(id,title,time,path) values({0},'{1}','{2}','{3}')", id, title, time, path);
count = DbHelperMySQL.ExecuteSql(sql);
if (count == 1)
{
getContent(path);
}
}
}
else
{
id = path.Split('/')[3].Split('.')[0];
string sql = string.Format("select count(*) from t_spider_bslc t where t.id={0}", id);
int count = Convert.ToInt32(DbHelperMySQL.GetSingle(sql));
if (count == 0)
{
sql = string.Format("insert into t_spider_bslc(id,title,time,path) values({0},'{1}','{2}','{3}')", id, title, time, path);
count = DbHelperMySQL.ExecuteSql(sql);
if (count == 1)
{
getContent(path);
}
}
}
}
}
public void VisibleTest()
{
var document = new JumonyParser().LoadDocument(Path.Combine(Environment.CurrentDirectory, "StyleTest1.html"));
document.DataBind(null);
Assert.AreEqual(document.Find(".invisible").Count(), 0);
}
void Webbrowser2_LoadCompleted(object sender, NavigationEventArgs e)
{
mshtml.HTMLDocument mhtml = (mshtml.HTMLDocument)Webbrowser2.Document;
string html = mhtml.body.innerHTML;
IHtmlDocument document = new JumonyParser().Parse(html);
IEnumerable <IHtmlElement> result = document.Find("ul").Where(x => x.Identity() == "houselist-mod-new");
IEnumerable <IHtmlElement> result_li = result.Find("li");
IEnumerable <IHtmlElement> result_li1 = document.Find("li>.list-item");
}
public void Test1()
{
var document = new JumonyParser().LoadDocument(Path.Combine(Environment.CurrentDirectory, "Test1.html"));
HtmlBinding.Create(document, null).DataBind();
Assert.AreEqual(document.FindFirst("title").InnerHtml(), "Test Title abc text", "对 title 元素内容的文本替换测试失败");
}
public static void getByPage(string cookie, string viewstate, int page)
{
HttpClient httpClient = new HttpClient();
HttpContent postContent = new FormUrlEncodedContent(new Dictionary <string, string>()
{
{ "__VIEWSTATE", viewstate },
{ "__VIEWSTATEGENERATOR", "7BE8FDE8" },
{ "__EVENTTARGET", "AspNetPager1" },
{ "__EVENTARGUMENT", page.ToString() },
{ "_keywords", "" },
{ "AspNetPager1_input", "1" },
});
httpClient
.PostAsync("http://hd.huachuan.gov.cn/aspx/gkml_list.aspx", postContent)
.ContinueWith((postTask) =>
{
HttpResponseMessage response = postTask.Result;
response.Content.ReadAsStringAsync().ContinueWith((readTask) =>
{
//Debug.WriteLine(readTask.Result);
IHtmlDocument source = new JumonyParser().Parse(readTask.Result);
var itemCount = source.Find(".listbox").Count();
for (int i = 1; i <= itemCount; i++)
{
try
{
string id = source.FindSingle("#four" + i).Attribute("href").Value().Split('=')[1];
string author = source.Find("#con_four_" + i).Find(".li1").Last().InnerText().Replace("发布机构：", "");
string time = source.Find("#con_four_" + i).Find(".li2").Last().InnerText().Replace("发文日期：", "");
string title = source.Find("#con_four_" + i).Find(".infoname").First().InnerText().Replace("名称：", "");
//判断第一条是否存在，如果存在，则说明新闻一直未更新，不需要继续下去了
//可以使用下边逻辑，continue换成return
//判断是否存在
string sql = string.Format("select count(*) from t_spider_zwgk t where t.id={0}", id);
int count = Convert.ToInt32(DbHelperMySQL.GetSingle(sql));
if (count > 0)
{
//continue;
return;
}
//不存在，插入数据库
sql = string.Format("insert into t_spider_zwgk(id,title,time,author) values({0},'{1}','{2}','{3}')", id, title.Replace('\'', '"'), time, author);
count = DbHelperMySQL.ExecuteSql(sql);
if (count == 1)
{
getContent(id);
}
}
catch (Exception e)
{
Debug.WriteLine("----->【" + page + "." + i + "】新闻创建异常<-----：" + e);
}
}
});
});
}
static void Main(string[] args)
{
//抓取小说
FileStream fs = null;
if (File.Exists("D\\c.txt"))
{
fs = new FileStream("D:\\c.txt", FileMode.Append);
}
else
{
fs = new FileStream("D:\\c.txt", FileMode.Create);
}
var sw = new StreamWriter(fs, Encoding.UTF8);
var baseUrl = "http://www.42xs.com/read/0/404/";
var nextUrl = "171271.html";
var url = "";
var txt = "";
var title = "";
while (nextUrl != "")
{
try
{
url = baseUrl + nextUrl;
var doc = new JumonyParser().LoadDocument(url);
var titleDom = doc.FindFirst("#center > div.title > h1");
title = titleDom.InnerText();
var dom = doc.FindFirst("#content");
txt = dom.InnerText();
var domNext = doc.FindFirst("#container > div:nth-child(3) > div > div.jump > a:nth-child(6)");
nextUrl = domNext.Attribute("href").Value();
}
catch
{
Console.WriteLine("{0}没有成功", url);
nextUrl = "";
}
Console.WriteLine(title);
sw.WriteLine("");
sw.WriteLine(title);
sw.WriteLine("");
sw.WriteLine(txt);
}
Console.Write("The End. Press any key to exit...");
Console.ReadKey();
sw.Close();
fs.Close();
}
public void css_class_has_hyphen()
{
var html = "<div class=\"css-class\"></div>";
var htmlParser = new JumonyParser();
var doc = htmlParser.Parse(html);
var css_class = doc.Find(".css-class");
Assert.AreEqual(1, css_class.Count());
}
public void SpecificationTest7()
{
var document = new JumonyParser().LoadDocument(Path.Combine(Environment.CurrentDirectory, "SpecificationTest7.html"));
var link = document.FindFirstOrDefault("a");
Assert.IsNotNull(link, "属性或内容包含特殊字符的标签解析失败");
Assert.AreEqual(link.Attribute("href").Value(), "#", "属性内容包含 > 时解析失败。");
Assert.AreEqual(link.Attribute("title").Value(), "this is a <a> tag", "属性内容包含 > 时解析失败。");
//Assert.AreEqual( link.Elements().Count(), 0, "错误的解析了以特殊字符为标签名的标签" );
}
public void SpecificationTest5()
{
var document = new JumonyParser().LoadDocument(Path.Combine(Environment.CurrentDirectory, "SpecificationTest5.html"));
//Assert.AreEqual( document.DocumentDeclaration, "<!DOCTYPE html>", "HTML 声明解析失败" );
var specials = document.DescendantNodes().OfType <IHtmlSpecial>().ToArray();
Assert.AreEqual(specials.Count(), 4, "特殊标签解析数量不对");
}
public void CompileTest()
{
var parser = new JumonyParser();
var document = parser.LoadDocument(Path.Combine(Environment.CurrentDirectory, "Test1.html"));
var method = document.Compile();
var document2 = method(parser.DomProvider);
Assert.IsTrue(document.DescendantNodes().SequenceEqual(document2.DescendantNodes(), new DomNodeComparer()), "编译还原测试失败");
}
