public void ParseProductTest()
{
SpiderRrxf target = new SpiderRrxf(); // TODO: 初始化为适当的值
ATag a = new ATag(); // TODO: 初始化为适当的值
a.Link = "http://rrxf.cn/product/lookcp.php?cpid=5041";
target.ParseProduct(a);
}
public void ParseProductsTest()
{
SpiderRrxf target = new SpiderRrxf(); // TODO: 初始化为适当的值
ATag a = new ATag(); // TODO: 初始化为适当的值
a.Link = "http://rrxf.cn/product/index.php?cplm=-252-";
target.ParseProducts(a);
}
public void ParseProductTest()
{
SpiderRrxf target = new SpiderRrxf(); // TODO: 初始化为适当的值
ATag a = new ATag(); // TODO: 初始化为适当的值
a.Link = "http://www.rrxf.cn/product/lookcp.php?cpid=4163";
target.ParseProduct(a);
Assert.Inconclusive("无法验证不返回值的方法。");
}
public void ParseProductsTest()
{
SpiderRrxf target = new SpiderRrxf(); // TODO: 初始化为适当的值
ATag a = new ATag(); // TODO: 初始化为适当的值
a.Link = "http://www.rrxf.cn/product/index.php?keyword=荷叶边蕾丝性感吊带睡裙";
target.ParseProducts(a);
Assert.Inconclusive("无法验证不返回值的方法。");
}
public void GetJobInfoParserTest()
{
Category catalog = new Category(); // TODO: 初始化为适当的值
Get1010Jobs target = new Get1010Jobs(catalog); // TODO: 初始化为适当的值
string url = "http://sh.1010jz.com/html/shanghai/011_1522688.html"; // TODO: 初始化为适当的值
Job expected = null; // TODO: 初始化为适当的值
Job actual;
ATag node = new ATag();
node.Link = url;
actual = target.GetJobInfoParser(url);
Console.WriteLine("title:{0}", actual.title);
Console.WriteLine("email:{0}", actual.poster_email);
}
private void ProcessLinkTag(ATag obTag)
{
LinkData obLinkData = new LinkData(this.m_obPageData, obTag);
m_obPageData.m_Outlinks.Add(obLinkData);
}
/// <summary>
///
/// </summary>
/// <param name="obPage"></param>
/// <param name="obTag"></param>
public LinkData(PageData obPage, ATag obTag)
:this(obPage)
{
this.ConvertFromLinkTag(obTag);
}
private void ConvertFromLinkTag(ATag obTag)
{
if (null == obTag)
{
throw new ArgumentNullException("obTag", "Null ATag object specified");
}
base.ConvertFromTag(obTag);
this.m_strText = obTag.LinkText;
this.m_strUrl = obTag.Link;
this.m_LinkType = LinkType.Outlink;
if (obTag.MailLink)
{
this.m_LinkType = LinkType.MailLink;
}
else if (obTag.FTPLink)
{
this.m_LinkType = LinkType.Ftp;
}
else if (obTag.IRCLink)
{
this.m_LinkType = LinkType.Other;
}
}
public void ParseProducts(ATag a)
{
string html = GetHtml(a.Link.Replace("../", "http://rrxf.cn/"));
Lexer lexer = new Lexer(html);
Parser parser = new Parser(lexer);
NodeFilter nav = new HasAttributeFilter("class", "photoyi");
NodeList navNodes = parser.Parse(nav);
if (navNodes == null)
return;
int length = navNodes.Count;
for (int i = 0; i < length; i++)
{
ATag link = ParseProductUrl(navNodes[i].ToHtml());
Console.WriteLine(link.Link);
ParseProduct(link);
}
}
public void ParseProduct(ATag a)
{
string html = GetHtml(a.Link);
Lexer lexer = new Lexer(html);
Parser parser = new Parser(lexer);
NodeFilter productArea = new HasAttributeFilter("id", "productyou");
NodeList nodes = parser.ExtractAllNodesThatMatch(productArea);
ParseProductTitle(nodes);
ParseProductShowPhoto(nodes);
ParseProductDemoPhoto(nodes);
ParsePorductDescribe(nodes);
NodeFilter productAttributeArea = new HasAttributeFilter("class", "chans");
NodeList productAttributeAreaNodes = nodes.ExtractAllNodesThatMatch(productAttributeArea,true);
NodeFilter productAttributes = new HasAttributeFilter("class", "cph");
NodeList productAttributeNodes = nodes.ExtractAllNodesThatMatch(productAttributes, true);
int length = productAttributeNodes.Count;
for (int i = 0; i < length; i++)
{
INode n = productAttributeNodes[i].Children[0];
string t =n.ToPlainTextString();
if (Regex.Match(t, @"^\s{0,}颜色", RegexOptions.IgnoreCase).Success)
{
ParseProductColors(n);
}
Console.WriteLine();
}
}
