public void ReadRule()
{
IPersistence.IRule rule = new Persistence.Rule();
rule.PersistencePath = @"d:\";
#region 写一个xml
#region 模拟2个ruleset
//第1个set
BaseRule rule1 = new BeginEndRule("<div id=\"jqlast_maincontent\" class=\"jqlast_main_title\"><h1>", "</h1>", false, false, true, true);
rule1.RuleNo = 10;
rule1.Name = "标题rule";
RuleSet ruleset = new RuleSet();
ruleset.Name = "标题";
ruleset.Rules.Add(rule1);
ruleset.Code = "title";
//第2个set
BaseRule rule2 = new BeginEndRule("<span class=\"grade\">", "</span>", false, false, true, true);
rule2.RuleNo = 10;
rule2.Name = "等级rule";
RuleSet ruleset2 = new RuleSet();
ruleset2.Name = "等级";
ruleset2.Code = "level";
ruleset2.Rules.Add(rule2);
//第3个set
string regexExp = @"id=""se_title_\d+"">.*?<span>(?<t_name>.*?)</span>.*?""parGd"">.?(?<t_price1>\d+)</span>.*?""Mne"">.</span>(?<price2>\d+)</dt>";
BaseRule rule3 = new RegexRule(regexExp);
rule3.RuleNo = 10;
rule3.Name = "价格rule";
RuleSet ruleset3 = new RuleSet();
ruleset3.Name = "价格";
ruleset3.Code = "price";
ruleset3.Rules.Add(rule3);
RuleAssembly assm = new RuleAssembly();
assm.CodeName = "Ass";
assm.Name = "tongchengcom";
ruleset.SetNo = 10;
ruleset2.SetNo = 11;
ruleset3.SetNo = 12;
assm.RuleSets.Add(ruleset);
assm.RuleSets.Add(ruleset2);
assm.RuleSets.Add(ruleset3);
#endregion
rule.SaveRule(assm);
#endregion
#region 读一个xml
CE.Domain.Rule.RuleAssembly ra = rule.ReadRule(assm.Name);
CE.Domain.Rule.BeginEndRule ber = (CE.Domain.Rule.BeginEndRule)ra.RuleSets[0].Rules[0];
Assert.AreEqual(ber.Name, "标题rule");
Assert.AreEqual(ber.RuleNo.ToString(), "10");
Assert.AreEqual(ber.Enabled.ToString(), "True");
Assert.AreEqual(ber.BeginMark, "<div id=\"jqlast_maincontent\" class=\"jqlast_main_title\"><h1>");
Assert.AreEqual(ber.EndMark, "</h1>");
CE.Domain.Rule.BeginEndRule ber1 = ra.RuleSets[0].Rules[0] as CE.Domain.Rule.BeginEndRule;
Assert.IsNotNull(ber1);
CE.Domain.Rule.RegexRule ber2 = ra.RuleSets[2].Rules[0] as CE.Domain.Rule.RegexRule;
Assert.IsNotNull(ber1);
#endregion
}
public void Persistence2Excel8Html(string htmlPath, string rulePath, string savePath,string savePricePath)
{
List<string> htmllist = new List<string>();
//查看是否存在html文件, 并添加到列表中
if (Directory.Exists(htmlPath))
{
foreach (string d in Directory.GetFileSystemEntries(htmlPath))
{
htmllist.Add(d);
}
}
else
{
return;
}
//查看
IRule rule = new Persistence.Rule();
rule.PersistencePath = Path.GetDirectoryName(rulePath);
ruleassembly = rule.ReadRule(Path.GetFileNameWithoutExtension(rulePath));
ExcelOpr.ExcelOpr excelopr = new ExcelOpr.ExcelOpr();
for (int i = 0; i < htmllist.Count; i++)
{
string html = htmlHandler.ReadHtml(htmllist[i]);
string result = ruleassembly.FilterUsingAssembly(html, false);
excelopr.Persistence2Excel(i + 2, result, savePath,savePricePath);
}
}