/// <summary>
/// 构造为Html完整的数据
/// </summary>
/// <param name="pRule"></param>
/// <returns></returns>
public string[] GetUrls(DrillRegularRule pRule, string pSourceUrl)
{
if (pRule.DrillType == 0)
{
string sRegionHtml = GetRegion(pRule.StartTag, pRule.EndTag);
if (sRegionHtml != null && sRegionHtml.Length > 0)
{
string sDomain;
string sUrlBase = GetUrlBase(pSourceUrl, out sDomain);
string[] sUrlList;
if (pRule.FeatureType == 0)
{
sUrlList = GetLinks(sRegionHtml, sDomain, sUrlBase);
}
else
{
sUrlList = GetImgs(sRegionHtml, sDomain, sUrlBase);
}
if (pRule.Feature != null && pRule.Feature.Length > 0)
{
List <string> sList = new List <string>();
for (int i = 0; i < sUrlList.Length; i++)
{
if (sUrlList[i].IndexOf(pRule.Feature, StringComparison.OrdinalIgnoreCase) >= 0)
{
sList.Add(sUrlList[i]);
}
}
return(sList.ToArray());
}
return(sUrlList);
}
}
return(null);
}
public DrillRegularRule ToDrillRegularRule()
{
DrillRegularRule sDrillRegularRule = new DrillRegularRule();
sDrillRegularRule.DbID=DbID;
sDrillRegularRule.DbName=DbName;
sDrillRegularRule.DrillType=DrillType;
sDrillRegularRule.EndTag=EndTag;
sDrillRegularRule.Feature=Feature;
sDrillRegularRule.FeatureType=FeatureType;
sDrillRegularRule.Feilds=Feilds;
sDrillRegularRule.MetaModalID=MetaModalID;
sDrillRegularRule.Name=Name;
sDrillRegularRule.Splitter=Splitter;
sDrillRegularRule.StartTag = StartTag;
return sDrillRegularRule;
}
public DrillRegularRule ToDrillRegularRule()
{
DrillRegularRule sDrillRegularRule = new DrillRegularRule();
sDrillRegularRule.DbID = DbID;
sDrillRegularRule.DbName = DbName;
sDrillRegularRule.DrillType = DrillType;
sDrillRegularRule.EndTag = EndTag;
sDrillRegularRule.Feature = Feature;
sDrillRegularRule.FeatureType = FeatureType;
sDrillRegularRule.Feilds = Feilds;
sDrillRegularRule.MetaModalID = MetaModalID;
sDrillRegularRule.Name = Name;
sDrillRegularRule.Splitter = Splitter;
sDrillRegularRule.StartTag = StartTag;
return(sDrillRegularRule);
}
/// <summary>
/// 获取记录区Html片段
/// </summary>
/// <param name="pRule"></param>
/// <returns></returns>
public string[] GetRecordHtmls(DrillRegularRule pRule)
{
if (pRule.DrillType == 2)
{
string sRegionHtml = GetRegion(pRule.StartTag, pRule.EndTag);
if (sRegionHtml != null && sRegionHtml.Length > 0)
{
if (pRule.Splitter.Length > 0)
{
return(sRegionHtml.Split(new string[] { pRule.Splitter }, StringSplitOptions.RemoveEmptyEntries));
}
else
{
return(new string[] { sRegionHtml });
}
}
}
return(null); //不符合要求
}
/// <summary>
/// 记录区-判断规则是否完备
/// </summary>
/// <param name="pDrillRule"></param>
/// <returns></returns>
public bool CanExe(DrillRegularRule pDrillRule)
{
if (pDrillRule.ConditionType == 0)
{
return true;
}
else if (pDrillRule.ConditionType == 1)
{
if (pDrillRule.ConditionTag == null || pDrillRule.ConditionTag.Length == 0 || mOriHtml.IndexOf(pDrillRule.ConditionTag, StringComparison.OrdinalIgnoreCase) >= 0)
{
return true;
}
}
else if (pDrillRule.ConditionType == 2)
{
if (pDrillRule.ConditionTag != null && pDrillRule.ConditionTag.Length > 0 && mOriHtml.IndexOf(pDrillRule.ConditionTag, StringComparison.OrdinalIgnoreCase) == -1)
{
return true;
}
}
return false;
}
/// <summary>
/// 记录区-判断规则是否完备
/// </summary>
/// <param name="pDrillRule"></param>
/// <returns></returns>
public bool CanExe(DrillRegularRule pDrillRule)
{
if (pDrillRule.ConditionType == 0)
{
return(true);
}
else if (pDrillRule.ConditionType == 1)
{
if (pDrillRule.ConditionTag == null || pDrillRule.ConditionTag.Length == 0 || mOriHtml.IndexOf(pDrillRule.ConditionTag, StringComparison.OrdinalIgnoreCase) >= 0)
{
return(true);
}
}
else if (pDrillRule.ConditionType == 2)
{
if (pDrillRule.ConditionTag != null && pDrillRule.ConditionTag.Length > 0 && mOriHtml.IndexOf(pDrillRule.ConditionTag, StringComparison.OrdinalIgnoreCase) == -1)
{
return(true);
}
}
return(false);
}
/// <summary>
/// 分隔记录
/// </summary>
/// <param name="sender"></param>
/// <param name="e"></param>
private void SplitRecord(object sender, System.EventArgs e)
{
try
{
DrillRegularRule sDrillRule = new DrillRegularRule();
sDrillRule.StartTag = TxtStartTag.Text.Trim();
sDrillRule.EndTag = TxtEndTag.Text.Trim();
sDrillRule.DrillType = 2;
sDrillRule.Splitter = TxtSpliter.Text.Trim();
mRecordHtmls = new RegScriptTransactor(mHtml).GetRecordHtmls(sDrillRule);
if (mRecordHtmls != null && mRecordHtmls.Length > 0)
{
TxtTotal.Text = mRecordHtmls.Length.ToString();
ShowRecord(0);
}
else
{
TxtTotal.Text = "0";
ShowRecord(-1);
}
}
catch (Exception E)
{
MessageBox.Show(E.Message, "提示信息");
}
}
/// <summary>
/// 加载提取规则
/// </summary>
private void LoadDrillRule(DrillRegularRule pRule)
{
if (pRule != null)
{
mDrillRule = pRule;
}
else
{
mDrillRule = new DrillRegularRule();
}
CmbConditionType.SelectedIndex = mDrillRule.ConditionType;
TxtConditionTag.Text = mDrillRule.ConditionTag;
TxtStartTag.Text = mDrillRule.StartTag;
TxtEndTag.Text = mDrillRule.EndTag;
if (mDrillRule.DrillType == 0)
{
tabControl3.SelectedIndex = 0; //提取类型
}
else
{
tabControl3.SelectedIndex = 1; //提取类型
}
#region 链接提取模式
CmbLinkType.SelectedIndex = mDrillRule.FeatureType;
TxtUrlFeatrue.Text = mDrillRule.Feature;
#endregion
#region 高级提取模式
CDbDefine sCDbDefine = CommonDbServer.GetModalDbByID(mDrillRule.MetaModalID);
TxtMeta.Text = sCDbDefine.DbName;
TxtMeta.Tag = sCDbDefine;
TxtSpliter.Text = mDrillRule.Splitter;
SplitRecord(null, null);
#endregion
}
private void btnTest_Click(object sender, EventArgs e)
{
RegScriptTransactor sRegScriptTransactor = new RegScriptTransactor(mTree.Html);
DrillRegularRule sDrillRule = new DrillRegularRule();
sDrillRule.StartTag = TxtStartTag.Text.Trim();
sDrillRule.EndTag = TxtEndTag.Text.Trim();
sDrillRule.DrillType = 0;
sDrillRule.MetaModalID = SysDbDefines.下载链接.DbModelID;
sDrillRule.FeatureType = CmbLinkType.SelectedIndex;
sDrillRule.Feature = TxtUrlFeatrue.Text;
string[] sUrls = sRegScriptTransactor.GetUrls(sDrillRule, mTree.URL);
List<string> sList = new List<string>();
for (int i = 0; i < sUrls.Length; i++)
{
sList.Add((i + 1).ToString() + "." + sUrls[i]);
}
TxtTestResult0.Lines = sList.ToArray();
}
/// <summary>
/// 构造为Html完整的数据
/// </summary>
/// <param name="pRule"></param>
/// <returns></returns>
public string[] GetUrls(DrillRegularRule pRule, string pSourceUrl)
{
if (pRule.DrillType == 0)
{
string sRegionHtml = GetRegion(pRule.StartTag, pRule.EndTag);
if (sRegionHtml != null && sRegionHtml.Length > 0)
{
string sDomain;
string sUrlBase = GetUrlBase(pSourceUrl, out sDomain);
string[] sUrlList;
if (pRule.FeatureType == 0)
{
sUrlList = GetLinks(sRegionHtml, sDomain, sUrlBase);
}
else
{
sUrlList = GetImgs(sRegionHtml, sDomain, sUrlBase);
}
if (pRule.Feature != null && pRule.Feature.Length > 0)
{
List<string> sList = new List<string>();
for (int i = 0; i < sUrlList.Length; i++)
{
if (sUrlList[i].IndexOf(pRule.Feature, StringComparison.OrdinalIgnoreCase) >= 0)
{
sList.Add(sUrlList[i]);
}
}
return sList.ToArray();
}
return sUrlList;
}
}
return null;
}
/// <summary>
/// 获取记录区Html片段
/// </summary>
/// <param name="pRule"></param>
/// <returns></returns>
public string[] GetRecordHtmls(DrillRegularRule pRule)
{
if (pRule.DrillType == 2)
{
string sRegionHtml = GetRegion(pRule.StartTag, pRule.EndTag);
if (sRegionHtml != null && sRegionHtml.Length > 0)
{
if (pRule.Splitter.Length > 0)
{
return sRegionHtml.Split(new string[] { pRule.Splitter }, StringSplitOptions.RemoveEmptyEntries);
}
else
{
return new string[] { sRegionHtml };
}
}
}
return null; //不符合要求
}