搜索引擎

开发平台：

C#

ClassModelBuilder.cs：源码内容

using System;
using System.IO;
using System.Collections.Generic;
using System.Text;
using System.Collections;
using System.Data.SqlClient;
using System.Threading;
using System.Net;
/*
' 迅龙中文分类搜索引擎 v0.6
'
' LGPL 许可发行
'
' 宁夏大学张冬康彩 zd4004@163.com
'
' 官网 http://blog.163.com/zd4004/
*/
namespace XunLong.ModelBuilder
{
/// <summary>
/// 建立模板
/// </summary>
public class ClassModelBuilder
{
private static Random raV = new Random();
/// <summary>
/// 取出的匹配字符串是否满足要求
/// </summary>
/// <param name="data"></param>
/// <returns>包含　且　《》成对 </returns>
private bool xStrItOK(string data)
{
if ((data.IndexOf("<") == -1) || (data.IndexOf(">") == -1) || (inStrNum(data, "<") != inStrNum(data, ">")))
{
return false;
}
return false;
}
/// <summary>
/// 某个字符在串中的个数
/// </summary>
/// <param name="data"></param>
/// <returns></returns>
public int inStrNum(string data, string one)
{
string c = data.Replace(one, "");
int cx = data.Length - c.Length;
return cx;
}
/// <summary>
/// 改变数据 <ZDKC0>" + data1 + "</ZDKC0> 在标题改变以便标题栏能在完全相同时被识别
/// 那么 0 位置就肯定为标题兰
/// </summary>
/// <param name="data"></param>
/// <returns></returns>
private static string RW_HTMLDB(string data)
{
int a1 = data.IndexOf("<ZDKC0>");
int a2 = data.IndexOf("</ZDKC0>");
int a3 = data.IndexOf("<ZDbody>");
int a4 = data.IndexOf("</ZDbody>");
int a5 = data.IndexOf(">", a3 + 1);
string data1 = "";
string data2 = "";
int nowTitle = raV.Next( 25,984562);
try
{
if (a1 > 0 & a2 > 0 & a2 > a1)
{
data1 = data.Substring(a1 + 7, a2 - a1 - 7);
}
if (a3 > 0 & a5 > 0 & a5 > a3)
{
data2 = data.Substring(a5 + 1, a4 - a5 - 1);
}
data1 = data1.Replace("*", "#");
data2 = data2.Replace("*", "#");
return " <ZDKC0>" + nowTitle.ToString()+"_"+data1.Length.ToString() + "</ZDKC0> <ZDbody>" + data2 + "</ZDbody>";
}
catch
{
return " <ZDKC0>" + nowTitle.ToString() + "_" + data.Length.ToString() + "</ZDKC0> <ZDbody>" + data + "</ZDbody> ";
}
}
/// <summary>
/// 根据数据建立一个网页模板
/// </summary>
/// <param name="mHTMs"></param>
/// <returns></returns>
public string BuilderModel(Hashtable mHTMs2)
{
Hashtable mHTMs = new Hashtable();
mHTMs.Clear();
foreach (System.Collections.DictionaryEntry de in mHTMs2)
{
//原始的数据需要修改以便可以在标题相同的情况下识别标题
string NewHTM = RW_HTMLDB(de.Value.ToString());
mHTMs.Add(de.Key, NewHTM);
}
Console.Write("..");
//得到最短的串
string oneModel = "";
foreach (System.Collections.DictionaryEntry de in mHTMs)
{
if (oneModel.Length == 0)
{
oneModel = de.Value.ToString();
}
else
{
if (oneModel.Length > de.Value.ToString().Length)
{
oneModel = de.Value.ToString();
}
}
}
//得到匹配临时模板的原始长度
int onemLong = oneModel.Length;
Hashtable tdict = new Hashtable();
int tdictI = 0; //字典序号
tdict.Clear();
Hashtable oneModels = new Hashtable();
oneModels.Clear();
oneModels.Add(oneModel,"0"); //碎片如果相同就不用继续寻找添加
//存放临时碎片
Hashtable newModels = new Hashtable();
newModels.Clear();
oneModel = "";
NewStart: //开始遍历匹配串
if (oneModel.Length > 0)
{
oneModels.Remove(oneModel); //移除旧的字符串压入分解后的字符串
}
foreach (System.Collections.DictionaryEntry de in newModels) //加入新的碎片项
{
if (oneModels.Contains(de.Key) == false)
{
oneModels.Add(de.Key, "0");
}
}
newModels.Clear();
foreach (System.Collections.DictionaryEntry de in oneModels)
{
oneModel =de.Key.ToString();
char[] oneModel_ss = oneModel.ToCharArray();
if (oneModel.Length > 4)
{
int h = oneModel.Length;
if ((5 * h>onemLong)&(onemLong>9000))
{
h = onemLong/5;
}
if (h > 512)
{
h = 512;
}
while (true)
{
if (h < 4) //最短字符不能小于5 >=四
{ goto nextcmd2; }
int st = 0;
while (true)
{
if (st + h > oneModel.Length)
{ goto nextcmd1; }
///////**********************************************
// string a1= oneModel.Substring(st, 1);
// string a2= oneModel.Substring(st+h-1, 1);
////////////*****************************************
char a1 = oneModel_ss[st];
if ( a1!= '<') //判断是否合法含有 < >
{
goto nextcmd3;
}
char a2 = oneModel_ss[st + h - 1];
if(a2 != '>') //判断是否合法含有 < >
{
goto nextcmd3;
}
string onestr = oneModel.Substring(st, h);
// if ((onestr.Substring(0, 1) != "<") | (onestr.Substring(onestr.Length - 1, 1) != ">") | (onestr.IndexOf(">") == -1) | (onestr.IndexOf("<") == -1))
// {
// goto nextcmd3; //取开头和末尾在<>中的数据
// }
//遍历数据是否符合全部数据
foreach (System.Collections.DictionaryEntry de1 in mHTMs)
{
// if ((de1.Value.ToString().IndexOf(onestr) == -1) & (de1.Value.ToString().Length > 0))
if (de1.Value.ToString().IndexOf(onestr) == -1)
{
goto nextcmd3;
}
}
//符合所有项都存在的条件压入字典
tdict.Add(tdictI, onestr);
////Console.WriteLine("M-Dict>> " + tdictI.ToString() + " || " + onestr);
//数据替换
Hashtable mtmpp = new Hashtable();
mtmpp.Clear();
foreach (System.Collections.DictionaryEntry de2 in mHTMs)
{
mtmpp.Add(de2.Key ,de2.Value.ToString().Replace(onestr, "*" + tdictI.ToString() + "*"));
}
mHTMs = mtmpp;
tdictI = tdictI + 1; //字典序号
//oneModel 截取取出碎片压入 newModels
string oneModelTmp = oneModel.Replace(onestr, "*");
string[] mykc = oneModelTmp.Split('*');
foreach (string ddee in mykc)
{
if ((ddee.Length > 4) & (newModels.Contains(ddee) == false))
{
newModels.Add(ddee, "0");
}
}
goto NewStart; //因为碎片变化所以重新开始扫描
nextcmd3:
st = st + 1;
}
nextcmd1:
h = h - 1;
}
}
nextcmd2: ;
goto NewStart; //本次遍历结束去掉本字符串
}
//mHTMs 处理新的网页只留下 *0*
Hashtable mtmppC = new Hashtable();
mtmppC.Clear();
foreach (System.Collections.DictionaryEntry de2 in mHTMs)
{
string cckc = de2.Value.ToString();
//把cckc中的*0*提取出来
mtmppC.Add(de2.Key,getXnX(de2.Value.ToString()));
//Console.WriteLine("M-DATA1>> "+cckc);
}
mHTMs= mtmppC;
string mmoo = GetComDataEND(mHTMs,tdict);
if (mmoo.Length == 0)
{
//从模板得到一个项目模板
mmoo = aassBuilderModel(mHTMs);
}
mmoo = mmoo.Replace("<","*");
mmoo = mmoo.Replace(">", "*");
string mo = mmoo;
if (mo.Length == 0)
{
return "";
}
else
{
Hashtable mytpp = new Hashtable();
mytpp.Clear();
foreach (System.Collections.DictionaryEntry dec in tdict)
{
mo = mo.Replace("*" + dec.Key.ToString() + "*", "<kc*Kc>" + dec.Value.ToString() + "<kc*Kc>" );
}
mo = mo.Replace("<kc*Kc><kc*Kc><kc*Kc>", "<kc*Kc>");
mo = mo.Replace("<kc*Kc><kc*Kc>", "<kc*Kc>");
mo = mo.Replace("<kc*Kc>", "*");
return mo;
// ArrayList mm = new ArrayList();
}
}
/// <summary>
/// 把cckc中的*0*提取出来
/// </summary>
/// <param name="data"></param>
/// <returns></returns>
private string getXnX(string data)
{
if (data.IndexOf("*") == -1)
{
return "";
}
string mkc = "";
int ii = 0;
while (true)
{
ii = ii + 1;
if (data.IndexOf("*") == -1)
{
// mkc = mkc.Replace("###", "#");
// mkc = mkc.Replace("##", "#");
for (int uu = 0; uu < 1000; uu++)
{
try
{
mkc = mkc.Replace("<" + uu.ToString() + "><" + uu.ToString() + "><" + uu.ToString() + ">", "<《3-" + uu.ToString() + "》>");
mkc = mkc.Replace("<" + uu.ToString() + "><" + uu.ToString() + ">", "<《2-" + uu.ToString() + "》>");
mkc = mkc.Replace("<《3-" + uu.ToString() + "》>", "<" + uu.ToString() + ">");
mkc = mkc.Replace("<《2-" + uu.ToString() + "》>", "<" + uu.ToString() + ">");
mkc = mkc.Replace("<" + uu.ToString() + "><" + uu.ToString() + "><" + uu.ToString() + ">", "<《3-" + uu.ToString() + "》>");
mkc = mkc.Replace("<" + uu.ToString() + "><" + uu.ToString() + ">", "<《2-" + uu.ToString() + "》>");
mkc = mkc.Replace("<《3-" + uu.ToString() + "》>", "<" + uu.ToString() + ">");
mkc = mkc.Replace("<《2-" + uu.ToString() + "》>", "<" + uu.ToString() + ">");
mkc = mkc.Replace("<" + uu.ToString() + "><" + uu.ToString() + "><" + uu.ToString() + ">", "<《3-" + uu.ToString() + "》>");
mkc = mkc.Replace("<" + uu.ToString() + "><" + uu.ToString() + ">", "<《2-" + uu.ToString() + "》>");
mkc = mkc.Replace("<《3-" + uu.ToString() + "》>", "<" + uu.ToString() + ">");
mkc = mkc.Replace("<《2-" + uu.ToString() + "》>", "<" + uu.ToString() + ">");
mkc = mkc.Replace("<" + uu.ToString() + "><" + uu.ToString() + "><" + uu.ToString() + ">", "<" + uu.ToString() + ">");
mkc = mkc.Replace("<" + uu.ToString() + "><" + uu.ToString() + ">", "<" + uu.ToString() + ">");
}
catch
{
for (int uu_i = 0; uu_i < 800; uu_i++)
{
mkc = mkc.Replace("<" + uu.ToString() + "><" + uu.ToString() + "><" + uu.ToString() + ">", "<《3-" + uu.ToString() + "》>");
mkc = mkc.Replace("<" + uu.ToString() + "><" + uu.ToString() + ">", "<《2-" + uu.ToString() + "》>");
mkc = mkc.Replace("<《3-" + uu.ToString() + "》>", "<" + uu.ToString() + ">");
mkc = mkc.Replace("<《2-" + uu.ToString() + "》>", "<" + uu.ToString() + ">");
}
}
}
return mkc;
}
//1 取得第一个* 取得第2个* 截取第一个和第2个之间的数据截断字符串
int a1 = data.IndexOf("*");
int a2 = data.IndexOf("*", a1 + 1);
// string aa = data.Substring(a1, a2 - a1);
// mkc = mkc + aa;
string aa = data.Substring(a1+1, a2 - a1-1);
// mkc = mkc + "#<"+aa+">#";
mkc = mkc + "<" + aa + ">";
data = data.Substring(a2 + 1, data.Length - a2-1);
}
}
/// <summary>
/// 《2》根据子模板建立一个超级模板
/// </summary>
/// <param name="mHTMs"></param>
/// <returns></returns>
private string aassBuilderModel(Hashtable mHTMs)
{
Console.Write("..");
//得到最短的串
string oneModel = "";
foreach (System.Collections.DictionaryEntry de in mHTMs)
{
if (oneModel.Length == 0)
{
oneModel = de.Value.ToString();
}
else
{
if (oneModel.Length > de.Value.ToString().Length)
{
oneModel = de.Value.ToString();
}
}
}
foreach (System.Collections.DictionaryEntry de in mHTMs)
{
if (oneModel != de.Value.ToString()) //不相同
{
if (de.Value.ToString().IndexOf(oneModel) == -1) //不包含
{
goto cXStart;
}
}
}
return oneModel;
cXStart:
//得到匹配临时模板的原始长度
int onemLong = oneModel.Length;
Hashtable tdict = new Hashtable();
int tdictI = 0; //字典序号
tdict.Clear();
Hashtable oneModels = new Hashtable();
oneModels.Clear();
oneModels.Add(oneModel, "0"); //碎片如果相同就不用继续寻找添加
//存放临时碎片
Hashtable newModels = new Hashtable();
newModels.Clear();
oneModel = "";
NewStart: //开始遍历匹配串
if (oneModel.Length > 0)
{
oneModels.Remove(oneModel); //移除旧的字符串压入分解后的字符串
}
foreach (System.Collections.DictionaryEntry de in newModels) //加入新的碎片项
{
if (oneModels.Contains(de.Key) == false)
{
oneModels.Add(de.Key, "0");
}
}
newModels.Clear();
foreach (System.Collections.DictionaryEntry de in oneModels)
{
oneModel = de.Key.ToString();
if (oneModel.Length > 2)
{
int h = oneModel.Length;
while (true)
{
if (h < 3) //最短字符不能小于5 >=四
{ goto nextcmd2; }
int st = 0;
while (true)
{
if (st + h > oneModel.Length)
{ goto nextcmd1; }
string a1 = oneModel.Substring(st, 1);
string a2 = oneModel.Substring(st + h - 1, 1);
if ((a1 != "<") | (a2 != ">")) //判断是否合法含有 < >
{
goto nextcmd3;
}
string onestr = oneModel.Substring(st, h);
if ((onestr.Substring(0, 1) != "<") | (onestr.Substring(onestr.Length - 1, 1) != ">") | (onestr.IndexOf(">") == -1) | (onestr.IndexOf("<") == -1))
{
goto nextcmd3; //取开头和末尾在<>中的数据
}
//遍历数据是否符合全部数据
foreach (System.Collections.DictionaryEntry de1 in mHTMs)
{
if ((de1.Value.ToString().IndexOf(onestr) == -1)&(de1.Value.ToString().Length>0))
{
goto nextcmd3;
}
}
//符合所有项都存在的条件压入字典
tdict.Add(tdictI, onestr);
//Console.WriteLine("M-Dict2>> " + tdictI.ToString() + " || " + onestr);
//数据替换
Hashtable mtmpp = new Hashtable();
mtmpp.Clear();
foreach (System.Collections.DictionaryEntry de2 in mHTMs)
{
// mtmpp.Add(de2.Key, de2.Value.ToString().Replace(onestr, "*" + tdictI.ToString() + "*"));
string tmp_str = de2.Value.ToString();
if (tmp_str == null)
{
}
else
{
mtmpp.Add(de2.Key, tmp_str.Replace(onestr, "*" + tdictI.ToString() + "*"));
}
}
mHTMs = mtmpp;
tdictI = tdictI + 1; //字典序号
//oneModel 截取取出碎片压入 newModels
string oneModelTmp = oneModel.Replace(onestr, "*");
string[] mykc = oneModelTmp.Split('*');
foreach (string ddee in mykc)
{
if ((ddee.Length > 4) & (newModels.Contains(ddee) == false))
{
newModels.Add(ddee, "0");
}
}
goto NewStart; //因为碎片变化所以重新开始扫描
nextcmd3:
st = st + 1;
}
nextcmd1:
h = h - 1;
}
}
nextcmd2: ;
goto NewStart; //本次遍历结束去掉本字符串
}
//mHTMs 处理新的网页只留下 *0*
Hashtable mtmppC = new Hashtable();
mtmppC.Clear();
foreach (System.Collections.DictionaryEntry de2 in mHTMs)
{
string cckc = de2.Value.ToString();
//把cckc中的*0*提取出来
mtmppC.Add(de2.Key, getXnX(de2.Value.ToString()));
//Console.WriteLine("M-DATA2>> " + cckc);
// mtmppC.Add(de2.Key, de2.Value.ToString());
}
mHTMs = mtmppC;
string mmoo = GetComDataEND(mHTMs,tdict);
if (mmoo.Length == 0)
{
//从模板得到一个项目模板
mmoo = aass3BuilderModel(mHTMs);
}
mmoo = mmoo.Replace("<", "*");
mmoo = mmoo.Replace(">", "*");
if (mmoo.Length == 0)
{
return "";
}
else
{
foreach (System.Collections.DictionaryEntry dec in tdict)
{
mmoo = mmoo.Replace("*" + dec.Key.ToString() + "*", dec.Value.ToString());
}
//Console.WriteLine("TRETURN2>> " + mmoo);
return mmoo;
}
}
/// <summary>
/// 《3》根据子模板建立一个超级模板
/// </summary>
/// <param name="mHTMs"></param>
/// <returns></returns>
private string aass3BuilderModel(Hashtable mHTMs)
{
Console.Write("..");
//得到最短的串
string oneModel = "";
foreach (System.Collections.DictionaryEntry de in mHTMs)
{
if (oneModel.Length == 0)
{
oneModel = de.Value.ToString();
}
else
{
if (oneModel.Length > de.Value.ToString().Length)
{
oneModel = de.Value.ToString();
}
}
}
foreach (System.Collections.DictionaryEntry de in mHTMs)
{
if (oneModel != de.Value.ToString()) //不相同
{
if (de.Value.ToString().IndexOf(oneModel) == -1) //不包含
{
goto cXStart;
}
}
}
return oneModel;
cXStart:
//得到匹配临时模板的原始长度
int onemLong = oneModel.Length;
Hashtable tdict = new Hashtable();
int tdictI = 0; //字典序号
tdict.Clear();
Hashtable oneModels = new Hashtable();
oneModels.Clear();
oneModels.Add(oneModel, "0"); //碎片如果相同就不用继续寻找添加
//存放临时碎片
Hashtable newModels = new Hashtable();
newModels.Clear();
oneModel = "";
NewStart: //开始遍历匹配串
if (oneModel.Length > 0)
{
oneModels.Remove(oneModel); //移除旧的字符串压入分解后的字符串
}
foreach (System.Collections.DictionaryEntry de in newModels) //加入新的碎片项
{
if (oneModels.Contains(de.Key) == false)
{
oneModels.Add(de.Key, "0");
}
}
newModels.Clear();
foreach (System.Collections.DictionaryEntry de in oneModels)
{
oneModel = de.Key.ToString();
if (oneModel.Length > 2)
{
int h = oneModel.Length;
while (true)
{
if (h < 3) //最短字符不能小于5 >=四
{ goto nextcmd2; }
int st = 0;
while (true)
{
if (st + h > oneModel.Length)
{ goto nextcmd1; }
string a1 = oneModel.Substring(st, 1);
string a2 = oneModel.Substring(st + h - 1, 1);
if ((a1 != "<") | (a2 != ">")) //判断是否合法含有 < >
{
goto nextcmd3;
}
string onestr = oneModel.Substring(st, h);
if ((onestr.Substring(0, 1) != "<") | (onestr.Substring(onestr.Length - 1, 1) != ">") | (onestr.IndexOf(">") == -1) | (onestr.IndexOf("<") == -1))
{
goto nextcmd3; //取开头和末尾在<>中的数据
}
//遍历数据是否符合全部数据
foreach (System.Collections.DictionaryEntry de1 in mHTMs)
{
if ((de1.Value.ToString().IndexOf(onestr) == -1)&(de1.Value.ToString().Length>0))
{
goto nextcmd3;
}
}
//符合所有项都存在的条件压入字典
tdict.Add(tdictI, onestr);
//Console.WriteLine("M-Dict3>> " + tdictI.ToString() + " || " + onestr);
//数据替换
Hashtable mtmpp = new Hashtable();
mtmpp.Clear();
foreach (System.Collections.DictionaryEntry de2 in mHTMs)
{
string tmp_str = de2.Value.ToString();
if (tmp_str == null)
{
}
else
{
try
{
mtmpp.Add(de2.Key, tmp_str.Replace(onestr, "*" + tdictI.ToString() + "*"));
}
catch
{
return "";
}
}
}
mHTMs = mtmpp;
tdictI = tdictI + 1; //字典序号
//oneModel 截取取出碎片压入 newModels
string oneModelTmp = oneModel.Replace(onestr, "*");
string[] mykc = oneModelTmp.Split('*');
foreach (string ddee in mykc)
{
if ((ddee.Length > 4) & (newModels.Contains(ddee) == false))
{
newModels.Add(ddee, "0");
}
}
goto NewStart; //因为碎片变化所以重新开始扫描
nextcmd3:
st = st + 1;
}
nextcmd1:
h = h - 1;
}
}
nextcmd2: ;
goto NewStart; //本次遍历结束去掉本字符串
}
//mHTMs 处理新的网页只留下 *0*
Hashtable mtmppC = new Hashtable();
mtmppC.Clear();
foreach (System.Collections.DictionaryEntry de2 in mHTMs)
{
string cckc = de2.Value.ToString();
//把cckc中的*0*提取出来
mtmppC.Add(de2.Key, getXnX(de2.Value.ToString()));
//Console.WriteLine("M-DATA3>> " + cckc);
// mtmppC.Add(de2.Key, de2.Value.ToString());
}
mHTMs = mtmppC;
string mmoo = GetComDataEND(mHTMs,tdict);
if (mmoo.Length == 0)
{
//从模板得到一个项目模板
mmoo = aass4BuilderModel(mHTMs);
}
mmoo = mmoo.Replace("<", "*");
mmoo = mmoo.Replace(">", "*");
if (mmoo.Length == 0)
{
return "";
}
else
{
foreach (System.Collections.DictionaryEntry dec in tdict)
{
mmoo = mmoo.Replace("*" + dec.Key.ToString() + "*", dec.Value.ToString());
}
//Console.WriteLine("TRETURN3>> " + mmoo);
return mmoo;
}
}
/// <summary>
/// 《4》根据子模板建立一个超级模板
/// </summary>
/// <param name="mHTMs"></param>
/// <returns></returns>
private string aass4BuilderModel(Hashtable mHTMs)
{
Console.Write("..");
//得到最短的串
string oneModel = "";
foreach (System.Collections.DictionaryEntry de in mHTMs)
{
if (oneModel.Length == 0)
{
oneModel = de.Value.ToString();
}
else
{
if (oneModel.Length > de.Value.ToString().Length)
{
oneModel = de.Value.ToString();
}
}
}
foreach (System.Collections.DictionaryEntry de in mHTMs)
{
if (oneModel != de.Value.ToString()) //不相同
{
if (de.Value.ToString().IndexOf(oneModel) == -1) //不包含
{
goto cXStart;
}
}
}
return oneModel;
cXStart:
//得到匹配临时模板的原始长度
int onemLong = oneModel.Length;
Hashtable tdict = new Hashtable();
int tdictI = 0; //字典序号
tdict.Clear();
Hashtable oneModels = new Hashtable();
oneModels.Clear();
oneModels.Add(oneModel, "0"); //碎片如果相同就不用继续寻找添加
//存放临时碎片
Hashtable newModels = new Hashtable();
newModels.Clear();
oneModel = "";
NewStart: //开始遍历匹配串
if (oneModel.Length > 0)
{
oneModels.Remove(oneModel); //移除旧的字符串压入分解后的字符串
}
foreach (System.Collections.DictionaryEntry de in newModels) //加入新的碎片项
{
if (oneModels.Contains(de.Key) == false)
{
oneModels.Add(de.Key, "0");
}
}
newModels.Clear();
foreach (System.Collections.DictionaryEntry de in oneModels)
{
oneModel = de.Key.ToString();
if (oneModel.Length > 2)
{
int h = oneModel.Length;
while (true)
{
if (h < 3) //最短字符不能小于5 >=四
{ goto nextcmd2; }
int st = 0;
while (true)
{
if (st + h > oneModel.Length)
{ goto nextcmd1; }
string a1 = oneModel.Substring(st, 1);
string a2 = oneModel.Substring(st + h - 1, 1);
if ((a1 != "<") | (a2 != ">")) //判断是否合法含有 < >
{
goto nextcmd3;
}
string onestr = oneModel.Substring(st, h);
if ((onestr.Substring(0, 1) != "<") | (onestr.Substring(onestr.Length - 1, 1) != ">") | (onestr.IndexOf(">") == -1) | (onestr.IndexOf("<") == -1))
{
goto nextcmd3; //取开头和末尾在<>中的数据
}
//遍历数据是否符合全部数据
foreach (System.Collections.DictionaryEntry de1 in mHTMs)
{
if ((de1.Value.ToString().IndexOf(onestr) == -1)&(de1.Value.ToString().Length>0))
{
goto nextcmd3;
}
}
//符合所有项都存在的条件压入字典
tdict.Add(tdictI, onestr);
//Console.WriteLine("M-Dict4>> " + tdictI.ToString() + " || " + onestr);
//数据替换
Hashtable mtmpp = new Hashtable();
mtmpp.Clear();
foreach (System.Collections.DictionaryEntry de2 in mHTMs)
{
// mtmpp.Add(de2.Key, de2.Value.ToString().Replace(onestr, "*" + tdictI.ToString() + "*"));
string tmp_str = de2.Value.ToString();
if (tmp_str == null)
{
}
else
{
mtmpp.Add(de2.Key, tmp_str.Replace(onestr, "*" + tdictI.ToString() + "*"));
}
}
mHTMs = mtmpp;
tdictI = tdictI + 1; //字典序号
//oneModel 截取取出碎片压入 newModels
string oneModelTmp = oneModel.Replace(onestr, "*");
string[] mykc = oneModelTmp.Split('*');
foreach (string ddee in mykc)
{
if ((ddee.Length > 4) & (newModels.Contains(ddee) == false))
{
newModels.Add(ddee, "0");
}
}
goto NewStart; //因为碎片变化所以重新开始扫描
nextcmd3:
st = st + 1;
}
nextcmd1:
h = h - 1;
}
}
nextcmd2: ;
goto NewStart; //本次遍历结束去掉本字符串
}
//mHTMs 处理新的网页只留下 *0*
Hashtable mtmppC = new Hashtable();
mtmppC.Clear();
foreach (System.Collections.DictionaryEntry de2 in mHTMs)
{
string cckc = de2.Value.ToString();
//把cckc中的*0*提取出来
mtmppC.Add(de2.Key, getXnX(de2.Value.ToString()));
//Console.WriteLine("M-DATA4>> " + cckc);
// mtmppC.Add(de2.Key, de2.Value.ToString());
}
mHTMs = mtmppC;
//从模板得到一个项目模板
//string mmoo = aass5BuilderModel(mHTMs);
string mmoo = GetComDataEND(mHTMs, tdict);
if (mmoo.Length == 0)
{
//从模板得到一个项目模板
mmoo = aass3BuilderModel(mHTMs);
}
mmoo = mmoo.Replace("<", "*");
mmoo = mmoo.Replace(">", "*");
if (mmoo.Length == 0)
{
return "";
}
else
{
foreach (System.Collections.DictionaryEntry dec in tdict)
{
mmoo = mmoo.Replace("*" + dec.Key.ToString() + "*", dec.Value.ToString());
}
//Console.WriteLine("TRETURN4>> " + mmoo);
return mmoo;
}
}
/// <summary>
/// 《5》根据子模板建立一个超级模板
/// </summary>
/// <param name="mHTMs"></param>
/// <returns></returns>
private string aass5BuilderModel(Hashtable mHTMs)
{
Console.Write("..");
//得到最短的串
string oneModel = "";
foreach (System.Collections.DictionaryEntry de in mHTMs)
{
if (oneModel.Length == 0)
{
oneModel = de.Value.ToString();
}
else
{
if (oneModel.Length > de.Value.ToString().Length)
{
oneModel = de.Value.ToString();
}
}
}
foreach (System.Collections.DictionaryEntry de in mHTMs)
{
if (oneModel != de.Value.ToString()) //不相同
{
if (de.Value.ToString().IndexOf(oneModel) == -1) //不包含
{
goto cXStart;
}
}
}
return oneModel;
cXStart:
//得到匹配临时模板的原始长度
int onemLong = oneModel.Length;
Hashtable tdict = new Hashtable();
int tdictI = 0; //字典序号
tdict.Clear();
Hashtable oneModels = new Hashtable();
oneModels.Clear();
oneModels.Add(oneModel, "0"); //碎片如果相同就不用继续寻找添加
//存放临时碎片
Hashtable newModels = new Hashtable();
newModels.Clear();
oneModel = "";
NewStart: //开始遍历匹配串
if (oneModel.Length > 0)
{
oneModels.Remove(oneModel); //移除旧的字符串压入分解后的字符串
}
foreach (System.Collections.DictionaryEntry de in newModels) //加入新的碎片项
{
if (oneModels.Contains(de.Key) == false)
{
oneModels.Add(de.Key, "0");
}
}
newModels.Clear();
foreach (System.Collections.DictionaryEntry de in oneModels)
{
oneModel = de.Key.ToString();
if (oneModel.Length > 2)
{
int h = oneModel.Length;
while (true)
{
if (h < 3) //最短字符不能小于5 >=四
{ goto nextcmd2; }
int st = 0;
while (true)
{
if (st + h > oneModel.Length)
{ goto nextcmd1; }
string a1 = oneModel.Substring(st, 1);
string a2 = oneModel.Substring(st + h - 1, 1);
if ((a1 != "<") | (a2 != ">")) //判断是否合法含有 < >
{
goto nextcmd3;
}
string onestr = oneModel.Substring(st, h);
if ((onestr.Substring(0, 1) != "<") | (onestr.Substring(onestr.Length - 1, 1) != ">") | (onestr.IndexOf(">") == -1) | (onestr.IndexOf("<") == -1))
{
goto nextcmd3; //取开头和末尾在<>中的数据
}
//遍历数据是否符合全部数据
foreach (System.Collections.DictionaryEntry de1 in mHTMs)
{
if ((de1.Value.ToString().IndexOf(onestr) == -1)&(de1.Value.ToString().Length>0))
{
goto nextcmd3;
}
}
//符合所有项都存在的条件压入字典
tdict.Add(tdictI, onestr);
//Console.WriteLine("M-Dict5>> " + tdictI.ToString() + " || " + onestr);
//数据替换
Hashtable mtmpp = new Hashtable();
mtmpp.Clear();
foreach (System.Collections.DictionaryEntry de2 in mHTMs)
{
//mtmpp.Add(de2.Key, de2.Value.ToString().Replace(onestr, "*" + tdictI.ToString() + "*"));
string tmp_str = de2.Value.ToString();
if (tmp_str == null)
{
}
else
{
mtmpp.Add(de2.Key, tmp_str.Replace(onestr, "*" + tdictI.ToString() + "*"));
}
}
mHTMs = mtmpp;
tdictI = tdictI + 1; //字典序号
//oneModel 截取取出碎片压入 newModels
string oneModelTmp = oneModel.Replace(onestr, "*");
string[] mykc = oneModelTmp.Split('*');
foreach (string ddee in mykc)
{
if ((ddee.Length > 4) & (newModels.Contains(ddee) == false))
{
newModels.Add(ddee, "0");
}
}
goto NewStart; //因为碎片变化所以重新开始扫描
nextcmd3:
st = st + 1;
}
nextcmd1:
h = h - 1;
}
}
nextcmd2: ;
goto NewStart; //本次遍历结束去掉本字符串
}
//mHTMs 处理新的网页只留下 *0*
Hashtable mtmppC = new Hashtable();
mtmppC.Clear();
foreach (System.Collections.DictionaryEntry de2 in mHTMs)
{
string cckc = de2.Value.ToString();
//把cckc中的*0*提取出来
mtmppC.Add(de2.Key, getXnX(de2.Value.ToString()));
//Console.WriteLine("M-DATA5>> " + cckc);
// mtmppC.Add(de2.Key, de2.Value.ToString());
}
mHTMs = mtmppC;
//从模板得到一个项目模板
//string mmoo = aass6BuilderModel(mHTMs);
string mmoo = GetComDataEND(mHTMs, tdict);
if (mmoo.Length == 0)
{
//从模板得到一个项目模板
mmoo = aass3BuilderModel(mHTMs);
}
mmoo = mmoo.Replace("<", "*");
mmoo = mmoo.Replace(">", "*");
if (mmoo.Length == 0)
{
return "";
}
else
{
foreach (System.Collections.DictionaryEntry dec in tdict)
{
mmoo = mmoo.Replace("*" + dec.Key.ToString() + "*", dec.Value.ToString());
}
//Console.WriteLine("TRETURN5>> " + mmoo);
return mmoo;
}
}
/// <summary>
/// 《6》根据子模板建立一个超级模板
/// </summary>
/// <param name="mHTMs"></param>
/// <returns></returns>
private string aass6BuilderModel(Hashtable mHTMs)
{
Console.Write("..");
//得到最短的串
string oneModel = "";
foreach (System.Collections.DictionaryEntry de in mHTMs)
{
if (oneModel.Length == 0)
{
oneModel = de.Value.ToString();
}
else
{
if (oneModel.Length > de.Value.ToString().Length)
{
oneModel = de.Value.ToString();
}
}
}
foreach (System.Collections.DictionaryEntry de in mHTMs)
{
if (oneModel != de.Value.ToString()) //不相同
{
if (de.Value.ToString().IndexOf(oneModel) == -1) //不包含
{
goto cXStart;
}
}
}
return oneModel;
cXStart:
//得到匹配临时模板的原始长度
int onemLong = oneModel.Length;
Hashtable tdict = new Hashtable();
int tdictI = 0; //字典序号
tdict.Clear();
Hashtable oneModels = new Hashtable();
oneModels.Clear();
oneModels.Add(oneModel, "0"); //碎片如果相同就不用继续寻找添加
//存放临时碎片
Hashtable newModels = new Hashtable();
newModels.Clear();
oneModel = "";
NewStart: //开始遍历匹配串
if (oneModel.Length > 0)
{
oneModels.Remove(oneModel); //移除旧的字符串压入分解后的字符串
}
foreach (System.Collections.DictionaryEntry de in newModels) //加入新的碎片项
{
if (oneModels.Contains(de.Key) == false)
{
oneModels.Add(de.Key, "0");
}
}
newModels.Clear();
foreach (System.Collections.DictionaryEntry de in oneModels)
{
oneModel = de.Key.ToString();
if (oneModel.Length > 2)
{
int h = oneModel.Length;
while (true)
{
if (h < 3) //最短字符不能小于5 >=四
{ goto nextcmd2; }
int st = 0;
while (true)
{
if (st + h > oneModel.Length)
{ goto nextcmd1; }
string a1 = oneModel.Substring(st, 1);
string a2 = oneModel.Substring(st + h - 1, 1);
if ((a1 != "<") | (a2 != ">")) //判断是否合法含有 < >
{
goto nextcmd3;
}
string onestr = oneModel.Substring(st, h);
if ((onestr.Substring(0, 1) != "<") | (onestr.Substring(onestr.Length - 1, 1) != ">") | (onestr.IndexOf(">") == -1) | (onestr.IndexOf("<") == -1))
{
goto nextcmd3; //取开头和末尾在<>中的数据
}
//遍历数据是否符合全部数据
foreach (System.Collections.DictionaryEntry de1 in mHTMs)
{
if ((de1.Value.ToString().IndexOf(onestr) == -1)&(de1.Value.ToString().Length>0))
{
goto nextcmd3;
}
}
//符合所有项都存在的条件压入字典
tdict.Add(tdictI, onestr);
//Console.WriteLine("M-Dict6>> " + tdictI.ToString() + " || " + onestr);
//数据替换
Hashtable mtmpp = new Hashtable();
mtmpp.Clear();
foreach (System.Collections.DictionaryEntry de2 in mHTMs)
{
// mtmpp.Add(de2.Key, de2.Value.ToString().Replace(onestr, "*" + tdictI.ToString() + "*"));
string tmp_str = de2.Value.ToString();
if (tmp_str == null)
{
}
else
{
mtmpp.Add(de2.Key, tmp_str.Replace(onestr, "*" + tdictI.ToString() + "*"));
}
}
mHTMs = mtmpp;
tdictI = tdictI + 1; //字典序号
//oneModel 截取取出碎片压入 newModels
string oneModelTmp = oneModel.Replace(onestr, "*");
string[] mykc = oneModelTmp.Split('*');
foreach (string ddee in mykc)
{
if ((ddee.Length > 4) & (newModels.Contains(ddee) == false))
{
newModels.Add(ddee, "0");
}
}
goto NewStart; //因为碎片变化所以重新开始扫描
nextcmd3:
st = st + 1;
}
nextcmd1:
h = h - 1;
}
}
nextcmd2: ;
goto NewStart; //本次遍历结束去掉本字符串
}
//mHTMs 处理新的网页只留下 *0*
Hashtable mtmppC = new Hashtable();
mtmppC.Clear();
foreach (System.Collections.DictionaryEntry de2 in mHTMs)
{
string cckc = de2.Value.ToString();
//把cckc中的*0*提取出来
mtmppC.Add(de2.Key, getXnX(de2.Value.ToString()));
//Console.WriteLine("M-DATA6>> " + cckc);
// mtmppC.Add(de2.Key, de2.Value.ToString());
}
mHTMs = mtmppC;
//从模板得到一个项目模板
//string mmoo = aass7BuilderModel(mHTMs);
string mmoo = GetComDataEND(mHTMs, tdict);
if (mmoo.Length == 0)
{
//从模板得到一个项目模板
mmoo = aass3BuilderModel(mHTMs);
}
mmoo = mmoo.Replace("<", "*");
mmoo = mmoo.Replace(">", "*");
if (mmoo.Length == 0)
{
return "";
}
else
{
foreach (System.Collections.DictionaryEntry dec in tdict)
{
mmoo = mmoo.Replace("*" + dec.Key.ToString() + "*", dec.Value.ToString());
}
//Console.WriteLine("TRETURN6>> " + mmoo);
return mmoo;
}
}
/// <summary>
/// 《7》根据子模板建立一个超级模板
/// </summary>
/// <param name="mHTMs"></param>
/// <returns></returns>
private string aass7BuilderModel(Hashtable mHTMs)
{
Console.Write("..");
//得到最短的串
string oneModel = "";
foreach (System.Collections.DictionaryEntry de in mHTMs)
{
if (oneModel.Length == 0)
{
oneModel = de.Value.ToString();
}
else
{
if (oneModel.Length > de.Value.ToString().Length)
{
oneModel = de.Value.ToString();
}
}
}
foreach (System.Collections.DictionaryEntry de in mHTMs)
{
if (oneModel != de.Value.ToString()) //不相同
{
if (de.Value.ToString().IndexOf(oneModel) == -1) //不包含
{
goto cXStart;
}
}
}
return oneModel;
cXStart:
//得到匹配临时模板的原始长度
int onemLong = oneModel.Length;
Hashtable tdict = new Hashtable();
int tdictI = 0; //字典序号
tdict.Clear();
Hashtable oneModels = new Hashtable();
oneModels.Clear();
oneModels.Add(oneModel, "0"); //碎片如果相同就不用继续寻找添加
//存放临时碎片
Hashtable newModels = new Hashtable();
newModels.Clear();
oneModel = "";
NewStart: //开始遍历匹配串
if (oneModel.Length > 0)
{
oneModels.Remove(oneModel); //移除旧的字符串压入分解后的字符串
}
foreach (System.Collections.DictionaryEntry de in newModels) //加入新的碎片项
{
if (oneModels.Contains(de.Key) == false)
{
oneModels.Add(de.Key, "0");
}
}
newModels.Clear();
foreach (System.Collections.DictionaryEntry de in oneModels)
{
oneModel = de.Key.ToString();
if (oneModel.Length > 2)
{
int h = oneModel.Length;
while (true)
{
if (h < 3) //最短字符不能小于5 >=四
{ goto nextcmd2; }
int st = 0;
while (true)
{
if (st + h > oneModel.Length)
{ goto nextcmd1; }
string a1 = oneModel.Substring(st, 1);
string a2 = oneModel.Substring(st + h - 1, 1);
if ((a1 != "<") | (a2 != ">")) //判断是否合法含有 < >
{
goto nextcmd3;
}
string onestr = oneModel.Substring(st, h);
if ((onestr.Substring(0, 1) != "<") | (onestr.Substring(onestr.Length - 1, 1) != ">") | (onestr.IndexOf(">") == -1) | (onestr.IndexOf("<") == -1))
{
goto nextcmd3; //取开头和末尾在<>中的数据
}
//遍历数据是否符合全部数据
foreach (System.Collections.DictionaryEntry de1 in mHTMs)
{
if ((de1.Value.ToString().IndexOf(onestr) == -1)&(de1.Value.ToString().Length>0))
{
goto nextcmd3;
}
}
//符合所有项都存在的条件压入字典
tdict.Add(tdictI, onestr);
//Console.WriteLine("M-Dict7>> " + tdictI.ToString() + " || " + onestr);
//数据替换
Hashtable mtmpp = new Hashtable();
mtmpp.Clear();
foreach (System.Collections.DictionaryEntry de2 in mHTMs)
{
// mtmpp.Add(de2.Key, de2.Value.ToString().Replace(onestr, "*" + tdictI.ToString() + "*"));
string tmp_str = de2.Value.ToString();
if (tmp_str == null)
{
}
else
{
mtmpp.Add(de2.Key, tmp_str.Replace(onestr, "*" + tdictI.ToString() + "*"));
}
}
mHTMs = mtmpp;
tdictI = tdictI + 1; //字典序号
//oneModel 截取取出碎片压入 newModels
string oneModelTmp = oneModel.Replace(onestr, "*");
string[] mykc = oneModelTmp.Split('*');
foreach (string ddee in mykc)
{
if ((ddee.Length > 4) & (newModels.Contains(ddee) == false))
{
newModels.Add(ddee, "0");
}
}
goto NewStart; //因为碎片变化所以重新开始扫描
nextcmd3:
st = st + 1;
}
nextcmd1:
h = h - 1;
}
}
nextcmd2: ;
goto NewStart; //本次遍历结束去掉本字符串
}
//mHTMs 处理新的网页只留下 *0*
Hashtable mtmppC = new Hashtable();
mtmppC.Clear();
foreach (System.Collections.DictionaryEntry de2 in mHTMs)
{
string cckc = de2.Value.ToString();
//把cckc中的*0*提取出来
mtmppC.Add(de2.Key, getXnX(de2.Value.ToString()));
//Console.WriteLine("M-DATA7>> " + cckc);
// mtmppC.Add(de2.Key, de2.Value.ToString());
}
mHTMs = mtmppC;
//从模板得到一个项目模板
// string mmoo = aass8BuilderModel(mHTMs);
string mmoo = GetComDataEND(mHTMs, tdict);
if (mmoo.Length == 0)
{
//从模板得到一个项目模板
mmoo = aass3BuilderModel(mHTMs);
}
mmoo = mmoo.Replace("<", "*");
mmoo = mmoo.Replace(">", "*");
if (mmoo.Length == 0)
{
return "";
}
else
{
foreach (System.Collections.DictionaryEntry dec in tdict)
{
mmoo = mmoo.Replace("*" + dec.Key.ToString() + "*", dec.Value.ToString());
}
//Console.WriteLine("TRETURN7>> " + mmoo);
return mmoo;
}
}
/// <summary>
/// 《8》根据子模板建立一个超级模板
/// </summary>
/// <param name="mHTMs"></param>
/// <returns></returns>
private string aass8BuilderModel(Hashtable mHTMs)
{
Console.Write("..");
//得到最短的串
string oneModel = "";
foreach (System.Collections.DictionaryEntry de in mHTMs)
{
if (oneModel.Length == 0)
{
oneModel = de.Value.ToString();
}
else
{
if (oneModel.Length > de.Value.ToString().Length)
{
oneModel = de.Value.ToString();
}
}
}
foreach (System.Collections.DictionaryEntry de in mHTMs)
{
if (oneModel != de.Value.ToString()) //不相同
{
if (de.Value.ToString().IndexOf(oneModel) == -1) //不包含
{
goto cXStart;
}
}
}
return oneModel;
cXStart:
//得到匹配临时模板的原始长度
int onemLong = oneModel.Length;
Hashtable tdict = new Hashtable();
int tdictI = 0; //字典序号
tdict.Clear();
Hashtable oneModels = new Hashtable();
oneModels.Clear();
oneModels.Add(oneModel, "0"); //碎片如果相同就不用继续寻找添加
//存放临时碎片
Hashtable newModels = new Hashtable();
newModels.Clear();
oneModel = "";
NewStart: //开始遍历匹配串
if (oneModel.Length > 0)
{
oneModels.Remove(oneModel); //移除旧的字符串压入分解后的字符串
}
foreach (System.Collections.DictionaryEntry de in newModels) //加入新的碎片项
{
if (oneModels.Contains(de.Key) == false)
{
oneModels.Add(de.Key, "0");
}
}
newModels.Clear();
foreach (System.Collections.DictionaryEntry de in oneModels)
{
oneModel = de.Key.ToString();
if (oneModel.Length > 2)
{
int h = oneModel.Length;
while (true)
{
if (h < 3) //最短字符不能小于5 >=四
{ goto nextcmd2; }
int st = 0;
while (true)
{
if (st + h > oneModel.Length)
{ goto nextcmd1; }
string a1 = oneModel.Substring(st, 1);
string a2 = oneModel.Substring(st + h - 1, 1);
if ((a1 != "<") | (a2 != ">")) //判断是否合法含有 < >
{
goto nextcmd3;
}
string onestr = oneModel.Substring(st, h);
if ((onestr.Substring(0, 1) != "<") | (onestr.Substring(onestr.Length - 1, 1) != ">") | (onestr.IndexOf(">") == -1) | (onestr.IndexOf("<") == -1))
{
goto nextcmd3; //取开头和末尾在<>中的数据
}
//遍历数据是否符合全部数据
foreach (System.Collections.DictionaryEntry de1 in mHTMs)
{
if ((de1.Value.ToString().IndexOf(onestr) == -1)&(de1.Value.ToString().Length>0))
{
goto nextcmd3;
}
}
//符合所有项都存在的条件压入字典
tdict.Add(tdictI, onestr);
//数据替换
Hashtable mtmpp = new Hashtable();
mtmpp.Clear();
foreach (System.Collections.DictionaryEntry de2 in mHTMs)
{
// mtmpp.Add(de2.Key, de2.Value.ToString().Replace(onestr, "*" + tdictI.ToString() + "*"));
string tmp_str = de2.Value.ToString();
if (tmp_str == null)
{
}
else
{
mtmpp.Add(de2.Key, tmp_str.Replace(onestr, "*" + tdictI.ToString() + "*"));
}
}
mHTMs = mtmpp;
tdictI = tdictI + 1; //字典序号
//oneModel 截取取出碎片压入 newModels
string oneModelTmp = oneModel.Replace(onestr, "*");
string[] mykc = oneModelTmp.Split('*');
foreach (string ddee in mykc)
{
if ((ddee.Length > 4) & (newModels.Contains(ddee) == false))
{
newModels.Add(ddee, "0");
}
}
goto NewStart; //因为碎片变化所以重新开始扫描
nextcmd3:
st = st + 1;
}
nextcmd1:
h = h - 1;
}
}
nextcmd2: ;
goto NewStart; //本次遍历结束去掉本字符串
}
//mHTMs 处理新的网页只留下 *0*
Hashtable mtmppC = new Hashtable();
mtmppC.Clear();
foreach (System.Collections.DictionaryEntry de2 in mHTMs)
{
string cckc = de2.Value.ToString();
//把cckc中的*0*提取出来
mtmppC.Add(de2.Key, getXnX(de2.Value.ToString()));
// mtmppC.Add(de2.Key, de2.Value.ToString());
}
mHTMs = mtmppC;
//从模板得到一个项目模板
//string mmoo = aass9BuilderModel(mHTMs);
string mmoo = GetComDataEND(mHTMs, tdict);
if (mmoo.Length == 0)
{
//从模板得到一个项目模板
mmoo = aass3BuilderModel(mHTMs);
}
mmoo = mmoo.Replace("<", "*");
mmoo = mmoo.Replace(">", "*");
if (mmoo.Length == 0)
{
return "";
}
else
{
foreach (System.Collections.DictionaryEntry dec in tdict)
{
mmoo = mmoo.Replace("*" + dec.Key.ToString() + "*", dec.Value.ToString());
}
//Console.WriteLine("TRETURN8>> " + mmoo);
return mmoo;
}
}
/// <summary>
/// 《9》根据子模板建立一个超级模板
/// </summary>
/// <param name="mHTMs"></param>
/// <returns></returns>
private string aass9BuilderModel(Hashtable mHTMs)
{
Console.Write("..");
//得到最短的串
string oneModel = "";
foreach (System.Collections.DictionaryEntry de in mHTMs)
{
if (oneModel.Length == 0)
{
oneModel = de.Value.ToString();
}
else
{
if (oneModel.Length > de.Value.ToString().Length)
{
oneModel = de.Value.ToString();
}
}
}
foreach (System.Collections.DictionaryEntry de in mHTMs)
{
if (oneModel != de.Value.ToString()) //不相同
{
if (de.Value.ToString().IndexOf(oneModel) == -1) //不包含
{
goto cXStart;
}
}
}
return oneModel;
cXStart:
//得到匹配临时模板的原始长度
int onemLong = oneModel.Length;
Hashtable tdict = new Hashtable();
int tdictI = 0; //字典序号
tdict.Clear();
Hashtable oneModels = new Hashtable();
oneModels.Clear();
oneModels.Add(oneModel, "0"); //碎片如果相同就不用继续寻找添加
//存放临时碎片
Hashtable newModels = new Hashtable();
newModels.Clear();
oneModel = "";
NewStart: //开始遍历匹配串
if (oneModel.Length > 0)
{
oneModels.Remove(oneModel); //移除旧的字符串压入分解后的字符串
}
foreach (System.Collections.DictionaryEntry de in newModels) //加入新的碎片项
{
if (oneModels.Contains(de.Key) == false)
{
oneModels.Add(de.Key, "0");
}
}
newModels.Clear();
foreach (System.Collections.DictionaryEntry de in oneModels)
{
oneModel = de.Key.ToString();
if (oneModel.Length > 2)
{
int h = oneModel.Length;
while (true)
{
if (h < 3) //最短字符不能小于5 >=四
{ goto nextcmd2; }
int st = 0;
while (true)
{
if (st + h > oneModel.Length)
{ goto nextcmd1; }
string a1 = oneModel.Substring(st, 1);
string a2 = oneModel.Substring(st + h - 1, 1);
if ((a1 != "<") | (a2 != ">")) //判断是否合法含有 < >
{
goto nextcmd3;
}
string onestr = oneModel.Substring(st, h);
if ((onestr.Substring(0, 1) != "<") | (onestr.Substring(onestr.Length - 1, 1) != ">") | (onestr.IndexOf(">") == -1) | (onestr.IndexOf("<") == -1))
{
goto nextcmd3; //取开头和末尾在<>中的数据
}
//遍历数据是否符合全部数据
foreach (System.Collections.DictionaryEntry de1 in mHTMs)
{
if ((de1.Value.ToString().IndexOf(onestr) == -1)&(de1.Value.ToString().Length>0))
{
goto nextcmd3;
}
}
//符合所有项都存在的条件压入字典
tdict.Add(tdictI, onestr);
//数据替换
Hashtable mtmpp = new Hashtable();
mtmpp.Clear();
foreach (System.Collections.DictionaryEntry de2 in mHTMs)
{
//mtmpp.Add(de2.Key, de2.Value.ToString().Replace(onestr, "*" + tdictI.ToString() + "*"));
string tmp_str = de2.Value.ToString();
if (tmp_str == null)
{
}
else
{
mtmpp.Add(de2.Key, tmp_str.Replace(onestr, "*" + tdictI.ToString() + "*"));
}
}
mHTMs = mtmpp;
tdictI = tdictI + 1; //字典序号
//oneModel 截取取出碎片压入 newModels
string oneModelTmp = oneModel.Replace(onestr, "*");
string[] mykc = oneModelTmp.Split('*');
foreach (string ddee in mykc)
{
if ((ddee.Length > 4) & (newModels.Contains(ddee) == false))
{
newModels.Add(ddee, "0");
}
}
goto NewStart; //因为碎片变化所以重新开始扫描
nextcmd3:
st = st + 1;
}
nextcmd1:
h = h - 1;
}
}
nextcmd2: ;
goto NewStart; //本次遍历结束去掉本字符串
}
//mHTMs 处理新的网页只留下 *0*
Hashtable mtmppC = new Hashtable();
mtmppC.Clear();
foreach (System.Collections.DictionaryEntry de2 in mHTMs)
{
string cckc = de2.Value.ToString();
//把cckc中的*0*提取出来
mtmppC.Add(de2.Key, getXnX(de2.Value.ToString()));
// mtmppC.Add(de2.Key, de2.Value.ToString());
}
mHTMs = mtmppC;
//从模板得到一个项目模板
// string mmoo = aass10BuilderModel(mHTMs);
string mmoo = GetComDataEND(mHTMs, tdict);
if (mmoo.Length == 0)
{
//从模板得到一个项目模板
mmoo = aass3BuilderModel(mHTMs);
}
mmoo = mmoo.Replace("<", "*");
mmoo = mmoo.Replace(">", "*");
if (mmoo.Length == 0)
{
return "";
}
else
{
foreach (System.Collections.DictionaryEntry dec in tdict)
{
mmoo = mmoo.Replace("*" + dec.Key.ToString() + "*", dec.Value.ToString());
}
return mmoo;
}
}
//10
/// <summary>
/// 《10》根据子模板建立一个超级模板
/// </summary>
/// <param name="mHTMs"></param>
/// <returns></returns>
private string aass10BuilderModel(Hashtable mHTMs)
{
Console.Write("..");
//得到最短的串
string oneModel = "";
foreach (System.Collections.DictionaryEntry de in mHTMs)
{
if (oneModel.Length == 0)
{
oneModel = de.Value.ToString();
}
else
{
if (oneModel.Length > de.Value.ToString().Length)
{
oneModel = de.Value.ToString();
}
}
}
foreach (System.Collections.DictionaryEntry de in mHTMs)
{
if (oneModel != de.Value.ToString()) //不相同
{
if (de.Value.ToString().IndexOf(oneModel) == -1) //不包含
{
goto cXStart;
}
}
}
return oneModel;
cXStart:
//得到匹配临时模板的原始长度
int onemLong = oneModel.Length;
Hashtable tdict = new Hashtable();
int tdictI = 0; //字典序号
tdict.Clear();
Hashtable oneModels = new Hashtable();
oneModels.Clear();
oneModels.Add(oneModel, "0"); //碎片如果相同就不用继续寻找添加
//存放临时碎片
Hashtable newModels = new Hashtable();
newModels.Clear();
oneModel = "";
NewStart: //开始遍历匹配串
if (oneModel.Length > 0)
{
oneModels.Remove(oneModel); //移除旧的字符串压入分解后的字符串
}
foreach (System.Collections.DictionaryEntry de in newModels) //加入新的碎片项
{
if (oneModels.Contains(de.Key) == false)
{
oneModels.Add(de.Key, "0");
}
}
newModels.Clear();
foreach (System.Collections.DictionaryEntry de in oneModels)
{
oneModel = de.Key.ToString();
if (oneModel.Length > 2)
{
int h = oneModel.Length;
while (true)
{
if (h < 3) //最短字符不能小于5 >=四
{ goto nextcmd2; }
int st = 0;
while (true)
{
if (st + h > oneModel.Length)
{ goto nextcmd1; }
string a1 = oneModel.Substring(st, 1);
string a2 = oneModel.Substring(st + h - 1, 1);
if ((a1 != "<") | (a2 != ">")) //判断是否合法含有 < >
{
goto nextcmd3;
}
string onestr = oneModel.Substring(st, h);
if ((onestr.Substring(0, 1) != "<") | (onestr.Substring(onestr.Length - 1, 1) != ">") | (onestr.IndexOf(">") == -1) | (onestr.IndexOf("<") == -1))
{
goto nextcmd3; //取开头和末尾在<>中的数据
}
//遍历数据是否符合全部数据
foreach (System.Collections.DictionaryEntry de1 in mHTMs)
{
if ((de1.Value.ToString().IndexOf(onestr) == -1)&(de1.Value.ToString().Length>0))
{
goto nextcmd3;
}
}
//符合所有项都存在的条件压入字典
tdict.Add(tdictI, onestr);
//数据替换
Hashtable mtmpp = new Hashtable();
mtmpp.Clear();
foreach (System.Collections.DictionaryEntry de2 in mHTMs)
{
// mtmpp.Add(de2.Key, de2.Value.ToString().Replace(onestr, "*" + tdictI.ToString() + "*"));
string tmp_str = de2.Value.ToString();
if (tmp_str == null)
{
}
else
{
mtmpp.Add(de2.Key, tmp_str.Replace(onestr, "*" + tdictI.ToString() + "*"));
}
}
mHTMs = mtmpp;
tdictI = tdictI + 1; //字典序号
//oneModel 截取取出碎片压入 newModels
string oneModelTmp = oneModel.Replace(onestr, "*");
string[] mykc = oneModelTmp.Split('*');
foreach (string ddee in mykc)
{
if ((ddee.Length > 4) & (newModels.Contains(ddee) == false))
{
newModels.Add(ddee, "0");
}
}
goto NewStart; //因为碎片变化所以重新开始扫描
nextcmd3:
st = st + 1;
}
nextcmd1:
h = h - 1;
}
}
nextcmd2: ;
goto NewStart; //本次遍历结束去掉本字符串
}
//mHTMs 处理新的网页只留下 *0*
Hashtable mtmppC = new Hashtable();
mtmppC.Clear();
foreach (System.Collections.DictionaryEntry de2 in mHTMs)
{
string cckc = de2.Value.ToString();
//把cckc中的*0*提取出来
mtmppC.Add(de2.Key, getXnX(de2.Value.ToString()));
// mtmppC.Add(de2.Key, de2.Value.ToString());
}
mHTMs = mtmppC;
//从模板得到一个项目模板
// string mmoo = aass11BuilderModel(mHTMs);
string mmoo = GetComDataEND(mHTMs, tdict);
if (mmoo.Length == 0)
{
//从模板得到一个项目模板
mmoo = aass3BuilderModel(mHTMs);
}
mmoo = mmoo.Replace("<", "*");
mmoo = mmoo.Replace(">", "*");
if (mmoo.Length == 0)
{
return "";
}
else
{
foreach (System.Collections.DictionaryEntry dec in tdict)
{
mmoo = mmoo.Replace("*" + dec.Key.ToString() + "*", dec.Value.ToString());
}
return mmoo;
}
}
//11
/// <summary>
/// 《11》根据子模板建立一个超级模板
/// </summary>
/// <param name="mHTMs"></param>
/// <returns></returns>
private string aass11BuilderModel(Hashtable mHTMs)
{
Console.Write("..");
//得到最短的串
string oneModel = "";
foreach (System.Collections.DictionaryEntry de in mHTMs)
{
if (oneModel.Length == 0)
{
oneModel = de.Value.ToString();
}
else
{
if (oneModel.Length > de.Value.ToString().Length)
{
oneModel = de.Value.ToString();
}
}
}
foreach (System.Collections.DictionaryEntry de in mHTMs)
{
if (oneModel != de.Value.ToString()) //不相同
{
if (de.Value.ToString().IndexOf(oneModel) == -1) //不包含
{
goto cXStart;
}
}
}
return oneModel;
cXStart:
//得到匹配临时模板的原始长度
int onemLong = oneModel.Length;
Hashtable tdict = new Hashtable();
int tdictI = 0; //字典序号
tdict.Clear();
Hashtable oneModels = new Hashtable();
oneModels.Clear();
oneModels.Add(oneModel, "0"); //碎片如果相同就不用继续寻找添加
//存放临时碎片
Hashtable newModels = new Hashtable();
newModels.Clear();
oneModel = "";
NewStart: //开始遍历匹配串
if (oneModel.Length > 0)
{
oneModels.Remove(oneModel); //移除旧的字符串压入分解后的字符串
}
foreach (System.Collections.DictionaryEntry de in newModels) //加入新的碎片项
{
if (oneModels.Contains(de.Key) == false)
{
oneModels.Add(de.Key, "0");
}
}
newModels.Clear();
foreach (System.Collections.DictionaryEntry de in oneModels)
{
oneModel = de.Key.ToString();
if (oneModel.Length > 2)
{
int h = oneModel.Length;
while (true)
{
if (h < 3) //最短字符不能小于5 >=四
{ goto nextcmd2; }
int st = 0;
while (true)
{
if (st + h > oneModel.Length)
{ goto nextcmd1; }
string a1 = oneModel.Substring(st, 1);
string a2 = oneModel.Substring(st + h - 1, 1);
if ((a1 != "<") | (a2 != ">")) //判断是否合法含有 < >
{
goto nextcmd3;
}
string onestr = oneModel.Substring(st, h);
if ((onestr.Substring(0, 1) != "<") | (onestr.Substring(onestr.Length - 1, 1) != ">") | (onestr.IndexOf(">") == -1) | (onestr.IndexOf("<") == -1))
{
goto nextcmd3; //取开头和末尾在<>中的数据
}
//遍历数据是否符合全部数据
foreach (System.Collections.DictionaryEntry de1 in mHTMs)
{
if ((de1.Value.ToString().IndexOf(onestr) == -1)&(de1.Value.ToString().Length>0))
{
goto nextcmd3;
}
}
//符合所有项都存在的条件压入字典
tdict.Add(tdictI, onestr);
//数据替换
Hashtable mtmpp = new Hashtable();
mtmpp.Clear();
foreach (System.Collections.DictionaryEntry de2 in mHTMs)
{
//mtmpp.Add(de2.Key, de2.Value.ToString().Replace(onestr, "*" + tdictI.ToString() + "*"));
string tmp_str = de2.Value.ToString();
if (tmp_str == null)
{
}
else
{
mtmpp.Add(de2.Key, tmp_str.Replace(onestr, "*" + tdictI.ToString() + "*"));
}
}
mHTMs = mtmpp;
tdictI = tdictI + 1; //字典序号
//oneModel 截取取出碎片压入 newModels
string oneModelTmp = oneModel.Replace(onestr, "*");
string[] mykc = oneModelTmp.Split('*');
foreach (string ddee in mykc)
{
if ((ddee.Length > 4) & (newModels.Contains(ddee) == false))
{
newModels.Add(ddee, "0");
}
}
goto NewStart; //因为碎片变化所以重新开始扫描
nextcmd3:
st = st + 1;
}
nextcmd1:
h = h - 1;
}
}
nextcmd2: ;
goto NewStart; //本次遍历结束去掉本字符串
}
//mHTMs 处理新的网页只留下 *0*
Hashtable mtmppC = new Hashtable();
mtmppC.Clear();
foreach (System.Collections.DictionaryEntry de2 in mHTMs)
{
string cckc = de2.Value.ToString();
//把cckc中的*0*提取出来
mtmppC.Add(de2.Key, getXnX(de2.Value.ToString()));
// mtmppC.Add(de2.Key, de2.Value.ToString());
}
mHTMs = mtmppC;
//从模板得到一个项目模板
// string mmoo = aass12BuilderModel(mHTMs);
string mmoo = GetComDataEND(mHTMs, tdict);
if (mmoo.Length == 0)
{
//从模板得到一个项目模板
mmoo = aass3BuilderModel(mHTMs);
}
mmoo = mmoo.Replace("<", "*");
mmoo = mmoo.Replace(">", "*");
if (mmoo.Length == 0)
{
return "";
}
else
{
foreach (System.Collections.DictionaryEntry dec in tdict)
{
mmoo = mmoo.Replace("*" + dec.Key.ToString() + "*", dec.Value.ToString());
}
return mmoo;
}
}
////////////////////////////////////////////////////////////////////////////////
//12
////////////////////////////////////////////////////////////////////////////////
//12
/// <summary>
/// 《12》根据子模板建立一个超级模板
/// </summary>
/// <param name="mHTMs"></param>
/// <returns></returns>
private string aass12BuilderModel(Hashtable mHTMs)
{
Console.Write("..");
//得到最短的串
string oneModel = "";
foreach (System.Collections.DictionaryEntry de in mHTMs)
{
if (oneModel.Length == 0)
{
oneModel = de.Value.ToString();
}
else
{
if (oneModel.Length > de.Value.ToString().Length)
{
oneModel = de.Value.ToString();
}
}
}
foreach (System.Collections.DictionaryEntry de in mHTMs)
{
if (oneModel != de.Value.ToString()) //不相同
{
if (de.Value.ToString().IndexOf(oneModel) == -1) //不包含
{
goto cXStart;
}
}
}
return oneModel;
cXStart:
//得到匹配临时模板的原始长度
int onemLong = oneModel.Length;
Hashtable tdict = new Hashtable();
int tdictI = 0; //字典序号
tdict.Clear();
Hashtable oneModels = new Hashtable();
oneModels.Clear();
oneModels.Add(oneModel, "0"); //碎片如果相同就不用继续寻找添加
//存放临时碎片
Hashtable newModels = new Hashtable();
newModels.Clear();
oneModel = "";
NewStart: //开始遍历匹配串
if (oneModel.Length > 0)
{
oneModels.Remove(oneModel); //移除旧的字符串压入分解后的字符串
}
foreach (System.Collections.DictionaryEntry de in newModels) //加入新的碎片项
{
if (oneModels.Contains(de.Key) == false)
{
oneModels.Add(de.Key, "0");
}
}
newModels.Clear();
foreach (System.Collections.DictionaryEntry de in oneModels)
{
oneModel = de.Key.ToString();
if (oneModel.Length > 2)
{
int h = oneModel.Length;
while (true)
{
if (h < 3) //最短字符不能小于5 >=四
{ goto nextcmd2; }
int st = 0;
while (true)
{
if (st + h > oneModel.Length)
{ goto nextcmd1; }
string a1 = oneModel.Substring(st, 1);
string a2 = oneModel.Substring(st + h - 1, 1);
if ((a1 != "<") | (a2 != ">")) //判断是否合法含有 < >
{
goto nextcmd3;
}
string onestr = oneModel.Substring(st, h);
if ((onestr.Substring(0, 1) != "<") | (onestr.Substring(onestr.Length - 1, 1) != ">") | (onestr.IndexOf(">") == -1) | (onestr.IndexOf("<") == -1))
{
goto nextcmd3; //取开头和末尾在<>中的数据
}
//遍历数据是否符合全部数据
foreach (System.Collections.DictionaryEntry de1 in mHTMs)
{
if ((de1.Value.ToString().IndexOf(onestr) == -1) & (de1.Value.ToString().Length > 0))
{
goto nextcmd3;
}
}
//符合所有项都存在的条件压入字典
tdict.Add(tdictI, onestr);
//数据替换
Hashtable mtmpp = new Hashtable();
mtmpp.Clear();
foreach (System.Collections.DictionaryEntry de2 in mHTMs)
{
// mtmpp.Add(de2.Key, de2.Value.ToString().Replace(onestr, "*" + tdictI.ToString() + "*"));
string tmp_str = de2.Value.ToString();
if (tmp_str == null)
{
}
else
{
mtmpp.Add(de2.Key, tmp_str.Replace(onestr, "*" + tdictI.ToString() + "*"));
}
}
mHTMs = mtmpp;
tdictI = tdictI + 1; //字典序号
//oneModel 截取取出碎片压入 newModels
string oneModelTmp = oneModel.Replace(onestr, "*");
string[] mykc = oneModelTmp.Split('*');
foreach (string ddee in mykc)
{
if ((ddee.Length > 4) & (newModels.Contains(ddee) == false))
{
newModels.Add(ddee, "0");
}
}
goto NewStart; //因为碎片变化所以重新开始扫描
nextcmd3:
st = st + 1;
}
nextcmd1:
h = h - 1;
}
}
nextcmd2: ;
goto NewStart; //本次遍历结束去掉本字符串
}
//mHTMs 处理新的网页只留下 *0*
Hashtable mtmppC = new Hashtable();
mtmppC.Clear();
foreach (System.Collections.DictionaryEntry de2 in mHTMs)
{
string cckc = de2.Value.ToString();
//把cckc中的*0*提取出来
mtmppC.Add(de2.Key, getXnX(de2.Value.ToString()));
// mtmppC.Add(de2.Key, de2.Value.ToString());
}
mHTMs = mtmppC;
//从模板得到一个项目模板
//string mmoo = aass13BuilderModel(mHTMs);
string mmoo = GetComDataEND(mHTMs, tdict);
if (mmoo.Length == 0)
{
//从模板得到一个项目模板
mmoo = aass3BuilderModel(mHTMs);
}
mmoo = mmoo.Replace("<", "*");
mmoo = mmoo.Replace(">", "*");
if (mmoo.Length == 0)
{
return "";
}
else
{
foreach (System.Collections.DictionaryEntry dec in tdict)
{
mmoo = mmoo.Replace("*" + dec.Key.ToString() + "*", dec.Value.ToString());
}
return mmoo;
}
}
////////////////////////////////////////////////////////////////////////////////
//13
////////////////////////////////////////////////////////////////////////////////
//13
/// <summary>
/// 《13》根据子模板建立一个超级模板
/// </summary>
/// <param name="mHTMs"></param>
/// <returns></returns>
private string aass13BuilderModel(Hashtable mHTMs)
{
//得到最短的串
string oneModel = "";
foreach (System.Collections.DictionaryEntry de in mHTMs)
{
if (oneModel.Length == 0)
{
oneModel = de.Value.ToString();
}
else
{
if (oneModel.Length > de.Value.ToString().Length)
{
oneModel = de.Value.ToString();
}
}
}
foreach (System.Collections.DictionaryEntry de in mHTMs)
{
if (oneModel != de.Value.ToString()) //不相同
{
if (de.Value.ToString().IndexOf(oneModel) == -1) //不包含
{
goto cXStart;
}
}
}
return oneModel;
cXStart:
//得到匹配临时模板的原始长度
int onemLong = oneModel.Length;
Hashtable tdict = new Hashtable();
int tdictI = 0; //字典序号
tdict.Clear();
Hashtable oneModels = new Hashtable();
oneModels.Clear();
oneModels.Add(oneModel, "0"); //碎片如果相同就不用继续寻找添加
//存放临时碎片
Hashtable newModels = new Hashtable();
newModels.Clear();
oneModel = "";
NewStart: //开始遍历匹配串
if (oneModel.Length > 0)
{
oneModels.Remove(oneModel); //移除旧的字符串压入分解后的字符串
}
foreach (System.Collections.DictionaryEntry de in newModels) //加入新的碎片项
{
if (oneModels.Contains(de.Key) == false)
{
oneModels.Add(de.Key, "0");
}
}
newModels.Clear();
foreach (System.Collections.DictionaryEntry de in oneModels)
{
oneModel = de.Key.ToString();
if (oneModel.Length > 2)
{
int h = oneModel.Length;
while (true)
{
if (h < 3) //最短字符不能小于5 >=四
{ goto nextcmd2; }
int st = 0;
while (true)
{
if (st + h > oneModel.Length)
{ goto nextcmd1; }
string a1 = oneModel.Substring(st, 1);
string a2 = oneModel.Substring(st + h - 1, 1);
if ((a1 != "<") | (a2 != ">")) //判断是否合法含有 < >
{
goto nextcmd3;
}
string onestr = oneModel.Substring(st, h);
if ((onestr.Substring(0, 1) != "<") | (onestr.Substring(onestr.Length - 1, 1) != ">") | (onestr.IndexOf(">") == -1) | (onestr.IndexOf("<") == -1))
{
goto nextcmd3; //取开头和末尾在<>中的数据
}
//遍历数据是否符合全部数据
foreach (System.Collections.DictionaryEntry de1 in mHTMs)
{
if ((de1.Value.ToString().IndexOf(onestr) == -1) & (de1.Value.ToString().Length > 0))
{
goto nextcmd3;
}
}
//符合所有项都存在的条件压入字典
tdict.Add(tdictI, onestr);
//数据替换
Hashtable mtmpp = new Hashtable();
mtmpp.Clear();
foreach (System.Collections.DictionaryEntry de2 in mHTMs)
{
//mtmpp.Add(de2.Key, de2.Value.ToString().Replace(onestr, "*" + tdictI.ToString() + "*"));
string tmp_str = de2.Value.ToString();
if (tmp_str == null)
{
}
else
{
mtmpp.Add(de2.Key, tmp_str.Replace(onestr, "*" + tdictI.ToString() + "*"));
}
}
mHTMs = mtmpp;
tdictI = tdictI + 1; //字典序号
//oneModel 截取取出碎片压入 newModels
string oneModelTmp = oneModel.Replace(onestr, "*");
string[] mykc = oneModelTmp.Split('*');
foreach (string ddee in mykc)
{
if ((ddee.Length > 4) & (newModels.Contains(ddee) == false))
{
newModels.Add(ddee, "0");
}
}
goto NewStart; //因为碎片变化所以重新开始扫描
nextcmd3:
st = st + 1;
}
nextcmd1:
h = h - 1;
}
}
nextcmd2: ;
goto NewStart; //本次遍历结束去掉本字符串
}
//mHTMs 处理新的网页只留下 *0*
Hashtable mtmppC = new Hashtable();
mtmppC.Clear();
foreach (System.Collections.DictionaryEntry de2 in mHTMs)
{
string cckc = de2.Value.ToString();
//把cckc中的*0*提取出来
mtmppC.Add(de2.Key, getXnX(de2.Value.ToString()));
// mtmppC.Add(de2.Key, de2.Value.ToString());
}
mHTMs = mtmppC;
//从模板得到一个项目模板
// string mmoo = aass14BuilderModel(mHTMs);
string mmoo = GetComDataEND(mHTMs, tdict);
if (mmoo.Length == 0)
{
//从模板得到一个项目模板
mmoo = aass3BuilderModel(mHTMs);
}
mmoo = mmoo.Replace("<", "*");
mmoo = mmoo.Replace(">", "*");
if (mmoo.Length == 0)
{
return "";
}
else
{
foreach (System.Collections.DictionaryEntry dec in tdict)
{
mmoo = mmoo.Replace("*" + dec.Key.ToString() + "*", dec.Value.ToString());
}
return mmoo;
}
}
////////////////////////////////////////////////////////////////////////////////
//14
////////////////////////////////////////////////////////////////////////////////
//14
/// <summary>
/// 《14》根据子模板建立一个超级模板
/// </summary>
/// <param name="mHTMs"></param>
/// <returns></returns>
private string aass14BuilderModel(Hashtable mHTMs)
{
//得到最短的串
string oneModel = "";
foreach (System.Collections.DictionaryEntry de in mHTMs)
{
if (oneModel.Length == 0)
{
oneModel = de.Value.ToString();
}
else
{
if (oneModel.Length > de.Value.ToString().Length)
{
oneModel = de.Value.ToString();
}
}
}
foreach (System.Collections.DictionaryEntry de in mHTMs)
{
if (oneModel != de.Value.ToString()) //不相同
{
if (de.Value.ToString().IndexOf(oneModel) == -1) //不包含
{
goto cXStart;
}
}
}
return oneModel;
cXStart:
//得到匹配临时模板的原始长度
int onemLong = oneModel.Length;
Hashtable tdict = new Hashtable();
int tdictI = 0; //字典序号
tdict.Clear();
Hashtable oneModels = new Hashtable();
oneModels.Clear();
oneModels.Add(oneModel, "0"); //碎片如果相同就不用继续寻找添加
//存放临时碎片
Hashtable newModels = new Hashtable();
newModels.Clear();
oneModel = "";
NewStart: //开始遍历匹配串
if (oneModel.Length > 0)
{
oneModels.Remove(oneModel); //移除旧的字符串压入分解后的字符串
}
foreach (System.Collections.DictionaryEntry de in newModels) //加入新的碎片项
{
if (oneModels.Contains(de.Key) == false)
{
oneModels.Add(de.Key, "0");
}
}
newModels.Clear();
foreach (System.Collections.DictionaryEntry de in oneModels)
{
oneModel = de.Key.ToString();
if (oneModel.Length > 2)
{
int h = oneModel.Length;
while (true)
{
if (h < 3) //最短字符不能小于5 >=四
{ goto nextcmd2; }
int st = 0;
while (true)
{
if (st + h > oneModel.Length)
{ goto nextcmd1; }
string a1 = oneModel.Substring(st, 1);
string a2 = oneModel.Substring(st + h - 1, 1);
if ((a1 != "<") | (a2 != ">")) //判断是否合法含有 < >
{
goto nextcmd3;
}
string onestr = oneModel.Substring(st, h);
if ((onestr.Substring(0, 1) != "<") | (onestr.Substring(onestr.Length - 1, 1) != ">") | (onestr.IndexOf(">") == -1) | (onestr.IndexOf("<") == -1))
{
goto nextcmd3; //取开头和末尾在<>中的数据
}
//遍历数据是否符合全部数据
foreach (System.Collections.DictionaryEntry de1 in mHTMs)
{
if ((de1.Value.ToString().IndexOf(onestr) == -1) & (de1.Value.ToString().Length > 0))
{
goto nextcmd3;
}
}
//符合所有项都存在的条件压入字典
tdict.Add(tdictI, onestr);
//数据替换
Hashtable mtmpp = new Hashtable();
mtmpp.Clear();
foreach (System.Collections.DictionaryEntry de2 in mHTMs)
{
// mtmpp.Add(de2.Key, de2.Value.ToString().Replace(onestr, "*" + tdictI.ToString() + "*"));
string tmp_str = de2.Value.ToString();
if (tmp_str == null)
{
}
else