Category

Development Platform
Home > SourceCode/Document > MultiLanguage > Information Retrieval > View Code
ParseHTML.cs
Package: HTML.Parser.zip　[view]
Upload User: sarage88
Upload Date: 2013-09-22
Package Size: 34k
Code Size: 4k
Category:

Information Retrieval

Development Platform:

C#

ParseHTML.cs：Code Content
  1. using System;
  3. namespace HTML
  4. {
  5.     /// <summary>
  6.     /// Summary description for ParseHTML.
  7.     /// </summary>
  9.     public class ParseHTML : Parse
  10.     {
  11.         public AttributeList GetTag()
  12.         {
  13.             AttributeList tag = new AttributeList();
  14.             tag.Name = m_tag;
  16.             foreach (Attribute x in List)
  17.             {
  18.                 tag.Add((Attribute)x.Clone());
  19.             }
  21.             return tag;
  22.         }
  24.         public String BuildTag()
  25.         {
  26.             String buffer = "<";
  27.             buffer += m_tag;
  28.             int i = 0;
  29.             while (this[i] != null)
  30.             {// has attributes
  31.                 buffer += " ";
  32.                 if (this[i].Value == null)
  33.                 {
  34.                     if (this[i].Delim != 0)
  35.                         buffer += this[i].Delim;
  36.                     buffer += this[i].Name;
  37.                     if (this[i].Delim != 0)
  38.                         buffer += this[i].Delim;
  39.                 }
  40.                 else
  41.                 {
  42.                     buffer += this[i].Name;
  43.                     if (this[i].Value != null)
  44.                     {
  45.                         buffer += "=";
  46.                         if (this[i].Delim != 0)
  47.                             buffer += this[i].Delim;
  48.                         buffer += this[i].Value;
  49.                         if (this[i].Delim != 0)
  50.                             buffer += this[i].Delim;
  51.                     }
  52.                 }
  53.                 i++;
  54.             }
  55.             buffer += ">";
  56.             return buffer;
  57.         }
  59.         protected void ParseTag()
  60.         {
  61.             m_tag = "";
  62.             Clear();
  64.             // Is it a comment?
  65.             if ((GetCurrentChar() == '!') &&
  66.               (GetCurrentChar(1) == '-') &&
  67.               (GetCurrentChar(2) == '-'))
  68.             {
  69.                 while (!Eof())
  70.                 {
  71.                     if ((GetCurrentChar() == '-') &&
  72.                       (GetCurrentChar(1) == '-') &&
  73.                       (GetCurrentChar(2) == '>'))
  74.                         break;
  75.                     if (GetCurrentChar() != 'r')
  76.                         m_tag += GetCurrentChar();
  77.                     Advance();
  78.                 }
  79.                 m_tag += "--";
  80.                 Advance();
  81.                 Advance();
  82.                 Advance();
  83.                 ParseDelim = (char)0;
  84.                 return;
  85.             }
  87.             // Find the tag name
  88.             while (!Eof())
  89.             {
  90.                 if (IsWhiteSpace(GetCurrentChar()) ||
  91.                                  (GetCurrentChar() == '>'))
  92.                     break;
  93.                 m_tag += GetCurrentChar();
  94.                 Advance();
  95.             }
  97.             EatWhiteSpace();
  99.             // Get the attributes
  100.             while (GetCurrentChar() != '>')
  101.             {
  102.                 ParseName = "";
  103.                 ParseValue = "";
  104.                 ParseDelim = (char)0;
  106.                 ParseAttributeName();
  108.                 if (GetCurrentChar() == '>')
  109.                 {
  110.                     AddAttribute();
  111.                     break;
  112.                 }
  114.                 // Get the value(if any)
  115.                 ParseAttributeValue();
  116.                 AddAttribute();
  117.             }
  118.             Advance();
  119.         }
  122.         public char Parse()
  123.         {
  124.             if (GetCurrentChar() == '<')
  125.             {
  126.                 Advance();
  128.                 char ch = char.ToUpper(GetCurrentChar());
  129.                 if ((ch >= 'A') && (ch <= 'Z') || (ch == '!') || (ch == '/'))
  130.                 {
  131.                     ParseTag();
  132.                     return (char)0;
  133.                 }
  135.                 else return (AdvanceCurrentChar());
  136.             }
  137.             else return (AdvanceCurrentChar());
  138.         }
  139.     }
  140. }