当前文件路径：NetCMSv15/NetCMS.Content/Collect/PageNews.cs 
1
//======================================================
2//== (c)2008 aspxcms inc by NeTCMS v1.0 ==
3//== Forum:bbs.aspxcms.com ==
4//== Website:www.aspxcms.com ==
5//======================================================
6using System;
7using System.Collections.Generic;
8using System.Text;
9using System.Text.RegularExpressions;
10using System.Collections;
11
12namespace NetCMS.Content.Collect
13
...{
14
public class PageNews : Page
15
...{
16 私有变量#region 私有变量
17 private string _title;
18 private string _content;
19 private string _author;
20 private string _source;
21 private string _titlerule = null;
22 private string _contentrule = null;
23 private string otherpgcon = "";
24 private DateTime _addtime;
25
#endregion 私有变量
26 public PageNews(string url) : base(url) ...{ }
27 public PageNews(string url, string encode) : base(url, encode) ...{ }
28 public string Title
29 ...{ get ...{ return _title; } }
30 public string Content
31 ...{ get ...{ return _content; } set ...{ _content = value; } }
32 public string Author ...{ get ...{ return _author; } }
33 public string Source ...{ get ...{ return _source; } }
34 public DateTime AddTime ...{ get ...{ return _addtime; } }
35 public string RuleOfTitle
36 ...{
37 set ...{ _titlerule = value; }
38 }
39 public string RuleOfContent
40 ...{
41 set ...{ _contentrule = value; }
42 }
43 /**//// <summary>
44 /// 过滤数据，提取作者
45 /// </summary>
46 /// <param name="pattern"></param>
47 /// <param name="settled"></param>
48 public void FigureAuthor(string pattern, bool settled)
49 ...{
50 if (!settled && pattern.IndexOf("[作者]") >= 0)
51 ...{
52 Match m = Utility.GetMatchRigid(_Doc, pattern, "[作者]");
53 if (m.Success)
54 ...{
55 _author = m.Groups["TARGET"].Value;
56 }
57 }
58 else
59 _author = pattern;
60 }
61 public void FigureSource(string pattern, bool settled)
62 ...{
63 if (!settled && pattern.IndexOf("[来源]") >= 0)
64 ...{
65 Match m = Utility.GetMatchRigid(_Doc, pattern, "[来源]");
66 if (m.Success)
67 ...{
68 _source = m.Groups["TARGET"].Value;
69 }
70 }
71 else
72 ...{
73 _source = pattern;
74 }
75 }
76 public void FigureAddTime(string pattern, bool settled)
77 ...{
78 string tm = "";
79 if (!settled && pattern.IndexOf("[加入时间]") >= 0)
80 ...{
81 Match m = Utility.GetMatchRigid(_Doc, pattern, "[加入时间]");
82 if (m.Success)
83 ...{
84 tm = m.Groups["TARGET"].Value;
85 }
86 }
87 else
88 ...{
89 tm = pattern;
90 }
91 try
92 ...{
93 _addtime = DateTime.Parse(tm);
94 }
95 catch
96 ...{
97 _addtime = DateTime.Now;
98 }
99 }
100 public void FigureTitle()
101 ...{
102 if (_titlerule == null || _titlerule.IndexOf("[标题]") < 0)
103 throw new Exception("采集新闻标题规则还没有设定!");
104 Match m = Utility.GetMatchRigid(_Doc, _titlerule, "[标题]");
105 if (m.Success)
106 ...{
107 _title = m.Groups["TARGET"].Value;
108 }
109 }
110 public void FigureContent()
111 ...{
112 if (_contentrule == null || _contentrule.IndexOf("[内容]") < 0)
113 throw new Exception("采集新闻内容规则还没有设定!");
114 Match m = Utility.GetMatch(_Doc, _contentrule, "[内容]");
115 if (m.Success)
116 ...{
117 _content = m.Groups["TARGET"].Value;
118 }
119 }
120 private void FilterHtml(string element, int type)
121 ...{
122 string pattern = "";
123 switch (type)
124 ...{
125 case 0:
126 pattern = element + "\\s?=\\s?(['\"][^'\"]*?['\"]|[^'\"]\\S*)";
127 break;
128 case 1:
129 pattern = "<" + element + "[^>]*>|</" + element + ">";
130 break;
131 case 2:
132 pattern = "<(?<tag>" + element + @")[^>]*>[\s\S]*</\k<tag>>";
133 break;
134 default:
135 return;
136 }
137 try
138 ...{
139 Regex reg = new Regex(pattern, RegexOptions.IgnoreCase | RegexOptions.Compiled);
140 _content = reg.Replace(_content, "");
141 }
142 catch
143 ...{ }
144 }
145 public void Replace(string profile, string newstr, bool bIgnoreCase)
146 ...{
147 string pattern = Regex.Escape(profile);
148 string instead = newstr.Replace("$", "$$");
149 pattern = pattern.Replace(@"\[变量]", @"[\s\S]*?");
150 string[] _pattern = pattern.Split(new char[] ...{ '[', '过', '滤', '字', '符', '串', ']' }, StringSplitOptions.RemoveEmptyEntries);
151 string p = "";
152 string n = "";
153 for (int i = 0; i < _pattern.Length; i++)
154 ...{
155 string s = _pattern[i];
156 if (!s.Equals(""))
157 ...{
158 p += "(?<ch" + i + ">" + s + @")[\s\S]+?";
159 n += "${ch" + i + "}" + instead;
160 }
161 }
162 Regex reg;
163 if (bIgnoreCase)
164 reg = new Regex(p, RegexOptions.Compiled | RegexOptions.IgnoreCase);
165 else
166 reg = new Regex(p, RegexOptions.Compiled);
167 _content = reg.Replace(_content, n);
168 }
169 public void Filter(bool ridhtml, bool ridstyle, bool riddiv, bool rida, bool ridclass, bool ridfont, bool ridspan, bool ridobject, bool ridiframe, bool ridscript)
170 ...{
171 //if(ridhtml);
172 if (ridstyle || ridhtml)
173 FilterHtml("style", 0);
174 if (riddiv || ridhtml)
175 FilterHtml("div", 1);
176 if (rida || ridhtml)
177 FilterHtml("a", 1);
178 if (ridclass || ridhtml)
179 FilterHtml("class", 0);
180 if (ridfont || ridhtml)
181 FilterHtml("font", 1);
182 if (ridspan || ridhtml)
183 FilterHtml("span", 1);
184 if (ridobject || ridhtml)
185 FilterHtml("object", 2);
186 if (ridiframe || ridhtml)
187 FilterHtml("iframe", 2);
188 if (ridscript || ridhtml)
189 FilterHtml("script", 2);
190 }
191 public string GetOtherPagination(string profile)
192 ...{
193 otherpgcon = "";
194 GetOtherPage(_Url, _Doc, profile);
195 return otherpgcon;
196 }
197 private void GetOtherPage(string otherurl, string PageDoc, string pattern)
198 ...{
199 Match m = Utility.GetMatchUrl(PageDoc, pattern, "[分页新闻]");
200 if (m.Success)
201 ...{
202 string obturl = Utility.StickUrl(otherurl, m.Groups["TARGET"].Value);
203 if (!obturl.Trim().Equals(otherurl.Trim()))
204 ...{
205 PageNews pgns = new PageNews(obturl, _Encode);
206 pgns.RuleOfContent = this._contentrule;
207 if (pgns.Fetch())
208 ...{
209 pgns.FigureContent();
210 otherpgcon += pgns.Content;
211 GetOtherPage(obturl, pgns._Doc, pattern);
212 }
213 }
214 }
215 }
216 public string GetIndexPagination(string profile)
217 ...{
218 string OtherContent = "";
219 Match m = Utility.GetMatchUrl(_Doc, profile, "[分页新闻]");
220 while (m.Success)
221 ...{
222 string otherurl = Utility.StickUrl(_Url, m.Groups["TARGET"].Value);
223 if (!otherurl.Trim().Equals(this._Url))
224 ...{
225 PageNews pgns = new PageNews(otherurl, _Encode);
226 if (pgns.Fetch())
227 ...{
228 pgns.FigureContent();
229 OtherContent += pgns.Content;
230 }
231 }
232 m = m.NextMatch();
233 }
234 return OtherContent;
235 }
236 public string AllDocument
237 ...
