1
//===========================================================
2//== (c)2007 Foosun Inc. by dotNETCMS 1.0 ==
3//== Forum:bbs.foosun.net ==
4//== website:www.foosun.net ==
5//== Address:NO.109 HuiMin ST.,Chengdu ,China ==
6//== TEL:86-28-85098980/66026180 ==
7//== qq:655071,MSN:ikoolls@gmail.com ==
8//== Code By JiangDong ==
9//===========================================================
10using System;
11using System.Collections;
12using System.Collections.Generic;
13using System.Text;
14using System.Text.RegularExpressions;
15using System.Net;
16using System.IO;
17
18namespace Foosun.CMS.Collect
19
...{
20
public class Utility
21
...{
22 /**//// <summary>
23 /// 取得网页的内容
24 /// </summary>
25 /// <param name="sUrl">url地址</param>
26 /// <param name="sEncode">编码名称</param>
27 /// <param name="sDocument">返回的网页内容或者是异常</param>
28
/// <returns>有异常返回false</returns>
29 public static string GetPageContent(Uri Url, string sEncode)
30 ...{
31 try
32 ...{
33
34 Encoding encoding = System.Text.Encoding.GetEncoding(sEncode);
35 return GetPageContent(Url, encoding);
36 }
37 catch (WebException ex)
38 ...{
39 throw ex;
40 }
41 catch (Exception ex)
42 ...{
43 throw ex;
44 }
45 }
46 /**//// <summary>
47 /// 取得网页的内容
48 /// </summary>
49 /// <param name="sUrl">url地址</param>
50 /// <param name="encoding">编码方式</param>
51 /// <param name="sDocument">返回的网页内容或者是异常</param>
52 /// <returns>有异常返回false</returns>
53 public static string GetPageContent(Uri Url, Encoding encoding)
54 ...{
55 WebClient webclient = new WebClient();
56 try
57 ...{
58
59 webclient.Encoding = encoding;
60 return webclient.DownloadString(Url);
61 }
62 catch (WebException ex)
63 ...{
64 throw ex;
65 }
66 catch (Exception ex)
67 ...{
68 throw ex;
69 }
70 finally
71 ...{
72 webclient.Dispose();
73 }
74 }
75 /**//// <summary>
76 /// 处理URL地址，当BranchUrl为一个全名的URL时则返回本身，否则恰当的衔接到BaseUrl后面
77 /// </summary>
78 /// <param name="BaseUrl">完整的URL</param>
79 /// <param name="BranchUrl">分支URL</param>
80 /// <returns></returns>
81 public static string StickUrl(string BaseUrl, string BranchUrl)
82 ...{
83 if (Regex.Match(BranchUrl, @"^(http|https|ftp|rtsp|mms)://", RegexOptions.IgnoreCase | RegexOptions.Compiled).Success)
84 ...{
85 return BranchUrl;
86 }
87 else
88 ...{
89 BaseUrl = BaseUrl.Replace("\\", "/");
90 BranchUrl = BranchUrl.Replace("\\", "/");
91 //2007-09-27 ken暂时修改
92 if (Foosun.Common.Input.GetSubString(BranchUrl, 1).ToString() == "/")
93 ...{
94 return GetLastUrl(BaseUrl, BranchUrl);
95 }
96 //--------------------------
97 BranchUrl = BranchUrl.TrimStart('/');
98 if (BranchUrl.IndexOf("../") != 0)
99 ...{
100 return UrlPlus(BaseUrl, BranchUrl);
101 }
102 else
103 ...{
104 if (Regex.Match(BaseUrl, @"/$", RegexOptions.Compiled).Success)
105 ...{
106 BaseUrl = BaseUrl.TrimEnd('/');
107 }
108 else if (Regex.Match(BaseUrl, @"/[^\./]+\.[^/]+$", RegexOptions.Compiled).Success)
109 ...{
110 BaseUrl = Regex.Replace(BaseUrl, @"/[^\./]+\.[^/]+$", "", RegexOptions.Compiled);
111 }
112 while (BranchUrl.IndexOf("../") >= 0)
113 ...{
114 BranchUrl = Regex.Replace(BranchUrl, @"^\.\./", "", RegexOptions.Compiled);
115 BaseUrl = Regex.Replace(BaseUrl, @"/[^/]*$", "", RegexOptions.Compiled);
116 }
117 return BaseUrl + "/" + BranchUrl;
118 }
119 }
120 }
121 //--------------------------------
122 private static string GetLastUrl(string BaseUrl, string BranchUrl)
123 ...{
124 BranchUrl = BranchUrl.TrimStart('/');
125 string Star_url = "";
126 string End_Url = BaseUrl;
127 if (BaseUrl.IndexOf("//") > 0)
128 ...{
129 BaseUrl = BaseUrl.Replace("//", "|");
130 string[] Url_Arr = BaseUrl.Split('|');
131 Star_url = Url_Arr[0].ToString();
132 End_Url = Url_Arr[1].ToString();
133 }
134 if (End_Url.IndexOf("/") > 0)
135 ...{
136 string[] End_Arr = End_Url.Split('/');
137 End_Url = End_Arr[0].ToString();
138 if (Star_url != string.Empty)
139 ...{
140 return Star_url + "//" + End_Url + "/" + BranchUrl;
141 }
142 else
143 ...{
144 return End_Url + "/" + BranchUrl;
145 }
146 }
147 else
148 ...{
149 if (Star_url != string.Empty)
150 ...{
151 return Star_url + "//" + End_Url + "/" + BranchUrl;
152 }
153 else
154 ...{
155 return End_Url + "/" + BranchUrl;
156 }
157 }
158 }
159
160
161
162 private static string UrlPlus(string front, string tail)
163 ...{
164 if (Regex.Match(front, "(http|https|ftp|rtsp|mms)://[^/]+$", RegexOptions.Compiled | RegexOptions.IgnoreCase).Success)
165 ...{
166 return front + "/" + tail;
167 }
168 else if (Regex.Match(front, "(http|https|ftp|rtsp|mms)://[^/]+/$", RegexOptions.Compiled | RegexOptions.IgnoreCase).Success)
169 ...{
170 return front + tail;
171 }
172 else if (Regex.Match(front, "(http|https|ftp|rtsp|mms)://.+/$", RegexOptions.Compiled | RegexOptions.IgnoreCase).Success)
173 ...{
174 return front + tail;
175 }
176 else if (Regex.Match(front, @"/[^/\.]+$", RegexOptions.Compiled | RegexOptions.IgnoreCase).Success)
177 ...{
178 return front + "/" + tail;
179 }
180 else if (Regex.Match(front, @"/[^/\.]+\.[^/]+$", RegexOptions.Compiled | RegexOptions.IgnoreCase).Success)
181 ...{
182 return Regex.Replace(front, @"/[^/\.]+\.[^/]+$", "", RegexOptions.IgnoreCase | RegexOptions.Compiled) + "/" + tail;
183 }
184 else
185 ...{
186 return front + "/" + tail;
187 }
188 }
189 /**//// <summary>
190 /// 获取一个目标的匹配结果
191 /// </summary>
192 /// <param name="input">要匹配的字符串</param>
193 /// <param name="pattern"></param>
194 /// <param name="find"></param>
195 /// <returns></returns>
196 public static Match GetMatch(string input, string pattern, string find)
197 ...{
198 string _pattn = Regex.Escape(pattern);
199 _pattn = _pattn.Replace(@"\[变量]", @"[\s\S]*?");
200 _pattn = Regex.Replace(_pattn, @"((\\r\\n)|(\\ ))+", @"\s*", RegexOptions.Compiled);
201 if (Regex.Match(pattern.TrimEnd(), Regex.Escape(find) + "$", RegexOptions.Compiled).Success)
202 _pattn = _pattn.Replace(@"\" + find, @"(?<TARGET>[\s\S]+)");
203 else
204 _pattn = _pattn.Replace(@"\" + find, @"(?<TARGET>[\s\S]+?)");
205 Regex r = new Regex(_pattn, RegexOptions.IgnoreCase | RegexOptions.Compiled);
206 Match m = r.Match(input);
207 return m;
208 }
209 /**//// <summary>
210 /// 按严格的匹配方式获取一个目标的匹配结果
211 /// </summary>
212 /// <param name="input"></param>
213 /// <param name="pattern"></param>
214 /// <param name="find"></param>
215 /// <returns></returns>
216 public static Match GetMatchRigid(string input, string pattern, string find)
217 ...{
218 string _pattn = Regex.Escape(pattern);
219 _pattn = _pattn.Replace(@"\[变量]", @"[\s\S]*?");
220 if (Regex.Match(pattern.TrimEnd(), Regex.Escape(find) + "$", RegexOptions.Compiled).Success)
221 _pattn = _pattn.Replace(@"\" + find, @"(?<TARGET>[\s\S]+)");
222 else
223 _pattn = _pattn.Replace(@"\" + find, @"(?<TARGET>[\s\S]+?)");
224 Regex r = new Regex(_pattn, RegexOptions.IgnoreCase | RegexOptions.Compiled);
225 Match m = r.Match(input);
226 return m;
227 }
228 /**//// <summary>
229 /// 匹配超级链接地址
230 /// </summary>
231 /// <param name="input"></param>
232 /// <param name="pattern"></param>
233 /// <param name="find"></param>
234 /// <returns></returns>
235 public static Match GetMatchUrl(string input, string pattern, string find)
236 ...{
237 string _pattn = Regex.Escape(pattern);
238 _pattn = _pattn.Replace(@"\[变量]", @"[\s\S]*?");
239 if (Regex.Match(pattern.TrimEnd(), Regex.Escape(find) + "$", RegexOptions.Compiled).Success)
240 _pattn = _pattn.Replace(@"\" + find, @"(?<TARGET>[^'""\ >]+)");
241 else
242 _pattn = _pattn.Replace(@"\" + find, @"(?<TARGET>[^'""\ >]+?)");
243 Regex r = new Regex(_pattn, RegexOptions.IgnoreCase | RegexOptions.Compiled);
244 Match m = r.Match(input);
245 return m;
246 }
247 }
248 /**//*
249 // The RequestState class passes data across async calls.
250 public class RequestState
251 {
252 const int BufferSize = 1024;
253 public StringBuilder RequestData;
254 public byte[] BufferRead;
255 public WebRequest Request;
256 public Stream ResponseStream;
257 // Create Decoder for appropriate enconding type.
258 public Decoder StreamDecode = Encoding.UTF8.GetDecoder();
259
260 public RequestState()
261 {
262 BufferRead = new byte[BufferSize];
263 RequestData = new StringBuilder(String.Empty);
264 Request = null;
265 ResponseStream = null;
266 }
267 }
268
269 // ClientGetAsync issues the async request.
270 class ClientGetAsync
271 {
272 public static ManualResetEvent allDone = new ManualResetEvent(false);
273 const int BUFFER_SIZE = 1024;
274 public static void BeginReq(string[] HttpURL)
275 {
276 foreach (string singleurl in HttpURL)
277 {
278 try
279 {
280 Uri httpSite = new Uri(singleurl);
281 HttpWebRequest wreq = (HttpWebRequest)WebRequest.Create(httpSite);
282 RequestState rs = new RequestState();
283 rs.Request = wreq;
284 IAsyncResult r = (IAsyncResult)wreq.BeginGetResponse(new AsyncCallback(RespCallback), rs);
285 allDone.WaitOne();
286 wreq.EndGetResponse(r);
287 }
288 catch (WebException e)
289 { }
290 catch (Exception e)
291 { }
292 }
293 }
294 private static void RespCallback(IAsyncResult ar)
295 {
296 RequestState rs = (RequestState)ar.AsyncState;
297 WebRequest req = rs.Request;
298 WebResponse resp = req.EndGetResponse(ar);
299 Stream ResponseStream = resp.GetResponseStream();
300 rs.ResponseStream = ResponseStream;
301 IAsyncResult iarRead = ResponseStream.BeginRead(rs.BufferRead, 0,
302 BUFFER_SIZE, new AsyncCallback(ReadCallBack), rs);
303 }
304
305
306 private static void ReadCallBack(IAsyncResult asyncResult)
307 {
308 RequestState rs = (RequestState)asyncResult.AsyncState;
309 Stream responseStream = rs.ResponseStream;
310 int read = responseStream.EndRead(asyncResult);
311 if (read > 0)
312 {
313 Char[] charBuffer = new Char[BUFFER_SIZE];
314 int len =
315 rs.StreamDecode.GetChars(rs.BufferRead, 0, read, charBuffer, 0);
316 String str = new String(charBuffer, 0, len);
317 rs.RequestData.Append(
318 Encoding.ASCII.GetString(rs.BufferRead, 0, read));
319 IAsyncResult ar = responseStream.BeginRead(
320 rs.BufferRead, 0, BUFFER_SIZE,
321 new AsyncCallback(ReadCallBack), rs);
322 }
323 else
324 {
325 if (rs.RequestData.Length > 0)
326 {
327 string strContent;
328 strContent = rs.RequestData.ToString();
329 }
330 responseStream.Close();
331 allDone.Set();
332 }
333 return;
334 }
335 }*/
336
337
}
338
FooSunCMS/Foosun.CMS/Collect/Utility.cs[13K，2009-6-12 11:42:42]，打开代码结构图
打包下载该项目完整源码：风讯dotNETCMSv1.0免安装版源码