]>
Commit | Line | Data |
---|---|---|
1 | ///////////////////////////////////////////////////////////////////////////// | |
2 | // Name: url.cpp | |
3 | // Purpose: URL parser | |
4 | // Author: Guilhem Lavaux | |
5 | // Modified by: | |
6 | // Created: 20/07/1997 | |
7 | // RCS-ID: $Id$ | |
8 | // Copyright: (c) 1997, 1998 Guilhem Lavaux | |
9 | // Licence: wxWindows licence | |
10 | ///////////////////////////////////////////////////////////////////////////// | |
11 | ||
12 | #if defined(__GNUG__) && !defined(NO_GCC_PRAGMA) | |
13 | #pragma implementation "url.h" | |
14 | #endif | |
15 | ||
16 | // For compilers that support precompilation, includes "wx.h". | |
17 | #include "wx/wxprec.h" | |
18 | ||
19 | #ifdef __BORLANDC__ | |
20 | #pragma hdrstop | |
21 | #endif | |
22 | ||
23 | #if wxUSE_URL | |
24 | ||
25 | #include "wx/string.h" | |
26 | #include "wx/list.h" | |
27 | #include "wx/utils.h" | |
28 | #include "wx/module.h" | |
29 | #include "wx/url.h" | |
30 | ||
31 | #include <string.h> | |
32 | #include <ctype.h> | |
33 | ||
34 | IMPLEMENT_CLASS(wxProtoInfo, wxObject) | |
35 | IMPLEMENT_CLASS(wxURL, wxObject) | |
36 | ||
37 | // Protocols list | |
38 | wxProtoInfo *wxURL::ms_protocols = NULL; | |
39 | ||
40 | // Enforce linking of protocol classes: | |
41 | USE_PROTOCOL(wxFileProto) | |
42 | ||
43 | #if wxUSE_SOCKETS | |
44 | USE_PROTOCOL(wxHTTP) | |
45 | USE_PROTOCOL(wxFTP) | |
46 | ||
47 | wxHTTP *wxURL::ms_proxyDefault = NULL; | |
48 | bool wxURL::ms_useDefaultProxy = FALSE; | |
49 | #endif | |
50 | ||
51 | // -------------------------------------------------------------- | |
52 | // wxURL | |
53 | // -------------------------------------------------------------- | |
54 | ||
55 | // -------------------------------------------------------------- | |
56 | // --------- wxURL CONSTRUCTOR DESTRUCTOR ----------------------- | |
57 | // -------------------------------------------------------------- | |
58 | ||
59 | wxURL::wxURL(const wxString& url) | |
60 | { | |
61 | m_protocol = NULL; | |
62 | m_error = wxURL_NOERR; | |
63 | m_url = url; | |
64 | #if wxUSE_URL_NATIVE | |
65 | m_nativeImp = CreateNativeImpObject(); | |
66 | #endif | |
67 | ||
68 | #if wxUSE_SOCKETS | |
69 | if ( ms_useDefaultProxy && !ms_proxyDefault ) | |
70 | { | |
71 | SetDefaultProxy( wxGetenv(wxT("HTTP_PROXY")) ); | |
72 | ||
73 | if ( !ms_proxyDefault ) | |
74 | { | |
75 | // don't try again | |
76 | ms_useDefaultProxy = FALSE; | |
77 | } | |
78 | } | |
79 | ||
80 | m_useProxy = ms_proxyDefault != NULL; | |
81 | m_proxy = ms_proxyDefault; | |
82 | #endif // wxUSE_SOCKETS | |
83 | ||
84 | ParseURL(); | |
85 | } | |
86 | ||
87 | bool wxURL::ParseURL() | |
88 | { | |
89 | wxString last_url = m_url; | |
90 | ||
91 | // If the URL was already parsed (m_protocol != NULL), pass this section. | |
92 | if (!m_protocol) | |
93 | { | |
94 | // Clean up | |
95 | CleanData(); | |
96 | ||
97 | // Extract protocol name | |
98 | if (!PrepProto(last_url)) | |
99 | { | |
100 | m_error = wxURL_SNTXERR; | |
101 | return FALSE; | |
102 | } | |
103 | ||
104 | // Find and create the protocol object | |
105 | if (!FetchProtocol()) | |
106 | { | |
107 | m_error = wxURL_NOPROTO; | |
108 | return FALSE; | |
109 | } | |
110 | ||
111 | // Do we need a host name ? | |
112 | if (m_protoinfo->m_needhost) | |
113 | { | |
114 | // Extract it | |
115 | if (!PrepHost(last_url)) | |
116 | { | |
117 | m_error = wxURL_SNTXERR; | |
118 | return FALSE; | |
119 | } | |
120 | } | |
121 | ||
122 | // Extract full path | |
123 | if (!PrepPath(last_url)) | |
124 | { | |
125 | m_error = wxURL_NOPATH; | |
126 | return FALSE; | |
127 | } | |
128 | } | |
129 | // URL parse finished. | |
130 | ||
131 | #if wxUSE_SOCKETS | |
132 | if (m_useProxy) | |
133 | { | |
134 | // destroy the previously created protocol as we'll be using m_proxy | |
135 | delete m_protocol; | |
136 | ||
137 | // Third, we rebuild the URL. | |
138 | m_url = m_protoname + wxT(":"); | |
139 | if (m_protoinfo->m_needhost) | |
140 | m_url = m_url + wxT("//") + m_hostname; | |
141 | ||
142 | m_url += m_path; | |
143 | ||
144 | // We initialize specific variables. | |
145 | m_protocol = m_proxy; // FIXME: we should clone the protocol | |
146 | } | |
147 | #endif | |
148 | ||
149 | m_error = wxURL_NOERR; | |
150 | return TRUE; | |
151 | } | |
152 | ||
153 | void wxURL::CleanData() | |
154 | { | |
155 | #if wxUSE_SOCKETS | |
156 | if (!m_useProxy) | |
157 | #endif | |
158 | delete m_protocol; | |
159 | } | |
160 | ||
161 | wxURL::~wxURL() | |
162 | { | |
163 | CleanData(); | |
164 | #if wxUSE_SOCKETS | |
165 | if (m_proxy && m_proxy != ms_proxyDefault) | |
166 | delete m_proxy; | |
167 | #endif | |
168 | #if wxUSE_URL_NATIVE | |
169 | delete m_nativeImp; | |
170 | #endif | |
171 | } | |
172 | ||
173 | // -------------------------------------------------------------- | |
174 | // --------- wxURL urls decoders -------------------------------- | |
175 | // -------------------------------------------------------------- | |
176 | ||
177 | bool wxURL::PrepProto(wxString& url) | |
178 | { | |
179 | int pos; | |
180 | ||
181 | // Find end | |
182 | pos = url.Find(wxT(':')); | |
183 | if (pos == -1) | |
184 | return FALSE; | |
185 | ||
186 | m_protoname = url(0, pos); | |
187 | ||
188 | url = url(pos+1, url.Length()); | |
189 | ||
190 | return TRUE; | |
191 | } | |
192 | ||
193 | bool wxURL::PrepHost(wxString& url) | |
194 | { | |
195 | wxString temp_url; | |
196 | int pos, pos2; | |
197 | ||
198 | if ((url.GetChar(0) != wxT('/')) || (url.GetChar(1) != wxT('/'))) | |
199 | return FALSE; | |
200 | ||
201 | url = url(2, url.Length()); | |
202 | ||
203 | pos = url.Find(wxT('/')); | |
204 | if (pos == -1) | |
205 | pos = url.Length(); | |
206 | ||
207 | if (pos == 0) | |
208 | return FALSE; | |
209 | ||
210 | temp_url = url(0, pos); | |
211 | url = url(url.Find(wxT('/')), url.Length()); | |
212 | ||
213 | // Retrieve service number | |
214 | pos2 = temp_url.Find(wxT(':'), TRUE); | |
215 | if (pos2 != -1 && pos2 < pos) | |
216 | { | |
217 | m_servname = temp_url(pos2+1, pos); | |
218 | if (!m_servname.IsNumber()) | |
219 | return FALSE; | |
220 | temp_url = temp_url(0, pos2); | |
221 | } | |
222 | ||
223 | // Retrieve user and password. | |
224 | pos2 = temp_url.Find(wxT('@')); | |
225 | // Even if pos2 equals -1, this code is right. | |
226 | m_hostname = temp_url(pos2+1, temp_url.Length()); | |
227 | ||
228 | m_user = wxT(""); | |
229 | m_password = wxT(""); | |
230 | ||
231 | if (pos2 == -1) | |
232 | return TRUE; | |
233 | ||
234 | temp_url = temp_url(0, pos2); | |
235 | pos2 = temp_url.Find(wxT(':')); | |
236 | ||
237 | if (pos2 == -1) | |
238 | return FALSE; | |
239 | ||
240 | m_user = temp_url(0, pos2); | |
241 | m_password = temp_url(pos2+1, url.Length()); | |
242 | ||
243 | return TRUE; | |
244 | } | |
245 | ||
246 | bool wxURL::PrepPath(wxString& url) | |
247 | { | |
248 | if (url.Length() != 0) | |
249 | m_path = ConvertToValidURI(url); | |
250 | else | |
251 | m_path = wxT("/"); | |
252 | return TRUE; | |
253 | } | |
254 | ||
255 | bool wxURL::FetchProtocol() | |
256 | { | |
257 | wxProtoInfo *info = ms_protocols; | |
258 | ||
259 | while (info) | |
260 | { | |
261 | if (m_protoname == info->m_protoname) | |
262 | { | |
263 | if (m_servname.IsNull()) | |
264 | m_servname = info->m_servname; | |
265 | ||
266 | m_protoinfo = info; | |
267 | m_protocol = (wxProtocol *)m_protoinfo->m_cinfo->CreateObject(); | |
268 | return TRUE; | |
269 | } | |
270 | info = info->next; | |
271 | } | |
272 | return FALSE; | |
273 | } | |
274 | ||
275 | // -------------------------------------------------------------- | |
276 | // --------- wxURL get ------------------------------------------ | |
277 | // -------------------------------------------------------------- | |
278 | ||
279 | wxInputStream *wxURL::GetInputStream() | |
280 | { | |
281 | if (!m_protocol) | |
282 | { | |
283 | m_error = wxURL_NOPROTO; | |
284 | return NULL; | |
285 | } | |
286 | ||
287 | m_error = wxURL_NOERR; | |
288 | if (m_user != wxT("")) | |
289 | { | |
290 | m_protocol->SetUser(m_user); | |
291 | m_protocol->SetPassword(m_password); | |
292 | } | |
293 | ||
294 | #if wxUSE_URL_NATIVE | |
295 | // give the native implementation to return a better stream | |
296 | // such as the native WinINet functionality under MS-Windows | |
297 | if (m_nativeImp) | |
298 | { | |
299 | wxInputStream *rc; | |
300 | rc = m_nativeImp->GetInputStream(this); | |
301 | if (rc != 0) | |
302 | return rc; | |
303 | } | |
304 | // else use the standard behaviour | |
305 | #endif // wxUSE_URL_NATIVE | |
306 | ||
307 | #if wxUSE_SOCKETS | |
308 | wxIPV4address addr; | |
309 | ||
310 | // m_protoinfo is NULL when we use a proxy | |
311 | if (!m_useProxy && m_protoinfo->m_needhost) | |
312 | { | |
313 | if (!addr.Hostname(m_hostname)) | |
314 | { | |
315 | m_error = wxURL_NOHOST; | |
316 | return NULL; | |
317 | } | |
318 | ||
319 | addr.Service(m_servname); | |
320 | ||
321 | if (!m_protocol->Connect(addr, TRUE)) // Watcom needs the 2nd arg for some reason | |
322 | { | |
323 | m_error = wxURL_CONNERR; | |
324 | return NULL; | |
325 | } | |
326 | } | |
327 | #endif | |
328 | ||
329 | // When we use a proxy, we have to pass the whole URL to it. | |
330 | wxInputStream *the_i_stream = | |
331 | (m_useProxy) ? m_protocol->GetInputStream(m_url) : | |
332 | m_protocol->GetInputStream(m_path); | |
333 | ||
334 | if (!the_i_stream) | |
335 | { | |
336 | m_error = wxURL_PROTOERR; | |
337 | return NULL; | |
338 | } | |
339 | ||
340 | return the_i_stream; | |
341 | } | |
342 | ||
343 | #if wxUSE_SOCKETS | |
344 | void wxURL::SetDefaultProxy(const wxString& url_proxy) | |
345 | { | |
346 | if ( !url_proxy ) | |
347 | { | |
348 | if ( ms_proxyDefault ) | |
349 | { | |
350 | ms_proxyDefault->Close(); | |
351 | delete ms_proxyDefault; | |
352 | ms_proxyDefault = NULL; | |
353 | } | |
354 | } | |
355 | else | |
356 | { | |
357 | wxString tmp_str = url_proxy; | |
358 | int pos = tmp_str.Find(wxT(':')); | |
359 | if (pos == -1) | |
360 | return; | |
361 | ||
362 | wxString hostname = tmp_str(0, pos), | |
363 | port = tmp_str(pos+1, tmp_str.Length()-pos); | |
364 | wxIPV4address addr; | |
365 | ||
366 | if (!addr.Hostname(hostname)) | |
367 | return; | |
368 | if (!addr.Service(port)) | |
369 | return; | |
370 | ||
371 | if (ms_proxyDefault) | |
372 | // Finally, when all is right, we connect the new proxy. | |
373 | ms_proxyDefault->Close(); | |
374 | else | |
375 | ms_proxyDefault = new wxHTTP(); | |
376 | ms_proxyDefault->Connect(addr, TRUE); // Watcom needs the 2nd arg for some reason | |
377 | } | |
378 | } | |
379 | ||
380 | void wxURL::SetProxy(const wxString& url_proxy) | |
381 | { | |
382 | if ( !url_proxy ) | |
383 | { | |
384 | if ( m_proxy && m_proxy != ms_proxyDefault ) | |
385 | { | |
386 | m_proxy->Close(); | |
387 | delete m_proxy; | |
388 | } | |
389 | ||
390 | m_useProxy = FALSE; | |
391 | } | |
392 | else | |
393 | { | |
394 | wxString tmp_str; | |
395 | wxString hostname, port; | |
396 | int pos; | |
397 | wxIPV4address addr; | |
398 | ||
399 | tmp_str = url_proxy; | |
400 | pos = tmp_str.Find(wxT(':')); | |
401 | // This is an invalid proxy name. | |
402 | if (pos == -1) | |
403 | return; | |
404 | ||
405 | hostname = tmp_str(0, pos); | |
406 | port = tmp_str(pos+1, tmp_str.Length()-pos); | |
407 | ||
408 | addr.Hostname(hostname); | |
409 | addr.Service(port); | |
410 | ||
411 | // Finally, create the whole stuff. | |
412 | if (m_proxy && m_proxy != ms_proxyDefault) | |
413 | delete m_proxy; | |
414 | m_proxy = new wxHTTP(); | |
415 | m_proxy->Connect(addr, TRUE); // Watcom needs the 2nd arg for some reason | |
416 | ||
417 | CleanData(); | |
418 | // Reparse url. | |
419 | m_useProxy = TRUE; | |
420 | ParseURL(); | |
421 | } | |
422 | } | |
423 | #endif // wxUSE_SOCKETS | |
424 | ||
425 | wxString wxURL::ConvertToValidURI(const wxString& uri, const wxChar* delims) | |
426 | { | |
427 | wxString out_str; | |
428 | wxString hexa_code; | |
429 | size_t i; | |
430 | ||
431 | for (i = 0; i < uri.Len(); i++) | |
432 | { | |
433 | wxChar c = uri.GetChar(i); | |
434 | ||
435 | if (c == wxT(' ')) | |
436 | { | |
437 | // GRG, Apr/2000: changed to "%20" instead of '+' | |
438 | ||
439 | out_str += wxT("%20"); | |
440 | } | |
441 | else | |
442 | { | |
443 | // GRG, Apr/2000: modified according to the URI definition (RFC 2396) | |
444 | // | |
445 | // - Alphanumeric characters are never escaped | |
446 | // - Unreserved marks are never escaped | |
447 | // - Delimiters must be escaped if they appear within a component | |
448 | // but not if they are used to separate components. Here we have | |
449 | // no clear way to distinguish between these two cases, so they | |
450 | // are escaped unless they are passed in the 'delims' parameter | |
451 | // (allowed delimiters). | |
452 | ||
453 | static const wxChar marks[] = wxT("-_.!~*()'"); | |
454 | ||
455 | if ( !wxIsalnum(c) && !wxStrchr(marks, c) && !wxStrchr(delims, c) ) | |
456 | { | |
457 | hexa_code.Printf(wxT("%%%02X"), c); | |
458 | out_str += hexa_code; | |
459 | } | |
460 | else | |
461 | { | |
462 | out_str += c; | |
463 | } | |
464 | } | |
465 | } | |
466 | ||
467 | return out_str; | |
468 | } | |
469 | ||
470 | wxString wxURL::ConvertFromURI(const wxString& uri) | |
471 | { | |
472 | wxString new_uri; | |
473 | ||
474 | size_t i = 0; | |
475 | while (i < uri.Len()) | |
476 | { | |
477 | int code; | |
478 | if (uri[i] == wxT('%')) | |
479 | { | |
480 | i++; | |
481 | if (uri[i] >= wxT('A') && uri[i] <= wxT('F')) | |
482 | code = (uri[i] - wxT('A') + 10) * 16; | |
483 | else if (uri[i] >= wxT('a') && uri[i] <= wxT('f')) | |
484 | code = (uri[i] - wxT('a') + 10) * 16; | |
485 | else | |
486 | code = (uri[i] - wxT('0')) * 16; | |
487 | ||
488 | i++; | |
489 | if (uri[i] >= wxT('A') && uri[i] <= wxT('F')) | |
490 | code += (uri[i] - wxT('A')) + 10; | |
491 | else if (uri[i] >= wxT('a') && uri[i] <= wxT('f')) | |
492 | code += (uri[i] - wxT('a')) + 10; | |
493 | else | |
494 | code += (uri[i] - wxT('0')); | |
495 | ||
496 | i++; | |
497 | new_uri += (wxChar)code; | |
498 | continue; | |
499 | } | |
500 | new_uri += uri[i]; | |
501 | i++; | |
502 | } | |
503 | return new_uri; | |
504 | } | |
505 | ||
506 | // ---------------------------------------------------------------------- | |
507 | // A module which deletes the default proxy if we created it | |
508 | // ---------------------------------------------------------------------- | |
509 | ||
510 | #if wxUSE_SOCKETS | |
511 | ||
512 | class wxURLModule : public wxModule | |
513 | { | |
514 | public: | |
515 | virtual bool OnInit(); | |
516 | virtual void OnExit(); | |
517 | ||
518 | private: | |
519 | DECLARE_DYNAMIC_CLASS(wxURLModule) | |
520 | }; | |
521 | ||
522 | IMPLEMENT_DYNAMIC_CLASS(wxURLModule, wxModule) | |
523 | ||
524 | bool wxURLModule::OnInit() | |
525 | { | |
526 | // env var HTTP_PROXY contains the address of the default proxy to use if | |
527 | // set, but don't try to create this proxy right now because it will slow | |
528 | // down the program startup (especially if there is no DNS server | |
529 | // available, in which case it may take up to 1 minute) | |
530 | ||
531 | if ( getenv("HTTP_PROXY") ) | |
532 | { | |
533 | wxURL::ms_useDefaultProxy = TRUE; | |
534 | } | |
535 | ||
536 | return TRUE; | |
537 | } | |
538 | ||
539 | void wxURLModule::OnExit() | |
540 | { | |
541 | delete wxURL::ms_proxyDefault; | |
542 | wxURL::ms_proxyDefault = NULL; | |
543 | } | |
544 | ||
545 | #endif // wxUSE_SOCKETS | |
546 | ||
547 | #endif // wxUSE_URL | |
548 |