]>
Commit | Line | Data |
---|---|---|
f4ada568 GL |
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 | |
55d99c7a | 9 | // Licence: wxWindows licence |
f4ada568 GL |
10 | ///////////////////////////////////////////////////////////////////////////// |
11 | ||
12 | #ifdef __GNUG__ | |
13 | #pragma implementation "url.h" | |
14 | #endif | |
fcc6dddd JS |
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 | ||
a5d46b73 | 23 | #if wxUSE_URL |
f4ada568 | 24 | |
3096bd2f VZ |
25 | #include "wx/string.h" |
26 | #include "wx/list.h" | |
27 | #include "wx/utils.h" | |
c092213d | 28 | #include "wx/module.h" |
3096bd2f | 29 | #include "wx/url.h" |
f4ada568 | 30 | |
a5d46b73 VZ |
31 | #include <string.h> |
32 | #include <ctype.h> | |
33 | ||
f4ada568 GL |
34 | IMPLEMENT_CLASS(wxProtoInfo, wxObject) |
35 | IMPLEMENT_CLASS(wxURL, wxObject) | |
f4ada568 GL |
36 | |
37 | // Protocols list | |
b2b35524 | 38 | wxProtoInfo *wxURL::ms_protocols = NULL; |
8a4df159 | 39 | |
f92f546c | 40 | // Enforce linking of protocol classes: |
f92f546c VS |
41 | USE_PROTOCOL(wxFileProto) |
42 | ||
8a4df159 | 43 | #if wxUSE_SOCKETS |
f80eabe5 JS |
44 | USE_PROTOCOL(wxHTTP) |
45 | USE_PROTOCOL(wxFTP) | |
46 | ||
b2b35524 VZ |
47 | wxHTTP *wxURL::ms_proxyDefault = NULL; |
48 | bool wxURL::ms_useDefaultProxy = FALSE; | |
8a4df159 | 49 | #endif |
f4ada568 | 50 | |
fae05df5 GL |
51 | // -------------------------------------------------------------- |
52 | // wxURL | |
53 | // -------------------------------------------------------------- | |
f4ada568 | 54 | |
fae05df5 GL |
55 | // -------------------------------------------------------------- |
56 | // --------- wxURL CONSTRUCTOR DESTRUCTOR ----------------------- | |
57 | // -------------------------------------------------------------- | |
f4ada568 GL |
58 | |
59 | wxURL::wxURL(const wxString& url) | |
60 | { | |
b2b35524 VZ |
61 | m_protocol = NULL; |
62 | m_error = wxURL_NOERR; | |
63 | m_url = url; | |
64 | ||
8a4df159 | 65 | #if wxUSE_SOCKETS |
b2b35524 VZ |
66 | if ( ms_useDefaultProxy && !ms_proxyDefault ) |
67 | { | |
2b5f62a0 | 68 | SetDefaultProxy( wxGetenv(wxT("HTTP_PROXY")) ); |
b2b35524 VZ |
69 | |
70 | if ( !ms_proxyDefault ) | |
71 | { | |
72 | // don't try again | |
73 | ms_useDefaultProxy = FALSE; | |
74 | } | |
75 | } | |
76 | ||
77 | m_useProxy = ms_proxyDefault != NULL; | |
78 | m_proxy = ms_proxyDefault; | |
79 | #endif // wxUSE_SOCKETS | |
80 | ||
81 | ParseURL(); | |
f4ada568 GL |
82 | } |
83 | ||
84 | bool wxURL::ParseURL() | |
85 | { | |
86 | wxString last_url = m_url; | |
87 | ||
f6bcfd97 BP |
88 | // If the URL was already parsed (m_protocol != NULL), pass this section. |
89 | if (!m_protocol) | |
90 | { | |
f61815af GL |
91 | // Clean up |
92 | CleanData(); | |
f4ada568 | 93 | |
f61815af | 94 | // Extract protocol name |
f6bcfd97 BP |
95 | if (!PrepProto(last_url)) |
96 | { | |
f4ada568 GL |
97 | m_error = wxURL_SNTXERR; |
98 | return FALSE; | |
99 | } | |
f61815af GL |
100 | |
101 | // Find and create the protocol object | |
f6bcfd97 BP |
102 | if (!FetchProtocol()) |
103 | { | |
f61815af GL |
104 | m_error = wxURL_NOPROTO; |
105 | return FALSE; | |
106 | } | |
107 | ||
108 | // Do we need a host name ? | |
f6bcfd97 BP |
109 | if (m_protoinfo->m_needhost) |
110 | { | |
f61815af | 111 | // Extract it |
f6bcfd97 BP |
112 | if (!PrepHost(last_url)) |
113 | { | |
f61815af GL |
114 | m_error = wxURL_SNTXERR; |
115 | return FALSE; | |
116 | } | |
117 | } | |
118 | ||
119 | // Extract full path | |
f6bcfd97 BP |
120 | if (!PrepPath(last_url)) |
121 | { | |
f61815af GL |
122 | m_error = wxURL_NOPATH; |
123 | return FALSE; | |
124 | } | |
f4ada568 | 125 | } |
f61815af | 126 | // URL parse finished. |
f4ada568 | 127 | |
8a4df159 | 128 | #if wxUSE_SOCKETS |
f6bcfd97 BP |
129 | if (m_useProxy) |
130 | { | |
f61815af GL |
131 | // We destroy the newly created protocol. |
132 | CleanData(); | |
133 | ||
134 | // Third, we rebuild the URL. | |
223d09f6 | 135 | m_url = m_protoname + wxT(":"); |
f61815af | 136 | if (m_protoinfo->m_needhost) |
223d09f6 | 137 | m_url = m_url + wxT("//") + m_hostname; |
f61815af GL |
138 | |
139 | m_url += m_path; | |
140 | ||
141 | // We initialize specific variables. | |
142 | m_protocol = m_proxy; // FIXME: we should clone the protocol | |
f4ada568 | 143 | } |
8a4df159 | 144 | #endif |
f4ada568 GL |
145 | |
146 | m_error = wxURL_NOERR; | |
147 | return TRUE; | |
148 | } | |
149 | ||
150 | void wxURL::CleanData() | |
151 | { | |
8a4df159 | 152 | #if wxUSE_SOCKETS |
f61815af | 153 | if (!m_useProxy) |
8a4df159 | 154 | #endif |
f4ada568 GL |
155 | delete m_protocol; |
156 | } | |
157 | ||
158 | wxURL::~wxURL() | |
159 | { | |
160 | CleanData(); | |
8a4df159 | 161 | #if wxUSE_SOCKETS |
b2b35524 | 162 | if (m_proxy && m_proxy != ms_proxyDefault) |
f61815af | 163 | delete m_proxy; |
8a4df159 | 164 | #endif |
f4ada568 GL |
165 | } |
166 | ||
fae05df5 GL |
167 | // -------------------------------------------------------------- |
168 | // --------- wxURL urls decoders -------------------------------- | |
169 | // -------------------------------------------------------------- | |
170 | ||
f4ada568 GL |
171 | bool wxURL::PrepProto(wxString& url) |
172 | { | |
173 | int pos; | |
174 | ||
175 | // Find end | |
223d09f6 | 176 | pos = url.Find(wxT(':')); |
f4ada568 GL |
177 | if (pos == -1) |
178 | return FALSE; | |
179 | ||
180 | m_protoname = url(0, pos); | |
181 | ||
182 | url = url(pos+1, url.Length()); | |
183 | ||
184 | return TRUE; | |
185 | } | |
186 | ||
187 | bool wxURL::PrepHost(wxString& url) | |
188 | { | |
856d2e52 | 189 | wxString temp_url; |
f4ada568 GL |
190 | int pos, pos2; |
191 | ||
58c837a4 | 192 | if ((url.GetChar(0) != wxT('/')) || (url.GetChar(1) != wxT('/'))) |
f4ada568 GL |
193 | return FALSE; |
194 | ||
195 | url = url(2, url.Length()); | |
196 | ||
223d09f6 | 197 | pos = url.Find(wxT('/')); |
f4ada568 | 198 | if (pos == -1) |
b7db6f0b | 199 | pos = url.Length(); |
f4ada568 | 200 | |
856d2e52 GL |
201 | if (pos == 0) |
202 | return FALSE; | |
203 | ||
204 | temp_url = url(0, pos); | |
223d09f6 | 205 | url = url(url.Find(wxT('/')), url.Length()); |
856d2e52 GL |
206 | |
207 | // Retrieve service number | |
223d09f6 | 208 | pos2 = temp_url.Find(wxT(':'), TRUE); |
f6bcfd97 BP |
209 | if (pos2 != -1 && pos2 < pos) |
210 | { | |
375abe3d | 211 | m_servname = temp_url(pos2+1, pos); |
f4ada568 GL |
212 | if (!m_servname.IsNumber()) |
213 | return FALSE; | |
856d2e52 | 214 | temp_url = temp_url(0, pos2); |
f4ada568 GL |
215 | } |
216 | ||
856d2e52 | 217 | // Retrieve user and password. |
223d09f6 | 218 | pos2 = temp_url.Find(wxT('@')); |
856d2e52 GL |
219 | // Even if pos2 equals -1, this code is right. |
220 | m_hostname = temp_url(pos2+1, temp_url.Length()); | |
f4ada568 | 221 | |
223d09f6 KB |
222 | m_user = wxT(""); |
223 | m_password = wxT(""); | |
856d2e52 GL |
224 | |
225 | if (pos2 == -1) | |
226 | return TRUE; | |
227 | ||
228 | temp_url = temp_url(0, pos2); | |
223d09f6 | 229 | pos2 = temp_url.Find(wxT(':')); |
856d2e52 GL |
230 | |
231 | if (pos2 == -1) | |
232 | return FALSE; | |
233 | ||
234 | m_user = temp_url(0, pos2); | |
235 | m_password = temp_url(pos2+1, url.Length()); | |
f4ada568 GL |
236 | |
237 | return TRUE; | |
238 | } | |
239 | ||
240 | bool wxURL::PrepPath(wxString& url) | |
241 | { | |
242 | if (url.Length() != 0) | |
f61815af | 243 | m_path = ConvertToValidURI(url); |
f4ada568 | 244 | else |
223d09f6 | 245 | m_path = wxT("/"); |
f4ada568 GL |
246 | return TRUE; |
247 | } | |
248 | ||
249 | bool wxURL::FetchProtocol() | |
250 | { | |
b2b35524 | 251 | wxProtoInfo *info = ms_protocols; |
f4ada568 | 252 | |
f6bcfd97 BP |
253 | while (info) |
254 | { | |
255 | if (m_protoname == info->m_protoname) | |
256 | { | |
f4ada568 GL |
257 | if (m_servname.IsNull()) |
258 | m_servname = info->m_servname; | |
259 | ||
260 | m_protoinfo = info; | |
261 | m_protocol = (wxProtocol *)m_protoinfo->m_cinfo->CreateObject(); | |
f4ada568 GL |
262 | return TRUE; |
263 | } | |
264 | info = info->next; | |
265 | } | |
266 | return FALSE; | |
267 | } | |
268 | ||
fae05df5 GL |
269 | // -------------------------------------------------------------- |
270 | // --------- wxURL get ------------------------------------------ | |
271 | // -------------------------------------------------------------- | |
272 | ||
58c837a4 | 273 | wxInputStream *wxURL::GetInputStream() |
f4ada568 | 274 | { |
f4ada568 GL |
275 | wxInputStream *the_i_stream = NULL; |
276 | ||
f6bcfd97 BP |
277 | if (!m_protocol) |
278 | { | |
f4ada568 GL |
279 | m_error = wxURL_NOPROTO; |
280 | return NULL; | |
281 | } | |
282 | ||
283 | m_error = wxURL_NOERR; | |
f6bcfd97 BP |
284 | if (m_user != wxT("")) |
285 | { | |
856d2e52 GL |
286 | m_protocol->SetUser(m_user); |
287 | m_protocol->SetPassword(m_password); | |
288 | } | |
289 | ||
8a4df159 | 290 | #if wxUSE_SOCKETS |
19e0e04b RD |
291 | wxIPV4address addr; |
292 | ||
f61815af | 293 | // m_protoinfo is NULL when we use a proxy |
f6bcfd97 BP |
294 | if (!m_useProxy && m_protoinfo->m_needhost) |
295 | { | |
296 | if (!addr.Hostname(m_hostname)) | |
297 | { | |
f4ada568 GL |
298 | m_error = wxURL_NOHOST; |
299 | return NULL; | |
300 | } | |
301 | ||
302 | addr.Service(m_servname); | |
303 | ||
8a2c6ef8 JS |
304 | if (!m_protocol->Connect(addr, TRUE)) // Watcom needs the 2nd arg for some reason |
305 | { | |
f4ada568 GL |
306 | m_error = wxURL_CONNERR; |
307 | return NULL; | |
308 | } | |
309 | } | |
8a4df159 | 310 | #endif |
f4ada568 | 311 | |
f61815af GL |
312 | // When we use a proxy, we have to pass the whole URL to it. |
313 | if (m_useProxy) | |
314 | the_i_stream = m_protocol->GetInputStream(m_url); | |
315 | else | |
316 | the_i_stream = m_protocol->GetInputStream(m_path); | |
317 | ||
f6bcfd97 BP |
318 | if (!the_i_stream) |
319 | { | |
f4ada568 GL |
320 | m_error = wxURL_PROTOERR; |
321 | return NULL; | |
322 | } | |
323 | ||
324 | return the_i_stream; | |
325 | } | |
326 | ||
8a4df159 | 327 | #if wxUSE_SOCKETS |
f4ada568 GL |
328 | void wxURL::SetDefaultProxy(const wxString& url_proxy) |
329 | { | |
b2b35524 VZ |
330 | if ( !url_proxy ) |
331 | { | |
332 | if ( ms_proxyDefault ) | |
333 | { | |
334 | ms_proxyDefault->Close(); | |
335 | delete ms_proxyDefault; | |
336 | ms_proxyDefault = NULL; | |
337 | } | |
f61815af | 338 | } |
f61815af | 339 | else |
b2b35524 VZ |
340 | { |
341 | wxString tmp_str = url_proxy; | |
342 | int pos = tmp_str.Find(wxT(':')); | |
343 | if (pos == -1) | |
344 | return; | |
345 | ||
346 | wxString hostname = tmp_str(0, pos), | |
347 | port = tmp_str(pos+1, tmp_str.Length()-pos); | |
348 | wxIPV4address addr; | |
349 | ||
350 | if (!addr.Hostname(hostname)) | |
351 | return; | |
352 | if (!addr.Service(port)) | |
353 | return; | |
354 | ||
355 | if (ms_proxyDefault) | |
356 | // Finally, when all is right, we connect the new proxy. | |
357 | ms_proxyDefault->Close(); | |
358 | else | |
359 | ms_proxyDefault = new wxHTTP(); | |
360 | ms_proxyDefault->Connect(addr, TRUE); // Watcom needs the 2nd arg for some reason | |
361 | } | |
f4ada568 GL |
362 | } |
363 | ||
364 | void wxURL::SetProxy(const wxString& url_proxy) | |
365 | { | |
b2b35524 VZ |
366 | if ( !url_proxy ) |
367 | { | |
368 | if ( m_proxy && m_proxy != ms_proxyDefault ) | |
369 | { | |
370 | m_proxy->Close(); | |
371 | delete m_proxy; | |
372 | } | |
f4ada568 | 373 | |
b2b35524 VZ |
374 | m_useProxy = FALSE; |
375 | } | |
376 | else | |
377 | { | |
378 | wxString tmp_str; | |
379 | wxString hostname, port; | |
380 | int pos; | |
381 | wxIPV4address addr; | |
382 | ||
383 | tmp_str = url_proxy; | |
384 | pos = tmp_str.Find(wxT(':')); | |
385 | // This is an invalid proxy name. | |
386 | if (pos == -1) | |
387 | return; | |
388 | ||
389 | hostname = tmp_str(0, pos); | |
bb80bb5b | 390 | port = tmp_str(pos+1, tmp_str.Length()-pos); |
b2b35524 VZ |
391 | |
392 | addr.Hostname(hostname); | |
393 | addr.Service(port); | |
394 | ||
395 | // Finally, create the whole stuff. | |
396 | if (m_proxy && m_proxy != ms_proxyDefault) | |
397 | delete m_proxy; | |
398 | m_proxy = new wxHTTP(); | |
399 | m_proxy->Connect(addr, TRUE); // Watcom needs the 2nd arg for some reason | |
400 | ||
401 | CleanData(); | |
402 | // Reparse url. | |
403 | m_useProxy = TRUE; | |
404 | ParseURL(); | |
405 | } | |
f4ada568 | 406 | } |
b2b35524 | 407 | #endif // wxUSE_SOCKETS |
35a4dab7 | 408 | |
f6bcfd97 | 409 | wxString wxURL::ConvertToValidURI(const wxString& uri, const wxChar* delims) |
14906731 | 410 | { |
fae05df5 GL |
411 | wxString out_str; |
412 | wxString hexa_code; | |
413 | size_t i; | |
414 | ||
f6bcfd97 BP |
415 | for (i = 0; i < uri.Len(); i++) |
416 | { | |
fae05df5 GL |
417 | wxChar c = uri.GetChar(i); |
418 | ||
223d09f6 | 419 | if (c == wxT(' ')) |
f6bcfd97 BP |
420 | { |
421 | // GRG, Apr/2000: changed to "%20" instead of '+' | |
422 | ||
423 | out_str += wxT("%20"); | |
424 | } | |
425 | else | |
426 | { | |
427 | // GRG, Apr/2000: modified according to the URI definition (RFC 2396) | |
428 | // | |
429 | // - Alphanumeric characters are never escaped | |
430 | // - Unreserved marks are never escaped | |
431 | // - Delimiters must be escaped if they appear within a component | |
432 | // but not if they are used to separate components. Here we have | |
433 | // no clear way to distinguish between these two cases, so they | |
434 | // are escaped unless they are passed in the 'delims' parameter | |
435 | // (allowed delimiters). | |
436 | ||
437 | static const wxChar marks[] = wxT("-_.!~*()'"); | |
438 | ||
439 | if ( !wxIsalnum(c) && !wxStrchr(marks, c) && !wxStrchr(delims, c) ) | |
440 | { | |
223d09f6 | 441 | hexa_code.Printf(wxT("%%%02X"), c); |
5a96d2f4 | 442 | out_str += hexa_code; |
f6bcfd97 BP |
443 | } |
444 | else | |
445 | { | |
5a96d2f4 | 446 | out_str += c; |
f6bcfd97 | 447 | } |
5a96d2f4 | 448 | } |
fae05df5 | 449 | } |
19e0e04b | 450 | |
fae05df5 | 451 | return out_str; |
14906731 GL |
452 | } |
453 | ||
aa6d9706 GL |
454 | wxString wxURL::ConvertFromURI(const wxString& uri) |
455 | { | |
aa6d9706 GL |
456 | wxString new_uri; |
457 | ||
74b31181 | 458 | size_t i = 0; |
f6bcfd97 BP |
459 | while (i < uri.Len()) |
460 | { | |
74b31181 | 461 | int code; |
f6bcfd97 BP |
462 | if (uri[i] == wxT('%')) |
463 | { | |
aa6d9706 | 464 | i++; |
223d09f6 KB |
465 | if (uri[i] >= wxT('A') && uri[i] <= wxT('F')) |
466 | code = (uri[i] - wxT('A') + 10) * 16; | |
a03c55c1 VZ |
467 | else if (uri[i] >= wxT('a') && uri[i] <= wxT('f')) |
468 | code = (uri[i] - wxT('a') + 10) * 16; | |
aa6d9706 | 469 | else |
223d09f6 | 470 | code = (uri[i] - wxT('0')) * 16; |
f6bcfd97 | 471 | |
aa6d9706 | 472 | i++; |
223d09f6 KB |
473 | if (uri[i] >= wxT('A') && uri[i] <= wxT('F')) |
474 | code += (uri[i] - wxT('A')) + 10; | |
a03c55c1 VZ |
475 | else if (uri[i] >= wxT('a') && uri[i] <= wxT('f')) |
476 | code += (uri[i] - wxT('a')) + 10; | |
aa6d9706 | 477 | else |
223d09f6 | 478 | code += (uri[i] - wxT('0')); |
f6bcfd97 | 479 | |
aa6d9706 GL |
480 | i++; |
481 | new_uri += (wxChar)code; | |
482 | continue; | |
483 | } | |
484 | new_uri += uri[i]; | |
485 | i++; | |
486 | } | |
487 | return new_uri; | |
488 | } | |
b2b35524 VZ |
489 | |
490 | // ---------------------------------------------------------------------- | |
491 | // A module which deletes the default proxy if we created it | |
492 | // ---------------------------------------------------------------------- | |
493 | ||
494 | #if wxUSE_SOCKETS | |
495 | ||
496 | class wxURLModule : public wxModule | |
497 | { | |
498 | public: | |
499 | virtual bool OnInit(); | |
500 | virtual void OnExit(); | |
501 | ||
502 | private: | |
503 | DECLARE_DYNAMIC_CLASS(wxURLModule) | |
504 | }; | |
505 | ||
506 | IMPLEMENT_DYNAMIC_CLASS(wxURLModule, wxModule) | |
507 | ||
508 | bool wxURLModule::OnInit() | |
509 | { | |
510 | // env var HTTP_PROXY contains the address of the default proxy to use if | |
511 | // set, but don't try to create this proxy right now because it will slow | |
512 | // down the program startup (especially if there is no DNS server | |
513 | // available, in which case it may take up to 1 minute) | |
f6bcfd97 | 514 | |
b2b35524 VZ |
515 | if ( getenv("HTTP_PROXY") ) |
516 | { | |
517 | wxURL::ms_useDefaultProxy = TRUE; | |
518 | } | |
519 | ||
520 | return TRUE; | |
521 | } | |
522 | ||
523 | void wxURLModule::OnExit() | |
524 | { | |
525 | delete wxURL::ms_proxyDefault; | |
526 | wxURL::ms_proxyDefault = NULL; | |
527 | } | |
528 | ||
529 | #endif // wxUSE_SOCKETS | |
a5d46b73 VZ |
530 | |
531 | #endif // wxUSE_URL | |
532 |