src/common/url.cpp

   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 license
  10 /////////////////////////////////////////////////////////////////////////////
  11
  12 #ifdef __GNUG__
  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
  65 #if wxUSE_SOCKETS
  66     if ( ms_useDefaultProxy && !ms_proxyDefault )
  67     {
  68         SetDefaultProxy(getenv("HTTP_PROXY"));
  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();
  82 }
  83
  84 bool wxURL::ParseURL()
  85 {
  86   wxString last_url = m_url;
  87
  88   // If the URL was already parsed (m_protocol != NULL), pass this section.
  89   if (!m_protocol)
  90   {
  91     // Clean up
  92     CleanData();
  93
  94     // Extract protocol name
  95     if (!PrepProto(last_url))
  96     {
  97       m_error = wxURL_SNTXERR;
  98       return FALSE;
  99     }
 100
 101     // Find and create the protocol object
 102     if (!FetchProtocol())
 103     {
 104       m_error = wxURL_NOPROTO;
 105       return FALSE;
 106     }
 107
 108     // Do we need a host name ?
 109     if (m_protoinfo->m_needhost)
 110     {
 111       // Extract it
 112       if (!PrepHost(last_url))
 113       {
 114         m_error = wxURL_SNTXERR;
 115         return FALSE;
 116       }
 117     }
 118
 119     // Extract full path
 120     if (!PrepPath(last_url))
 121     {
 122       m_error = wxURL_NOPATH;
 123       return FALSE;
 124     }
 125   }
 126   // URL parse finished.
 127
 128 #if wxUSE_SOCKETS
 129   if (m_useProxy)
 130   {
 131     // We destroy the newly created protocol.
 132     CleanData();
 133
 134     // Third, we rebuild the URL.
 135     m_url = m_protoname + wxT(":");
 136     if (m_protoinfo->m_needhost)
 137       m_url = m_url + wxT("//") + m_hostname;
 138
 139     m_url += m_path;
 140
 141     // We initialize specific variables.
 142     m_protocol = m_proxy; // FIXME: we should clone the protocol
 143   }
 144 #endif
 145
 146   m_error = wxURL_NOERR;
 147   return TRUE;
 148 }
 149
 150 void wxURL::CleanData()
 151 {
 152 #if wxUSE_SOCKETS
 153   if (!m_useProxy)
 154 #endif
 155     delete m_protocol;
 156 }
 157
 158 wxURL::~wxURL()
 159 {
 160   CleanData();
 161 #if wxUSE_SOCKETS
 162   if (m_proxy && m_proxy != ms_proxyDefault)
 163     delete m_proxy;
 164 #endif
 165 }
 166
 167 // --------------------------------------------------------------
 168 // --------- wxURL urls decoders --------------------------------
 169 // --------------------------------------------------------------
 170
 171 bool wxURL::PrepProto(wxString& url)
 172 {
 173   int pos;
 174
 175   // Find end
 176   pos = url.Find(wxT(':'));
 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 {
 189   wxString temp_url;
 190   int pos, pos2;
 191
 192   if ((url.GetChar(0) != wxT('/')) || (url.GetChar(1) != wxT('/')))
 193     return FALSE;
 194
 195   url = url(2, url.Length());
 196
 197   pos = url.Find(wxT('/'));
 198   if (pos == -1)
 199     pos = url.Length();
 200
 201   if (pos == 0)
 202     return FALSE;
 203
 204   temp_url = url(0, pos);
 205   url = url(url.Find(wxT('/')), url.Length());
 206
 207   // Retrieve service number
 208   pos2 = temp_url.Find(wxT(':'), TRUE);
 209   if (pos2 != -1 && pos2 < pos)
 210   {
 211     m_servname = temp_url(pos2+1, pos);
 212     if (!m_servname.IsNumber())
 213       return FALSE;
 214     temp_url = temp_url(0, pos2);
 215   }
 216
 217   // Retrieve user and password.
 218   pos2 = temp_url.Find(wxT('@'));
 219   // Even if pos2 equals -1, this code is right.
 220   m_hostname = temp_url(pos2+1, temp_url.Length());
 221
 222   m_user = wxT("");
 223   m_password = wxT("");
 224
 225   if (pos2 == -1)
 226     return TRUE;
 227
 228   temp_url = temp_url(0, pos2);
 229   pos2 = temp_url.Find(wxT(':'));
 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());
 236
 237   return TRUE;
 238 }
 239
 240 bool wxURL::PrepPath(wxString& url)
 241 {
 242   if (url.Length() != 0)
 243     m_path = ConvertToValidURI(url);
 244   else
 245     m_path = wxT("/");
 246   return TRUE;
 247 }
 248
 249 bool wxURL::FetchProtocol()
 250 {
 251   wxProtoInfo *info = ms_protocols;
 252
 253   while (info)
 254   {
 255     if (m_protoname == info->m_protoname)
 256     {
 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();
 262       return TRUE;
 263     }
 264     info = info->next;
 265   }
 266   return FALSE;
 267 }
 268
 269 // --------------------------------------------------------------
 270 // --------- wxURL get ------------------------------------------
 271 // --------------------------------------------------------------
 272
 273 wxInputStream *wxURL::GetInputStream()
 274 {
 275   wxInputStream *the_i_stream = NULL;
 276
 277   if (!m_protocol)
 278   {
 279     m_error = wxURL_NOPROTO;
 280     return NULL;
 281   }
 282
 283   m_error = wxURL_NOERR;
 284   if (m_user != wxT(""))
 285   {
 286     m_protocol->SetUser(m_user);
 287     m_protocol->SetPassword(m_password);
 288   }
 289
 290 #if wxUSE_SOCKETS
 291     wxIPV4address addr;
 292
 293   // m_protoinfo is NULL when we use a proxy
 294   if (!m_useProxy && m_protoinfo->m_needhost)
 295   {
 296     if (!addr.Hostname(m_hostname))
 297     {
 298       m_error = wxURL_NOHOST;
 299       return NULL;
 300     }
 301
 302     addr.Service(m_servname);
 303
 304     if (!m_protocol->Connect(addr, TRUE)) // Watcom needs the 2nd arg for some reason
 305     {
 306       m_error = wxURL_CONNERR;
 307       return NULL;
 308     }
 309   }
 310 #endif
 311
 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
 318   if (!the_i_stream)
 319   {
 320     m_error = wxURL_PROTOERR;
 321     return NULL;
 322   }
 323
 324   return the_i_stream;
 325 }
 326
 327 #if wxUSE_SOCKETS
 328 void wxURL::SetDefaultProxy(const wxString& url_proxy)
 329 {
 330   if ( !url_proxy )
 331   {
 332       if ( ms_proxyDefault )
 333       {
 334           ms_proxyDefault->Close();
 335           delete ms_proxyDefault;
 336           ms_proxyDefault = NULL;
 337       }
 338   }
 339   else
 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   }
 362 }
 363
 364 void wxURL::SetProxy(const wxString& url_proxy)
 365 {
 366     if ( !url_proxy )
 367     {
 368         if ( m_proxy && m_proxy != ms_proxyDefault )
 369         {
 370             m_proxy->Close();
 371             delete m_proxy;
 372         }
 373
 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);
 390         port = tmp_str(pos, tmp_str.Length()-pos);
 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     }
 406 }
 407 #endif // wxUSE_SOCKETS
 408
 409 wxString wxURL::ConvertToValidURI(const wxString& uri, const wxChar* delims)
 410 {
 411   wxString out_str;
 412   wxString hexa_code;
 413   size_t i;
 414
 415   for (i = 0; i < uri.Len(); i++)
 416   {
 417     wxChar c = uri.GetChar(i);
 418
 419     if (c == wxT(' '))
 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       {
 441         hexa_code.Printf(wxT("%%%02X"), c);
 442         out_str += hexa_code;
 443       }
 444       else
 445       {
 446         out_str += c;
 447       }
 448     }
 449   }
 450
 451   return out_str;
 452 }
 453
 454 wxString wxURL::ConvertFromURI(const wxString& uri)
 455 {
 456   wxString new_uri;
 457
 458   size_t i = 0;
 459   while (i < uri.Len())
 460   {
 461     int code;
 462     if (uri[i] == wxT('%'))
 463     {
 464       i++;
 465       if (uri[i] >= wxT('A') && uri[i] <= wxT('F'))
 466         code = (uri[i] - wxT('A') + 10) * 16;
 467       else
 468         code = (uri[i] - wxT('0')) * 16;
 469
 470       i++;
 471       if (uri[i] >= wxT('A') && uri[i] <= wxT('F'))
 472         code += (uri[i] - wxT('A')) + 10;
 473       else
 474         code += (uri[i] - wxT('0'));
 475
 476       i++;
 477       new_uri += (wxChar)code;
 478       continue;
 479     }
 480     new_uri += uri[i];
 481     i++;
 482   }
 483   return new_uri;
 484 }
 485
 486 // ----------------------------------------------------------------------
 487 // A module which deletes the default proxy if we created it
 488 // ----------------------------------------------------------------------
 489
 490 #if wxUSE_SOCKETS
 491
 492 class wxURLModule : public wxModule
 493 {
 494 public:
 495     virtual bool OnInit();
 496     virtual void OnExit();
 497
 498 private:
 499     DECLARE_DYNAMIC_CLASS(wxURLModule)
 500 };
 501
 502 IMPLEMENT_DYNAMIC_CLASS(wxURLModule, wxModule)
 503
 504 bool wxURLModule::OnInit()
 505 {
 506     // env var HTTP_PROXY contains the address of the default proxy to use if
 507     // set, but don't try to create this proxy right now because it will slow
 508     // down the program startup (especially if there is no DNS server
 509     // available, in which case it may take up to 1 minute)
 510
 511     if ( getenv("HTTP_PROXY") )
 512     {
 513         wxURL::ms_useDefaultProxy = TRUE;
 514     }
 515
 516     return TRUE;
 517 }
 518
 519 void wxURLModule::OnExit()
 520 {
 521     delete wxURL::ms_proxyDefault;
 522     wxURL::ms_proxyDefault = NULL;
 523 }
 524
 525 #endif // wxUSE_SOCKETS
 526
 527 #endif // wxUSE_URL
 528