src/xrc/xml.cpp

   1 /////////////////////////////////////////////////////////////////////////////
   2 // Name:        xml.cpp
   3 // Purpose:     wxXmlDocument - XML parser & data holder class
   4 // Author:      Vaclav Slavik
   5 // Created:     2000/03/05
   6 // RCS-ID:      $Id$
   7 // Copyright:   (c) 2000 Vaclav Slavik
   8 // Licence:     wxWindows licence
   9 /////////////////////////////////////////////////////////////////////////////
  10
  11 #ifdef __GNUG__
  12 #pragma implementation "xml.h"
  13 #pragma implementation "xmlio.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
  24 #include "wx/wfstream.h"
  25 #include "wx/datstrm.h"
  26 #include "wx/zstream.h"
  27 #include "wx/log.h"
  28 #include "wx/intl.h"
  29 #include "wx/strconv.h"
  30
  31 #include "wx/xrc/xml.h"
  32
  33 #include "xmlparse.h" // from Expat
  34
  35 //-----------------------------------------------------------------------------
  36 //  wxXmlNode
  37 //-----------------------------------------------------------------------------
  38
  39 wxXmlNode::wxXmlNode(wxXmlNode *parent,wxXmlNodeType type,
  40                      const wxString& name, const wxString& content,
  41                      wxXmlProperty *props, wxXmlNode *next)
  42     : m_type(type), m_name(name), m_content(content),
  43       m_properties(props), m_parent(parent),
  44       m_children(NULL), m_next(next)
  45 {
  46     if (m_parent)
  47     {
  48         if (m_parent->m_children)
  49         {
  50             m_next = m_parent->m_children;
  51             m_parent->m_children = this;
  52         }
  53         else
  54             m_parent->m_children = this;
  55     }
  56 }
  57
  58 wxXmlNode::wxXmlNode(wxXmlNodeType type, const wxString& name,
  59                      const wxString& content)
  60     : m_type(type), m_name(name), m_content(content),
  61       m_properties(NULL), m_parent(NULL),
  62       m_children(NULL), m_next(NULL)
  63 {}
  64
  65 wxXmlNode::wxXmlNode(const wxXmlNode& node)
  66 {
  67     m_next = NULL;
  68     m_parent = NULL;
  69     DoCopy(node);
  70 }
  71
  72 wxXmlNode::~wxXmlNode()
  73 {
  74     wxXmlNode *c, *c2;
  75     for (c = m_children; c; c = c2)
  76     {
  77         c2 = c->m_next;
  78         delete c;
  79     }
  80
  81     wxXmlProperty *p, *p2;
  82     for (p = m_properties; p; p = p2)
  83     {
  84         p2 = p->GetNext();
  85         delete p;
  86     }
  87 }
  88
  89 wxXmlNode& wxXmlNode::operator=(const wxXmlNode& node)
  90 {
  91     wxDELETE(m_properties);
  92     wxDELETE(m_children);
  93     DoCopy(node);
  94     return *this;
  95 }
  96
  97 void wxXmlNode::DoCopy(const wxXmlNode& node)
  98 {
  99     m_type = node.m_type;
 100     m_name = node.m_name;
 101     m_content = node.m_content;
 102     m_children = NULL;
 103
 104     wxXmlNode *n = node.m_children;
 105     while (n)
 106     {
 107         AddChild(new wxXmlNode(*n));
 108         n = n->GetNext();
 109     }
 110
 111     m_properties = NULL;
 112     wxXmlProperty *p = node.m_properties;
 113     while (p)
 114     {
 115        AddProperty(p->GetName(), p->GetValue());
 116        p = p->GetNext();
 117     }
 118 }
 119
 120 bool wxXmlNode::HasProp(const wxString& propName) const
 121 {
 122     wxXmlProperty *prop = GetProperties();
 123
 124     while (prop)
 125     {
 126         if (prop->GetName() == propName) return TRUE;
 127         prop = prop->GetNext();
 128     }
 129
 130     return FALSE;
 131 }
 132
 133 bool wxXmlNode::GetPropVal(const wxString& propName, wxString *value) const
 134 {
 135     wxXmlProperty *prop = GetProperties();
 136
 137     while (prop)
 138     {
 139         if (prop->GetName() == propName)
 140         {
 141             *value = prop->GetValue();
 142             return TRUE;
 143         }
 144         prop = prop->GetNext();
 145     }
 146
 147     return FALSE;
 148 }
 149
 150 wxString wxXmlNode::GetPropVal(const wxString& propName, const wxString& defaultVal) const
 151 {
 152     wxString tmp;
 153     if (GetPropVal(propName, &tmp))
 154         return tmp;
 155     else
 156         return defaultVal;
 157 }
 158
 159 void wxXmlNode::AddChild(wxXmlNode *child)
 160 {
 161     if (m_children == NULL)
 162         m_children = child;
 163     else
 164     {
 165         wxXmlNode *ch = m_children;
 166         while (ch->m_next) ch = ch->m_next;
 167         ch->m_next = child;
 168     }
 169     child->m_next = NULL;
 170     child->m_parent = this;
 171 }
 172
 173 void wxXmlNode::InsertChild(wxXmlNode *child, wxXmlNode *before_node)
 174 {
 175     wxASSERT_MSG(before_node->GetParent() == this, wxT("wxXmlNode::InsertChild - the node has incorrect parent"));
 176
 177     if (m_children == before_node)
 178        m_children = child;
 179     else
 180     {
 181         wxXmlNode *ch = m_children;
 182         while (ch->m_next != before_node) ch = ch->m_next;
 183         ch->m_next = child;
 184     }
 185
 186     child->m_parent = this;
 187     child->m_next = before_node;
 188 }
 189
 190 bool wxXmlNode::RemoveChild(wxXmlNode *child)
 191 {
 192     if (m_children == NULL)
 193         return FALSE;
 194     else if (m_children == child)
 195     {
 196         m_children = child->m_next;
 197         child->m_parent = NULL;
 198         child->m_next = NULL;
 199         return TRUE;
 200     }
 201     else
 202     {
 203         wxXmlNode *ch = m_children;
 204         while (ch->m_next)
 205         {
 206             if (ch->m_next == child)
 207             {
 208                 ch->m_next = child->m_next;
 209                 child->m_parent = NULL;
 210                 child->m_next = NULL;
 211                 return TRUE;
 212             }
 213             ch = ch->m_next;
 214         }
 215         return FALSE;
 216     }
 217 }
 218
 219 void wxXmlNode::AddProperty(const wxString& name, const wxString& value)
 220 {
 221     AddProperty(new wxXmlProperty(name, value, NULL));
 222 }
 223
 224 void wxXmlNode::AddProperty(wxXmlProperty *prop)
 225 {
 226     if (m_properties == NULL)
 227         m_properties = prop;
 228     else
 229     {
 230         wxXmlProperty *p = m_properties;
 231         while (p->GetNext()) p = p->GetNext();
 232         p->SetNext(prop);
 233     }
 234 }
 235
 236 bool wxXmlNode::DeleteProperty(const wxString& name)
 237 {
 238     wxXmlProperty *prop;
 239
 240     if (m_properties == NULL)
 241         return FALSE;
 242
 243     else if (m_properties->GetName() == name)
 244     {
 245         prop = m_properties;
 246         m_properties = prop->GetNext();
 247         prop->SetNext(NULL);
 248         delete prop;
 249         return TRUE;
 250     }
 251
 252     else
 253     {
 254         wxXmlProperty *p = m_properties;
 255         while (p->GetNext())
 256         {
 257             if (p->GetNext()->GetName() == name)
 258             {
 259                 prop = p->GetNext();
 260                 p->SetNext(prop->GetNext());
 261                 prop->SetNext(NULL);
 262                 delete prop;
 263                 return TRUE;
 264             }
 265             p = p->GetNext();
 266         }
 267         return FALSE;
 268     }
 269 }
 270
 271
 272
 273 //-----------------------------------------------------------------------------
 274 //  wxXmlDocument
 275 //-----------------------------------------------------------------------------
 276
 277 wxXmlDocument::wxXmlDocument(const wxString& filename, const wxString& encoding)
 278                           : wxObject(), m_root(NULL)
 279 {
 280     if ( !Load(filename, encoding) )
 281     {
 282         wxDELETE(m_root);
 283     }
 284 }
 285
 286 wxXmlDocument::wxXmlDocument(wxInputStream& stream, const wxString& encoding)
 287                           : wxObject(), m_root(NULL)
 288 {
 289     if ( !Load(stream, encoding) )
 290     {
 291         wxDELETE(m_root);
 292     }
 293 }
 294
 295 wxXmlDocument::wxXmlDocument(const wxXmlDocument& doc)
 296 {
 297     DoCopy(doc);
 298 }
 299
 300 wxXmlDocument& wxXmlDocument::operator=(const wxXmlDocument& doc)
 301 {
 302     wxDELETE(m_root);
 303     DoCopy(doc);
 304     return *this;
 305 }
 306
 307 void wxXmlDocument::DoCopy(const wxXmlDocument& doc)
 308 {
 309     m_version = doc.m_version;
 310 #if !wxUSE_UNICODE
 311     m_encoding = doc.m_encoding;
 312 #endif
 313     m_fileEncoding = doc.m_fileEncoding;
 314     m_root = new wxXmlNode(*doc.m_root);
 315 }
 316
 317 bool wxXmlDocument::Load(const wxString& filename, const wxString& encoding)
 318 {
 319     wxFileInputStream stream(filename);
 320     return Load(stream, encoding);
 321 }
 322
 323 bool wxXmlDocument::Save(const wxString& filename) const
 324 {
 325     wxFileOutputStream stream(filename);
 326     return Save(stream);
 327 }
 328
 329
 330
 331 //-----------------------------------------------------------------------------
 332 //  wxXmlDocument loading routines
 333 //-----------------------------------------------------------------------------
 334
 335 /*
 336     FIXME:
 337        - process all elements, including CDATA
 338  */
 339
 340 // converts Expat-produced string in UTF-8 into wxString.
 341 inline static wxString CharToString(wxMBConv *conv,
 342                                     const char *s, size_t len = wxSTRING_MAXLEN)
 343 {
 344 #if wxUSE_UNICODE
 345     (void)conv;
 346     return wxString(s, wxConvUTF8, len);
 347 #else
 348     if ( conv )
 349     {
 350         size_t nLen = (len != wxSTRING_MAXLEN) ? len :
 351                           nLen = wxConvUTF8.MB2WC((wchar_t*) NULL, s, 0);
 352
 353         wchar_t *buf = new wchar_t[nLen+1];
 354         wxConvUTF8.MB2WC(buf, s, nLen);
 355         buf[nLen] = 0;
 356         wxString s(buf, *conv, len);
 357         delete[] buf;
 358         return s;
 359     }
 360     else
 361         return wxString(s, len);
 362 #endif
 363 }
 364
 365 struct wxXmlParsingContext
 366 {
 367     wxMBConv  *conv;
 368     wxXmlNode *root;
 369     wxXmlNode *node;
 370     wxXmlNode *lastAsText;
 371     wxString   encoding;
 372     wxString   version;
 373 };
 374
 375 static void StartElementHnd(void *userData, const char *name, const char **atts)
 376 {
 377     wxXmlParsingContext *ctx = (wxXmlParsingContext*)userData;
 378     wxXmlNode *node = new wxXmlNode(wxXML_ELEMENT_NODE, CharToString(ctx->conv, name));
 379     const char **a = atts;
 380     while (*a)
 381     {
 382         node->AddProperty(CharToString(ctx->conv, a[0]), CharToString(ctx->conv, a[1]));
 383         a += 2;
 384     }
 385     if (ctx->root == NULL)
 386         ctx->root = node;
 387     else
 388         ctx->node->AddChild(node);
 389     ctx->node = node;
 390     ctx->lastAsText = NULL;
 391 }
 392
 393 static void EndElementHnd(void *userData, const char* WXUNUSED(name))
 394 {
 395     wxXmlParsingContext *ctx = (wxXmlParsingContext*)userData;
 396
 397     ctx->node = ctx->node->GetParent();
 398     ctx->lastAsText = NULL;
 399 }
 400
 401 static void TextHnd(void *userData, const char *s, int len)
 402 {
 403     wxXmlParsingContext *ctx = (wxXmlParsingContext*)userData;
 404     char *buf = new char[len + 1];
 405
 406     buf[len] = '\0';
 407     memcpy(buf, s, (size_t)len);
 408
 409     if (ctx->lastAsText)
 410     {
 411         ctx->lastAsText->SetContent(ctx->lastAsText->GetContent() +
 412                                     CharToString(ctx->conv, buf));
 413     }
 414     else
 415     {
 416         bool whiteOnly = TRUE;
 417         for (char *c = buf; *c != '\0'; c++)
 418             if (*c != ' ' && *c != '\t' && *c != '\n' && *c != '\r')
 419             {
 420                 whiteOnly = FALSE;
 421                 break;
 422             }
 423         if (!whiteOnly)
 424         {
 425             ctx->lastAsText = new wxXmlNode(wxXML_TEXT_NODE, wxT("text"),
 426                                             CharToString(ctx->conv, buf));
 427             ctx->node->AddChild(ctx->lastAsText);
 428         }
 429     }
 430
 431     delete[] buf;
 432 }
 433
 434 static void CommentHnd(void *userData, const char *data)
 435 {
 436     wxXmlParsingContext *ctx = (wxXmlParsingContext*)userData;
 437
 438     if (ctx->node)
 439     {
 440         // VS: ctx->node == NULL happens if there is a comment before
 441         //     the root element (e.g. wxDesigner's output). We ignore such
 442         //     comments, no big deal...
 443         ctx->node->AddChild(new wxXmlNode(wxXML_COMMENT_NODE,
 444                             wxT("comment"), CharToString(ctx->conv, data)));
 445     }
 446     ctx->lastAsText = NULL;
 447 }
 448
 449 static void DefaultHnd(void *userData, const char *s, int len)
 450 {
 451     // XML header:
 452     if (len > 6 && memcmp(s, "<?xml ", 6) == 0)
 453     {
 454         wxXmlParsingContext *ctx = (wxXmlParsingContext*)userData;
 455
 456         wxString buf = CharToString(ctx->conv, s, (size_t)len);
 457         int pos;
 458         pos = buf.Find(wxT("encoding="));
 459         if (pos != wxNOT_FOUND)
 460             ctx->encoding = buf.Mid(pos + 10).BeforeFirst(buf[(size_t)pos+9]);
 461         pos = buf.Find(wxT("version="));
 462         if (pos != wxNOT_FOUND)
 463             ctx->version = buf.Mid(pos + 9).BeforeFirst(buf[(size_t)pos+8]);
 464     }
 465 }
 466
 467 static int UnknownEncodingHnd(void * WXUNUSED(encodingHandlerData),
 468                                const XML_Char *name, XML_Encoding *info)
 469 {
 470     // We must build conversion table for expat. The easiest way to do so
 471     // is to let wxCSConv convert as string containing all characters to
 472     // wide character representation:
 473     wxCSConv conv(wxString(name, wxConvLibc));
 474     char mbBuf[255];
 475     wchar_t wcBuf[255];
 476     size_t i;
 477
 478     for (i = 0; i < 255; i++)
 479         mbBuf[i] = (char) (i+1);
 480     mbBuf[255] = 0;
 481     conv.MB2WC(wcBuf, mbBuf, 255);
 482     wcBuf[255] = 0;
 483
 484     info->map[0] = 0;
 485     for (i = 0; i < 255; i++)
 486         info->map[i+1] = (int)wcBuf[i];
 487
 488     info->data = NULL;
 489     info->convert = NULL;
 490     info->release = NULL;
 491
 492     return 1;
 493 }
 494
 495 bool wxXmlDocument::Load(wxInputStream& stream, const wxString& encoding)
 496 {
 497 #if wxUSE_UNICODE
 498     (void)encoding;
 499 #else
 500     m_encoding = encoding;
 501 #endif
 502
 503     const size_t BUFSIZE = 1024;
 504     char buf[BUFSIZE];
 505     wxXmlParsingContext ctx;
 506     bool done;
 507     XML_Parser parser = XML_ParserCreate(NULL);
 508
 509     ctx.root = ctx.node = NULL;
 510     ctx.encoding = wxT("UTF-8"); // default in absence of encoding=""
 511     ctx.conv = NULL;
 512 #if !wxUSE_UNICODE
 513     if ( encoding != wxT("UTF-8") && encoding != wxT("utf-8") )
 514         ctx.conv = new wxCSConv(encoding);
 515 #endif
 516
 517     XML_SetUserData(parser, (void*)&ctx);
 518     XML_SetElementHandler(parser, StartElementHnd, EndElementHnd);
 519     XML_SetCharacterDataHandler(parser, TextHnd);
 520     XML_SetCommentHandler(parser, CommentHnd);
 521     XML_SetDefaultHandler(parser, DefaultHnd);
 522     XML_SetUnknownEncodingHandler(parser, UnknownEncodingHnd, NULL);
 523
 524     do
 525     {
 526         size_t len = stream.Read(buf, BUFSIZE).LastRead();
 527         done = (len < BUFSIZE);
 528         if (!XML_Parse(parser, buf, len, done))
 529         {
 530             wxLogError(_("XML parsing error: '%s' at line %d"),
 531                        XML_ErrorString(XML_GetErrorCode(parser)),
 532                        XML_GetCurrentLineNumber(parser));
 533           return FALSE;
 534         }
 535     } while (!done);
 536
 537     SetVersion(ctx.version);
 538     SetFileEncoding(ctx.encoding);
 539     SetRoot(ctx.root);
 540
 541     XML_ParserFree(parser);
 542 #if !wxUSE_UNICODE
 543     if ( ctx.conv )
 544         delete ctx.conv;
 545 #endif
 546
 547     return TRUE;
 548
 549 }
 550
 551
 552
 553 //-----------------------------------------------------------------------------
 554 //  wxXmlDocument saving routines
 555 //-----------------------------------------------------------------------------
 556
 557 // write string to output:
 558 inline static void OutputString(wxOutputStream& stream, const wxString& str,
 559                                 wxMBConv *convMem, wxMBConv *convFile)
 560 {
 561     if (str.IsEmpty()) return;
 562 #if wxUSE_UNICODE
 563     const wxWX2MBbuf buf(str.mb_str(convFile ? *convFile : wxConvUTF8));
 564     stream.Write((const char*)buf, strlen((const char*)buf));
 565 #else
 566     if ( convFile == NULL )
 567         stream.Write(str.mb_str(), str.Len());
 568     else
 569     {
 570         wxString str2(str.wc_str(*convMem), *convFile);
 571         stream.Write(str2.mb_str(), str2.Len());
 572     }
 573 #endif
 574 }
 575
 576 // Same as above, but create entities first.
 577 // Translates '<' to "&lt;", '>' to "&gt;" and '&' to "&amp;"
 578 static void OutputStringEnt(wxOutputStream& stream, const wxString& str,
 579                             wxMBConv *convMem, wxMBConv *convFile)
 580 {
 581     wxString buf;
 582     size_t i, last, len;
 583     wxChar c;
 584
 585     len = str.Len();
 586     last = 0;
 587     for (i = 0; i < len; i++)
 588     {
 589         c = str.GetChar(i);
 590         if (c == wxT('<') || c == wxT('>') ||
 591             (c == wxT('&') && str.Mid(i+1, 4) != wxT("amp;")))
 592         {
 593             OutputString(stream, str.Mid(last, i - last), convMem, convFile);
 594             switch (c)
 595             {
 596                 case wxT('<'):
 597                     OutputString(stream, wxT("&lt;"), NULL, NULL);
 598                     break;
 599                 case wxT('>'):
 600                     OutputString(stream, wxT("&gt;"), NULL, NULL);
 601                     break;
 602                 case wxT('&'):
 603                     OutputString(stream, wxT("&amp;"), NULL, NULL);
 604                     break;
 605                 default: break;
 606             }
 607             last = i + 1;
 608         }
 609     }
 610     OutputString(stream, str.Mid(last, i - last), convMem, convFile);
 611 }
 612
 613 inline static void OutputIndentation(wxOutputStream& stream, int indent)
 614 {
 615     wxString str = wxT("\n");
 616     for (int i = 0; i < indent; i++)
 617         str << wxT(' ') << wxT(' ');
 618     OutputString(stream, str, NULL, NULL);
 619 }
 620
 621 static void OutputNode(wxOutputStream& stream, wxXmlNode *node, int indent,
 622                        wxMBConv *convMem, wxMBConv *convFile)
 623 {
 624     wxXmlNode *n, *prev;
 625     wxXmlProperty *prop;
 626
 627     switch (node->GetType())
 628     {
 629         case wxXML_TEXT_NODE:
 630             OutputStringEnt(stream, node->GetContent(), convMem, convFile);
 631             break;
 632
 633         case wxXML_ELEMENT_NODE:
 634             OutputString(stream, wxT("<"), NULL, NULL);
 635             OutputString(stream, node->GetName(), NULL, NULL);
 636
 637             prop = node->GetProperties();
 638             while (prop)
 639             {
 640                 OutputString(stream, wxT(" ") + prop->GetName() +
 641                              wxT("=\"") + prop->GetValue() + wxT("\""),
 642                              NULL, NULL);
 643                 // FIXME - what if prop contains '"'?
 644                 prop = prop->GetNext();
 645             }
 646
 647             if (node->GetChildren())
 648             {
 649                 OutputString(stream, wxT(">"), NULL, NULL);
 650                 prev = NULL;
 651                 n = node->GetChildren();
 652                 while (n)
 653                 {
 654                     if (n && n->GetType() != wxXML_TEXT_NODE)
 655                         OutputIndentation(stream, indent + 1);
 656                     OutputNode(stream, n, indent + 1, convMem, convFile);
 657                     prev = n;
 658                     n = n->GetNext();
 659                 }
 660                 if (prev && prev->GetType() != wxXML_TEXT_NODE)
 661                     OutputIndentation(stream, indent);
 662                 OutputString(stream, wxT("</"), NULL, NULL);
 663                 OutputString(stream, node->GetName(), NULL, NULL);
 664                 OutputString(stream, wxT(">"), NULL, NULL);
 665             }
 666             else
 667                 OutputString(stream, wxT("/>"), NULL, NULL);
 668             break;
 669
 670         case wxXML_COMMENT_NODE:
 671             OutputString(stream, wxT("<!--"), NULL, NULL);
 672             OutputString(stream, node->GetContent(), convMem, convFile);
 673             OutputString(stream, wxT("-->"), NULL, NULL);
 674             break;
 675
 676         default:
 677             wxFAIL_MSG(wxT("unsupported node type"));
 678     }
 679 }
 680
 681 bool wxXmlDocument::Save(wxOutputStream& stream) const
 682 {
 683     if ( !IsOk() )
 684         return FALSE;
 685
 686     wxString s;
 687
 688     wxMBConv *convMem = NULL, *convFile = NULL;
 689 #if wxUSE_UNICODE
 690     convFile = new wxCSConv(GetFileEncoding());
 691 #else
 692     if ( GetFileEncoding() != GetEncoding() )
 693     {
 694         convFile = new wxCSConv(GetFileEncoding());
 695         convMem = new wxCSConv(GetEncoding());
 696     }
 697 #endif
 698
 699     s.Printf(wxT("<?xml version=\"%s\" encoding=\"%s\"?>\n"),
 700              GetVersion().c_str(), GetFileEncoding().c_str());
 701     OutputString(stream, s, NULL, NULL);
 702
 703     OutputNode(stream, GetRoot(), 0, convMem, convFile);
 704     OutputString(stream, wxT("\n"), NULL, NULL);
 705
 706     if ( convFile )
 707         delete convFile;
 708     if ( convMem )
 709         delete convMem;
 710
 711     return TRUE;
 712 }