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 #endif
  14
  15 // For compilers that support precompilation, includes "wx.h".
  16 #include "wx/wxprec.h"
  17
  18 #ifdef __BORLANDC__
  19     #pragma hdrstop
  20 #endif
  21
  22
  23 #include "wx/wfstream.h"
  24 #include "wx/datstrm.h"
  25 #include "wx/zstream.h"
  26 #include "wx/log.h"
  27 #include "wx/intl.h"
  28 #include "wx/strconv.h"
  29
  30 #include "wx/xrc/xml.h"
  31
  32 #include "xmlparse.h" // from Expat
  33
  34 //-----------------------------------------------------------------------------
  35 //  wxXmlNode
  36 //-----------------------------------------------------------------------------
  37
  38 wxXmlNode::wxXmlNode(wxXmlNode *parent,wxXmlNodeType type,
  39                      const wxString& name, const wxString& content,
  40                      wxXmlProperty *props, wxXmlNode *next)
  41     : m_type(type), m_name(name), m_content(content),
  42       m_properties(props), m_parent(parent),
  43       m_children(NULL), m_next(next)
  44 {
  45     if (m_parent)
  46     {
  47         if (m_parent->m_children)
  48         {
  49             m_next = m_parent->m_children;
  50             m_parent->m_children = this;
  51         }
  52         else
  53             m_parent->m_children = this;
  54     }
  55 }
  56
  57 wxXmlNode::wxXmlNode(wxXmlNodeType type, const wxString& name,
  58                      const wxString& content)
  59     : m_type(type), m_name(name), m_content(content),
  60       m_properties(NULL), m_parent(NULL),
  61       m_children(NULL), m_next(NULL)
  62 {}
  63
  64 wxXmlNode::wxXmlNode(const wxXmlNode& node)
  65 {
  66     m_next = NULL;
  67     m_parent = NULL;
  68     DoCopy(node);
  69 }
  70
  71 wxXmlNode::~wxXmlNode()
  72 {
  73     wxXmlNode *c, *c2;
  74     for (c = m_children; c; c = c2)
  75     {
  76         c2 = c->m_next;
  77         delete c;
  78     }
  79
  80     wxXmlProperty *p, *p2;
  81     for (p = m_properties; p; p = p2)
  82     {
  83         p2 = p->GetNext();
  84         delete p;
  85     }
  86 }
  87
  88 wxXmlNode& wxXmlNode::operator=(const wxXmlNode& node)
  89 {
  90     wxDELETE(m_properties);
  91     wxDELETE(m_children);
  92     DoCopy(node);
  93     return *this;
  94 }
  95
  96 void wxXmlNode::DoCopy(const wxXmlNode& node)
  97 {
  98     m_type = node.m_type;
  99     m_name = node.m_name;
 100     m_content = node.m_content;
 101     m_children = NULL;
 102
 103     wxXmlNode *n = node.m_children;
 104     while (n)
 105     {
 106         AddChild(new wxXmlNode(*n));
 107         n = n->GetNext();
 108     }
 109
 110     m_properties = NULL;
 111     wxXmlProperty *p = node.m_properties;
 112     while (p)
 113     {
 114        AddProperty(p->GetName(), p->GetValue());
 115        p = p->GetNext();
 116     }
 117 }
 118
 119 bool wxXmlNode::HasProp(const wxString& propName) const
 120 {
 121     wxXmlProperty *prop = GetProperties();
 122
 123     while (prop)
 124     {
 125         if (prop->GetName() == propName) return TRUE;
 126         prop = prop->GetNext();
 127     }
 128
 129     return FALSE;
 130 }
 131
 132 bool wxXmlNode::GetPropVal(const wxString& propName, wxString *value) const
 133 {
 134     wxXmlProperty *prop = GetProperties();
 135
 136     while (prop)
 137     {
 138         if (prop->GetName() == propName)
 139         {
 140             *value = prop->GetValue();
 141             return TRUE;
 142         }
 143         prop = prop->GetNext();
 144     }
 145
 146     return FALSE;
 147 }
 148
 149 wxString wxXmlNode::GetPropVal(const wxString& propName, const wxString& defaultVal) const
 150 {
 151     wxString tmp;
 152     if (GetPropVal(propName, &tmp))
 153         return tmp;
 154     else
 155         return defaultVal;
 156 }
 157
 158 void wxXmlNode::AddChild(wxXmlNode *child)
 159 {
 160     if (m_children == NULL)
 161         m_children = child;
 162     else
 163     {
 164         wxXmlNode *ch = m_children;
 165         while (ch->m_next) ch = ch->m_next;
 166         ch->m_next = child;
 167     }
 168     child->m_next = NULL;
 169     child->m_parent = this;
 170 }
 171
 172 void wxXmlNode::InsertChild(wxXmlNode *child, wxXmlNode *before_node)
 173 {
 174     wxASSERT_MSG(before_node->GetParent() == this, wxT("wxXmlNode::InsertChild - the node has incorrect parent"));
 175
 176     if (m_children == before_node)
 177        m_children = child;
 178     else
 179     {
 180         wxXmlNode *ch = m_children;
 181         while (ch->m_next != before_node) ch = ch->m_next;
 182         ch->m_next = child;
 183     }
 184
 185     child->m_parent = this;
 186     child->m_next = before_node;
 187 }
 188
 189 bool wxXmlNode::RemoveChild(wxXmlNode *child)
 190 {
 191     if (m_children == NULL)
 192         return FALSE;
 193     else if (m_children == child)
 194     {
 195         m_children = child->m_next;
 196         child->m_parent = NULL;
 197         child->m_next = NULL;
 198         return TRUE;
 199     }
 200     else
 201     {
 202         wxXmlNode *ch = m_children;
 203         while (ch->m_next)
 204         {
 205             if (ch->m_next == child)
 206             {
 207                 ch->m_next = child->m_next;
 208                 child->m_parent = NULL;
 209                 child->m_next = NULL;
 210                 return TRUE;
 211             }
 212             ch = ch->m_next;
 213         }
 214         return FALSE;
 215     }
 216 }
 217
 218 void wxXmlNode::AddProperty(const wxString& name, const wxString& value)
 219 {
 220     AddProperty(new wxXmlProperty(name, value, NULL));
 221 }
 222
 223 void wxXmlNode::AddProperty(wxXmlProperty *prop)
 224 {
 225     if (m_properties == NULL)
 226         m_properties = prop;
 227     else
 228     {
 229         wxXmlProperty *p = m_properties;
 230         while (p->GetNext()) p = p->GetNext();
 231         p->SetNext(prop);
 232     }
 233 }
 234
 235 bool wxXmlNode::DeleteProperty(const wxString& name)
 236 {
 237     wxXmlProperty *prop;
 238
 239     if (m_properties == NULL)
 240         return FALSE;
 241
 242     else if (m_properties->GetName() == name)
 243     {
 244         prop = m_properties;
 245         m_properties = prop->GetNext();
 246         prop->SetNext(NULL);
 247         delete prop;
 248         return TRUE;
 249     }
 250
 251     else
 252     {
 253         wxXmlProperty *p = m_properties;
 254         while (p->GetNext())
 255         {
 256             if (p->GetNext()->GetName() == name)
 257             {
 258                 prop = p->GetNext();
 259                 p->SetNext(prop->GetNext());
 260                 prop->SetNext(NULL);
 261                 delete prop;
 262                 return TRUE;
 263             }
 264             p = p->GetNext();
 265         }
 266         return FALSE;
 267     }
 268 }
 269
 270
 271
 272 //-----------------------------------------------------------------------------
 273 //  wxXmlDocument
 274 //-----------------------------------------------------------------------------
 275
 276 wxXmlDocument::wxXmlDocument()
 277     : m_version(wxT("1.0")), m_fileEncoding(wxT("utf-8")), m_root(NULL)
 278 {
 279 #if !wxUSE_UNICODE
 280     m_encoding = wxT("UTF-8");
 281 #endif
 282 }
 283
 284 wxXmlDocument::wxXmlDocument(const wxString& filename, const wxString& encoding)
 285                           : wxObject(), m_root(NULL)
 286 {
 287     if ( !Load(filename, encoding) )
 288     {
 289         wxDELETE(m_root);
 290     }
 291 }
 292
 293 wxXmlDocument::wxXmlDocument(wxInputStream& stream, const wxString& encoding)
 294                           : wxObject(), m_root(NULL)
 295 {
 296     if ( !Load(stream, encoding) )
 297     {
 298         wxDELETE(m_root);
 299     }
 300 }
 301
 302 wxXmlDocument::wxXmlDocument(const wxXmlDocument& doc)
 303 {
 304     DoCopy(doc);
 305 }
 306
 307 wxXmlDocument& wxXmlDocument::operator=(const wxXmlDocument& doc)
 308 {
 309     wxDELETE(m_root);
 310     DoCopy(doc);
 311     return *this;
 312 }
 313
 314 void wxXmlDocument::DoCopy(const wxXmlDocument& doc)
 315 {
 316     m_version = doc.m_version;
 317 #if !wxUSE_UNICODE
 318     m_encoding = doc.m_encoding;
 319 #endif
 320     m_fileEncoding = doc.m_fileEncoding;
 321     m_root = new wxXmlNode(*doc.m_root);
 322 }
 323
 324 bool wxXmlDocument::Load(const wxString& filename, const wxString& encoding)
 325 {
 326     wxFileInputStream stream(filename);
 327     return Load(stream, encoding);
 328 }
 329
 330 bool wxXmlDocument::Save(const wxString& filename) const
 331 {
 332     wxFileOutputStream stream(filename);
 333     return Save(stream);
 334 }
 335
 336
 337
 338 //-----------------------------------------------------------------------------
 339 //  wxXmlDocument loading routines
 340 //-----------------------------------------------------------------------------
 341
 342 /*
 343     FIXME:
 344        - process all elements, including CDATA
 345  */
 346
 347 // converts Expat-produced string in UTF-8 into wxString.
 348 inline static wxString CharToString(wxMBConv *conv,
 349                                     const char *s, size_t len = wxSTRING_MAXLEN)
 350 {
 351 #if wxUSE_UNICODE
 352     (void)conv;
 353     return wxString(s, wxConvUTF8, len);
 354 #else
 355     if ( conv )
 356     {
 357         size_t nLen = (len != wxSTRING_MAXLEN) ? len :
 358                           nLen = wxConvUTF8.MB2WC((wchar_t*) NULL, s, 0);
 359
 360         wchar_t *buf = new wchar_t[nLen+1];
 361         wxConvUTF8.MB2WC(buf, s, nLen);
 362         buf[nLen] = 0;
 363         wxString str(buf, *conv, len);
 364         delete[] buf;
 365         return str;
 366     }
 367     else
 368         return wxString(s, len);
 369 #endif
 370 }
 371
 372 struct wxXmlParsingContext
 373 {
 374     wxMBConv  *conv;
 375     wxXmlNode *root;
 376     wxXmlNode *node;
 377     wxXmlNode *lastAsText;
 378     wxString   encoding;
 379     wxString   version;
 380 };
 381
 382 static void StartElementHnd(void *userData, const char *name, const char **atts)
 383 {
 384     wxXmlParsingContext *ctx = (wxXmlParsingContext*)userData;
 385     wxXmlNode *node = new wxXmlNode(wxXML_ELEMENT_NODE, CharToString(ctx->conv, name));
 386     const char **a = atts;
 387     while (*a)
 388     {
 389         node->AddProperty(CharToString(ctx->conv, a[0]), CharToString(ctx->conv, a[1]));
 390         a += 2;
 391     }
 392     if (ctx->root == NULL)
 393         ctx->root = node;
 394     else
 395         ctx->node->AddChild(node);
 396     ctx->node = node;
 397     ctx->lastAsText = NULL;
 398 }
 399
 400 static void EndElementHnd(void *userData, const char* WXUNUSED(name))
 401 {
 402     wxXmlParsingContext *ctx = (wxXmlParsingContext*)userData;
 403
 404     ctx->node = ctx->node->GetParent();
 405     ctx->lastAsText = NULL;
 406 }
 407
 408 static void TextHnd(void *userData, const char *s, int len)
 409 {
 410     wxXmlParsingContext *ctx = (wxXmlParsingContext*)userData;
 411     char *buf = new char[len + 1];
 412
 413     buf[len] = '\0';
 414     memcpy(buf, s, (size_t)len);
 415
 416     if (ctx->lastAsText)
 417     {
 418         ctx->lastAsText->SetContent(ctx->lastAsText->GetContent() +
 419                                     CharToString(ctx->conv, buf));
 420     }
 421     else
 422     {
 423         bool whiteOnly = TRUE;
 424         for (char *c = buf; *c != '\0'; c++)
 425             if (*c != ' ' && *c != '\t' && *c != '\n' && *c != '\r')
 426             {
 427                 whiteOnly = FALSE;
 428                 break;
 429             }
 430         if (!whiteOnly)
 431         {
 432             ctx->lastAsText = new wxXmlNode(wxXML_TEXT_NODE, wxT("text"),
 433                                             CharToString(ctx->conv, buf));
 434             ctx->node->AddChild(ctx->lastAsText);
 435         }
 436     }
 437
 438     delete[] buf;
 439 }
 440
 441 static void CommentHnd(void *userData, const char *data)
 442 {
 443     wxXmlParsingContext *ctx = (wxXmlParsingContext*)userData;
 444
 445     if (ctx->node)
 446     {
 447         // VS: ctx->node == NULL happens if there is a comment before
 448         //     the root element (e.g. wxDesigner's output). We ignore such
 449         //     comments, no big deal...
 450         ctx->node->AddChild(new wxXmlNode(wxXML_COMMENT_NODE,
 451                             wxT("comment"), CharToString(ctx->conv, data)));
 452     }
 453     ctx->lastAsText = NULL;
 454 }
 455
 456 static void DefaultHnd(void *userData, const char *s, int len)
 457 {
 458     // XML header:
 459     if (len > 6 && memcmp(s, "<?xml ", 6) == 0)
 460     {
 461         wxXmlParsingContext *ctx = (wxXmlParsingContext*)userData;
 462
 463         wxString buf = CharToString(ctx->conv, s, (size_t)len);
 464         int pos;
 465         pos = buf.Find(wxT("encoding="));
 466         if (pos != wxNOT_FOUND)
 467             ctx->encoding = buf.Mid(pos + 10).BeforeFirst(buf[(size_t)pos+9]);
 468         pos = buf.Find(wxT("version="));
 469         if (pos != wxNOT_FOUND)
 470             ctx->version = buf.Mid(pos + 9).BeforeFirst(buf[(size_t)pos+8]);
 471     }
 472 }
 473
 474 static int UnknownEncodingHnd(void * WXUNUSED(encodingHandlerData),
 475                               const XML_Char *name, XML_Encoding *info)
 476 {
 477     // We must build conversion table for expat. The easiest way to do so
 478     // is to let wxCSConv convert as string containing all characters to
 479     // wide character representation:
 480     wxCSConv conv(wxString(name, wxConvLibc));
 481     char mbBuf[2];
 482     wchar_t wcBuf[10];
 483     size_t i;
 484
 485     mbBuf[1] = 0;
 486     info->map[0] = 0;
 487     for (i = 0; i < 255; i++)
 488     {
 489         mbBuf[0] = (char)(i+1);
 490         if (conv.MB2WC(wcBuf, mbBuf, 2) == (size_t)-1)
 491         {
 492             // invalid/undefined byte in the encoding:
 493             info->map[i+1] = -1;
 494         }
 495         info->map[i+1] = (int)wcBuf[0];
 496     }
 497
 498     info->data = NULL;
 499     info->convert = NULL;
 500     info->release = NULL;
 501
 502     return 1;
 503 }
 504
 505 bool wxXmlDocument::Load(wxInputStream& stream, const wxString& encoding)
 506 {
 507 #if wxUSE_UNICODE
 508     (void)encoding;
 509 #else
 510     m_encoding = encoding;
 511 #endif
 512
 513     const size_t BUFSIZE = 1024;
 514     char buf[BUFSIZE];
 515     wxXmlParsingContext ctx;
 516     bool done;
 517     XML_Parser parser = XML_ParserCreate(NULL);
 518
 519     ctx.root = ctx.node = NULL;
 520     ctx.encoding = wxT("UTF-8"); // default in absence of encoding=""
 521     ctx.conv = NULL;
 522 #if !wxUSE_UNICODE
 523     if ( encoding != wxT("UTF-8") && encoding != wxT("utf-8") )
 524         ctx.conv = new wxCSConv(encoding);
 525 #endif
 526
 527     XML_SetUserData(parser, (void*)&ctx);
 528     XML_SetElementHandler(parser, StartElementHnd, EndElementHnd);
 529     XML_SetCharacterDataHandler(parser, TextHnd);
 530     XML_SetCommentHandler(parser, CommentHnd);
 531     XML_SetDefaultHandler(parser, DefaultHnd);
 532     XML_SetUnknownEncodingHandler(parser, UnknownEncodingHnd, NULL);
 533
 534     do
 535     {
 536         size_t len = stream.Read(buf, BUFSIZE).LastRead();
 537         done = (len < BUFSIZE);
 538         if (!XML_Parse(parser, buf, len, done))
 539         {
 540             wxLogError(_("XML parsing error: '%s' at line %d"),
 541                        XML_ErrorString(XML_GetErrorCode(parser)),
 542                        XML_GetCurrentLineNumber(parser));
 543           return FALSE;
 544         }
 545     } while (!done);
 546
 547     SetVersion(ctx.version);
 548     SetFileEncoding(ctx.encoding);
 549     SetRoot(ctx.root);
 550
 551     XML_ParserFree(parser);
 552 #if !wxUSE_UNICODE
 553     if ( ctx.conv )
 554         delete ctx.conv;
 555 #endif
 556
 557     return TRUE;
 558
 559 }
 560
 561
 562
 563 //-----------------------------------------------------------------------------
 564 //  wxXmlDocument saving routines
 565 //-----------------------------------------------------------------------------
 566
 567 // write string to output:
 568 inline static void OutputString(wxOutputStream& stream, const wxString& str,
 569                                 wxMBConv *convMem, wxMBConv *convFile)
 570 {
 571     if (str.IsEmpty()) return;
 572 #if wxUSE_UNICODE
 573     const wxWX2MBbuf buf(str.mb_str(convFile ? *convFile : wxConvUTF8));
 574     stream.Write((const char*)buf, strlen((const char*)buf));
 575 #else
 576     if ( convFile == NULL )
 577         stream.Write(str.mb_str(), str.Len());
 578     else
 579     {
 580         wxString str2(str.wc_str(*convMem), *convFile);
 581         stream.Write(str2.mb_str(), str2.Len());
 582     }
 583 #endif
 584 }
 585
 586 // Same as above, but create entities first.
 587 // Translates '<' to "&lt;", '>' to "&gt;" and '&' to "&amp;"
 588 static void OutputStringEnt(wxOutputStream& stream, const wxString& str,
 589                             wxMBConv *convMem, wxMBConv *convFile)
 590 {
 591     wxString buf;
 592     size_t i, last, len;
 593     wxChar c;
 594
 595     len = str.Len();
 596     last = 0;
 597     for (i = 0; i < len; i++)
 598     {
 599         c = str.GetChar(i);
 600         if (c == wxT('<') || c == wxT('>') ||
 601             (c == wxT('&') && str.Mid(i+1, 4) != wxT("amp;")))
 602         {
 603             OutputString(stream, str.Mid(last, i - last), convMem, convFile);
 604             switch (c)
 605             {
 606                 case wxT('<'):
 607                     OutputString(stream, wxT("&lt;"), NULL, NULL);
 608                     break;
 609                 case wxT('>'):
 610                     OutputString(stream, wxT("&gt;"), NULL, NULL);
 611                     break;
 612                 case wxT('&'):
 613                     OutputString(stream, wxT("&amp;"), NULL, NULL);
 614                     break;
 615                 default: break;
 616             }
 617             last = i + 1;
 618         }
 619     }
 620     OutputString(stream, str.Mid(last, i - last), convMem, convFile);
 621 }
 622
 623 inline static void OutputIndentation(wxOutputStream& stream, int indent)
 624 {
 625     wxString str = wxT("\n");
 626     for (int i = 0; i < indent; i++)
 627         str << wxT(' ') << wxT(' ');
 628     OutputString(stream, str, NULL, NULL);
 629 }
 630
 631 static void OutputNode(wxOutputStream& stream, wxXmlNode *node, int indent,
 632                        wxMBConv *convMem, wxMBConv *convFile)
 633 {
 634     wxXmlNode *n, *prev;
 635     wxXmlProperty *prop;
 636
 637     switch (node->GetType())
 638     {
 639         case wxXML_TEXT_NODE:
 640             OutputStringEnt(stream, node->GetContent(), convMem, convFile);
 641             break;
 642
 643         case wxXML_ELEMENT_NODE:
 644             OutputString(stream, wxT("<"), NULL, NULL);
 645             OutputString(stream, node->GetName(), NULL, NULL);
 646
 647             prop = node->GetProperties();
 648             while (prop)
 649             {
 650                 OutputString(stream, wxT(" ") + prop->GetName() +
 651                              wxT("=\"") + prop->GetValue() + wxT("\""),
 652                              NULL, NULL);
 653                 // FIXME - what if prop contains '"'?
 654                 prop = prop->GetNext();
 655             }
 656
 657             if (node->GetChildren())
 658             {
 659                 OutputString(stream, wxT(">"), NULL, NULL);
 660                 prev = NULL;
 661                 n = node->GetChildren();
 662                 while (n)
 663                 {
 664                     if (n && n->GetType() != wxXML_TEXT_NODE)
 665                         OutputIndentation(stream, indent + 1);
 666                     OutputNode(stream, n, indent + 1, convMem, convFile);
 667                     prev = n;
 668                     n = n->GetNext();
 669                 }
 670                 if (prev && prev->GetType() != wxXML_TEXT_NODE)
 671                     OutputIndentation(stream, indent);
 672                 OutputString(stream, wxT("</"), NULL, NULL);
 673                 OutputString(stream, node->GetName(), NULL, NULL);
 674                 OutputString(stream, wxT(">"), NULL, NULL);
 675             }
 676             else
 677                 OutputString(stream, wxT("/>"), NULL, NULL);
 678             break;
 679
 680         case wxXML_COMMENT_NODE:
 681             OutputString(stream, wxT("<!--"), NULL, NULL);
 682             OutputString(stream, node->GetContent(), convMem, convFile);
 683             OutputString(stream, wxT("-->"), NULL, NULL);
 684             break;
 685
 686         default:
 687             wxFAIL_MSG(wxT("unsupported node type"));
 688     }
 689 }
 690
 691 bool wxXmlDocument::Save(wxOutputStream& stream) const
 692 {
 693     if ( !IsOk() )
 694         return FALSE;
 695
 696     wxString s;
 697
 698     wxMBConv *convMem = NULL, *convFile = NULL;
 699 #if wxUSE_UNICODE
 700     convFile = new wxCSConv(GetFileEncoding());
 701 #else
 702     if ( GetFileEncoding() != GetEncoding() )
 703     {
 704         convFile = new wxCSConv(GetFileEncoding());
 705         convMem = new wxCSConv(GetEncoding());
 706     }
 707 #endif
 708
 709     s.Printf(wxT("<?xml version=\"%s\" encoding=\"%s\"?>\n"),
 710              GetVersion().c_str(), GetFileEncoding().c_str());
 711     OutputString(stream, s, NULL, NULL);
 712
 713     OutputNode(stream, GetRoot(), 0, convMem, convFile);
 714     OutputString(stream, wxT("\n"), NULL, NULL);
 715
 716     if ( convFile )
 717         delete convFile;
 718     if ( convMem )
 719         delete convMem;
 720
 721     return TRUE;
 722 }