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         return wxString(buf, *conv, len);
 357         delete[] buf;
 358     }
 359     else
 360         return wxString(s, len);
 361 #endif
 362 }
 363
 364 struct wxXmlParsingContext
 365 {
 366     wxMBConv  *conv;
 367     wxXmlNode *root;
 368     wxXmlNode *node;
 369     wxXmlNode *lastAsText;
 370     wxString   encoding;
 371     wxString   version;
 372 };
 373
 374 static void StartElementHnd(void *userData, const char *name, const char **atts)
 375 {
 376     wxXmlParsingContext *ctx = (wxXmlParsingContext*)userData;
 377     wxXmlNode *node = new wxXmlNode(wxXML_ELEMENT_NODE, CharToString(ctx->conv, name));
 378     const char **a = atts;
 379     while (*a)
 380     {
 381         node->AddProperty(CharToString(ctx->conv, a[0]), CharToString(ctx->conv, a[1]));
 382         a += 2;
 383     }
 384     if (ctx->root == NULL)
 385         ctx->root = node;
 386     else
 387         ctx->node->AddChild(node);
 388     ctx->node = node;
 389     ctx->lastAsText = NULL;
 390 }
 391
 392 static void EndElementHnd(void *userData, const char* WXUNUSED(name))
 393 {
 394     wxXmlParsingContext *ctx = (wxXmlParsingContext*)userData;
 395
 396     ctx->node = ctx->node->GetParent();
 397     ctx->lastAsText = NULL;
 398 }
 399
 400 static void TextHnd(void *userData, const char *s, int len)
 401 {
 402     wxXmlParsingContext *ctx = (wxXmlParsingContext*)userData;
 403     char *buf = new char[len + 1];
 404
 405     buf[len] = '\0';
 406     memcpy(buf, s, (size_t)len);
 407
 408     if (ctx->lastAsText)
 409     {
 410         ctx->lastAsText->SetContent(ctx->lastAsText->GetContent() +
 411                                     CharToString(ctx->conv, buf));
 412     }
 413     else
 414     {
 415         bool whiteOnly = TRUE;
 416         for (char *c = buf; *c != '\0'; c++)
 417             if (*c != ' ' && *c != '\t' && *c != '\n' && *c != '\r')
 418             {
 419                 whiteOnly = FALSE;
 420                 break;
 421             }
 422         if (!whiteOnly)
 423         {
 424             ctx->lastAsText = new wxXmlNode(wxXML_TEXT_NODE, wxT("text"),
 425                                             CharToString(ctx->conv, buf));
 426             ctx->node->AddChild(ctx->lastAsText);
 427         }
 428     }
 429
 430     delete[] buf;
 431 }
 432
 433 static void CommentHnd(void *userData, const char *data)
 434 {
 435     wxXmlParsingContext *ctx = (wxXmlParsingContext*)userData;
 436
 437     if (ctx->node)
 438     {
 439         // VS: ctx->node == NULL happens if there is a comment before
 440         //     the root element (e.g. wxDesigner's output). We ignore such
 441         //     comments, no big deal...
 442         ctx->node->AddChild(new wxXmlNode(wxXML_COMMENT_NODE,
 443                             wxT("comment"), CharToString(ctx->conv, data)));
 444     }
 445     ctx->lastAsText = NULL;
 446 }
 447
 448 static void DefaultHnd(void *userData, const char *s, int len)
 449 {
 450     // XML header:
 451     if (len > 6 && memcmp(s, "<?xml ", 6) == 0)
 452     {
 453         wxXmlParsingContext *ctx = (wxXmlParsingContext*)userData;
 454
 455         wxString buf = CharToString(ctx->conv, s, (size_t)len);
 456         int pos;
 457         pos = buf.Find(wxT("encoding="));
 458         if (pos != wxNOT_FOUND)
 459             ctx->encoding = buf.Mid(pos + 10).BeforeFirst(buf[(size_t)pos+9]);
 460         pos = buf.Find(wxT("version="));
 461         if (pos != wxNOT_FOUND)
 462             ctx->version = buf.Mid(pos + 9).BeforeFirst(buf[(size_t)pos+8]);
 463     }
 464 }
 465
 466 static int UnknownEncodingHnd(void * WXUNUSED(encodingHandlerData),
 467                                const XML_Char *name, XML_Encoding *info)
 468 {
 469     // We must build conversion table for expat. The easiest way to do so
 470     // is to let wxCSConv convert as string containing all characters to
 471     // wide character representation:
 472     wxCSConv conv(wxString(name, wxConvLibc));
 473     char mbBuf[255];
 474     wchar_t wcBuf[255];
 475     size_t i;
 476
 477     for (i = 0; i < 255; i++)
 478         mbBuf[i] = i+1;
 479     mbBuf[255] = 0;
 480     conv.MB2WC(wcBuf, mbBuf, 255);
 481     wcBuf[255] = 0;
 482
 483     info->map[0] = 0;
 484     for (i = 0; i < 255; i++)
 485         info->map[i+1] = (int)wcBuf[i];
 486
 487     info->data = NULL;
 488     info->convert = NULL;
 489     info->release = NULL;
 490
 491     return 1;
 492 }
 493
 494 bool wxXmlDocument::Load(wxInputStream& stream, const wxString& encoding)
 495 {
 496 #if wxUSE_UNICODE
 497     (void)encoding;
 498 #else
 499     m_encoding = encoding;
 500 #endif
 501
 502     const size_t BUFSIZE = 1024;
 503     char buf[BUFSIZE];
 504     wxXmlParsingContext ctx;
 505     bool done;
 506     XML_Parser parser = XML_ParserCreate(NULL);
 507
 508     ctx.root = ctx.node = NULL;
 509     ctx.encoding = wxT("UTF-8"); // default in absence of encoding=""
 510     ctx.conv = NULL;
 511 #if !wxUSE_UNICODE
 512     if ( encoding != wxT("UTF-8") && encoding != wxT("utf-8") )
 513         ctx.conv = new wxCSConv(encoding);
 514 #endif
 515
 516     XML_SetUserData(parser, (void*)&ctx);
 517     XML_SetElementHandler(parser, StartElementHnd, EndElementHnd);
 518     XML_SetCharacterDataHandler(parser, TextHnd);
 519     XML_SetCommentHandler(parser, CommentHnd);
 520     XML_SetDefaultHandler(parser, DefaultHnd);
 521     XML_SetUnknownEncodingHandler(parser, UnknownEncodingHnd, NULL);
 522
 523     do
 524     {
 525         size_t len = stream.Read(buf, BUFSIZE).LastRead();
 526         done = (len < BUFSIZE);
 527         if (!XML_Parse(parser, buf, len, done))
 528         {
 529             wxLogError(_("XML parsing error: '%s' at line %d"),
 530                        XML_ErrorString(XML_GetErrorCode(parser)),
 531                        XML_GetCurrentLineNumber(parser));
 532           return FALSE;
 533         }
 534     } while (!done);
 535
 536     SetVersion(ctx.version);
 537     SetFileEncoding(ctx.encoding);
 538     SetRoot(ctx.root);
 539
 540     XML_ParserFree(parser);
 541 #if !wxUSE_UNICODE
 542     if ( ctx.conv )
 543         delete ctx.conv;
 544 #endif
 545
 546     return TRUE;
 547
 548 }
 549
 550
 551
 552 //-----------------------------------------------------------------------------
 553 //  wxXmlDocument saving routines
 554 //-----------------------------------------------------------------------------
 555
 556 // write string to output:
 557 inline static void OutputString(wxOutputStream& stream, const wxString& str,
 558                                 wxMBConv *convMem, wxMBConv *convFile)
 559 {
 560     if (str.IsEmpty()) return;
 561 #if wxUSE_UNICODE
 562     const wxWX2MBbuf buf(str.mb_str(convFile ? *convFile : wxConvUTF8));
 563     stream.Write((const char*)buf, strlen((const char*)buf));
 564 #else
 565     if ( convFile == NULL )
 566         stream.Write(str.mb_str(), str.Len());
 567     else
 568     {
 569         wxString str2(str.wc_str(*convMem), *convFile);
 570         stream.Write(str2.mb_str(), str2.Len());
 571     }
 572 #endif
 573 }
 574
 575 // Same as above, but create entities first.
 576 // Translates '<' to "&lt;", '>' to "&gt;" and '&' to "&amp;"
 577 static void OutputStringEnt(wxOutputStream& stream, const wxString& str,
 578                             wxMBConv *convMem, wxMBConv *convFile)
 579 {
 580     wxString buf;
 581     size_t i, last, len;
 582     wxChar c;
 583
 584     len = str.Len();
 585     last = 0;
 586     for (i = 0; i < len; i++)
 587     {
 588         c = str.GetChar(i);
 589         if (c == wxT('<') || c == wxT('>') ||
 590             (c == wxT('&') && str.Mid(i+1, 4) != wxT("amp;")))
 591         {
 592             OutputString(stream, str.Mid(last, i - last), convMem, convFile);
 593             switch (c)
 594             {
 595                 case wxT('<'):
 596                     OutputString(stream, wxT("&lt;"), NULL, NULL);
 597                     break;
 598                 case wxT('>'):
 599                     OutputString(stream, wxT("&gt;"), NULL, NULL);
 600                     break;
 601                 case wxT('&'):
 602                     OutputString(stream, wxT("&amp;"), NULL, NULL);
 603                     break;
 604                 default: break;
 605             }
 606             last = i + 1;
 607         }
 608     }
 609     OutputString(stream, str.Mid(last, i - last), convMem, convFile);
 610 }
 611
 612 inline static void OutputIndentation(wxOutputStream& stream, int indent)
 613 {
 614     wxString str = wxT("\n");
 615     for (int i = 0; i < indent; i++)
 616         str << wxT(' ') << wxT(' ');
 617     OutputString(stream, str, NULL, NULL);
 618 }
 619
 620 static void OutputNode(wxOutputStream& stream, wxXmlNode *node, int indent,
 621                        wxMBConv *convMem, wxMBConv *convFile)
 622 {
 623     wxXmlNode *n, *prev;
 624     wxXmlProperty *prop;
 625
 626     switch (node->GetType())
 627     {
 628         case wxXML_TEXT_NODE:
 629             OutputStringEnt(stream, node->GetContent(), convMem, convFile);
 630             break;
 631
 632         case wxXML_ELEMENT_NODE:
 633             OutputString(stream, wxT("<"), NULL, NULL);
 634             OutputString(stream, node->GetName(), NULL, NULL);
 635
 636             prop = node->GetProperties();
 637             while (prop)
 638             {
 639                 OutputString(stream, wxT(" ") + prop->GetName() +
 640                              wxT("=\"") + prop->GetValue() + wxT("\""),
 641                              NULL, NULL);
 642                 // FIXME - what if prop contains '"'?
 643                 prop = prop->GetNext();
 644             }
 645
 646             if (node->GetChildren())
 647             {
 648                 OutputString(stream, wxT(">"), NULL, NULL);
 649                 prev = NULL;
 650                 n = node->GetChildren();
 651                 while (n)
 652                 {
 653                     if (n && n->GetType() != wxXML_TEXT_NODE)
 654                         OutputIndentation(stream, indent + 1);
 655                     OutputNode(stream, n, indent + 1, convMem, convFile);
 656                     prev = n;
 657                     n = n->GetNext();
 658                 }
 659                 if (prev && prev->GetType() != wxXML_TEXT_NODE)
 660                     OutputIndentation(stream, indent);
 661                 OutputString(stream, wxT("</"), NULL, NULL);
 662                 OutputString(stream, node->GetName(), NULL, NULL);
 663                 OutputString(stream, wxT(">"), NULL, NULL);
 664             }
 665             else
 666                 OutputString(stream, wxT("/>"), NULL, NULL);
 667             break;
 668
 669         case wxXML_COMMENT_NODE:
 670             OutputString(stream, wxT("<!--"), NULL, NULL);
 671             OutputString(stream, node->GetContent(), convMem, convFile);
 672             OutputString(stream, wxT("-->"), NULL, NULL);
 673             break;
 674
 675         default:
 676             wxFAIL_MSG(wxT("unsupported node type"));
 677     }
 678 }
 679
 680 bool wxXmlDocument::Save(wxOutputStream& stream) const
 681 {
 682     if ( !IsOk() )
 683         return FALSE;
 684
 685     wxString s;
 686
 687     wxMBConv *convMem = NULL, *convFile = NULL;
 688 #if wxUSE_UNICODE
 689     convFile = new wxCSConv(GetFileEncoding());
 690 #else
 691     if ( GetFileEncoding() != GetEncoding() )
 692     {
 693         convFile = new wxCSConv(GetFileEncoding());
 694         convMem = new wxCSConv(GetEncoding());
 695     }
 696 #endif
 697
 698     s.Printf(wxT("<?xml version=\"%s\" encoding=\"%s\"?>\n"),
 699              GetVersion().c_str(), GetFileEncoding().c_str());
 700     OutputString(stream, s, NULL, NULL);
 701
 702     OutputNode(stream, GetRoot(), 0, convMem, convFile);
 703     OutputString(stream, wxT("\n"), NULL, NULL);
 704
 705     if ( convFile )
 706         delete convFile;
 707     if ( convMem )
 708         delete convMem;
 709
 710     return TRUE;
 711 }