src/xml/xml.cpp

   1 /////////////////////////////////////////////////////////////////////////////
   2 // Name:        src/xml/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 // For compilers that support precompilation, includes "wx.h".
  12 #include "wx/wxprec.h"
  13
  14 #ifdef __BORLANDC__
  15     #pragma hdrstop
  16 #endif
  17
  18 #if wxUSE_XML
  19
  20 #include "wx/xml/xml.h"
  21
  22 #ifndef WX_PRECOMP
  23     #include "wx/intl.h"
  24     #include "wx/log.h"
  25     #include "wx/app.h"
  26 #endif
  27
  28 #include "wx/wfstream.h"
  29 #include "wx/datstrm.h"
  30 #include "wx/zstream.h"
  31 #include "wx/strconv.h"
  32 #include "wx/scopedptr.h"
  33 #include "wx/versioninfo.h"
  34
  35 #include "expat.h" // from Expat
  36
  37 // DLL options compatibility check:
  38 WX_CHECK_BUILD_OPTIONS("wxXML")
  39
  40
  41 IMPLEMENT_CLASS(wxXmlDocument, wxObject)
  42
  43
  44 // a private utility used by wxXML
  45 static bool wxIsWhiteOnly(const wxString& buf);
  46
  47
  48 //-----------------------------------------------------------------------------
  49 //  wxXmlNode
  50 //-----------------------------------------------------------------------------
  51
  52 wxXmlNode::wxXmlNode(wxXmlNode *parent,wxXmlNodeType type,
  53                      const wxString& name, const wxString& content,
  54                      wxXmlAttribute *attrs, wxXmlNode *next, int lineNo)
  55     : m_type(type), m_name(name), m_content(content),
  56       m_attrs(attrs), m_parent(parent),
  57       m_children(NULL), m_next(next),
  58       m_lineNo(lineNo),
  59       m_noConversion(false)
  60 {
  61     if (m_parent)
  62     {
  63         if (m_parent->m_children)
  64         {
  65             m_next = m_parent->m_children;
  66             m_parent->m_children = this;
  67         }
  68         else
  69             m_parent->m_children = this;
  70     }
  71 }
  72
  73 wxXmlNode::wxXmlNode(wxXmlNodeType type, const wxString& name,
  74                      const wxString& content,
  75                      int lineNo)
  76     : m_type(type), m_name(name), m_content(content),
  77       m_attrs(NULL), m_parent(NULL),
  78       m_children(NULL), m_next(NULL),
  79       m_lineNo(lineNo), m_noConversion(false)
  80 {}
  81
  82 wxXmlNode::wxXmlNode(const wxXmlNode& node)
  83 {
  84     m_next = NULL;
  85     m_parent = NULL;
  86     DoCopy(node);
  87 }
  88
  89 wxXmlNode::~wxXmlNode()
  90 {
  91     wxXmlNode *c, *c2;
  92     for (c = m_children; c; c = c2)
  93     {
  94         c2 = c->m_next;
  95         delete c;
  96     }
  97
  98     wxXmlAttribute *p, *p2;
  99     for (p = m_attrs; p; p = p2)
 100     {
 101         p2 = p->GetNext();
 102         delete p;
 103     }
 104 }
 105
 106 wxXmlNode& wxXmlNode::operator=(const wxXmlNode& node)
 107 {
 108     wxDELETE(m_attrs);
 109     wxDELETE(m_children);
 110     DoCopy(node);
 111     return *this;
 112 }
 113
 114 void wxXmlNode::DoCopy(const wxXmlNode& node)
 115 {
 116     m_type = node.m_type;
 117     m_name = node.m_name;
 118     m_content = node.m_content;
 119     m_lineNo = node.m_lineNo;
 120     m_noConversion = node.m_noConversion;
 121     m_children = NULL;
 122
 123     wxXmlNode *n = node.m_children;
 124     while (n)
 125     {
 126         AddChild(new wxXmlNode(*n));
 127         n = n->GetNext();
 128     }
 129
 130     m_attrs = NULL;
 131     wxXmlAttribute *p = node.m_attrs;
 132     while (p)
 133     {
 134        AddAttribute(p->GetName(), p->GetValue());
 135        p = p->GetNext();
 136     }
 137 }
 138
 139 bool wxXmlNode::HasAttribute(const wxString& attrName) const
 140 {
 141     wxXmlAttribute *attr = GetAttributes();
 142
 143     while (attr)
 144     {
 145         if (attr->GetName() == attrName) return true;
 146         attr = attr->GetNext();
 147     }
 148
 149     return false;
 150 }
 151
 152 bool wxXmlNode::GetAttribute(const wxString& attrName, wxString *value) const
 153 {
 154     wxCHECK_MSG( value, false, "value argument must not be NULL" );
 155
 156     wxXmlAttribute *attr = GetAttributes();
 157
 158     while (attr)
 159     {
 160         if (attr->GetName() == attrName)
 161         {
 162             *value = attr->GetValue();
 163             return true;
 164         }
 165         attr = attr->GetNext();
 166     }
 167
 168     return false;
 169 }
 170
 171 wxString wxXmlNode::GetAttribute(const wxString& attrName, const wxString& defaultVal) const
 172 {
 173     wxString tmp;
 174     if (GetAttribute(attrName, &tmp))
 175         return tmp;
 176
 177     return defaultVal;
 178 }
 179
 180 void wxXmlNode::AddChild(wxXmlNode *child)
 181 {
 182     if (m_children == NULL)
 183         m_children = child;
 184     else
 185     {
 186         wxXmlNode *ch = m_children;
 187         while (ch->m_next) ch = ch->m_next;
 188         ch->m_next = child;
 189     }
 190     child->m_next = NULL;
 191     child->m_parent = this;
 192 }
 193
 194 // inserts a new node in front of 'followingNode'
 195 bool wxXmlNode::InsertChild(wxXmlNode *child, wxXmlNode *followingNode)
 196 {
 197     wxCHECK_MSG( child, false, "cannot insert a NULL node!" );
 198     wxCHECK_MSG( child->m_parent == NULL, false, "node already has a parent" );
 199     wxCHECK_MSG( child->m_next == NULL, false, "node already has m_next" );
 200     wxCHECK_MSG( followingNode == NULL || followingNode->GetParent() == this,
 201                  false,
 202                  "wxXmlNode::InsertChild - followingNode has incorrect parent" );
 203
 204     // this is for backward compatibility, NULL was allowed here thanks to
 205     // the confusion about followingNode's meaning
 206     if ( followingNode == NULL )
 207         followingNode = m_children;
 208
 209     if ( m_children == followingNode )
 210     {
 211         child->m_next = m_children;
 212         m_children = child;
 213     }
 214     else
 215     {
 216         wxXmlNode *ch = m_children;
 217         while ( ch && ch->m_next != followingNode )
 218             ch = ch->m_next;
 219         if ( !ch )
 220         {
 221             wxFAIL_MSG( "followingNode has this node as parent, but couldn't be found among children" );
 222             return false;
 223         }
 224
 225         child->m_next = followingNode;
 226         ch->m_next = child;
 227     }
 228
 229     child->m_parent = this;
 230     return true;
 231 }
 232
 233 // inserts a new node right after 'precedingNode'
 234 bool wxXmlNode::InsertChildAfter(wxXmlNode *child, wxXmlNode *precedingNode)
 235 {
 236     wxCHECK_MSG( child, false, "cannot insert a NULL node!" );
 237     wxCHECK_MSG( child->m_parent == NULL, false, "node already has a parent" );
 238     wxCHECK_MSG( child->m_next == NULL, false, "node already has m_next" );
 239     wxCHECK_MSG( precedingNode == NULL || precedingNode->m_parent == this, false,
 240                  "precedingNode has wrong parent" );
 241
 242     if ( precedingNode )
 243     {
 244         child->m_next = precedingNode->m_next;
 245         precedingNode->m_next = child;
 246     }
 247     else // precedingNode == NULL
 248     {
 249         wxCHECK_MSG( m_children == NULL, false,
 250                      "NULL precedingNode only makes sense when there are no children" );
 251
 252         child->m_next = m_children;
 253         m_children = child;
 254     }
 255
 256     child->m_parent = this;
 257     return true;
 258 }
 259
 260 bool wxXmlNode::RemoveChild(wxXmlNode *child)
 261 {
 262     if (m_children == NULL)
 263         return false;
 264     else if (m_children == child)
 265     {
 266         m_children = child->m_next;
 267         child->m_parent = NULL;
 268         child->m_next = NULL;
 269         return true;
 270     }
 271     else
 272     {
 273         wxXmlNode *ch = m_children;
 274         while (ch->m_next)
 275         {
 276             if (ch->m_next == child)
 277             {
 278                 ch->m_next = child->m_next;
 279                 child->m_parent = NULL;
 280                 child->m_next = NULL;
 281                 return true;
 282             }
 283             ch = ch->m_next;
 284         }
 285         return false;
 286     }
 287 }
 288
 289 void wxXmlNode::AddAttribute(const wxString& name, const wxString& value)
 290 {
 291     AddProperty(name, value);
 292 }
 293
 294 void wxXmlNode::AddAttribute(wxXmlAttribute *attr)
 295 {
 296     AddProperty(attr);
 297 }
 298
 299 bool wxXmlNode::DeleteAttribute(const wxString& name)
 300 {
 301     return DeleteProperty(name);
 302 }
 303
 304 void wxXmlNode::AddProperty(const wxString& name, const wxString& value)
 305 {
 306     AddProperty(new wxXmlAttribute(name, value, NULL));
 307 }
 308
 309 void wxXmlNode::AddProperty(wxXmlAttribute *attr)
 310 {
 311     if (m_attrs == NULL)
 312         m_attrs = attr;
 313     else
 314     {
 315         wxXmlAttribute *p = m_attrs;
 316         while (p->GetNext()) p = p->GetNext();
 317         p->SetNext(attr);
 318     }
 319 }
 320
 321 bool wxXmlNode::DeleteProperty(const wxString& name)
 322 {
 323     wxXmlAttribute *attr;
 324
 325     if (m_attrs == NULL)
 326         return false;
 327
 328     else if (m_attrs->GetName() == name)
 329     {
 330         attr = m_attrs;
 331         m_attrs = attr->GetNext();
 332         attr->SetNext(NULL);
 333         delete attr;
 334         return true;
 335     }
 336
 337     else
 338     {
 339         wxXmlAttribute *p = m_attrs;
 340         while (p->GetNext())
 341         {
 342             if (p->GetNext()->GetName() == name)
 343             {
 344                 attr = p->GetNext();
 345                 p->SetNext(attr->GetNext());
 346                 attr->SetNext(NULL);
 347                 delete attr;
 348                 return true;
 349             }
 350             p = p->GetNext();
 351         }
 352         return false;
 353     }
 354 }
 355
 356 wxString wxXmlNode::GetNodeContent() const
 357 {
 358     wxXmlNode *n = GetChildren();
 359
 360     while (n)
 361     {
 362         if (n->GetType() == wxXML_TEXT_NODE ||
 363             n->GetType() == wxXML_CDATA_SECTION_NODE)
 364             return n->GetContent();
 365         n = n->GetNext();
 366     }
 367     return wxEmptyString;
 368 }
 369
 370 int wxXmlNode::GetDepth(wxXmlNode *grandparent) const
 371 {
 372     const wxXmlNode *n = this;
 373     int ret = -1;
 374
 375     do
 376     {
 377         ret++;
 378         n = n->GetParent();
 379         if (n == grandparent)
 380             return ret;
 381
 382     } while (n);
 383
 384     return wxNOT_FOUND;
 385 }
 386
 387 bool wxXmlNode::IsWhitespaceOnly() const
 388 {
 389     return wxIsWhiteOnly(m_content);
 390 }
 391
 392
 393
 394 //-----------------------------------------------------------------------------
 395 //  wxXmlDocument
 396 //-----------------------------------------------------------------------------
 397
 398 wxXmlDocument::wxXmlDocument()
 399     : m_version(wxS("1.0")), m_fileEncoding(wxS("UTF-8")), m_docNode(NULL)
 400 {
 401 #if !wxUSE_UNICODE
 402     m_encoding = wxS("UTF-8");
 403 #endif
 404 }
 405
 406 wxXmlDocument::wxXmlDocument(const wxString& filename, const wxString& encoding)
 407               :wxObject(), m_docNode(NULL)
 408 {
 409     if ( !Load(filename, encoding) )
 410     {
 411         wxDELETE(m_docNode);
 412     }
 413 }
 414
 415 wxXmlDocument::wxXmlDocument(wxInputStream& stream, const wxString& encoding)
 416               :wxObject(), m_docNode(NULL)
 417 {
 418     if ( !Load(stream, encoding) )
 419     {
 420         wxDELETE(m_docNode);
 421     }
 422 }
 423
 424 wxXmlDocument::wxXmlDocument(const wxXmlDocument& doc)
 425               :wxObject()
 426 {
 427     DoCopy(doc);
 428 }
 429
 430 wxXmlDocument& wxXmlDocument::operator=(const wxXmlDocument& doc)
 431 {
 432     wxDELETE(m_docNode);
 433     DoCopy(doc);
 434     return *this;
 435 }
 436
 437 void wxXmlDocument::DoCopy(const wxXmlDocument& doc)
 438 {
 439     m_version = doc.m_version;
 440 #if !wxUSE_UNICODE
 441     m_encoding = doc.m_encoding;
 442 #endif
 443     m_fileEncoding = doc.m_fileEncoding;
 444
 445     if (doc.m_docNode)
 446         m_docNode = new wxXmlNode(*doc.m_docNode);
 447     else
 448         m_docNode = NULL;
 449 }
 450
 451 bool wxXmlDocument::Load(const wxString& filename, const wxString& encoding, int flags)
 452 {
 453     wxFileInputStream stream(filename);
 454     if (!stream.Ok())
 455         return false;
 456     return Load(stream, encoding, flags);
 457 }
 458
 459 bool wxXmlDocument::Save(const wxString& filename, int indentstep) const
 460 {
 461     wxFileOutputStream stream(filename);
 462     if (!stream.Ok())
 463         return false;
 464     return Save(stream, indentstep);
 465 }
 466
 467 wxXmlNode *wxXmlDocument::GetRoot() const
 468 {
 469     wxXmlNode *node = m_docNode;
 470     if (node)
 471     {
 472         node = m_docNode->GetChildren();
 473         while (node != NULL && node->GetType() != wxXML_ELEMENT_NODE)
 474             node = node->GetNext();
 475     }
 476     return node;
 477 }
 478
 479 wxXmlNode *wxXmlDocument::DetachRoot()
 480 {
 481     wxXmlNode *node = m_docNode;
 482     if (node)
 483     {
 484         node = m_docNode->GetChildren();
 485         wxXmlNode *prev = NULL;
 486         while (node != NULL && node->GetType() != wxXML_ELEMENT_NODE)
 487         {
 488             prev = node;
 489             node = node->GetNext();
 490         }
 491         if (node)
 492         {
 493             if (node == m_docNode->GetChildren())
 494                 m_docNode->SetChildren(node->GetNext());
 495
 496             if (prev)
 497                 prev->SetNext(node->GetNext());
 498
 499             node->SetParent(NULL);
 500             node->SetNext(NULL);
 501         }
 502     }
 503     return node;
 504 }
 505
 506 void wxXmlDocument::SetRoot(wxXmlNode *root)
 507 {
 508     if (root)
 509     {
 510         wxASSERT_MSG( root->GetType() == wxXML_ELEMENT_NODE,
 511                       "Can only set an element type node as root" );
 512     }
 513
 514     wxXmlNode *node = m_docNode;
 515     if (node)
 516     {
 517         node = m_docNode->GetChildren();
 518         wxXmlNode *prev = NULL;
 519         while (node != NULL && node->GetType() != wxXML_ELEMENT_NODE)
 520         {
 521             prev = node;
 522             node = node->GetNext();
 523         }
 524         if (node && root)
 525         {
 526             root->SetNext( node->GetNext() );
 527             wxDELETE(node);
 528         }
 529         if (prev)
 530             prev->SetNext(root);
 531         else
 532             m_docNode->SetChildren(root);
 533     }
 534     else
 535     {
 536         m_docNode = new wxXmlNode(wxXML_DOCUMENT_NODE, wxEmptyString);
 537         m_docNode->SetChildren(root);
 538     }
 539     if (root)
 540         root->SetParent(m_docNode);
 541 }
 542
 543 void wxXmlDocument::AppendToProlog(wxXmlNode *node)
 544 {
 545     if (!m_docNode)
 546         m_docNode = new wxXmlNode(wxXML_DOCUMENT_NODE, wxEmptyString);
 547     if (IsOk())
 548         m_docNode->InsertChild( node, GetRoot() );
 549     else
 550         m_docNode->AddChild( node );
 551 }
 552
 553 //-----------------------------------------------------------------------------
 554 //  wxXmlDocument loading routines
 555 //-----------------------------------------------------------------------------
 556
 557 // converts Expat-produced string in UTF-8 into wxString using the specified
 558 // conv or keep in UTF-8 if conv is NULL
 559 static wxString CharToString(wxMBConv *conv,
 560                              const char *s, size_t len = wxString::npos)
 561 {
 562 #if !wxUSE_UNICODE
 563     if ( conv )
 564     {
 565         // there can be no embedded NULs in this string so we don't need the
 566         // output length, it will be NUL-terminated
 567         const wxWCharBuffer wbuf(
 568             wxConvUTF8.cMB2WC(s, len == wxString::npos ? wxNO_LEN : len, NULL));
 569
 570         return wxString(wbuf, *conv);
 571     }
 572     // else: the string is wanted in UTF-8
 573 #endif // !wxUSE_UNICODE
 574
 575     wxUnusedVar(conv);
 576     return wxString::FromUTF8Unchecked(s, len);
 577 }
 578
 579 // returns true if the given string contains only whitespaces
 580 bool wxIsWhiteOnly(const wxString& buf)
 581 {
 582     for ( wxString::const_iterator i = buf.begin(); i != buf.end(); ++i )
 583     {
 584         wxChar c = *i;
 585         if ( c != wxS(' ') && c != wxS('\t') && c != wxS('\n') && c != wxS('\r'))
 586             return false;
 587     }
 588     return true;
 589 }
 590
 591
 592 struct wxXmlParsingContext
 593 {
 594     wxXmlParsingContext()
 595         : conv(NULL),
 596           node(NULL),
 597           lastChild(NULL),
 598           lastAsText(NULL),
 599           removeWhiteOnlyNodes(false)
 600     {}
 601
 602     XML_Parser parser;
 603     wxMBConv  *conv;
 604     wxXmlNode *node;                    // the node being parsed
 605     wxXmlNode *lastChild;               // the last child of "node"
 606     wxXmlNode *lastAsText;              // the last _text_ child of "node"
 607     wxString   encoding;
 608     wxString   version;
 609     bool       removeWhiteOnlyNodes;
 610 };
 611
 612 // checks that ctx->lastChild is in consistent state
 613 #define ASSERT_LAST_CHILD_OK(ctx)                                   \
 614     wxASSERT( ctx->lastChild == NULL ||                             \
 615               ctx->lastChild->GetNext() == NULL );                  \
 616     wxASSERT( ctx->lastChild == NULL ||                             \
 617               ctx->lastChild->GetParent() == ctx->node )
 618
 619 extern "C" {
 620 static void StartElementHnd(void *userData, const char *name, const char **atts)
 621 {
 622     wxXmlParsingContext *ctx = (wxXmlParsingContext*)userData;
 623     wxXmlNode *node = new wxXmlNode(wxXML_ELEMENT_NODE,
 624                                     CharToString(ctx->conv, name),
 625                                     wxEmptyString,
 626                                     XML_GetCurrentLineNumber(ctx->parser));
 627     const char **a = atts;
 628
 629     // add node attributes
 630     while (*a)
 631     {
 632         node->AddAttribute(CharToString(ctx->conv, a[0]), CharToString(ctx->conv, a[1]));
 633         a += 2;
 634     }
 635
 636     ASSERT_LAST_CHILD_OK(ctx);
 637     ctx->node->InsertChildAfter(node, ctx->lastChild);
 638     ctx->lastAsText = NULL;
 639     ctx->lastChild = NULL; // our new node "node" has no children yet
 640
 641     ctx->node = node;
 642 }
 643
 644 static void EndElementHnd(void *userData, const char* WXUNUSED(name))
 645 {
 646     wxXmlParsingContext *ctx = (wxXmlParsingContext*)userData;
 647
 648     // we're exiting the last children of ctx->node->GetParent() and going
 649     // back one level up, so current value of ctx->node points to the last
 650     // child of ctx->node->GetParent()
 651     ctx->lastChild = ctx->node;
 652
 653     ctx->node = ctx->node->GetParent();
 654     ctx->lastAsText = NULL;
 655 }
 656
 657 static void TextHnd(void *userData, const char *s, int len)
 658 {
 659     wxXmlParsingContext *ctx = (wxXmlParsingContext*)userData;
 660     wxString str = CharToString(ctx->conv, s, len);
 661
 662     if (ctx->lastAsText)
 663     {
 664         ctx->lastAsText->SetContent(ctx->lastAsText->GetContent() + str);
 665     }
 666     else
 667     {
 668         bool whiteOnly = false;
 669         if (ctx->removeWhiteOnlyNodes)
 670             whiteOnly = wxIsWhiteOnly(str);
 671
 672         if (!whiteOnly)
 673         {
 674             wxXmlNode *textnode =
 675                 new wxXmlNode(wxXML_TEXT_NODE, wxS("text"), str,
 676                               XML_GetCurrentLineNumber(ctx->parser));
 677
 678             ASSERT_LAST_CHILD_OK(ctx);
 679             ctx->node->InsertChildAfter(textnode, ctx->lastChild);
 680             ctx->lastChild= ctx->lastAsText = textnode;
 681         }
 682     }
 683 }
 684
 685 static void StartCdataHnd(void *userData)
 686 {
 687     wxXmlParsingContext *ctx = (wxXmlParsingContext*)userData;
 688
 689     wxXmlNode *textnode =
 690         new wxXmlNode(wxXML_CDATA_SECTION_NODE, wxS("cdata"), wxS(""),
 691                       XML_GetCurrentLineNumber(ctx->parser));
 692
 693     ASSERT_LAST_CHILD_OK(ctx);
 694     ctx->node->InsertChildAfter(textnode, ctx->lastChild);
 695     ctx->lastChild= ctx->lastAsText = textnode;
 696 }
 697
 698 static void EndCdataHnd(void *userData)
 699 {
 700     wxXmlParsingContext *ctx = (wxXmlParsingContext*)userData;
 701
 702     // we need to reset this pointer so that subsequent text nodes don't append
 703     // their contents to this one but create new wxXML_TEXT_NODE objects (or
 704     // not create anything at all if only white space follows the CDATA section
 705     // and wxXMLDOC_KEEP_WHITESPACE_NODES is not used as is commonly the case)
 706     ctx->lastAsText = NULL;
 707 }
 708
 709 static void CommentHnd(void *userData, const char *data)
 710 {
 711     wxXmlParsingContext *ctx = (wxXmlParsingContext*)userData;
 712
 713     wxXmlNode *commentnode =
 714         new wxXmlNode(wxXML_COMMENT_NODE,
 715                       wxS("comment"), CharToString(ctx->conv, data),
 716                       XML_GetCurrentLineNumber(ctx->parser));
 717
 718     ASSERT_LAST_CHILD_OK(ctx);
 719     ctx->node->InsertChildAfter(commentnode, ctx->lastChild);
 720     ctx->lastChild = commentnode;
 721     ctx->lastAsText = NULL;
 722 }
 723
 724 static void PIHnd(void *userData, const char *target, const char *data)
 725 {
 726     wxXmlParsingContext *ctx = (wxXmlParsingContext*)userData;
 727
 728     wxXmlNode *pinode =
 729         new wxXmlNode(wxXML_PI_NODE, CharToString(ctx->conv, target),
 730                       CharToString(ctx->conv, data),
 731                       XML_GetCurrentLineNumber(ctx->parser));
 732
 733     ASSERT_LAST_CHILD_OK(ctx);
 734     ctx->node->InsertChildAfter(pinode, ctx->lastChild);
 735     ctx->lastChild = pinode;
 736     ctx->lastAsText = NULL;
 737 }
 738
 739 static void DefaultHnd(void *userData, const char *s, int len)
 740 {
 741     // XML header:
 742     if (len > 6 && memcmp(s, "<?xml ", 6) == 0)
 743     {
 744         wxXmlParsingContext *ctx = (wxXmlParsingContext*)userData;
 745
 746         wxString buf = CharToString(ctx->conv, s, (size_t)len);
 747         int pos;
 748         pos = buf.Find(wxS("encoding="));
 749         if (pos != wxNOT_FOUND)
 750             ctx->encoding = buf.Mid(pos + 10).BeforeFirst(buf[(size_t)pos+9]);
 751         pos = buf.Find(wxS("version="));
 752         if (pos != wxNOT_FOUND)
 753             ctx->version = buf.Mid(pos + 9).BeforeFirst(buf[(size_t)pos+8]);
 754     }
 755 }
 756
 757 static int UnknownEncodingHnd(void * WXUNUSED(encodingHandlerData),
 758                               const XML_Char *name, XML_Encoding *info)
 759 {
 760     // We must build conversion table for expat. The easiest way to do so
 761     // is to let wxCSConv convert as string containing all characters to
 762     // wide character representation:
 763     wxCSConv conv(name);
 764     char mbBuf[2];
 765     wchar_t wcBuf[10];
 766     size_t i;
 767
 768     mbBuf[1] = 0;
 769     info->map[0] = 0;
 770     for (i = 0; i < 255; i++)
 771     {
 772         mbBuf[0] = (char)(i+1);
 773         if (conv.MB2WC(wcBuf, mbBuf, 2) == (size_t)-1)
 774         {
 775             // invalid/undefined byte in the encoding:
 776             info->map[i+1] = -1;
 777         }
 778         info->map[i+1] = (int)wcBuf[0];
 779     }
 780
 781     info->data = NULL;
 782     info->convert = NULL;
 783     info->release = NULL;
 784
 785     return 1;
 786 }
 787
 788 } // extern "C"
 789
 790 bool wxXmlDocument::Load(wxInputStream& stream, const wxString& encoding, int flags)
 791 {
 792 #if wxUSE_UNICODE
 793     (void)encoding;
 794 #else
 795     m_encoding = encoding;
 796 #endif
 797
 798     const size_t BUFSIZE = 1024;
 799     char buf[BUFSIZE];
 800     wxXmlParsingContext ctx;
 801     bool done;
 802     XML_Parser parser = XML_ParserCreate(NULL);
 803     wxXmlNode *root = new wxXmlNode(wxXML_DOCUMENT_NODE, wxEmptyString);
 804
 805     ctx.encoding = wxS("UTF-8"); // default in absence of encoding=""
 806     ctx.conv = NULL;
 807 #if !wxUSE_UNICODE
 808     if ( encoding.CmpNoCase(wxS("UTF-8")) != 0 )
 809         ctx.conv = new wxCSConv(encoding);
 810 #endif
 811     ctx.removeWhiteOnlyNodes = (flags & wxXMLDOC_KEEP_WHITESPACE_NODES) == 0;
 812     ctx.parser = parser;
 813     ctx.node = root;
 814
 815     XML_SetUserData(parser, (void*)&ctx);
 816     XML_SetElementHandler(parser, StartElementHnd, EndElementHnd);
 817     XML_SetCharacterDataHandler(parser, TextHnd);
 818     XML_SetCdataSectionHandler(parser, StartCdataHnd, EndCdataHnd);;
 819     XML_SetCommentHandler(parser, CommentHnd);
 820     XML_SetProcessingInstructionHandler(parser, PIHnd);
 821     XML_SetDefaultHandler(parser, DefaultHnd);
 822     XML_SetUnknownEncodingHandler(parser, UnknownEncodingHnd, NULL);
 823
 824     bool ok = true;
 825     do
 826     {
 827         size_t len = stream.Read(buf, BUFSIZE).LastRead();
 828         done = (len < BUFSIZE);
 829         if (!XML_Parse(parser, buf, len, done))
 830         {
 831             wxString error(XML_ErrorString(XML_GetErrorCode(parser)),
 832                            *wxConvCurrent);
 833             wxLogError(_("XML parsing error: '%s' at line %d"),
 834                        error.c_str(),
 835                        (int)XML_GetCurrentLineNumber(parser));
 836             ok = false;
 837             break;
 838         }
 839     } while (!done);
 840
 841     if (ok)
 842     {
 843         if (!ctx.version.empty())
 844             SetVersion(ctx.version);
 845         if (!ctx.encoding.empty())
 846             SetFileEncoding(ctx.encoding);
 847         SetDocumentNode(root);
 848     }
 849     else
 850     {
 851         delete root;
 852     }
 853
 854     XML_ParserFree(parser);
 855 #if !wxUSE_UNICODE
 856     if ( ctx.conv )
 857         delete ctx.conv;
 858 #endif
 859
 860     return ok;
 861
 862 }
 863
 864
 865
 866 //-----------------------------------------------------------------------------
 867 //  wxXmlDocument saving routines
 868 //-----------------------------------------------------------------------------
 869
 870 // helpers for XML generation
 871 namespace
 872 {
 873
 874 // write string to output:
 875 bool OutputString(wxOutputStream& stream,
 876                   const wxString& str,
 877                   wxMBConv *convMem,
 878                   wxMBConv *convFile)
 879 {
 880     if (str.empty())
 881         return true;
 882
 883 #if wxUSE_UNICODE
 884     wxUnusedVar(convMem);
 885     if ( !convFile )
 886         convFile = &wxConvUTF8;
 887
 888     const wxScopedCharBuffer buf(str.mb_str(*convFile));
 889     if ( !buf.length() )
 890     {
 891         // conversion failed, can't write this string in an XML file in this
 892         // (presumably non-UTF-8) encoding
 893         return false;
 894     }
 895
 896     stream.Write(buf, buf.length());
 897 #else // !wxUSE_UNICODE
 898     if ( convFile && convMem )
 899     {
 900         wxString str2(str.wc_str(*convMem), *convFile);
 901         stream.Write(str2.mb_str(), str2.length());
 902     }
 903     else // no conversions to do
 904     {
 905         stream.Write(str.mb_str(), str.length());
 906     }
 907 #endif // wxUSE_UNICODE/!wxUSE_UNICODE
 908
 909     return stream.IsOk();
 910 }
 911
 912 enum EscapingMode
 913 {
 914     Escape_Text,
 915     Escape_Attribute
 916 };
 917
 918 // Same as above, but create entities first.
 919 // Translates '<' to "&lt;", '>' to "&gt;" and so on, according to the spec:
 920 // http://www.w3.org/TR/2000/WD-xml-c14n-20000119.html#charescaping
 921 bool OutputEscapedString(wxOutputStream& stream,
 922                          const wxString& str,
 923                          wxMBConv *convMem,
 924                          wxMBConv *convFile,
 925                          EscapingMode mode)
 926 {
 927     wxString escaped;
 928     escaped.reserve(str.length());
 929
 930     for ( wxString::const_iterator i = str.begin(); i != str.end(); ++i )
 931     {
 932         const wxChar c = *i;
 933
 934         switch ( c )
 935         {
 936             case wxS('<'):
 937                 escaped.append(wxS("&lt;"));
 938                 break;
 939             case wxS('>'):
 940                 escaped.append(wxS("&gt;"));
 941                 break;
 942             case wxS('&'):
 943                 escaped.append(wxS("&amp;"));
 944                 break;
 945             case wxS('\r'):
 946                 escaped.append(wxS("&#xD;"));
 947                 break;
 948             default:
 949                 if ( mode == Escape_Attribute )
 950                 {
 951                     switch ( c )
 952                     {
 953                         case wxS('"'):
 954                             escaped.append(wxS("&quot;"));
 955                             break;
 956                         case wxS('\t'):
 957                             escaped.append(wxS("&#x9;"));
 958                             break;
 959                         case wxS('\n'):
 960                             escaped.append(wxS("&#xA;"));
 961                             break;
 962                         default:
 963                             escaped.append(c);
 964                     }
 965
 966                 }
 967                 else
 968                 {
 969                     escaped.append(c);
 970                 }
 971         }
 972     }
 973
 974     return OutputString(stream, escaped, convMem, convFile);
 975 }
 976
 977 bool OutputIndentation(wxOutputStream& stream,
 978                        int indent,
 979                        wxMBConv *convMem,
 980                        wxMBConv *convFile)
 981 {
 982     wxString str(wxS("\n"));
 983     str += wxString(indent, wxS(' '));
 984     return OutputString(stream, str, convMem, convFile);
 985 }
 986
 987 bool OutputNode(wxOutputStream& stream,
 988                 wxXmlNode *node,
 989                 int indent,
 990                 wxMBConv *convMem,
 991                 wxMBConv *convFile,
 992                 int indentstep)
 993 {
 994     bool rc;
 995     switch (node->GetType())
 996     {
 997         case wxXML_CDATA_SECTION_NODE:
 998             rc = OutputString(stream, wxS("<![CDATA["), convMem, convFile) &&
 999                  OutputString(stream, node->GetContent(), convMem, convFile) &&
1000                  OutputString(stream, wxS("]]>"), convMem, convFile);
1001             break;
1002
1003         case wxXML_TEXT_NODE:
1004             if (node->GetNoConversion())
1005             {
1006                 stream.Write(node->GetContent().c_str(), node->GetContent().Length());
1007                 rc = true;
1008             }
1009             else
1010                 rc = OutputEscapedString(stream, node->GetContent(),
1011                                      convMem, convFile,
1012                                      Escape_Text);
1013             break;
1014
1015         case wxXML_ELEMENT_NODE:
1016             rc = OutputString(stream, wxS("<"), convMem, convFile) &&
1017                  OutputString(stream, node->GetName(), convMem, convFile);
1018
1019             if ( rc )
1020             {
1021                 for ( wxXmlAttribute *attr = node->GetAttributes();
1022                       attr && rc;
1023                       attr = attr->GetNext() )
1024                 {
1025                     rc = OutputString(stream,
1026                                       wxS(" ") + attr->GetName() +  wxS("=\""),
1027                                       convMem, convFile) &&
1028                          OutputEscapedString(stream, attr->GetValue(),
1029                                              convMem, convFile,
1030                                              Escape_Attribute) &&
1031                          OutputString(stream, wxS("\""), convMem, convFile);
1032                 }
1033             }
1034
1035             if ( node->GetChildren() )
1036             {
1037                 rc = OutputString(stream, wxS(">"), convMem, convFile);
1038
1039                 wxXmlNode *prev = NULL;
1040                 for ( wxXmlNode *n = node->GetChildren();
1041                       n && rc;
1042                       n = n->GetNext() )
1043                 {
1044                     if ( indentstep >= 0 && n->GetType() != wxXML_TEXT_NODE )
1045                     {
1046                         rc = OutputIndentation(stream, indent + indentstep,
1047                                                convMem, convFile);
1048                     }
1049
1050                     if ( rc )
1051                         rc = OutputNode(stream, n, indent + indentstep,
1052                                         convMem, convFile, indentstep);
1053
1054                     prev = n;
1055                 }
1056
1057                 if ( rc && indentstep >= 0 &&
1058                         prev && prev->GetType() != wxXML_TEXT_NODE )
1059                 {
1060                     rc = OutputIndentation(stream, indent, convMem, convFile);
1061                 }
1062
1063                 if ( rc )
1064                 {
1065                     rc = OutputString(stream, wxS("</"), convMem, convFile) &&
1066                          OutputString(stream, node->GetName(),
1067                                       convMem, convFile) &&
1068                          OutputString(stream, wxS(">"), convMem, convFile);
1069                 }
1070             }
1071             else // no children, output "<foo/>"
1072             {
1073                 rc = OutputString(stream, wxS("/>"), convMem, convFile);
1074             }
1075             break;
1076
1077         case wxXML_COMMENT_NODE:
1078             rc = OutputString(stream, wxS("<!--"), convMem, convFile) &&
1079                  OutputString(stream, node->GetContent(), convMem, convFile) &&
1080                  OutputString(stream, wxS("-->"), convMem, convFile);
1081             break;
1082
1083         case wxXML_PI_NODE:
1084             rc = OutputString(stream, wxT("<?"), convMem, convFile) &&
1085                  OutputString(stream, node->GetName(), convMem, convFile) &&
1086                  OutputString(stream, wxT(" "), convMem, convFile) &&
1087                  OutputString(stream, node->GetContent(), convMem, convFile) &&
1088                  OutputString(stream, wxT("?>"), convMem, convFile);
1089             break;
1090
1091         default:
1092             wxFAIL_MSG("unsupported node type");
1093             rc = false;
1094     }
1095
1096     return rc;
1097 }
1098
1099 } // anonymous namespace
1100
1101 bool wxXmlDocument::Save(wxOutputStream& stream, int indentstep) const
1102 {
1103     if ( !IsOk() )
1104         return false;
1105
1106     wxScopedPtr<wxMBConv> convMem, convFile;
1107
1108 #if wxUSE_UNICODE
1109     convFile.reset(new wxCSConv(GetFileEncoding()));
1110 #else
1111     if ( GetFileEncoding().CmpNoCase(GetEncoding()) != 0 )
1112     {
1113         convFile.reset(new wxCSConv(GetFileEncoding()));
1114         convMem.reset(new wxCSConv(GetEncoding()));
1115     }
1116     //else: file and in-memory encodings are the same, no conversion needed
1117 #endif
1118
1119     wxString dec = wxString::Format(
1120                                     wxS("<?xml version=\"%s\" encoding=\"%s\"?>\n"),
1121                                     GetVersion(), GetFileEncoding()
1122                                    );
1123     bool rc = OutputString(stream, dec, convMem.get(), convFile.get());
1124
1125     wxXmlNode *node = GetDocumentNode();
1126     if ( node )
1127         node = node->GetChildren();
1128
1129     while( rc && node )
1130     {
1131         rc = OutputNode(stream, node, 0, convMem.get(),
1132                         convFile.get(), indentstep) &&
1133              OutputString(stream, wxS("\n"), convMem.get(), convFile.get());
1134         node = node->GetNext();
1135     }
1136     return rc;
1137 }
1138
1139 /*static*/ wxVersionInfo wxXmlDocument::GetLibraryVersionInfo()
1140 {
1141     return wxVersionInfo("expat",
1142                          XML_MAJOR_VERSION,
1143                          XML_MINOR_VERSION,
1144                          XML_MICRO_VERSION);
1145 }
1146
1147 #endif // wxUSE_XML