| 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 | |
| 34 | #include "expat.h" // from Expat |
| 35 | |
| 36 | // DLL options compatibility check: |
| 37 | WX_CHECK_BUILD_OPTIONS("wxXML") |
| 38 | |
| 39 | |
| 40 | IMPLEMENT_CLASS(wxXmlDocument, wxObject) |
| 41 | |
| 42 | |
| 43 | // a private utility used by wxXML |
| 44 | static bool wxIsWhiteOnly(const wxString& buf); |
| 45 | |
| 46 | |
| 47 | //----------------------------------------------------------------------------- |
| 48 | // wxXmlNode |
| 49 | //----------------------------------------------------------------------------- |
| 50 | |
| 51 | wxXmlNode::wxXmlNode(wxXmlNode *parent,wxXmlNodeType type, |
| 52 | const wxString& name, const wxString& content, |
| 53 | wxXmlAttribute *attrs, wxXmlNode *next, int lineNo) |
| 54 | : m_type(type), m_name(name), m_content(content), |
| 55 | m_attrs(attrs), m_parent(parent), |
| 56 | m_children(NULL), m_next(next), |
| 57 | m_lineNo(lineNo) |
| 58 | { |
| 59 | if (m_parent) |
| 60 | { |
| 61 | if (m_parent->m_children) |
| 62 | { |
| 63 | m_next = m_parent->m_children; |
| 64 | m_parent->m_children = this; |
| 65 | } |
| 66 | else |
| 67 | m_parent->m_children = this; |
| 68 | } |
| 69 | } |
| 70 | |
| 71 | wxXmlNode::wxXmlNode(wxXmlNodeType type, const wxString& name, |
| 72 | const wxString& content, |
| 73 | int lineNo) |
| 74 | : m_type(type), m_name(name), m_content(content), |
| 75 | m_attrs(NULL), m_parent(NULL), |
| 76 | m_children(NULL), m_next(NULL), |
| 77 | m_lineNo(lineNo) |
| 78 | {} |
| 79 | |
| 80 | wxXmlNode::wxXmlNode(const wxXmlNode& node) |
| 81 | { |
| 82 | m_next = NULL; |
| 83 | m_parent = NULL; |
| 84 | DoCopy(node); |
| 85 | } |
| 86 | |
| 87 | wxXmlNode::~wxXmlNode() |
| 88 | { |
| 89 | wxXmlNode *c, *c2; |
| 90 | for (c = m_children; c; c = c2) |
| 91 | { |
| 92 | c2 = c->m_next; |
| 93 | delete c; |
| 94 | } |
| 95 | |
| 96 | wxXmlAttribute *p, *p2; |
| 97 | for (p = m_attrs; p; p = p2) |
| 98 | { |
| 99 | p2 = p->GetNext(); |
| 100 | delete p; |
| 101 | } |
| 102 | } |
| 103 | |
| 104 | wxXmlNode& wxXmlNode::operator=(const wxXmlNode& node) |
| 105 | { |
| 106 | wxDELETE(m_attrs); |
| 107 | wxDELETE(m_children); |
| 108 | DoCopy(node); |
| 109 | return *this; |
| 110 | } |
| 111 | |
| 112 | void wxXmlNode::DoCopy(const wxXmlNode& node) |
| 113 | { |
| 114 | m_type = node.m_type; |
| 115 | m_name = node.m_name; |
| 116 | m_content = node.m_content; |
| 117 | m_lineNo = node.m_lineNo; |
| 118 | m_children = NULL; |
| 119 | |
| 120 | wxXmlNode *n = node.m_children; |
| 121 | while (n) |
| 122 | { |
| 123 | AddChild(new wxXmlNode(*n)); |
| 124 | n = n->GetNext(); |
| 125 | } |
| 126 | |
| 127 | m_attrs = NULL; |
| 128 | wxXmlAttribute *p = node.m_attrs; |
| 129 | while (p) |
| 130 | { |
| 131 | AddAttribute(p->GetName(), p->GetValue()); |
| 132 | p = p->GetNext(); |
| 133 | } |
| 134 | } |
| 135 | |
| 136 | bool wxXmlNode::HasAttribute(const wxString& attrName) const |
| 137 | { |
| 138 | wxXmlAttribute *attr = GetAttributes(); |
| 139 | |
| 140 | while (attr) |
| 141 | { |
| 142 | if (attr->GetName() == attrName) return true; |
| 143 | attr = attr->GetNext(); |
| 144 | } |
| 145 | |
| 146 | return false; |
| 147 | } |
| 148 | |
| 149 | bool wxXmlNode::GetAttribute(const wxString& attrName, wxString *value) const |
| 150 | { |
| 151 | wxCHECK_MSG( value, false, "value argument must not be NULL" ); |
| 152 | |
| 153 | wxXmlAttribute *attr = GetAttributes(); |
| 154 | |
| 155 | while (attr) |
| 156 | { |
| 157 | if (attr->GetName() == attrName) |
| 158 | { |
| 159 | *value = attr->GetValue(); |
| 160 | return true; |
| 161 | } |
| 162 | attr = attr->GetNext(); |
| 163 | } |
| 164 | |
| 165 | return false; |
| 166 | } |
| 167 | |
| 168 | wxString wxXmlNode::GetAttribute(const wxString& attrName, const wxString& defaultVal) const |
| 169 | { |
| 170 | wxString tmp; |
| 171 | if (GetAttribute(attrName, &tmp)) |
| 172 | return tmp; |
| 173 | |
| 174 | return defaultVal; |
| 175 | } |
| 176 | |
| 177 | void wxXmlNode::AddChild(wxXmlNode *child) |
| 178 | { |
| 179 | if (m_children == NULL) |
| 180 | m_children = child; |
| 181 | else |
| 182 | { |
| 183 | wxXmlNode *ch = m_children; |
| 184 | while (ch->m_next) ch = ch->m_next; |
| 185 | ch->m_next = child; |
| 186 | } |
| 187 | child->m_next = NULL; |
| 188 | child->m_parent = this; |
| 189 | } |
| 190 | |
| 191 | // inserts a new node in front of 'followingNode' |
| 192 | bool wxXmlNode::InsertChild(wxXmlNode *child, wxXmlNode *followingNode) |
| 193 | { |
| 194 | wxCHECK_MSG( child, false, "cannot insert a NULL node!" ); |
| 195 | wxCHECK_MSG( child->m_parent == NULL, false, "node already has a parent" ); |
| 196 | wxCHECK_MSG( child->m_next == NULL, false, "node already has m_next" ); |
| 197 | wxCHECK_MSG( followingNode == NULL || followingNode->GetParent() == this, |
| 198 | false, |
| 199 | "wxXmlNode::InsertChild - followingNode has incorrect parent" ); |
| 200 | |
| 201 | // this is for backward compatibility, NULL was allowed here thanks to |
| 202 | // the confusion about followingNode's meaning |
| 203 | if ( followingNode == NULL ) |
| 204 | followingNode = m_children; |
| 205 | |
| 206 | if ( m_children == followingNode ) |
| 207 | { |
| 208 | child->m_next = m_children; |
| 209 | m_children = child; |
| 210 | } |
| 211 | else |
| 212 | { |
| 213 | wxXmlNode *ch = m_children; |
| 214 | while ( ch && ch->m_next != followingNode ) |
| 215 | ch = ch->m_next; |
| 216 | if ( !ch ) |
| 217 | { |
| 218 | wxFAIL_MSG( "followingNode has this node as parent, but couldn't be found among children" ); |
| 219 | return false; |
| 220 | } |
| 221 | |
| 222 | child->m_next = followingNode; |
| 223 | ch->m_next = child; |
| 224 | } |
| 225 | |
| 226 | child->m_parent = this; |
| 227 | return true; |
| 228 | } |
| 229 | |
| 230 | // inserts a new node right after 'precedingNode' |
| 231 | bool wxXmlNode::InsertChildAfter(wxXmlNode *child, wxXmlNode *precedingNode) |
| 232 | { |
| 233 | wxCHECK_MSG( child, false, "cannot insert a NULL node!" ); |
| 234 | wxCHECK_MSG( child->m_parent == NULL, false, "node already has a parent" ); |
| 235 | wxCHECK_MSG( child->m_next == NULL, false, "node already has m_next" ); |
| 236 | wxCHECK_MSG( precedingNode == NULL || precedingNode->m_parent == this, false, |
| 237 | "precedingNode has wrong parent" ); |
| 238 | |
| 239 | if ( precedingNode ) |
| 240 | { |
| 241 | child->m_next = precedingNode->m_next; |
| 242 | precedingNode->m_next = child; |
| 243 | } |
| 244 | else // precedingNode == NULL |
| 245 | { |
| 246 | wxCHECK_MSG( m_children == NULL, false, |
| 247 | "NULL precedingNode only makes sense when there are no children" ); |
| 248 | |
| 249 | child->m_next = m_children; |
| 250 | m_children = child; |
| 251 | } |
| 252 | |
| 253 | child->m_parent = this; |
| 254 | return true; |
| 255 | } |
| 256 | |
| 257 | bool wxXmlNode::RemoveChild(wxXmlNode *child) |
| 258 | { |
| 259 | if (m_children == NULL) |
| 260 | return false; |
| 261 | else if (m_children == child) |
| 262 | { |
| 263 | m_children = child->m_next; |
| 264 | child->m_parent = NULL; |
| 265 | child->m_next = NULL; |
| 266 | return true; |
| 267 | } |
| 268 | else |
| 269 | { |
| 270 | wxXmlNode *ch = m_children; |
| 271 | while (ch->m_next) |
| 272 | { |
| 273 | if (ch->m_next == child) |
| 274 | { |
| 275 | ch->m_next = child->m_next; |
| 276 | child->m_parent = NULL; |
| 277 | child->m_next = NULL; |
| 278 | return true; |
| 279 | } |
| 280 | ch = ch->m_next; |
| 281 | } |
| 282 | return false; |
| 283 | } |
| 284 | } |
| 285 | |
| 286 | void wxXmlNode::AddAttribute(const wxString& name, const wxString& value) |
| 287 | { |
| 288 | AddProperty(name, value); |
| 289 | } |
| 290 | |
| 291 | void wxXmlNode::AddAttribute(wxXmlAttribute *attr) |
| 292 | { |
| 293 | AddProperty(attr); |
| 294 | } |
| 295 | |
| 296 | bool wxXmlNode::DeleteAttribute(const wxString& name) |
| 297 | { |
| 298 | return DeleteProperty(name); |
| 299 | } |
| 300 | |
| 301 | void wxXmlNode::AddProperty(const wxString& name, const wxString& value) |
| 302 | { |
| 303 | AddProperty(new wxXmlAttribute(name, value, NULL)); |
| 304 | } |
| 305 | |
| 306 | void wxXmlNode::AddProperty(wxXmlAttribute *attr) |
| 307 | { |
| 308 | if (m_attrs == NULL) |
| 309 | m_attrs = attr; |
| 310 | else |
| 311 | { |
| 312 | wxXmlAttribute *p = m_attrs; |
| 313 | while (p->GetNext()) p = p->GetNext(); |
| 314 | p->SetNext(attr); |
| 315 | } |
| 316 | } |
| 317 | |
| 318 | bool wxXmlNode::DeleteProperty(const wxString& name) |
| 319 | { |
| 320 | wxXmlAttribute *attr; |
| 321 | |
| 322 | if (m_attrs == NULL) |
| 323 | return false; |
| 324 | |
| 325 | else if (m_attrs->GetName() == name) |
| 326 | { |
| 327 | attr = m_attrs; |
| 328 | m_attrs = attr->GetNext(); |
| 329 | attr->SetNext(NULL); |
| 330 | delete attr; |
| 331 | return true; |
| 332 | } |
| 333 | |
| 334 | else |
| 335 | { |
| 336 | wxXmlAttribute *p = m_attrs; |
| 337 | while (p->GetNext()) |
| 338 | { |
| 339 | if (p->GetNext()->GetName() == name) |
| 340 | { |
| 341 | attr = p->GetNext(); |
| 342 | p->SetNext(attr->GetNext()); |
| 343 | attr->SetNext(NULL); |
| 344 | delete attr; |
| 345 | return true; |
| 346 | } |
| 347 | p = p->GetNext(); |
| 348 | } |
| 349 | return false; |
| 350 | } |
| 351 | } |
| 352 | |
| 353 | wxString wxXmlNode::GetNodeContent() const |
| 354 | { |
| 355 | wxXmlNode *n = GetChildren(); |
| 356 | |
| 357 | while (n) |
| 358 | { |
| 359 | if (n->GetType() == wxXML_TEXT_NODE || |
| 360 | n->GetType() == wxXML_CDATA_SECTION_NODE) |
| 361 | return n->GetContent(); |
| 362 | n = n->GetNext(); |
| 363 | } |
| 364 | return wxEmptyString; |
| 365 | } |
| 366 | |
| 367 | int wxXmlNode::GetDepth(wxXmlNode *grandparent) const |
| 368 | { |
| 369 | const wxXmlNode *n = this; |
| 370 | int ret = -1; |
| 371 | |
| 372 | do |
| 373 | { |
| 374 | ret++; |
| 375 | n = n->GetParent(); |
| 376 | if (n == grandparent) |
| 377 | return ret; |
| 378 | |
| 379 | } while (n); |
| 380 | |
| 381 | return wxNOT_FOUND; |
| 382 | } |
| 383 | |
| 384 | bool wxXmlNode::IsWhitespaceOnly() const |
| 385 | { |
| 386 | return wxIsWhiteOnly(m_content); |
| 387 | } |
| 388 | |
| 389 | |
| 390 | |
| 391 | //----------------------------------------------------------------------------- |
| 392 | // wxXmlDocument |
| 393 | //----------------------------------------------------------------------------- |
| 394 | |
| 395 | wxXmlDocument::wxXmlDocument() |
| 396 | : m_version(wxS("1.0")), m_fileEncoding(wxS("utf-8")), m_root(NULL) |
| 397 | { |
| 398 | #if !wxUSE_UNICODE |
| 399 | m_encoding = wxS("UTF-8"); |
| 400 | #endif |
| 401 | } |
| 402 | |
| 403 | wxXmlDocument::wxXmlDocument(const wxString& filename, const wxString& encoding) |
| 404 | :wxObject(), m_root(NULL) |
| 405 | { |
| 406 | if ( !Load(filename, encoding) ) |
| 407 | { |
| 408 | wxDELETE(m_root); |
| 409 | } |
| 410 | } |
| 411 | |
| 412 | wxXmlDocument::wxXmlDocument(wxInputStream& stream, const wxString& encoding) |
| 413 | :wxObject(), m_root(NULL) |
| 414 | { |
| 415 | if ( !Load(stream, encoding) ) |
| 416 | { |
| 417 | wxDELETE(m_root); |
| 418 | } |
| 419 | } |
| 420 | |
| 421 | wxXmlDocument::wxXmlDocument(const wxXmlDocument& doc) |
| 422 | :wxObject() |
| 423 | { |
| 424 | DoCopy(doc); |
| 425 | } |
| 426 | |
| 427 | wxXmlDocument& wxXmlDocument::operator=(const wxXmlDocument& doc) |
| 428 | { |
| 429 | wxDELETE(m_root); |
| 430 | DoCopy(doc); |
| 431 | return *this; |
| 432 | } |
| 433 | |
| 434 | void wxXmlDocument::DoCopy(const wxXmlDocument& doc) |
| 435 | { |
| 436 | m_version = doc.m_version; |
| 437 | #if !wxUSE_UNICODE |
| 438 | m_encoding = doc.m_encoding; |
| 439 | #endif |
| 440 | m_fileEncoding = doc.m_fileEncoding; |
| 441 | |
| 442 | if (doc.m_root) |
| 443 | m_root = new wxXmlNode(*doc.m_root); |
| 444 | else |
| 445 | m_root = NULL; |
| 446 | } |
| 447 | |
| 448 | bool wxXmlDocument::Load(const wxString& filename, const wxString& encoding, int flags) |
| 449 | { |
| 450 | wxFileInputStream stream(filename); |
| 451 | if (!stream.Ok()) |
| 452 | return false; |
| 453 | return Load(stream, encoding, flags); |
| 454 | } |
| 455 | |
| 456 | bool wxXmlDocument::Save(const wxString& filename, int indentstep) const |
| 457 | { |
| 458 | wxFileOutputStream stream(filename); |
| 459 | if (!stream.Ok()) |
| 460 | return false; |
| 461 | return Save(stream, indentstep); |
| 462 | } |
| 463 | |
| 464 | |
| 465 | |
| 466 | //----------------------------------------------------------------------------- |
| 467 | // wxXmlDocument loading routines |
| 468 | //----------------------------------------------------------------------------- |
| 469 | |
| 470 | // converts Expat-produced string in UTF-8 into wxString using the specified |
| 471 | // conv or keep in UTF-8 if conv is NULL |
| 472 | static wxString CharToString(wxMBConv *conv, |
| 473 | const char *s, size_t len = wxString::npos) |
| 474 | { |
| 475 | #if !wxUSE_UNICODE |
| 476 | if ( conv ) |
| 477 | { |
| 478 | // there can be no embedded NULs in this string so we don't need the |
| 479 | // output length, it will be NUL-terminated |
| 480 | const wxWCharBuffer wbuf( |
| 481 | wxConvUTF8.cMB2WC(s, len == wxString::npos ? wxNO_LEN : len, NULL)); |
| 482 | |
| 483 | return wxString(wbuf, *conv); |
| 484 | } |
| 485 | // else: the string is wanted in UTF-8 |
| 486 | #endif // !wxUSE_UNICODE |
| 487 | |
| 488 | wxUnusedVar(conv); |
| 489 | return wxString::FromUTF8Unchecked(s, len); |
| 490 | } |
| 491 | |
| 492 | // returns true if the given string contains only whitespaces |
| 493 | bool wxIsWhiteOnly(const wxString& buf) |
| 494 | { |
| 495 | for ( wxString::const_iterator i = buf.begin(); i != buf.end(); ++i ) |
| 496 | { |
| 497 | wxChar c = *i; |
| 498 | if ( c != wxS(' ') && c != wxS('\t') && c != wxS('\n') && c != wxS('\r')) |
| 499 | return false; |
| 500 | } |
| 501 | return true; |
| 502 | } |
| 503 | |
| 504 | |
| 505 | struct wxXmlParsingContext |
| 506 | { |
| 507 | wxXmlParsingContext() |
| 508 | : conv(NULL), |
| 509 | root(NULL), |
| 510 | node(NULL), |
| 511 | lastChild(NULL), |
| 512 | lastAsText(NULL), |
| 513 | removeWhiteOnlyNodes(false) |
| 514 | {} |
| 515 | |
| 516 | XML_Parser parser; |
| 517 | wxMBConv *conv; |
| 518 | wxXmlNode *root; |
| 519 | wxXmlNode *node; // the node being parsed |
| 520 | wxXmlNode *lastChild; // the last child of "node" |
| 521 | wxXmlNode *lastAsText; // the last _text_ child of "node" |
| 522 | wxString encoding; |
| 523 | wxString version; |
| 524 | bool removeWhiteOnlyNodes; |
| 525 | }; |
| 526 | |
| 527 | // checks that ctx->lastChild is in consistent state |
| 528 | #define ASSERT_LAST_CHILD_OK(ctx) \ |
| 529 | wxASSERT( ctx->lastChild == NULL || \ |
| 530 | ctx->lastChild->GetNext() == NULL ); \ |
| 531 | wxASSERT( ctx->lastChild == NULL || \ |
| 532 | ctx->lastChild->GetParent() == ctx->node ) |
| 533 | |
| 534 | extern "C" { |
| 535 | static void StartElementHnd(void *userData, const char *name, const char **atts) |
| 536 | { |
| 537 | wxXmlParsingContext *ctx = (wxXmlParsingContext*)userData; |
| 538 | wxXmlNode *node = new wxXmlNode(wxXML_ELEMENT_NODE, |
| 539 | CharToString(ctx->conv, name), |
| 540 | wxEmptyString, |
| 541 | XML_GetCurrentLineNumber(ctx->parser)); |
| 542 | const char **a = atts; |
| 543 | |
| 544 | // add node attributes |
| 545 | while (*a) |
| 546 | { |
| 547 | node->AddAttribute(CharToString(ctx->conv, a[0]), CharToString(ctx->conv, a[1])); |
| 548 | a += 2; |
| 549 | } |
| 550 | |
| 551 | if (ctx->root == NULL) |
| 552 | { |
| 553 | ctx->root = node; |
| 554 | } |
| 555 | else |
| 556 | { |
| 557 | ASSERT_LAST_CHILD_OK(ctx); |
| 558 | ctx->node->InsertChildAfter(node, ctx->lastChild); |
| 559 | } |
| 560 | |
| 561 | ctx->lastAsText = NULL; |
| 562 | ctx->lastChild = NULL; // our new node "node" has no children yet |
| 563 | |
| 564 | ctx->node = node; |
| 565 | } |
| 566 | |
| 567 | static void EndElementHnd(void *userData, const char* WXUNUSED(name)) |
| 568 | { |
| 569 | wxXmlParsingContext *ctx = (wxXmlParsingContext*)userData; |
| 570 | |
| 571 | // we're exiting the last children of ctx->node->GetParent() and going |
| 572 | // back one level up, so current value of ctx->node points to the last |
| 573 | // child of ctx->node->GetParent() |
| 574 | ctx->lastChild = ctx->node; |
| 575 | |
| 576 | ctx->node = ctx->node->GetParent(); |
| 577 | ctx->lastAsText = NULL; |
| 578 | } |
| 579 | |
| 580 | static void TextHnd(void *userData, const char *s, int len) |
| 581 | { |
| 582 | wxXmlParsingContext *ctx = (wxXmlParsingContext*)userData; |
| 583 | wxString str = CharToString(ctx->conv, s, len); |
| 584 | |
| 585 | if (ctx->lastAsText) |
| 586 | { |
| 587 | ctx->lastAsText->SetContent(ctx->lastAsText->GetContent() + str); |
| 588 | } |
| 589 | else |
| 590 | { |
| 591 | bool whiteOnly = false; |
| 592 | if (ctx->removeWhiteOnlyNodes) |
| 593 | whiteOnly = wxIsWhiteOnly(str); |
| 594 | |
| 595 | if (!whiteOnly) |
| 596 | { |
| 597 | wxXmlNode *textnode = |
| 598 | new wxXmlNode(wxXML_TEXT_NODE, wxS("text"), str, |
| 599 | XML_GetCurrentLineNumber(ctx->parser)); |
| 600 | |
| 601 | ASSERT_LAST_CHILD_OK(ctx); |
| 602 | ctx->node->InsertChildAfter(textnode, ctx->lastChild); |
| 603 | ctx->lastChild= ctx->lastAsText = textnode; |
| 604 | } |
| 605 | } |
| 606 | } |
| 607 | |
| 608 | static void StartCdataHnd(void *userData) |
| 609 | { |
| 610 | wxXmlParsingContext *ctx = (wxXmlParsingContext*)userData; |
| 611 | |
| 612 | wxXmlNode *textnode = |
| 613 | new wxXmlNode(wxXML_CDATA_SECTION_NODE, wxS("cdata"), wxS(""), |
| 614 | XML_GetCurrentLineNumber(ctx->parser)); |
| 615 | |
| 616 | ASSERT_LAST_CHILD_OK(ctx); |
| 617 | ctx->node->InsertChildAfter(textnode, ctx->lastChild); |
| 618 | ctx->lastChild= ctx->lastAsText = textnode; |
| 619 | } |
| 620 | |
| 621 | static void EndCdataHnd(void *userData) |
| 622 | { |
| 623 | wxXmlParsingContext *ctx = (wxXmlParsingContext*)userData; |
| 624 | |
| 625 | // we need to reset this pointer so that subsequent text nodes don't append |
| 626 | // their contents to this one but create new wxXML_TEXT_NODE objects (or |
| 627 | // not create anything at all if only white space follows the CDATA section |
| 628 | // and wxXMLDOC_KEEP_WHITESPACE_NODES is not used as is commonly the case) |
| 629 | ctx->lastAsText = NULL; |
| 630 | } |
| 631 | |
| 632 | static void CommentHnd(void *userData, const char *data) |
| 633 | { |
| 634 | wxXmlParsingContext *ctx = (wxXmlParsingContext*)userData; |
| 635 | |
| 636 | if (ctx->node) |
| 637 | { |
| 638 | wxXmlNode *commentnode = |
| 639 | new wxXmlNode(wxXML_COMMENT_NODE, |
| 640 | wxS("comment"), CharToString(ctx->conv, data), |
| 641 | XML_GetCurrentLineNumber(ctx->parser)); |
| 642 | |
| 643 | ASSERT_LAST_CHILD_OK(ctx); |
| 644 | ctx->node->InsertChildAfter(commentnode, ctx->lastChild); |
| 645 | ctx->lastChild = commentnode; |
| 646 | } |
| 647 | //else: ctx->node == NULL happens if there is a comment before |
| 648 | // the root element. We current don't have a way to represent |
| 649 | // these in wxXmlDocument (FIXME). |
| 650 | |
| 651 | ctx->lastAsText = NULL; |
| 652 | } |
| 653 | |
| 654 | static void DefaultHnd(void *userData, const char *s, int len) |
| 655 | { |
| 656 | // XML header: |
| 657 | if (len > 6 && memcmp(s, "<?xml ", 6) == 0) |
| 658 | { |
| 659 | wxXmlParsingContext *ctx = (wxXmlParsingContext*)userData; |
| 660 | |
| 661 | wxString buf = CharToString(ctx->conv, s, (size_t)len); |
| 662 | int pos; |
| 663 | pos = buf.Find(wxS("encoding=")); |
| 664 | if (pos != wxNOT_FOUND) |
| 665 | ctx->encoding = buf.Mid(pos + 10).BeforeFirst(buf[(size_t)pos+9]); |
| 666 | pos = buf.Find(wxS("version=")); |
| 667 | if (pos != wxNOT_FOUND) |
| 668 | ctx->version = buf.Mid(pos + 9).BeforeFirst(buf[(size_t)pos+8]); |
| 669 | } |
| 670 | } |
| 671 | |
| 672 | static int UnknownEncodingHnd(void * WXUNUSED(encodingHandlerData), |
| 673 | const XML_Char *name, XML_Encoding *info) |
| 674 | { |
| 675 | // We must build conversion table for expat. The easiest way to do so |
| 676 | // is to let wxCSConv convert as string containing all characters to |
| 677 | // wide character representation: |
| 678 | wxCSConv conv(name); |
| 679 | char mbBuf[2]; |
| 680 | wchar_t wcBuf[10]; |
| 681 | size_t i; |
| 682 | |
| 683 | mbBuf[1] = 0; |
| 684 | info->map[0] = 0; |
| 685 | for (i = 0; i < 255; i++) |
| 686 | { |
| 687 | mbBuf[0] = (char)(i+1); |
| 688 | if (conv.MB2WC(wcBuf, mbBuf, 2) == (size_t)-1) |
| 689 | { |
| 690 | // invalid/undefined byte in the encoding: |
| 691 | info->map[i+1] = -1; |
| 692 | } |
| 693 | info->map[i+1] = (int)wcBuf[0]; |
| 694 | } |
| 695 | |
| 696 | info->data = NULL; |
| 697 | info->convert = NULL; |
| 698 | info->release = NULL; |
| 699 | |
| 700 | return 1; |
| 701 | } |
| 702 | |
| 703 | } // extern "C" |
| 704 | |
| 705 | bool wxXmlDocument::Load(wxInputStream& stream, const wxString& encoding, int flags) |
| 706 | { |
| 707 | #if wxUSE_UNICODE |
| 708 | (void)encoding; |
| 709 | #else |
| 710 | m_encoding = encoding; |
| 711 | #endif |
| 712 | |
| 713 | const size_t BUFSIZE = 1024; |
| 714 | char buf[BUFSIZE]; |
| 715 | wxXmlParsingContext ctx; |
| 716 | bool done; |
| 717 | XML_Parser parser = XML_ParserCreate(NULL); |
| 718 | |
| 719 | ctx.encoding = wxS("UTF-8"); // default in absence of encoding="" |
| 720 | ctx.conv = NULL; |
| 721 | #if !wxUSE_UNICODE |
| 722 | if ( encoding.CmpNoCase(wxS("UTF-8")) != 0 ) |
| 723 | ctx.conv = new wxCSConv(encoding); |
| 724 | #endif |
| 725 | ctx.removeWhiteOnlyNodes = (flags & wxXMLDOC_KEEP_WHITESPACE_NODES) == 0; |
| 726 | ctx.parser = parser; |
| 727 | |
| 728 | XML_SetUserData(parser, (void*)&ctx); |
| 729 | XML_SetElementHandler(parser, StartElementHnd, EndElementHnd); |
| 730 | XML_SetCharacterDataHandler(parser, TextHnd); |
| 731 | XML_SetCdataSectionHandler(parser, StartCdataHnd, EndCdataHnd);; |
| 732 | XML_SetCommentHandler(parser, CommentHnd); |
| 733 | XML_SetDefaultHandler(parser, DefaultHnd); |
| 734 | XML_SetUnknownEncodingHandler(parser, UnknownEncodingHnd, NULL); |
| 735 | |
| 736 | bool ok = true; |
| 737 | do |
| 738 | { |
| 739 | size_t len = stream.Read(buf, BUFSIZE).LastRead(); |
| 740 | done = (len < BUFSIZE); |
| 741 | if (!XML_Parse(parser, buf, len, done)) |
| 742 | { |
| 743 | wxString error(XML_ErrorString(XML_GetErrorCode(parser)), |
| 744 | *wxConvCurrent); |
| 745 | wxLogError(_("XML parsing error: '%s' at line %d"), |
| 746 | error.c_str(), |
| 747 | XML_GetCurrentLineNumber(parser)); |
| 748 | ok = false; |
| 749 | break; |
| 750 | } |
| 751 | } while (!done); |
| 752 | |
| 753 | if (ok) |
| 754 | { |
| 755 | if (!ctx.version.empty()) |
| 756 | SetVersion(ctx.version); |
| 757 | if (!ctx.encoding.empty()) |
| 758 | SetFileEncoding(ctx.encoding); |
| 759 | SetRoot(ctx.root); |
| 760 | } |
| 761 | else |
| 762 | { |
| 763 | delete ctx.root; |
| 764 | } |
| 765 | |
| 766 | XML_ParserFree(parser); |
| 767 | #if !wxUSE_UNICODE |
| 768 | if ( ctx.conv ) |
| 769 | delete ctx.conv; |
| 770 | #endif |
| 771 | |
| 772 | return ok; |
| 773 | |
| 774 | } |
| 775 | |
| 776 | |
| 777 | |
| 778 | //----------------------------------------------------------------------------- |
| 779 | // wxXmlDocument saving routines |
| 780 | //----------------------------------------------------------------------------- |
| 781 | |
| 782 | // helpers for XML generation |
| 783 | namespace |
| 784 | { |
| 785 | |
| 786 | // write string to output: |
| 787 | bool OutputString(wxOutputStream& stream, |
| 788 | const wxString& str, |
| 789 | wxMBConv *convMem, |
| 790 | wxMBConv *convFile) |
| 791 | { |
| 792 | if (str.empty()) |
| 793 | return true; |
| 794 | |
| 795 | #if wxUSE_UNICODE |
| 796 | wxUnusedVar(convMem); |
| 797 | if ( !convFile ) |
| 798 | convFile = &wxConvUTF8; |
| 799 | |
| 800 | const wxScopedCharBuffer buf(str.mb_str(*convFile)); |
| 801 | if ( !buf.length() ) |
| 802 | { |
| 803 | // conversion failed, can't write this string in an XML file in this |
| 804 | // (presumably non-UTF-8) encoding |
| 805 | return false; |
| 806 | } |
| 807 | |
| 808 | stream.Write(buf, buf.length()); |
| 809 | #else // !wxUSE_UNICODE |
| 810 | if ( convFile && convMem ) |
| 811 | { |
| 812 | wxString str2(str.wc_str(*convMem), *convFile); |
| 813 | stream.Write(str2.mb_str(), str2.length()); |
| 814 | } |
| 815 | else // no conversions to do |
| 816 | { |
| 817 | stream.Write(str.mb_str(), str.length()); |
| 818 | } |
| 819 | #endif // wxUSE_UNICODE/!wxUSE_UNICODE |
| 820 | |
| 821 | return stream.IsOk(); |
| 822 | } |
| 823 | |
| 824 | // flags for OutputStringEnt() |
| 825 | enum |
| 826 | { |
| 827 | XML_ESCAPE_QUOTES = 1 |
| 828 | }; |
| 829 | |
| 830 | // Same as above, but create entities first. |
| 831 | // Translates '<' to "<", '>' to ">" and '&' to "&" |
| 832 | bool OutputStringEnt(wxOutputStream& stream, |
| 833 | const wxString& str, |
| 834 | wxMBConv *convMem, |
| 835 | wxMBConv *convFile, |
| 836 | int flags = 0) |
| 837 | { |
| 838 | const size_t len = str.length(); |
| 839 | size_t i, |
| 840 | last = 0; |
| 841 | for (i = 0; i < len; i++) |
| 842 | { |
| 843 | wxChar c = str.GetChar(i); |
| 844 | if (c == wxS('<') || c == wxS('>') || |
| 845 | (c == wxS('&') && str.substr(i+1, 4) != wxS("amp;")) || |
| 846 | ((flags & XML_ESCAPE_QUOTES) && c == wxS('"'))) |
| 847 | { |
| 848 | if ( !OutputString(stream, str.substr(last, i - last), |
| 849 | convMem, convFile) ) |
| 850 | return false; |
| 851 | |
| 852 | const char *escaped; |
| 853 | switch ( c ) |
| 854 | { |
| 855 | case wxS('<'): |
| 856 | escaped = "<"; |
| 857 | break; |
| 858 | case wxS('>'): |
| 859 | escaped = ">"; |
| 860 | break; |
| 861 | case wxS('&'): |
| 862 | escaped = "&"; |
| 863 | break; |
| 864 | case wxS('"'): |
| 865 | escaped = """; |
| 866 | break; |
| 867 | default: |
| 868 | wxFAIL_MSG( "logic error in the code" ); |
| 869 | return false; |
| 870 | } |
| 871 | |
| 872 | if ( !OutputString(stream, escaped, convMem, convFile) ) |
| 873 | return false; |
| 874 | |
| 875 | last = i + 1; |
| 876 | } |
| 877 | } |
| 878 | |
| 879 | return OutputString(stream, str.substr(last, i - last), convMem, convFile); |
| 880 | } |
| 881 | |
| 882 | bool OutputIndentation(wxOutputStream& stream, |
| 883 | int indent, |
| 884 | wxMBConv *convMem, |
| 885 | wxMBConv *convFile) |
| 886 | { |
| 887 | wxString str(wxS("\n")); |
| 888 | str += wxString(indent, wxS(' ')); |
| 889 | return OutputString(stream, str, convMem, convFile); |
| 890 | } |
| 891 | |
| 892 | bool OutputNode(wxOutputStream& stream, |
| 893 | wxXmlNode *node, |
| 894 | int indent, |
| 895 | wxMBConv *convMem, |
| 896 | wxMBConv *convFile, |
| 897 | int indentstep) |
| 898 | { |
| 899 | bool rc; |
| 900 | switch (node->GetType()) |
| 901 | { |
| 902 | case wxXML_CDATA_SECTION_NODE: |
| 903 | rc = OutputString(stream, wxS("<![CDATA["), convMem, convFile) && |
| 904 | OutputString(stream, node->GetContent(), convMem, convFile) && |
| 905 | OutputString(stream, wxS("]]>"), convMem, convFile); |
| 906 | break; |
| 907 | |
| 908 | case wxXML_TEXT_NODE: |
| 909 | rc = OutputStringEnt(stream, node->GetContent(), convMem, convFile); |
| 910 | break; |
| 911 | |
| 912 | case wxXML_ELEMENT_NODE: |
| 913 | rc = OutputString(stream, wxS("<"), convMem, convFile) && |
| 914 | OutputString(stream, node->GetName(), convMem, convFile); |
| 915 | |
| 916 | if ( rc ) |
| 917 | { |
| 918 | for ( wxXmlAttribute *attr = node->GetAttributes(); |
| 919 | attr && rc; |
| 920 | attr = attr->GetNext() ) |
| 921 | { |
| 922 | rc = OutputString(stream, |
| 923 | wxS(" ") + attr->GetName() + wxS("=\""), |
| 924 | convMem, convFile) && |
| 925 | OutputStringEnt(stream, attr->GetValue(), |
| 926 | convMem, convFile, |
| 927 | XML_ESCAPE_QUOTES) && |
| 928 | OutputString(stream, wxS("\""), convMem, convFile); |
| 929 | } |
| 930 | } |
| 931 | |
| 932 | if ( node->GetChildren() ) |
| 933 | { |
| 934 | rc = OutputString(stream, wxS(">"), convMem, convFile); |
| 935 | |
| 936 | wxXmlNode *prev = NULL; |
| 937 | for ( wxXmlNode *n = node->GetChildren(); |
| 938 | n && rc; |
| 939 | n = n->GetNext() ) |
| 940 | { |
| 941 | if ( indentstep >= 0 && n->GetType() != wxXML_TEXT_NODE ) |
| 942 | { |
| 943 | rc = OutputIndentation(stream, indent + indentstep, |
| 944 | convMem, convFile); |
| 945 | } |
| 946 | |
| 947 | if ( rc ) |
| 948 | rc = OutputNode(stream, n, indent + indentstep, |
| 949 | convMem, convFile, indentstep); |
| 950 | |
| 951 | prev = n; |
| 952 | } |
| 953 | |
| 954 | if ( rc && indentstep >= 0 && |
| 955 | prev && prev->GetType() != wxXML_TEXT_NODE ) |
| 956 | { |
| 957 | rc = OutputIndentation(stream, indent, convMem, convFile); |
| 958 | } |
| 959 | |
| 960 | if ( rc ) |
| 961 | { |
| 962 | rc = OutputString(stream, wxS("</"), convMem, convFile) && |
| 963 | OutputString(stream, node->GetName(), |
| 964 | convMem, convFile) && |
| 965 | OutputString(stream, wxS(">"), convMem, convFile); |
| 966 | } |
| 967 | } |
| 968 | else // no children, output "<foo/>" |
| 969 | { |
| 970 | rc = OutputString(stream, wxS("/>"), convMem, convFile); |
| 971 | } |
| 972 | break; |
| 973 | |
| 974 | case wxXML_COMMENT_NODE: |
| 975 | rc = OutputString(stream, wxS("<!--"), convMem, convFile) && |
| 976 | OutputString(stream, node->GetContent(), convMem, convFile) && |
| 977 | OutputString(stream, wxS("-->"), convMem, convFile); |
| 978 | break; |
| 979 | |
| 980 | default: |
| 981 | wxFAIL_MSG("unsupported node type"); |
| 982 | rc = false; |
| 983 | } |
| 984 | |
| 985 | return rc; |
| 986 | } |
| 987 | |
| 988 | } // anonymous namespace |
| 989 | |
| 990 | bool wxXmlDocument::Save(wxOutputStream& stream, int indentstep) const |
| 991 | { |
| 992 | if ( !IsOk() ) |
| 993 | return false; |
| 994 | |
| 995 | wxScopedPtr<wxMBConv> convMem, convFile; |
| 996 | |
| 997 | #if wxUSE_UNICODE |
| 998 | convFile.reset(new wxCSConv(GetFileEncoding())); |
| 999 | #else |
| 1000 | if ( GetFileEncoding().CmpNoCase(GetEncoding()) != 0 ) |
| 1001 | { |
| 1002 | convFile.reset(new wxCSConv(GetFileEncoding())); |
| 1003 | convMem.reset(new wxCSConv(GetEncoding())); |
| 1004 | } |
| 1005 | //else: file and in-memory encodings are the same, no conversion needed |
| 1006 | #endif |
| 1007 | |
| 1008 | return OutputString(stream, |
| 1009 | wxString::Format |
| 1010 | ( |
| 1011 | wxS("<?xml version=\"%s\" encoding=\"%s\"?>\n"), |
| 1012 | GetVersion(), GetFileEncoding() |
| 1013 | ), |
| 1014 | convMem.get(), |
| 1015 | convFile.get()) && |
| 1016 | OutputNode(stream, GetRoot(), 0, |
| 1017 | convMem.get(), convFile.get(), indentstep) && |
| 1018 | OutputString(stream, wxS("\n"), convMem.get(), convFile.get()); |
| 1019 | } |
| 1020 | |
| 1021 | #endif // wxUSE_XML |