WebCore/ChangeLog

   1 2006-07-29  Alexey Proskuryakov  <ap@nypop.com>
   2
   3         Reviewed by Darin.
   4
   5         Fix http://bugzilla.opendarwin.org/show_bug.cgi?id=10147
   6         REGRESSION: custom attribute values set via javascript are not persistent
   7
   8         Test: fast/dom/Element/setAttribute-case-insensitivity.html
   9
  10         * dom/Element.cpp:
  11         (WebCore::Element::setAttribute): Use the lowercased localName for retrieving the old value, too.
  12
  13 2006-07-28  Timothy Hatcher  <timothy@apple.com>
  14
  15         Suggested by Darin.
  16
  17         Moving the implementation of isContentEditable to the
  18         the WebPrivate category.
  19
  20         * bindings/objc/DOM.mm:
  21         (-[DOMNode isContentEditable]):
  22
  23 2006-07-28  Justin Garcia  <justin.garcia@apple.com>
  24
  25         Reviewed by levi
  26
  27         First part of:
  28         <rdar://problem/4515463>
  29         REGRESSION: Blot and Mail both do a very poor job of pasting the main www.apple.com page
  30
  31         Migrate to isBlock and enclosingBlock.
  32         Changed RendereringInfo::isBlock and ReplacementFragment::isBlock
  33         to wasBlock so that they don't conflict with isBlock and because
  34         it's more descriptive.
  35
  36         * editing/ReplaceSelectionCommand.cpp:
  37         (WebCore::ReplacementFragment::mergeStartNode):
  38         (WebCore::ReplacementFragment::wasBlock):
  39         (WebCore::ReplacementFragment::saveRenderingInfo):
  40         (WebCore::ReplacementFragment::renderedBlocks):
  41         (WebCore::RenderingInfo::RenderingInfo):
  42         (WebCore::ReplaceSelectionCommand::shouldMergeStart):
  43         (WebCore::ReplaceSelectionCommand::doApply):
  44         * editing/ReplaceSelectionCommand.h:
  45         (WebCore::RenderingInfo::wasBlock):
  46
  47 2006-07-27  Justin Garcia  <justin.garcia@apple.com>
  48
  49         Reviewed by levi
  50
  51         More changes for:
  52         <rdar://problem/4242293>
  53         Poor paste fidelity of http://www.google.com/
  54
  55         * editing/markup.cpp:
  56         (WebCore::needInterchangeNewlineAfter): Added.
  57         (WebCore::createMarkup):
  58         Only add an interchangeNewline if we're annotating for interchange.  Can't test
  59         this until we expose createMarkup in non-annotate mode (I'd expose Range::toHTML).
  60         Remove the FIXME about prepending the interchangeNewline before creating the rest
  61         of the markup.  Its correct to surround the interchangeNewline in ancestors of the
  62         adjusted startNode.
  63         Fixed the check to see if an interchangeNewline is needed (!inSameBlock).  This would
  64         add an interchangeNewline for a selection that started before a table.
  65         Removed the workaround for the bug where markup for a table was not included when
  66         selecting the contents of a table, since this change fixes the general problem of
  67         markup for ancestors of the startNode being left out.
  68         Don't leave out ancestors of the startNode as we leave their subtrees.  Execute the
  69         code to include them even if we closed ancestors in the ancestorsToClose list.
  70
  71 2006-07-28  Timothy Hatcher  <timothy@apple.com>
  72
  73         Reviewed by John.
  74
  75         Adding isContentEditable to DOMNode's private category.
  76
  77         * bindings/objc/DOMPrivate.h:
  78
  79 2006-07-28  Alexey Proskuryakov  <ap@nypop.com>
  80
  81         Rolling out http://bugzilla.opendarwin.org/show_bug.cgi?id=6010 because of failing layout tests.
  82         Add better SVGLoad event support.
  83
  84         * dom/EventTargetNode.cpp:
  85         (WebCore::EventTargetNode::dispatchGenericEvent):
  86         * html/HTMLImageLoader.cpp:
  87         (WebCore::HTMLImageLoader::dispatchLoadEvent):
  88         * html/HTMLImageLoader.h:
  89         * ksvg2/misc/SVGImageLoader.cpp:
  90         (WebCore::SVGImageLoader::SVGImageLoader):
  91         * ksvg2/misc/SVGImageLoader.h:
  92         * ksvg2/svg/SVGAElement.cpp:
  93         (WebCore::SVGAElement::parseMappedAttribute):
  94         * ksvg2/svg/SVGAnimatedBoolean.h:
  95         * ksvg2/svg/SVGElement.cpp:
  96         (WebCore::SVGElement::SVGElement):
  97         (WebCore::SVGElement::parseMappedAttribute):
  98         * ksvg2/svg/SVGElement.h:
  99         (WebCore::SVGElement::adjustViewportClipping):
 100         (WebCore::SVGElement::closeRenderer):
 101         (WebCore::SVGElement::rendererIsNeeded):
 102         (WebCore::SVGElement::isClosed):
 103         (WebCore::svg_dynamic_cast):
 104         * ksvg2/svg/SVGExternalResourcesRequired.h:
 105         * ksvg2/svg/SVGImageElement.cpp:
 106         (SVGImageElement::SVGImageElement):
 107         * ksvg2/svg/SVGImageElement.h:
 108         * ksvg2/svg/SVGSVGElement.cpp:
 109         (WebCore::SVGSVGElement::parseMappedAttribute):
 110         * ksvg2/svg/SVGScriptElement.cpp:
 111         (SVGScriptElement::SVGScriptElement):
 112         (SVGScriptElement::parseMappedAttribute):
 113         (SVGScriptElement::executeScript):
 114         * xml/xmlhttprequest.cpp:
 115         (WebCore::XMLHttpRequest::callReadyStateChangeListener):
 116
 117 2006-07-28  Adele Peterson  <adele@apple.com>
 118
 119         Rolling out http://bugzilla.opendarwin.org/show_bug.cgi?id=9753 because of failing layout tests.
 120         SVG with width and height 100% leaves room for scrollbar on the right.
 121
 122         * kcanvas/RenderPath.cpp:
 123         * kcanvas/RenderPath.h:
 124         * kcanvas/RenderSVGContainer.cpp:
 125         (WebCore::RenderSVGContainer::layout):
 126         * ksvg2/svg/SVGCircleElement.cpp:
 127         * ksvg2/svg/SVGCircleElement.h:
 128         * ksvg2/svg/SVGEllipseElement.cpp:
 129         * ksvg2/svg/SVGEllipseElement.h:
 130         * ksvg2/svg/SVGLineElement.cpp:
 131         * ksvg2/svg/SVGLineElement.h:
 132         * ksvg2/svg/SVGRectElement.cpp:
 133         * ksvg2/svg/SVGRectElement.h:
 134         * ksvg2/svg/SVGStyledElement.h:
 135
 136 2006-07-27  Eric Seidel  <eric@eseidel.com>
 137
 138         Reviewed by darin.
 139
 140         Add better SVGLoad event support.
 141         http://bugzilla.opendarwin.org/show_bug.cgi?id=6010
 142         There will still need to be additional support added for <link> and <script> elements.
 143         Not all of the error -> loaded -> error transition cases work yet.
 144
 145         * dom/EventTargetNode.cpp:
 146         (WebCore::EventTargetNode::dispatchGenericEvent):
 147         * html/HTMLImageLoader.cpp:
 148         (WebCore::HTMLImageLoader::dispatchLoadEvent):
 149         * html/HTMLImageLoader.h:
 150         (WebCore::HTMLImageLoader::haveFiredLoadEvent):
 151         (WebCore::HTMLImageLoader::setHaveFiredLoadEvent):
 152         * ksvg2/misc/SVGImageLoader.cpp:
 153         (WebCore::SVGImageLoader::SVGImageLoader):
 154         (WebCore::SVGImageLoader::dispatchLoadEvent):
 155         * ksvg2/misc/SVGImageLoader.h:
 156         * ksvg2/svg/SVGAElement.cpp:
 157         (WebCore::SVGAElement::parseMappedAttribute):
 158         * ksvg2/svg/SVGAnimatedBoolean.h:
 159         * ksvg2/svg/SVGElement.cpp:
 160         (WebCore::SVGElement::SVGElement):
 161         (WebCore::SVGElement::parseMappedAttribute):
 162         (WebCore::SVGElement::haveLoadedRequiredResources):
 163         (WebCore::SVGElement::sendSVGLoadEventIfPossible):
 164         (WebCore::SVGElement::closeRenderer):
 165         * ksvg2/svg/SVGElement.h:
 166         (WebCore::SVGElement::rendererIsNeeded):
 167         (WebCore::svg_dynamic_cast):
 168         * ksvg2/svg/SVGExternalResourcesRequired.h:
 169         * ksvg2/svg/SVGImageElement.cpp:
 170         (SVGImageElement::SVGImageElement):
 171         (SVGImageElement::haveLoadedRequiredResources):
 172         * ksvg2/svg/SVGImageElement.h:
 173         * ksvg2/svg/SVGSVGElement.cpp:
 174         (WebCore::SVGSVGElement::parseMappedAttribute):
 175         * ksvg2/svg/SVGScriptElement.cpp:
 176         (WebCore::SVGScriptElement::SVGScriptElement):
 177         (WebCore::SVGScriptElement::parseMappedAttribute):
 178         (WebCore::SVGScriptElement::executeScript):
 179         * xml/xmlhttprequest.cpp:
 180         (WebCore::XMLHttpRequest::callReadyStateChangeListener): use new Event() instead of createEvent
 181
 182 2006-07-27  Mitz Pettel  <opendarwin.org@mitzpettel.com>
 183
 184         Reviewed by Anders.
 185
 186         - http://bugzilla.opendarwin.org/show_bug.cgi?id=10135
 187           Cleanup: remove useless parameters from RenderObject::position()
 188
 189         No test possible (no functionality change)
 190
 191         * rendering/RenderBox.cpp:
 192         (WebCore::RenderBox::position):
 193         * rendering/RenderBox.h:
 194         * rendering/RenderObject.h:
 195         (WebCore::RenderObject::position):
 196         * rendering/RenderTableRow.h:
 197         (WebCore::RenderTableRow::position):
 198         * rendering/RenderTableSection.h:
 199         (WebCore::RenderTableSection::position):
 200         * rendering/RenderText.cpp:
 201         (WebCore::RenderText::position):
 202         * rendering/RenderText.h:
 203         * rendering/bidi.cpp:
 204         (WebCore::RenderBlock::constructLine):
 205         (WebCore::RenderBlock::computeVerticalPositionsForLine):
 206
 207 2006-07-27  Rob Buis  <buis@kde.org>
 208
 209         Reviewed by Maciej.
 210
 211         http://bugzilla.opendarwin.org/show_bug.cgi?id=9753
 212         SVG with width and height 100% leaves room for scrollbar on the right.
 213
 214         Relayout children of svg containers, except svg shapes that do not depend on percentages.
 215
 216         * kcanvas/RenderPath.cpp: (WebCore::RenderPath::hasPercentageValues):
 217         * kcanvas/RenderPath.h:
 218         * kcanvas/RenderSVGContainer.cpp: (WebCore::RenderSVGContainer::layout):
 219         * ksvg2/svg/SVGCircleElement.cpp: (SVGCircleElement::hasPercentageValues):
 220         * ksvg2/svg/SVGCircleElement.h:
 221         * ksvg2/svg/SVGEllipseElement.cpp: (WebCore::SVGEllipseElement::hasPercentageValues):
 222         * ksvg2/svg/SVGEllipseElement.h:
 223         * ksvg2/svg/SVGLineElement.cpp: (SVGLineElement::hasPercentageValues):
 224         * ksvg2/svg/SVGLineElement.h:
 225         * ksvg2/svg/SVGRectElement.cpp: (WebCore::SVGRectElement::hasPercentageValues):
 226         * ksvg2/svg/SVGRectElement.h:
 227         * ksvg2/svg/SVGStyledElement.h: (WebCore::SVGStyledElement::hasPercentageValues):
 228
 229 2006-07-27  Justin Garcia  <justin.garcia@apple.com>
 230
 231         Reviewed by levi
 232
 233         <rdar://problem/4242293>
 234         Poor paste fidelity of http://www.google.com/
 235
 236         * editing/ApplyStyleCommand.cpp:
 237         (WebCore::ApplyStyleCommand::applyInlineStyle):  Fixed crasher with nil-check.
 238         * editing/markup.cpp:
 239         (WebCore::startMarkup): Style information wasn't added to an element if it had
 240         attributes but no style attribute.
 241
 242 2006-07-27  Alexander Kellett  <lypanov@kde.org>
 243
 244         Reviewed by Darin.
 245
 246         - fix for http://bugzilla.opendarwin.org/show_bug.cgi?id=9867
 247           REGRESSION: code that modifies form element in Firefox modifies form attribute in WebKit
 248
 249         * bindings/scripts/CodeGeneratorJS.pm: add new attribute - HasOverridingNameGetter
 250         * html/HTMLFormElement.idl:
 251
 252 2006-07-27  Rob Buis  <buis@kde.org>
 253
 254         Reviewed by Anders.
 255
 256         http://bugzilla.opendarwin.org/show_bug.cgi?id=8992
 257         Some HTML DOM attributes have incorrect types
 258
 259         Adjust idl and dom source code for HTMLImageElement and
 260         HTMLBaseFontElement to choose correct type for
 261         HTMLImageElement::border and HTMLBaseFontElement::size
 262         attributes.
 263
 264         * html/HTMLBaseFontElement.cpp:
 265         (WebCore::HTMLBaseFontElement::size):
 266         (WebCore::HTMLBaseFontElement::setSize):
 267         * html/HTMLBaseFontElement.h:
 268         * html/HTMLBaseFontElement.idl:
 269         * html/HTMLImageElement.cpp:
 270         (WebCore::HTMLImageElement::border):
 271         (WebCore::HTMLImageElement::setBorder):
 272         * html/HTMLImageElement.h:
 273         * html/HTMLImageElement.idl:
 274
 275 2006-07-27  Justin Garcia  <justin.garcia@apple.com>
 276
 277         Reviewed by levi
 278
 279         <rdar://problem/4628409>
 280         Safari crashes in CompositeEditCommand::removeBlockPlaceholder entering text in a web page form
 281
 282         * editing/CompositeEditCommand.cpp:
 283         (WebCore::CompositeEditCommand::removeBlockPlaceholder):
 284         A script on the page shrinks the textarea down to zero rows just
 285         before insertion.  That gives the shadow div a zero height, removing
 286         [div, 0] as a candidate for a VisiblePosition.  Then
 287         InsertTextCommand::input() tries to create a VisiblePosition from
 288         [div, 0] and fails.  A nil check is a quick fix.  Another might
 289         be to make positions where there is a caret selection candidates.
 290         Another might be to make any [div, 0] where div is a shadow div inside
 291         a form element a candidate.
 292
 293 2006-07-27  Graham Dennis  <graham.dennis@gmail.com>
 294
 295         Reviewed by Darin.
 296
 297         - fix for http://bugzilla.opendarwin.org/show_bug.cgi?id=10094
 298           Japanese characters improperly rendering in TOT
 299
 300         Test: fast/encoding/denormalised-voiced-japanese-chars.html
 301
 302         * platform/Font.cpp:
 303         (WebCore::WidthIterator::advance): pass currentCharacter to normalizeVoicingMarks
 304         (WebCore::WidthIterator::normalizeVoicingMarks): turn the currentCharacter variable
 305         into an argument as WidthIterator::advance doesn't update m_currentCharacter while
 306         looping
 307
 308 2006-07-27  Eric Seidel  <eric@eseidel.com>
 309
 310         Reviewed by ap.
 311
 312         Complex entities seem to fail on TOT
 313         http://bugzilla.opendarwin.org/show_bug.cgi?id=6390
 314         Use an alternative work-around to an XML Entity parsing bug.
 315
 316         * dom/XMLTokenizer.cpp:
 317         (WebCore::XMLTokenizer::write):
 318         (WebCore::hackAroundLibXMLEntityBug): added.
 319         (WebCore::startElementNsHandler):
 320         (WebCore::endElementNsHandler):
 321         (WebCore::charactersHandler):
 322         (WebCore::processingInstructionHandler):
 323         (WebCore::cdataBlockHandler):
 324         (WebCore::commentHandler):
 325         (WebCore::getEntityHandler):
 326
 327 2006-07-27  Rob Buis  <buis@kde.org>
 328
 329         Reviewed by Darin.
 330
 331         http://bugzilla.opendarwin.org/show_bug.cgi?id=9938
 332         Fix viewBox issues in Hixie's test
 333
 334         Do a more accurate and quicker job of parsing the viewBox
 335         attribute. Handle properly on negative width/height and
 336         invalid syntax in general.
 337
 338         * ksvg2/svg/SVGFitToViewBox.cpp:
 339         (SVGFitToViewBox::parseViewBox):
 340
 341 2006-07-27  Eric Seidel  <eric@eseidel.com>
 342
 343         Reviewed by ap.
 344
 345         Enables getSVGDocument() support for HTMLObjectElement and HTMLEmbedElement
 346         http://bugzilla.opendarwin.org/show_bug.cgi?id=9237
 347         This patch does not try to autogenerate anything, but enables the SVG functionality.
 348         Also fix HTMLObjectElement vspace and hspace to be numbers instead of strings to match IE & Firefox.
 349
 350         * bindings/js/kjs_dom.cpp:
 351         (KJS::toJS):
 352         * bindings/js/kjs_html.cpp:
 353         (KJS::JSHTMLElement::objectGetter):
 354         (KJS::HTMLElementFunction::callAsFunction):
 355         (KJS::JSHTMLElement::objectSetter):
 356         * bindings/js/kjs_html.h:
 357         (KJS::JSHTMLElement::):
 358         * dom/Document.h:
 359         (WebCore::Document::isSVGDocument):
 360         * html/HTMLEmbedElement.cpp:
 361         (WebCore::HTMLEmbedElement::contentDocument):
 362         (WebCore::HTMLEmbedElement::getSVGDocument):
 363         * html/HTMLEmbedElement.h:
 364         * html/HTMLObjectElement.cpp:
 365         (WebCore::HTMLObjectElement::HTMLObjectElement):
 366         (WebCore::HTMLObjectElement::getInstance):
 367         (WebCore::HTMLObjectElement::parseMappedAttribute):
 368         (WebCore::HTMLObjectElement::contentDocument):
 369         (WebCore::HTMLObjectElement::rendererIsNeeded):
 370         (WebCore::HTMLObjectElement::createRenderer):
 371         (WebCore::HTMLObjectElement::setCode):
 372         (WebCore::HTMLObjectElement::setArchive):
 373         (WebCore::HTMLObjectElement::setBorder):
 374         (WebCore::HTMLObjectElement::setCodeBase):
 375         (WebCore::HTMLObjectElement::setCodeType):
 376         (WebCore::HTMLObjectElement::setData):
 377         (WebCore::HTMLObjectElement::hspace):
 378         (WebCore::HTMLObjectElement::setHspace):
 379         (WebCore::HTMLObjectElement::setStandby):
 380         (WebCore::HTMLObjectElement::setType):
 381         (WebCore::HTMLObjectElement::setUseMap):
 382         (WebCore::HTMLObjectElement::vspace):
 383         (WebCore::HTMLObjectElement::setVspace):
 384         (WebCore::HTMLObjectElement::getSVGDocument):
 385         * html/HTMLObjectElement.h:
 386         * ksvg2/svg/SVGDocument.h:
 387         (WebCore::SVGDocument::isSVGDocument):
 388         * rendering/RenderWidget.h:
 389         (WebCore::RenderWidget::frameView):
 390
 391 2006-07-26  Alexey Proskuryakov  <ap@nypop.com>
 392
 393         Reviewed by Darin.
 394
 395         Fix http://bugzilla.opendarwin.org/show_bug.cgi?id=9901
 396         XSLTProcessor cannot parse text nodes for INPUT ELEMENTS in an XSLT stylesheet
 397
 398         Test: fast/js/xhtml-serialize.html
 399
 400         * editing/markup.cpp:
 401         (WebCore::endMarkup): If the element has child nodes, write a closing tag even if it
 402         is not permitted in HTML, to make it valid XML.
 403
 404 2006-07-26  Alice Liu  <alice.liu@apple.com>
 405
 406         Reviewed by John Sullivan (and Kevin and Timo).
 407
 408         landing this patch for John et al.
 409         Fixed <rdar://problem/4651380> REGRESSION (9A232->9A234): existing web clips don't render if you restart Dashboard
 410
 411         * page/FrameView.cpp:
 412         (WebCore::FrameView::layout):
 413         moved the call to didFirstLayout to above the check for needsLayout
 414         because we need to call didFirstLayout for every webview, not just
 415         ones we think need layout.
 416
 417 2006-07-26  Justin Garcia  <justin.garcia@apple.com>
 418
 419         Reviewed by darin
 420
 421         <rdar://problem/4652788>
 422         REGRESSION: Can't change size or style of any text in message body
 423
 424         * dom/Node.cpp:
 425         (WebCore::Node::isContentRichlyEditable): Mimic Node::isContentEditable.
 426         * html/HTMLElement.cpp:
 427         (WebCore::HTMLElement::isContentRichlyEditable): First check
 428         the frame for editability, then userModify, like what
 429         HTMLElement::isContentEditable() does.
 430         * html/HTMLElement.h:
 431
 432 2006-07-26  Mitz Pettel  <opendarwin.org@mitzpettel.com>
 433
 434         Reviewed by ap.
 435
 436         - fix debug-only information() string for table cells
 437
 438         * rendering/RenderObject.cpp:
 439         (WebCore::RenderObject::information):
 440
 441 2006-07-25  David Harrison  <harrison@apple.com>
 442
 443         Reviewed by timo and Darin.
 444
 445         <rdar://problem/4618584> "Paste and Match Style" is not working in Mail (add SPI)
 446
 447         * WebCore.xcodeproj/project.pbxproj:
 448         * bridge/mac/WebCoreFrameBridge.h:
 449         * bridge/mac/WebCoreFrameBridge.mm:
 450         (-[WebCoreFrameBridge replaceSelectionWithNode:selectReplacement:smartReplace:matchStyle:]):
 451         Add matchStyle parameter for use by WebKit.
 452
 453 2006-07-25  Justin Garcia  <justin.garcia@apple.com>
 454
 455         Reviewed by adele
 456
 457         <rdar://problem/4649560>
 458         REGRESSION: Can't Select All when the body is contentEditable="true"
 459
 460         * editing/VisiblePosition.cpp:
 461         (WebCore::VisiblePosition::canonicalPosition): Allow descent from a non-editable html
 462         element into an editable body as a convenience.
 463
 464 2006-07-25  Justin Garcia  <justin.garcia@apple.com>
 465
 466         Reviewed by darin
 467
 468         <rdar://problem/4468458> Incoming HTML message is editable <body contenteditable="true">
 469
 470         * dom/Document.cpp:
 471         (WebCore::Document::inDesignMode): Removed some spaces.
 472         * page/Frame.cpp:
 473         (WebCore::Frame::applyEditingStyleToElement): Just add CSS properties needed for editing.
 474         (WebCore::Frame::removeEditingStyleFromElement): Made this a no-op, we'll remove
 475         it eventually.
 476
 477 2006-07-25  David Kilzer  <ddkilzer@kilzer.net>
 478
 479         Reviewed by Darin.
 480
 481         - fix http://bugzilla.opendarwin.org/show_bug.cgi?id=9717
 482           Replace IDL ConvertUndefinedToTrue parameter attribute with Optional attribute
 483
 484         No test added because there is no change in functionality.
 485
 486         * bindings/js/kjs_binding.cpp: Removed valueToBooleanTreatUndefinedAsTrue() method.
 487         * bindings/js/kjs_binding.h: Ditto.
 488         * bindings/scripts/CodeGeneratorJS.pm: Removed support for ConvertUndefinedToTrue parameter attribute.
 489         * dom/Element.h: Set default value to true for scrollIntoView() and scrollIntoViewIfNeeded() arguments.
 490         * dom/Element.idl: Replaced ConvertUndefinedToTrue attribute with Optional attribute.
 491           Fixed name of scrollIntoViewIfNeeded() argument to match implementation.
 492         * page/DOMWindow.idl: Replaced ConvertUndefinedToTrue attribute with Optional attribute.
 493
 494 2006-07-24  Dan Waylonis  <waylonis@google.com>
 495
 496         Reviewed and tweaked a bunch by Darin.
 497
 498         Test: plugins/bindings-test.html
 499
 500         - fix http://bugzilla.opendarwin.org/show_bug.cgi?id=9902
 501           jsNull and NSNull not properly converted between JS and ObjC
 502
 503         * bindings/objc/WebScriptObject.mm:
 504         (+[WebScriptObject _convertValueToObjcValue:originExecutionContext:executionContext:]):
 505         Added case that converts jsNull to NSNull.
 506
 507 2006-07-24  Darin Adler  <darin@apple.com>
 508
 509         Reviewed by Adele and Justin.
 510
 511         - fix <rdar://problem/4613616> REGRESSION: some spaces typed in <textarea> are posted as non-breaking spaces (9630)
 512           http://bugzilla.opendarwin.org/show_bug.cgi?id=9630
 513         - also fixes http://bugzilla.opendarwin.org/show_bug.cgi?id=9939
 514           REGRESSION: Pasting text into native text area with newline at end does not preserve newline
 515         - removed some uses of DeprecatedPtrList in the markup code
 516
 517         Test: fast/forms/textarea-type-spaces.html
 518         Test: fast/forms/textarea-paste-newline.html
 519
 520         * bindings/objc/DOMHTML.mm: (-[DOMHTMLDocument createDocumentFragmentWithText:]):
 521         Updated call to pass a range -- in this case it is the range of the entire document,
 522         so this will not handle the whitespace properly.
 523
 524         * bridge/mac/WebCoreFrameBridge.h: Added range context parameter to the
 525         documentFragmentWithText: method, so we can handle whitespace properly.
 526         * bridge/mac/WebCoreFrameBridge.mm:
 527         (-[WebCoreFrameBridge nodesFromList:]): Changed from DeprecatedPtrList to Vector.
 528         (-[WebCoreFrameBridge markupStringFromNode:nodes:]): Ditto.
 529         (-[WebCoreFrameBridge markupStringFromRange:nodes:]): Ditto.
 530         (-[WebCoreFrameBridge documentFragmentWithText:inContext:]): Added range context
 531         parameter -- pass it on to createFragmentFromText.
 532         (-[WebCoreFrameBridge documentFragmentWithNodesAsParagraphs:]): Changed from
 533         DeprecatedPtrList to Vector.
 534         (-[WebCoreFrameBridge replaceSelectionWithText:selectReplacement:smartReplace:]):
 535         Pass the range of the current selection as context when creating the fragment.
 536
 537         * dom/Position.cpp: (WebCore::Position::inRenderedText): Replace range check with
 538         a call to the new containsCaretOffset function -- helps make the caret work right when
 539         it is past the end of the last line in a textarea.
 540
 541         * editing/CompositeEditCommand.cpp:
 542         (WebCore::CompositeEditCommand::rebalanceWhitespaceAt): Don't do anything if the
 543         style does not call for collapsing whitespace.
 544         (WebCore::CompositeEditCommand::rebalanceWhitespace): Call replaceWhitespaceAt
 545         to share code, including the new logic mentioned above.
 546
 547         * editing/InsertLineBreakCommand.cpp: (WebCore::InsertLineBreakCommand::doApply):
 548         Use a text node instead of a break element when inserting and the style is preserveNewline.
 549
 550         * editing/JSEditor.cpp: (WebCore::execRemoveFormat): Pass the selection range
 551         to createFragmentFromText.
 552
 553         * editing/RebalanceWhitespaceCommand.cpp: (WebCore::RebalanceWhitespaceCommand::doApply):
 554         Assert that we're in a style that collapses whitespace. It's the caller's responsibility
 555         not to call otherwise.
 556
 557         * editing/ReplaceSelectionCommand.h: Removed unused destructor, type, isSingleTextNode,
 558         isTreeFragment, m_type, and added a context parameter to inertFragmentForTestRendering.
 559         Also changed the constructor to take a selection rather than a pointer to the root
 560         editable element, replaced removeEndBRIfNeeded with shouldRemoveEndBR and removed an
 561         unused parameter from shouldMergeEnd.
 562         * editing/ReplaceSelectionCommand.cpp:
 563         (WebCore::ReplacementFragment::ReplacementFragment): Removed code to set up m_type.
 564         Compute root editable element from passed-in selection. Used the start of the selection
 565         as a base node for style purposes for the test rendering. Removed the special case
 566         "single text node" alternative to createFragmentFromText in the plain text case, since
 567         createFragmentFromText now handles that correctly.
 568         (WebCore::ReplacementFragment::insertFragmentForTestRendering): Copy the whitespace
 569         property from the source location when creating a temporary element for test rendering.
 570         (WebCore::ReplacementFragment::shouldMergeEnd): Removed unneeded boolean
 571         parameter fragmentHadInterchangeNewlineAtEnd, which is always false.
 572         (WebCore::ReplaceSelectionCommand::doApply): Update for ReplacementFragment changes,
 573         change code to not remove end BR when it can be re-used instead, don't call the
 574         paragraph separator insertion when the position is at the start of a paragraph already,
 575         removed redundant computation of identical "next" value, removed unneeded boolean
 576         parameter to shouldMergeEnd, add case for merging when all we need to do is to delete
 577         a newline character, removed unneeded code to set insertionPos after all code that uses
 578         it, and  use spaces instead of non-breaking spaces when doing smart paste if the
 579         context is one where we do not collapse white space.
 580         (WebCore::ReplaceSelectionCommand::shouldRemoveEndBR): Renamed and changed to return
 581         a boolean instead of doing the removal.
 582
 583         * editing/markup.h: Use Vector instead of DeprecatedPtrList. Change the
 584         createFragmentFromText function to take a range for context instead of a document.
 585         * editing/markup.cpp:
 586         (WebCore::markup): Use Vector instead of DeprecatedPtrList.
 587         (WebCore::createMarkup): Ditto.
 588         (WebCore::createParagraphContentsFromString): Remove unneeded document parameter
 589         and changed a couple places to use isEmpty instead of comparing with "".
 590         (WebCore::createFragmentFromText): Given the new context parameter, if the context
 591         is one that preserves newlines, then use "\n" instead of <br> elements.
 592         (WebCore::createFragmentFromNodes): Use Vector instead of DeprecatedPtrList.
 593
 594         * html/HTMLElement.cpp: (WebCore::HTMLElement::setInnerText): Do not use <br>
 595         elements if the context of this node is one where we preserve newlines.
 596
 597         * rendering/InlineTextBox.h:
 598         * rendering/InlineTextBox.cpp: (WebCore::InlineTextBox::containsCaretOffset):
 599         Added. Implements the appropriate rule for determining if a caret position is
 600         in this line or not. Returns true for both one line and the next in cases where
 601         affinity must be considered to determine which line the caret is on.
 602
 603         * rendering/RenderText.h: Make atLineWrap no longer be a member function.
 604         * rendering/RenderText.cpp:
 605         (WebCore::atLineWrap): Remove special rule about preserveNewline and isLineBreak,
 606         which will no longer apply due to the new containsCaretOffset function logic.
 607         (WebCore::RenderText::caretRect): Use containsCaretOffset.
 608         (WebCore::RenderText::inlineBox): Ditto.
 609
 610         * rendering/RenderTextControl.cpp:
 611         (WebCore::RenderTextControl::updateFromElement): Make a placeholder <br> element
 612         after calling setInnerText so that the last newline in the string has the effect
 613         we expect outside the HTML world (an additional line).
 614         (WebCore::RenderTextControl::setSelectionRange): Set granularity of the selection
 615         too. The layout tests caught this problem, which needs a better solution long term.
 616         (WebCore::RenderTextControl::text): Call textContent with the parameter false
 617         so it will not include newlines for <br> elements. Now the only <br> element
 618         that will ever be in the shadow DOM tree is the one to prevent collapsing, and
 619         that one should not show up in the text value.
 620
 621         * rendering/bidi.cpp: (WebCore::RenderBlock::findNextLineBreak): Took a rule
 622         that specifically called out the pre whitespace mode and made it work for all
 623         the modes that preserve newlines. This makes sure we get a last line box for
 624         text after the last "\n" even in cases where there is no <br> afterward.
 625
 626         * editing/DeleteSelectionCommand.cpp:
 627         (WebCore::DeleteSelectionCommand::fixupWhitespace):
 628         * editing/InsertParagraphSeparatorCommand.cpp:
 629         (WebCore::InsertParagraphSeparatorCommand::doApply):
 630         Added assertions to make sure we don't do anything when we're not collapsing
 631         whitespace.
 632
 633         * html/HTMLTextAreaElement.cpp: (WebCore::HTMLTextAreaElement::setDefaultValue):
 634         Changed to use Vector instead of DeprecatedPtrList.
 635
 636         * editing/HTMLInterchange.cpp: Removed obsolete comment.
 637
 638         * loader/Cache.h: Removed a stray include.
 639
 640 2006-07-24  Adele Peterson  <adele@apple.com>
 641
 642         Reviewed by Justin.
 643
 644         - Fix for <rdar://problem/4632132> Changing style of content with mixed editability fails
 645
 646         Test: * editing/style/apple-style-editable-mix.html
 647
 648         * dom/Node.cpp: (WebCore::Node::isContentRichlyEditable): Doesn't need to check the editable root.  It now just
 649           checks the user modify property of its own renderer, since that is inherited.
 650         * editing/ApplyStyleCommand.cpp: (WebCore::ApplyStyleCommand::applyInlineStyle): If the end position is in a table,
 651         Adjust the end node to the last descendant of the table, so we don't skip over any runs.
 652
 653 2006-07-24  Darin Adler  <darin@apple.com>
 654
 655         Reviewed by Hyatt.
 656
 657         - implement String::createCFString
 658
 659         * WebCore.xcodeproj/project.pbxproj: Added StringImplMac.mm and StringMac.mm.
 660
 661         * bindings/objc/DOMInternal.mm: Moved NSString conversion functions out of here.
 662
 663         * platform/mac/StringImplMac.mm: Added. Moved NSString conversion members here
 664         and added the CFStringRef ones.
 665         * platform/mac/StringMac.mm: Added. Ditto.
 666
 667 2006-07-24  David Hyatt  <hyatt@apple.com>
 668
 669         Two bug fixes.
 670
 671         (1) Make sure to mark positioned objects that depend on line position to
 672         mark themselves as needing layout when their line box placeholder gets
 673         repositioned.
 674
 675         * rendering/RenderBox.cpp:
 676         (WebCore::RenderBox::position):
 677
 678         (2) Improve highlighting to factor the inflation into overflow so that
 679         invalidation and repainting will work properly.
 680
 681         * rendering/RootInlineBox.cpp:
 682         (WebCore::RootInlineBox::addHighlightOverflow):
 683         * rendering/RootInlineBox.h:
 684         * rendering/bidi.cpp:
 685         (WebCore::RenderBlock::layoutInlineChildren):
 686
 687 2006-07-24  Anders Carlsson  <acarlsson@apple.com>
 688
 689         Reviewed by Adele.
 690
 691         <rdar://problem/4641824> REGRESSION: Wikipedia widget fails to complete a search after pressing return key
 692
 693         * html/HTMLInputElement.cpp:
 694         (WebCore::HTMLInputElement::defaultEventHandler):
 695         Don't eat enter events for search fields that have no enclosing form; they need to be passed to AppKit so
 696         that onsearch can be invoked.
 697
 698         * manual-tests/onsearch-enter.html: Added.
 699
 700 2006-07-24  Dave MacLachlan  <dmaclach@mac.com>
 701
 702         Reviewed by Darin and Alexey.
 703
 704         Fix for: <http://bugzilla.opendarwin.org/show_bug.cgi?id=8425>
 705         and <http://bugzilla.opendarwin.org/show_bug.cgi?id=6947>
 706
 707         Test: svg/custom/non-opaque-filters.svg
 708
 709         * kcanvas/device/quartz/KCanvasFilterQuartz.mm:
 710         (WebCore::KCanvasFilterQuartz::prepareFilter):
 711         We create an autorelease pool so we can control the deallocation of the
 712         CIContext that we're creating. The CIContext retains the CGContext that
 713         you pass it internally so when the CIContext is released, the
 714         CGContext is released as well.
 715         This is all fine and dandy unless you wrap the creation of the CIFilter
 716         with a pair of CGBegin/EndTransparencyLayer calls which swap the context
 717         out from underneath you. So if you start with context A,
 718         CGBeginTransparencyLayer swaps it out and gives you B. You create a CIFilter
 719         with it and add a reference to B. CGEndTransparencyLayer swaps out B and
 720         gives you back A. Autorelease pool comes and cleans up the Filter, and calls
 721         release on A, but A never got the refcount in the first place. B did. BOOM!
 722         So we create a pool, then do a retain, then release the pool so that we
 723         don't have to worry about the pool releasing it at a later time.
 724         See <rdar://problem/4647735> for reduction of CGEndTransparencyLayer case
 725
 726 2006-07-24  Mitz Pettel  <opendarwin.org@mitzpettel.com>
 727
 728         Reviewed by Hyatt.
 729
 730         - fix http://bugzilla.opendarwin.org/show_bug.cgi?id=9669
 731           Incomplete repaint when changing an inline's border
 732
 733         Test: fast/repaint/line-overflow.html
 734
 735         * rendering/RenderBlock.h:
 736         * rendering/bidi.cpp:
 737         (WebCore::RenderBlock::layoutInlineChildren): Added repaintTop and repaintBottom
 738         variables to track the vertical edges of the area that changed, accounting for
 739         lines that were deleted, inserted or moved. Removed unnecessary updating of
 740         m_overflowHeight.
 741         (WebCore::RenderBlock::determineStartPosition): Removed unnecessary updating of
 742         m_overflowHeight.
 743         (WebCore::RenderBlock::determineEndPosition):
 744         (WebCore::RenderBlock::matchedEndLine): Added repaintTop and repaintBottom
 745         arguments, which this method updates to account for deleted lines.
 746         (WebCore::RenderBlock::checkLinesForOverflow): Removed outdated FIXME.
 747
 748 2006-07-24  Mitz Pettel  <opendarwin.org@mitzpettel.com>
 749
 750         Reviewed by Darin.
 751
 752         - fix http://bugzilla.opendarwin.org/show_bug.cgi?id=10075
 753           REGRESSION: Reversed pop-up text in visually-ordered Hebrew pages
 754
 755         Test: fast/forms/select-visual-hebrew.html
 756
 757         * css/html4.css: Set "-webkit-rtl-ordering: logical" on select elements.
 758
 759 2006-07-24  Alexey Proskuryakov  <ap@nypop.com>
 760
 761         Reviewed by Darin.
 762
 763         Fix http://bugzilla.opendarwin.org/show_bug.cgi?id=10009
 764         REGRESSION: Schubert-IT PDF Plug-in not working for full page (works in frames)
 765
 766         * page/Frame.cpp:
 767         (WebCore::Frame::begin): Give PDF plugins a chance to handle frame content, before ImageDocument
 768         claims it.
 769
 770 2006-07-24  Mitz Pettel  <opendarwin.org@mitzpettel.com>
 771
 772         Reviewed by Darin.
 773
 774         - fix http://bugzilla.opendarwin.org/show_bug.cgi?id=10083
 775           REGRESSION (r15584): editing/selection/select-from-textfield-outwards failing
 776
 777         * rendering/RenderBlock.cpp:
 778         (WebCore::RenderBlock::positionForCoordinates): Added back change from the fix
 779         for bug 9312: For coordinates outside a replaced object, return the position
 780         just before (after) the element if the coordinates are above or to the left (below
 781         or to the right).
 782
 783 2006-07-23  Mark Rowe  <opendarwin.org@bdash.net.nz>
 784
 785         Reviewed by maciej.
 786
 787         Bug 9686: [Drosera] Need the ability to break into Drosera on Javascript exceptions
 788         http://bugzilla.opendarwin.org/show_bug.cgi?id=9686
 789
 790         WebCore portion of the fix.
 791
 792         * bridge/mac/WebCoreScriptDebugger.h:
 793         (-[WebScriptDebugger exceptionRaised:sourceId:line::]): Add delegate method.
 794         * bridge/mac/WebCoreScriptDebugger.mm:
 795         (WebCoreScriptDebuggerImp::exception): Call delegate method when an exception is raised.
 796
 797 2006-07-23  Alice Liu  <alice.liu@apple.com>
 798
 799         Reviewed by Adele.
 800         Actually Adele figured out how to fix the layout test failures.  Landing this patch for her.  Layout test failures exposed but not caused by  r15584 (my patch earlier today).
 801
 802         * editing/visible_units.cpp:
 803         (WebCore::previousLinePosition):
 804         When setting selection endpoints, don't traverse down into nodes where editing would ignore its contents. Use the parent node instead.
 805         (WebCore::nextLinePosition):
 806         same as above
 807
 808 2006-07-23  David Harrison  <harrison@apple.com>
 809
 810         Reviewed by timo.
 811
 812         <rdar://problem/4646759> Mail crash editing To Do - WebCore::InsertTextCommand::prepareForTextInsertion
 813
 814         * Test: editing/deleting/delete-mixed-editable-content-001.html
 815
 816         * editing/visible_units.cpp:
 817         (WebCore::startOfParagraph):
 818         Respect editable boundary the same way endOfParagraph does.
 819
 820 2006-07-23  Beth Dakin  <bdakin@apple.com>
 821
 822         Reviewed by Maciej. (Patch by me, Maciej, and Harrison.)
 823
 824         Fix for <rdar://problem/4529398> WebCore crashes when pasting rich
 825         text - WebCore::InlineBox::root()
 826
 827         The initial rendering crash was due to a render object having a
 828         stale reference to an inline box that had already been deleted and
 829         then recreated in the exact same location in memory. (Crazy, I
 830         know.) The situation seemed pretty specific to list markers
 831         according to Hyatt according to Maciej, so that is what I patched
 832         specifically. Fixing this crash unearthed a separate editing crash
 833         where we were trying to insert a block into itself. I worked on
 834         that with Maciej and Harrison, and Harrison came up with a fix.
 835
 836         * editing/CompositeEditCommand.cpp:
 837         (WebCore::CompositeEditCommand::moveParagraphContentsToNewBlockIfNecessary): This is the fix for the editing crash. If paragraphStart is an atomic
 838         node, insert the new block into the parent instead.
 839         * rendering/InlineBox.cpp:
 840         (WebCore::InlineBox::isChildOfParent): This function is for
 841         posterity. It will help keep the linebox tree in check.
 842         * rendering/InlineBox.h:
 843         * rendering/InlineFlowBox.cpp:
 844         (WebCore::InlineFlowBox::addToLine): Added assert.
 845         (WebCore::InlineFlowBox::deleteLine): Added assert.
 846         * rendering/ListMarkerBox.cpp:
 847         (WebCore::ListMarkerBox::destroy): If this has a parent, call
 848         removeChild on this.
 849         (WebCore::ListMarkerBox::operator delete):
 850         * rendering/ListMarkerBox.h:
 851
 852 2006-07-23  Alice Liu  <alice.liu@apple.com>
 853
 854         Reviewed by Darin.
 855
 856         fixed <rdar://problem/4617841> REGRESSION (NativeTextField): You can move keyboard focus to a field without getting insertion point
 857
 858         * rendering/RenderBlock.cpp:
 859         (WebCore::RenderBlock::positionForCoordinates):
 860         removed some unnecessary vertical position checks and added a fudge factor to be more forgiving for clicks near lines.
 861         * rendering/RenderTextControl.cpp:
 862         (WebCore::RenderTextControl::nodeAtPoint):
 863         added implementation
 864         * rendering/RenderTextControl.h:
 865         added function protocol
 866
 867 2006-07-23  Adele Peterson  <adele@apple.com>
 868
 869         Reviewed by Darin.
 870
 871         Updating fix from last checkin.
 872
 873         * rendering/RenderFlexibleBox.cpp: (WebCore::RenderFlexibleBox::layoutVerticalBox):
 874         Only tell the view about this flexbox if there isn't another flexbox already cached.
 875
 876 2006-07-23  Adele Peterson  <adele@apple.com>
 877
 878         Reviewed by Maciej.
 879
 880         - Fix for <rdar://problem/4644614> REGRESSION: Typing, pasting or dragging in new text areas causes unnecessary scrolling
 881
 882         * rendering/RenderBlock.cpp: (WebCore::RenderBlock::layoutBlock): Don't updateScrollInfoAfterLayout if an ancestor flexible box is just on
 883         its first layout.
 884         * rendering/RenderFlexibleBox.cpp:
 885         (WebCore::RenderFlexibleBox::layoutBlock): ditto.
 886         (WebCore::RenderFlexibleBox::layoutVerticalBox): Let the view know if this flex box is doing its first pass at layout.
 887
 888         * rendering/RenderView.cpp: (WebCore::RenderView::RenderView): Caches a flexible box that's doing its first layout.
 889         * rendering/RenderView.h:
 890         (WebCore::RenderView::setFlexBoxInFirstLayout):
 891         (WebCore::RenderView::flexBoxInFirstLayout):
 892
 893 2006-07-22  Alexey Proskuryakov  <ap@nypop.com>
 894
 895         Reviewed by Darin.
 896
 897         Fix http://bugzilla.opendarwin.org/show_bug.cgi?id=10038
 898         REGRESSION: Length of navigator.mimeTypes collection returns number of installed plugins, not number
 899         of registered mime types
 900
 901         * bindings/js/kjs_navigator.cpp:
 902         (KJS::MimeTypes::getValueProperty): Return the length of the mime types vector, not the plugins one.
 903         (KJS::MimeTypes::getOwnPropertySlot): Fixed another typo, which prevented MimeTypes::getValueProperty()
 904         from even being called.
 905
 906 2006-07-21  Maciej Stachowiak  <mjs@apple.com>
 907
 908         Reviewed by Adele and Tim Omernick.
 909
 910         <rdar://problem/4641004> 9A224: Safari crashes in WebCore::RenderMenuList::showPopup when changing a input select field to input text field
 911
 912         Couldn't figure out how to make layout test, here's a manual test:
 913         * manual-tests/remove-select-onchange.html: Added.
 914
 915         * rendering/RenderMenuList.cpp:
 916         (WebCore::RenderMenuList::showPopup): Don't add the RenderPopupMenu to the render
 917         tree so it doesn't get destroyed if we do.
 918
 919         The remaining changes are to let the RenderPopupMenu know its menu list w/o having
 920         to be in the render tree:
 921
 922         * rendering/RenderPopupMenu.cpp:
 923         (WebCore::RenderPopupMenu::RenderPopupMenu):
 924         * rendering/RenderPopupMenu.h:
 925         (WebCore::RenderPopupMenu::menuList):
 926         * rendering/RenderPopupMenuMac.h:
 927         * rendering/RenderPopupMenuMac.mm:
 928         (WebCore::RenderPopupMenuMac::RenderPopupMenuMac):
 929         * rendering/RenderPopupMenuWin.h:
 930         (WebCore::RenderPopupMenuWin::RenderPopupMenuWin):
 931         * rendering/RenderTheme.h:
 932         * rendering/RenderThemeMac.h:
 933         * rendering/RenderThemeMac.mm:
 934         (WebCore::RenderThemeMac::createPopupMenu):
 935         * rendering/RenderThemeWin.cpp:
 936         (WebCore::RenderThemeWin::createPopupMenu):
 937         * rendering/RenderThemeWin.h:
 938
 939 2006-07-21  Adele Peterson  <adele@apple.com>
 940
 941         Reviewed by Darin.
 942
 943         Fix for: <rdar://problem/4643238> REGRESSION: Can't set insertion point at the end of a line of text
 944
 945         Test: fast/forms/textarea-scrolled-endline-caret.html
 946
 947         * rendering/RenderBlock.cpp: (WebCore::RenderBlock::positionForCoordinates):
 948         When looking for the closest line box, take the scroll offset into account.
 949
 950 2006-07-21  Tim Omernick  <timo@apple.com>
 951
 952         Reviewed by Geoff & Maciej.
 953
 954         <rdar://problem/4632505> REGRESSION: Crash at WebCore::Widget::getView() const + 6
 955
 956         Geoff is working on a layout test for this.
 957
 958         * bridge/mac/FrameMac.mm:
 959         (WebCore::FrameMac::focusWindow):
 960         (WebCore::FrameMac::unfocusWindow):
 961         Check for a NULL view.  The view can be NULL if the frame has not yet loaded any data.  This fixes the crash, but the behavior is still
 962         wrong -- we should focus the frame once it gets a view.  I've filed <rdar://problem/4645685> to track that.  It's not as urgent as this
 963         crasher.
 964
 965 2006-07-21  Maciej Stachowiak  <mjs@apple.com>
 966
 967         Reviewed by Darin.
 968
 969         <rdar://problem/4523976> REGRESSION (NativeTextField): Crash occurs when choosing "Undo Typing" after typing and setting the value
 970
 971         * rendering/RenderTextControl.cpp:
 972         (WebCore::RenderTextControl::updateFromElement): Clear the undo
 973         chain when the text control contents have been set
 974         programmatically.
 975         * bridge/mac/FrameMac.mm:
 976         (WebCore::FrameMac::clearUndoRedoOperations): Before clearing undo
 977         stack, close all open undo groups and then open an equal number,
 978         since otherwise NSUndoManager ends up in an inconsistent state
 979         leading to uncaught ObjC exceptions.
 980
 981 2006-07-21  Beth Dakin  <bdakin@apple.com>
 982
 983         Reviewed by Darin.
 984
 985         Just moving this assertion down a couple of lines. It was hitting
 986         every time you try to print a page because we put it too early in
 987         the function.
 988
 989         * bridge/mac/WebCoreFrameBridge.mm:
 990         (-[WebCoreFrameBridge drawRect:]):
 991
 992 2006-07-21  Justin Garcia  <justin.garcia@apple.com>
 993
 994         Reviewed by levi
 995
 996         <rdar://problem/4548238>
 997         REGRESSION: Can't remove the first OL/UL list item in a Mail's compose window
 998
 999         * editing/CompositeEditCommand.cpp:
1000         (WebCore::CompositeEditCommand::breakOutOfEmptyListItem):
1001         Moved from InsertParagraphSeparator.  Does its own typing style restoration.
1002         * editing/CompositeEditCommand.h:
1003         * editing/InsertParagraphSeparatorCommand.cpp:
1004         (WebCore::InsertParagraphSeparatorCommand::doApply): Call breakOutOfEmptyListItem.
1005         * editing/TypingCommand.cpp:
1006         (WebCore::TypingCommand::deleteKeyPressed): Call breakOutOfEmptyListItem if
1007         the endingSelection is at the start of an editable region.
1008         * editing/htmlediting.cpp:
1009         (WebCore::embeddedSublist): Moved from InsertParagraphSeparator.
1010         (WebCore::appendedSublist): Ditto.
1011         (WebCore::enclosingEmptyListItem): Ditto.
1012         * editing/htmlediting.h:
1013
1014 === Safari-521.20 ===
1015
1016 2006-07-21  Tim Omernick  <timo@apple.com>
1017
1018         Reviewed by Beth Dakin & John Sullivan.
1019
1020         <rdar://problem/4633717> 9A218: Reproducible crash in -[NSScroller mouseDown:]
1021
1022         Beth is in the middle of making a layout test for this; she will land it soon.
1023
1024         * platform/mac/ScrollBarMac.mm:
1025         (ScrollBar::~ScrollBar):
1026         Call Widget::removeFromSuperview() rather than -removeFromSuperview directly on the scroll bar view.
1027         Widget::removeFromSuperview() obeys the "mustStayInWindow" flag, which is set while tracking the mouse
1028         in view-based widgets to prevent their destruction while the tracking is in progress.
1029         I searched around WebCore and this is the only Widget subclass that directly removes its view rather
1030         than using removeFromSuperview().
1031
1032 2006-07-21  Mitz Pettel  <opendarwin.org@mitzpettel.com>
1033
1034         Reviewed by Darin.
1035
1036         - fix http://bugzilla.opendarwin.org/show_bug.cgi?id=9312
1037           REGRESSION: Selection bug in new text fields when selecting past the first letter
1038
1039         Test: editing/selection/select-from-textfield-outwards.html
1040
1041         * editing/Selection.cpp:
1042         (WebCore::Selection::adjustForEditableContent): Added code to handle the case
1043         where the selection starts (resp. ends) in the last (resp. first) visual position
1044         inside an editable root.
1045         * editing/htmlediting.cpp:
1046         (WebCore::comparePositions): Fixed the case of comparing a position inside a shadow
1047         tree with a position in the shadow ancestor.
1048         * rendering/RenderBlock.cpp:
1049         (WebCore::RenderBlock::positionForCoordinates): For coordinates outside a replaced
1050         object, return the position just before (after) the element if the coordinates are above or
1051         to the left (below or to the right).
1052         * rendering/RenderObject.cpp:
1053         (WebCore::RenderObject::caretMaxOffset): Changed to return 1 for replaced objects.
1054         * rendering/RenderText.cpp:
1055         (WebCore::RenderText::positionForCoordinates): Changed to return the last position
1056         in the lowest text box if the y coordinate is below all text boxes.
1057
1058 2006-07-21  Geoffrey Garen  <ggaren@apple.com>
1059
1060         Reviewed by Darin.
1061
1062         - Fixed <rdar://problem/4507265> REGRESSION: overlays don't work on
1063         HousingMaps.com (Google Maps-based site)
1064
1065         - Made style.filter undetectable, like document.all.
1066
1067         Unfortunately, the SVG spec-makers invented a CSS attribute named 'filter',
1068         which conflicts with IE's custom CSS attribute by the same name. Web programs
1069         like the Google maps API test for style.filter, and assume it's the IE
1070         style.filter if they find it, so we need to make style.filter undetectable
1071         to avoid breaking them.
1072
1073         An alternative solution would be to hotwire a delorean, go back in time,
1074         and beg the web standards makers to make standards that work on the web.
1075
1076         * bindings/js/kjs_css.cpp:
1077         (KJS::DOMCSSStyleDeclaration::cssPropertyGetter):
1078
1079 2006-07-20  Justin Garcia  <justin.garcia@apple.com>
1080
1081         Reviewed by john
1082
1083         <rdar://problem/4641033>
1084         REGRESSION: Pasting from web pages into Mail (or Blot) often loses most of the content
1085
1086         * editing/markup.cpp:
1087         (WebCore::createMarkup): We surround the currently accumulated markup with markup
1088         for ancestors of the startNode when the pre-order traversal leaves the trees rooted
1089         at those ancestors.  We assumed that any ancestors of the current node not in the
1090         ancestorsToClose list were those kind of ancestors.  But we don't add renderer-less
1091         containers to the ancestorsToClose list.  So, we were incorrectly surrounding the
1092         currently accumulated markup with markup for render-less containers.  Most of the
1093         content at apple.com was being put inside an <optgroup> inside a <select> element.
1094         Then on Paste createContextualFragment would drop it.
1095
1096 2006-07-21  Anders Carlsson  <acarlsson@apple.com>
1097
1098         Reviewed by Geoff.
1099
1100         * dom/Document.h:
1101         (WebCore::Document::):
1102         * dom/EventTargetNode.cpp:
1103         (WebCore::EventTargetNode::addEventListener):
1104         * page/FrameView.cpp:
1105         (WebCore::FrameView::layout):
1106         * rendering/RenderLayer.cpp:
1107         (WebCore::RenderLayer::updateScrollInfoAfterLayout):
1108         Only dispatch overflowchanged events if overflowchanged event listeners have been registered on the document.
1109
1110 2006-07-20  Anders Carlsson  <acarlsson@apple.com>
1111
1112         Reviewed by Geoff.
1113
1114         <rdar://problem/4637807> REGRESSION: "Features & Options" page at volvocars.us fails
1115
1116         * xml/XSLTProcessor.cpp:
1117         (WebCore::xsltParamArrayFromParameterMap):
1118         Turns out parameters never worked. String parameters need to be escaped so we create a transform context,
1119         add the parameters to it quoted and then use xsltApplyStylesheetUser which lets us pass the transform context to it.
1120         This also works around a bug in libxslt where a hash table isn't allocated.
1121
1122 2006-07-20  Alice Liu  <alice.liu@apple.com>
1123
1124         Reviewed by Tim Omernick.
1125
1126         Fixed <rdar://problem/4621649> repro crash: Upload link causes crash on pages.google.com in Frame::nodeForWidget
1127         added manual test because there is no way to change the value of an <input type=file> via the DOM.  It's not supported because it's a security risk.
1128
1129         * manual-tests/remove-input-file-onchange.html: Added.
1130         * platform/mac/FileButtonMac.mm:
1131         (-[WebFileChooserButton chooseFilename:]):
1132         swapped the calls to changeFilename and bridgeForWidget because changeFilename will destroy the widget that is accessed in bridgeForWidget
1133
1134 2006-07-20  Alice Liu  <alice.liu@apple.com>
1135
1136         Reviewed by Adele.
1137
1138         Fixed <rdar://problem/4532113> REGRESSION (NativeTextField): Crash occurs after modifying field then reloading page -[FormDelegate frameLayoutHasChanged:]
1139         added a manual test because of the need to use AutoFill.
1140
1141         * dom/Node.h:
1142         (WebCore::Node::aboutToUnload):
1143         added virtual function prototype
1144         * html/HTMLInputElement.cpp:
1145         (WebCore::HTMLInputElement::aboutToUnload):
1146         added implementation that sends textFieldDidEndEditing notification
1147         * html/HTMLInputElement.h:
1148         added virtual function prototype
1149         * manual-tests/input-type-file-autocomplete-frame-1.html: Added.
1150         * manual-tests/input-type-file-autocomplete-frame-2.html: Added.
1151         * manual-tests/input-type-file-autocomplete-refresh.html: Added.
1152         * page/Frame.cpp:
1153         (WebCore::Frame::stopLoading):
1154         before unloading, call aboutToUnload on the current focused node
1155
1156 2006-07-20  Brady Eidson  <beidson@apple.com>
1157
1158         Reviewed by Maciej
1159
1160         fixed <rdar://problem/4611303> REGRESSION: repro crash in WebCore::EventTargetNode::dispatchWindowEvent at aeropostale.com
1161         There was an attempt to deref an EventListener that got cleaned up in GC.  Changing Document's EventListener list from
1162         vanilla ptrs to refptrs to prevent GC, following EventTargetNode's proven example.
1163
1164         * dom/Document.cpp:
1165         (WebCore::Document::Document):
1166         (WebCore::Document::clear):
1167         (WebCore::Document::handleWindowEvent):
1168         (WebCore::Document::getHTMLWindowEventListener):
1169         (WebCore::Document::removeHTMLWindowEventListener):
1170         (WebCore::Document::removeWindowEventListener):
1171         (WebCore::Document::hasWindowEventListener):
1172         Changed all of the list iterators to work with RefPtrs instead of vanilla ptrs
1173
1174         * dom/Document.h:
1175         Changed the EventListener list to be a vanilla pointer list to a refptr list
1176
1177 2006-07-20  John Sullivan  <sullivan@apple.com>
1178
1179         Reviewed by Maciej
1180
1181         - WebCore part of fix for:
1182         <rdar://problem/4557386> REGRESSION (419.3-521.19): repro Safari world leak involving
1183         closing tabs after clicking in a web page
1184
1185         * bridge/mac/WebCoreFrameBridge.h:
1186         (-[WebCoreFrameBridge textViewWasFirstResponderAtMouseDownTime:])
1187         renamed to be more specific (formerly wasFirstResponderAtMouseDownTime:)
1188         * bridge/mac/FrameMac.mm:
1189         (WebCore::FrameMac::passMouseDownEventToWidget):
1190         updated for name change
1191
1192 2006-07-19  Alexey Proskuryakov  <ap@nypop.com>
1193
1194         Reviewed by Darin.
1195
1196         Fix http://bugzilla.opendarwin.org/show_bug.cgi?id=8272
1197         Use of window.open & window.close can cause crash
1198
1199         * platform/PlatformMouseEvent.h:
1200         (WebCore::PlatformMouseEvent::PlatformMouseEvent):
1201         * platform/mac/PlatformMouseEventMac.mm:
1202         (WebCore::PlatformMouseEvent::PlatformMouseEvent):
1203         The default constructor now creates a "zero" event, and a new one was added to create the "current" one.
1204
1205         * bridge/mac/FrameMac.mm:
1206         (WebCore::FrameMac::handleMouseMoveEvent):
1207         * rendering/RenderFormElement.cpp:
1208         (WebCore::RenderFormElement::clicked):
1209         Updated for the above changes.
1210
1211         * platform/gdk/MouseEventGdk.cpp:
1212         * platform/gdk/TemporaryLinkStubs.cpp:
1213         (PlatformMouseEvent::PlatformMouseEvent):
1214         * platform/win/MouseEventWin.cpp:
1215         * platform/win/TemporaryLinkStubs.cpp:
1216         (PlatformMouseEvent::PlatformMouseEvent):
1217         Trying not to break other platforms.
1218
1219         * manual-tests/invalid-mouse-event.html: Added.
1220
1221 2006-07-20  Maciej Stachowiak  <mjs@apple.com>
1222
1223         Reviewed by Anders.
1224
1225         - delete line that I only commented in the last patch - meant to do this before landing
1226
1227         * html/HTMLFrameElement.cpp:
1228         (WebCore::HTMLFrameElement::close):
1229
1230 2006-07-19  Maciej Stachowiak  <mjs@apple.com>
1231
1232         Reviewed by Geoff.
1233
1234         - fixed <rdar://problem/4634484> REGRESSION: Project Change Request form should have vertical scroll bar, but doesn't
1235
1236         I couldn't figure out how to make an automated test case for this,
1237         or even a manual one. It seems to be timing-related in some way.
1238
1239         * html/HTMLFrameElement.cpp:
1240         (WebCore::HTMLFrameElement::close): detach the child frame from
1241         its element, not the frame containing this element.
1242
1243 2006-07-20  Anders Carlsson  <acarlsson@apple.com>
1244
1245         Reviewed by Darin.
1246
1247         * bridge/mac/FrameMac.mm:
1248         (WebCore::FrameMac::nextKeyViewInFrame):
1249         Hold a ref to the node in case it gets destroyed by an event handler.
1250
1251 2006-07-19  Mark Rowe  <opendarwin.org@bdash.net.nz>
1252
1253         Reviewed by Darin.
1254
1255         http://bugzilla.opendarwin.org/show_bug.cgi?id=10021
1256         Bug 10021: REGRESSION: Stack overflow due to infinite recursion in
1257         Image::checkForSolidColor
1258
1259         * platform/Image.cpp:
1260         (WebCore::Image::cacheFrame): Don't call checkForSolidColor unless
1261         the frame was decoded successfully.
1262
1263 2006-07-19  Adele Peterson  <adele@apple.com>
1264
1265         Reviewed by Maciej.
1266
1267         - Fix for:
1268         <rdar://problem/4422657> REGRESSION: member name field or password field should have focus after loading webmail.mac.com (7405)
1269         http://bugzilla.opendarwin.org/show_bug.cgi?id=7405
1270
1271         <rdar://problem/4614181> REGRESSION: Crash in WebCore::RenderTextField::text() when quoting post at the Ars Technica forum (9707)
1272         http://bugzilla.opendarwin.org/show_bug.cgi?id=9707
1273
1274         Needs an http test. (http://bugzilla.opendarwin.org/show_bug.cgi?id=10020)
1275
1276         These bugs were both cases where focus() was called on an element which didn't have a renderer yet because stylesheets hadn't finished loading yet.
1277         Now, we detect this case and let setFocusNode be called.  And when the stylesheet finishes loading, and the element attaches, a timer will fire,
1278         which will cause the correct selection & scrolling behavior to occur.
1279
1280         This fix removes selection and scrolling behavior from the focus method.  This code is now in a new method, updateFocusAppearance.
1281         updateFocusAppearance can now be called directly from focus(), but it can also be called when a timer fires.  This timer gets set
1282         up when an element attaches, and its already been focused by the focus method.  We have to use a timer, because updateFocusAppearance can cause
1283         a layout to happen, and we don't want that to happen in the middle of attach().
1284
1285         * bindings/objc/DOM.mm: (-[DOMElement isFocused]): Added SPI for autofill.
1286         * bindings/objc/DOMPrivate.h:
1287
1288         * dom/Element.cpp:
1289         (WebCore::Element::Element): Initializes timer and needFocusAppearanceUpdate bool.
1290         (WebCore::Element::attach): Checks needsFocusAppearanceUpdate, and if the node is focused, then starts the timer.
1291         (WebCore::Element::focus): Updated to check supportsFocus before calling setFocusNode,
1292         and only requiring the element to be focusable now before updating focus appearance.
1293         (WebCore::Element::updateFocusAppearance): Added. Separates the selection, and the scrolling from focusing the node.
1294         (WebCore::Element::updateFocusAppearanceTimerFired): Stops the timer, and if the element is focusable, calls updateFocusAppearance.
1295         (WebCore::Element::stopUpdateFocusAppearanceTimer):  Cancels timer, and setsNeedsFocusAppearanceUpdate(false).
1296         (WebCore::Element::detach): Calls stopUpdateFocusAppearanceTimer.
1297         (WebCore::Element::blur): ditto.
1298         * dom/Element.h:
1299         (WebCore::Element::needsFocusAppearanceUpdate): Added so the timer only fires when focus() methods have caused an element to be focused.
1300         (WebCore::Element::setNeedsFocusAppearanceUpdate): Added so focus methods can set this flag.
1301
1302         * dom/Node.h: (WebCore::Node::supportsFocus): Added.  Base class just calls isFocusable.
1303         * html/HTMLAnchorElement.h: Added supportsFocus.
1304         * html/HTMLAnchorElement.cpp: (WebCore::HTMLAnchorElement::supportsFocus): Added. Checks for the case where stylesheets haven't loaded yet,
1305         so we can still focus the node without a renderer, and when it gets a renderer, we'll update the focus appearance.
1306         * html/HTMLGenericFormElement.h: (WebCore::HTMLGenericFormElement::supportsFocus): ditto.
1307         * html/HTMLGenericFormElement.cpp: Removed include of Document.h since this is now in the header.
1308
1309         * html/HTMLInputElement.cpp:
1310         (WebCore::HTMLInputElement::focus): Updated to check supportsFocus before calling setFocusNode,
1311         and only requiring the element to be focusable now before updating focus appearance.
1312         (WebCore::HTMLInputElement::updateFocusAppearance): Added. Separates the selection, and the scrolling from focusing the node.
1313         * html/HTMLInputElement.h:
1314
1315         * html/HTMLTextAreaElement.cpp:
1316         (WebCore::HTMLTextAreaElement::focus): Updated to check supportsFocus before calling setFocusNode,
1317         and only requiring the element to be focusable now before updating focus appearance.
1318         (WebCore::HTMLTextAreaElement::updateFocusAppearance): Added. Separates the selection, and the scrolling from focusing the node.
1319         * html/HTMLTextAreaElement.h:
1320
1321 2006-07-19  Justin Garcia  <justin.garcia@apple.com>
1322
1323         Reviewed by levi
1324
1325         <rdar://problem/4631972>
1326         REGRESSION: Mail crashes when pasting entire contents of http://www.apple.com/support/ into a new mail message
1327
1328         * dom/Position.cpp:
1329         (WebCore::Position::upstream): Deployed isBlock and enclosingBlock.
1330         (WebCore::Position::downstream): Ditto.
1331         * editing/CompositeEditCommand.cpp:
1332         (WebCore::CompositeEditCommand::moveParagraphContentsToNewBlockIfNecessary): Ditto.
1333         * editing/htmlediting.cpp:
1334         (WebCore::canHaveChildrenForEditing): Added !iframe.
1335         (WebCore::isBlock): Added, returns !node->renderer()->isInline()
1336         (WebCore::enclosingBlock): Added.
1337         * editing/htmlediting.h:
1338
1339 2006-07-19  Anders Carlsson  <acarlsson@apple.com>
1340
1341         Reviewed by Adele.
1342
1343         <rdar://problem/4631272> REGRESSION: Crash tabbing out of hostname field at autorestore.apple.com
1344
1345         * bridge/mac/FrameMac.mm:
1346         (WebCore::FrameMac::nextKeyViewInFrame):
1347         Take into account that focus handlers can cause a node's renderer to be destroyed.
1348
1349 2006-07-19  David Hyatt  <hyatt@apple.com>
1350
1351         Fix for 4638376.  The 1x1 solid color image optimization was broken when
1352         I re-architected image.  This patch restores the optimization for CG.  Cairo
1353         will still need the optimization.
1354
1355         Reviewed by darin
1356
1357         WARNING: NO TEST CASES ADDED OR CHANGED (need test cases still)
1358
1359         * platform/Image.cpp:
1360         (WebCore::Image::Image):
1361         (WebCore::Image::invalidateData):
1362         (WebCore::Image::cacheFrame):
1363         * platform/Image.h:
1364         (WebCore::Image::setIsPDF):
1365         * platform/cairo/ImageCairo.cpp:
1366         (WebCore::Image::checkForSolidColor):
1367         * platform/mac/ImageMac.mm:
1368         (WebCore::Image::initNativeData):
1369         (WebCore::Image::invalidateNativeData):
1370         (WebCore::Image::checkForSolidColor):
1371         (WebCore::Image::draw):
1372         (WebCore::Image::drawTiled):
1373
1374 2006-07-19  Justin Garcia  <justin.garcia@apple.com>
1375
1376         Reviewed by levi
1377
1378         <rdar://problem/4613519>
1379         REGRESSION: Pasting text in native text area inserts text one character before it should (9527)
1380
1381         * editing/InsertParagraphSeparatorCommand.cpp:
1382         (WebCore::InsertParagraphSeparatorCommand::doApply): Turn into an InsertLineBreakCommand instead of bailing
1383         if the enclosingBlockFlowElement doesn't have a parent.
1384
1385 2006-07-19  David Harrison  <harrison@apple.com>
1386
1387         Reviewed by Tim Hatcher.
1388
1389         <rdar://problem/4629307> -[DOMRange markupString] does not include the initial table element if it is at the beginning of the range
1390         <rdar://problem/4062218> pasting in contents of web.apple.com strips off the <table> element from the first table
1391
1392         Better patch than previous checkin.
1393
1394         * test: editing/pasteboard/paste-table-003.html
1395
1396         * editing/markup.cpp:
1397         (WebCore::createMarkup):
1398         Make sure to include the table when including a tbody.
1399
1400 === Safari-521.19 ===
1401
1402 2006-07-19  David Harrison  <harrison@apple.com>
1403
1404         Reviewed by Tim Hatcher.
1405
1406         <rdar://problem/4629307> -[DOMRange markupString] does not include the initial table element if it is at the beginning of the range
1407
1408         No test case as this can only be reproduced through ObjC APIs.
1409
1410         * editing/markup.cpp:
1411         (WebCore::createMarkup):
1412         Adjust the range for ancestor markup handling when main loop skips first node.
1413
1414 2006-07-18  David Hyatt  <hyatt@apple.com>
1415
1416         Fix for radar 4611287, make resizable text fields opt-in rather than
1417         opt-out.  Open source users will need to write the pref into their defaults
1418         to see the resizers on textareas now.
1419
1420         Reviewed by mjs and adele
1421
1422         * WebCore.xcodeproj/project.pbxproj:
1423         * bridge/mac/WebCoreSettings.mm:
1424         (-[WebCoreSettings setTextAreasAreResizable:]):
1425         * css/cssparser.cpp:
1426         (WebCore::CSSParser::parseValue):
1427         * css/cssstyleselector.cpp:
1428         (WebCore::CSSStyleSelector::applyProperty):
1429         * css/html4.css:
1430
1431 2006-07-18  Anders Carlsson  <acarlsson@apple.com>
1432
1433         Reviewed by Darin.
1434
1435         http://bugzilla.opendarwin.org/show_bug.cgi?id=9959
1436         REGRESSION: iframes stop rendering after 200th one on successive reloads
1437
1438         * html/HTMLFrameElement.cpp:
1439         (WebCore::HTMLFrameElement::attach):
1440         (WebCore::HTMLFrameElement::close):
1441         * html/HTMLIFrameElement.cpp:
1442         (WebCore::HTMLIFrameElement::insertedIntoDocument):
1443         (WebCore::HTMLIFrameElement::willRemove):
1444         Remove calls to incrementFrameCount and decrementFrameCount.
1445
1446         * page/Frame.cpp:
1447         (WebCore::Frame::Frame):
1448         Call incrementFrameCount here.
1449
1450         (WebCore::Frame::~Frame):
1451         Call disconnectOwnerElement.
1452
1453         (WebCore::Frame::disconnectOwnerElement):
1454         Call decrementFrameCount here.
1455
1456 2006-07-18  Alexey Proskuryakov  <ap@nypop.com>
1457
1458         Reviewed by Darin.
1459
1460         Fix http://bugzilla.opendarwin.org/show_bug.cgi?id=9013
1461         Let correct CSS custom cursor declarations parse
1462
1463         * css/cssparser.cpp:
1464         (WebCore::CSSParser::parseValue): Ignore any values following a custom CSS cursor URI, instead of
1465         requiring that it is the only one in the list (which is actually illegal, according to the spec).
1466         This is only a temporary hack, with a real implementation to follow in bug 6001/6002.
1467
1468         * manual-tests/custom-cursors.html: Added a test for this issue.
1469         * manual-tests/resources/helpCursor.tiff: A cursor image used in the test.
1470
1471 2006-07-18  Timothy Hatcher  <timothy@apple.com>
1472
1473         Reviewed by Anders.
1474
1475         <rdar://problem/4636216> NetNewsWire fails to launch on 9A224 - missing symbols
1476
1477         WebCore needs to add -sub_library libobjc to maintain
1478         backwards compatibility with binaries linked with WebKit
1479         before JavaScriptCore moved out of WebKit.
1480
1481         * WebCore.xcodeproj/project.pbxproj:
1482
1483 2006-07-18  Mitz Pettel  <opendarwin.org@mitzpettel.com>
1484
1485         Reviewed by Justin.
1486
1487         - fix http://bugzilla.opendarwin.org/show_bug.cgi?id=9978
1488           REGRESSION (r12949-r12988): Clicking the first letter on a line places the caret at the end of the previous line
1489
1490         Test: editing/selection/click-start-of-line.html
1491
1492         * rendering/RenderText.cpp:
1493         (WebCore::RenderText::positionForCoordinates): Changed to return downstream
1494         affinity if the x coordinate is to the left of the middle of the first character
1495         in the text box.
1496
1497 2006-07-18  Anders Carlsson  <acarlsson@apple.com>
1498
1499         Reviewed by Darin.
1500
1501         http://bugzilla.opendarwin.org/show_bug.cgi?id=9695
1502         <rdar://problem/4614085>
1503         TOT REGRESSION: NativeTextArea: Text area does not respond to Cmd-UpArrow/Cmd-DownArrow (9695)
1504
1505         * editing/SelectionController.cpp:
1506         (WebCore::SelectionController::modifyExtendingRightForward):
1507         (WebCore::SelectionController::modifyMovingRightForward):
1508         (WebCore::SelectionController::modifyExtendingLeftBackward):
1509         (WebCore::SelectionController::modifyMovingLeftBackward):
1510         If the caret is inside an editable region, the beginning/end of the document should actually be the
1511         beginning/end of the editable region.
1512
1513 2006-07-17  Tim Omernick  <timo@apple.com>
1514
1515         Reviewed by Maciej.
1516
1517         <rdar://problem/4612079> need a way to prevent pages from scrolling to reveal elements that are focused
1518         by script
1519
1520         * bridge/mac/WebCoreFrameBridge.h:
1521         * bridge/mac/WebCoreFrameBridge.mm:
1522         (-[WebCoreFrameBridge setProhibitsScrolling:]):
1523         * page/Frame.cpp:
1524         (WebCore::Frame::prohibitsScrolling):
1525         (WebCore::Frame::setProhibitsScrolling):
1526         * page/Frame.h:
1527         * page/FramePrivate.h:
1528         (WebCore::FramePrivate::FramePrivate):
1529         * page/FrameView.cpp:
1530         (WebCore::FrameView::scrollPointRecursively):
1531         (WebCore::FrameView::setContentsPos):
1532         * page/FrameView.h:
1533         * platform/ScrollView.h:
1534
1535 2006-07-17  Maciej Stachowiak  <mjs@apple.com>
1536
1537         Reviewed by Dave Harrison.
1538
1539         - fixed <rdar://problem/4618089> Blot crashes when I paste in all the contents of http://www.apple.com/startpage/
1540
1541         * editing/htmlediting.cpp:
1542         (WebCore::editingIgnoresContent):
1543
1544 2006-07-17  Justin Garcia  <justin.garcia@apple.com>
1545
1546         Reviewed by levi
1547
1548         <rdar://problem/4621728>
1549         REGRESSION: Selecting by dragging down creates selection in wrong direction, with certain steps
1550         <rdar://problem/4604932>
1551         REGRESSION: Dragging too far left on a line will select all lines above it.
1552
1553         No layout test because of 9980.
1554
1555         * rendering/RenderBlock.cpp:
1556         (WebCore::RenderBlock::positionForCoordinates): Don't return positions inside editable roots
1557         for coordinates outside those roots, except for coordinates outside a document that is entirely
1558         editable.
1559
1560 2006-07-17  Timothy Hatcher  <timothy@apple.com>
1561
1562         Reviewed by Darin.
1563
1564         <rdar://problem/4635281> JSCanvasRenderingContext2D::drawImage needs to initialize the exception code to zero
1565
1566         * bindings/js/JSCanvasRenderingContext2DCustom.cpp:
1567         (WebCore::JSCanvasRenderingContext2D::drawImage): initialize ec to zero
1568         * html/CanvasRenderingContext2D.cpp:
1569         (WebCore::CanvasRenderingContext2D::drawImage): initialize ec before earlier return
1570
1571 2006-07-17  Timothy Hatcher  <timothy@apple.com>
1572
1573         Reviewed by Maciej.
1574
1575         <rdar://problem/4634874> WebScriptObject and WebUndefined are no longer defined by WebKit
1576
1577         Moves WebScriptObject and WebUndefined to WebCore.
1578
1579         * WebCore.exp:
1580         * WebCore.xcodeproj/project.pbxproj:
1581         * bindings/objc/DOM.mm:
1582         * bindings/objc/DOMCore.h:
1583         * bindings/objc/DOMInternal.mm:
1584         * bindings/objc/DOMUtility.mm:
1585         * bindings/objc/WebScriptObject.mm: Added.
1586         (+[WebUndefined allocWithZone:]):
1587         (-[WebUndefined initWithCoder:]):
1588         (-[WebUndefined encodeWithCoder:]):
1589         (-[WebUndefined copyWithZone:]):
1590         * bindings/objc/WebScriptObjectPrivate.h: Added.
1591         * bridge/mac/FrameMac.mm:
1592         * bridge/mac/WebCoreScriptDebugger.mm:
1593
1594 2006-07-17  David Hyatt  <hyatt@apple.com>
1595
1596         Back out the fix for 5564, since it turns out font:x-small; is a pretty
1597         prominent IE-specific CSS hack.  Because Web sites rely on IE's incorrect
1598         font parsing as a means of also correcting IE's incorrect font size rules.
1599
1600         This fixes Yahoo.com.
1601
1602         Reviewed by darin
1603
1604         * css/cssparser.cpp:
1605         (WebCore::CSSParser::parseFont):
1606
1607 2006-07-17  Justin Garcia  <justin.garcia@apple.com>
1608
1609         Reviewed by levi
1610
1611         <rdar://problem/4618389> REGRESSION: After applying a link to the last word of a sentence, a new selection is created at the start of the sentence
1612
1613         * dom/Document.cpp:
1614         (WebCore::Document::updateSelection):
1615
1616 2006-07-17  Maciej Stachowiak  <mjs@apple.com>
1617
1618         Reviewed by Beth.
1619
1620         <rdar://problem/4604946> REGRESSION: 'checkboxRef.checked = true' fires onchange event in Leopard
1621
1622         * html/HTMLInputElement.cpp:
1623         (WebCore::HTMLInputElement::setChecked):
1624         (WebCore::HTMLInputElement::preDispatchEventHandler):
1625         * html/HTMLInputElement.h:
1626
1627 2006-07-16  David Kilzer  <ddkilzer@kilzer.net>
1628
1629         Reviewed by Darin.
1630
1631         - fix for http://bugzilla.opendarwin.org/show_bug.cgi?id=9954
1632           REGRESSION: document.dir should return empty string in <head>
1633
1634         Test: fast/dom/document-dir-property.html
1635
1636         * bindings/js/kjs_html.cpp:
1637         (KJS::JSHTMLDocument::getValueProperty): Return an empty string instead of an
1638         undefined value for an unset document.dir property.
1639
1640 2006-07-16  David Kilzer  <ddkilzer@kilzer.net>
1641
1642         Reviewed by Darin.
1643
1644         * dom/NodeFilter.idl:
1645         Removed "No newline at end of file" that snuck in as part of the "apply patch" process.
1646         * manual-tests/input-empty-on-focus.html:
1647         Removed "Property changes" that snuck in as part of the "apply patch" process.
1648
1649 2006-07-16  Darin Adler  <darin@apple.com>
1650
1651         Reviewed by John Sullivan.
1652
1653         - fix http://bugzilla.opendarwin.org/show_bug.cgi?id=9943
1654           <rdar://problem/4590613>
1655           REGRESSION (Tiger-TOT): menus are offset to the right at http://movies.aol.com/movie-photo-bts/superman-returns
1656
1657         The problem turns out to be the fact that body.offsetLeft is returning a non-zero value.
1658         I have no idea why the menus worked in Tiger Safari, because body.offsetLeft behavior has not changed.
1659         Presumably some other bug that we fixed had been compensating.
1660
1661         Test: fast/html/body-offset-properties.html
1662
1663         * rendering/RenderObject.cpp:
1664         (WebCore::RenderObject::offsetLeft): Always return 0 for any object without an offsetParent.
1665         (WebCore::RenderObject::offsetTop): Ditto.
1666         (WebCore::RenderObject::offsetParent): Return 0 for the body (treat as a special case).
1667
1668 2006-07-16  Beth Dakin  <bdakin@apple.com>
1669
1670         Reviewed by Maciej.
1671
1672         Fix for <rdar://problem/4616595> REGRESSION: Problems with world
1673         clock widget clock hand motion on 9A211 + 4604574
1674
1675         The second hand on the widget was jiggling because the rotation was
1676         messing up the use of the affine transformation matrix while
1677         rounding to pixel boundaries in device space. We are mainly
1678         concerned with rounding to pixel boundaries with the scale in mind,
1679         so this patch extracts the scale to device space from the matrix,
1680         and rounds to pixel boundaries using only the scale. This doesn't
1681         seem like it is a perfect solution, but it definitely solves the
1682         immediate problem. We will probably need to re-address what should
1683         happen to avoid pixel cracks with rotations at non-integral scale
1684         factors.
1685
1686         * platform/cg/GraphicsContextCG.cpp:
1687         (WebCore::GraphicsContext::roundToDevicePixels):
1688
1689 2006-07-15  Darin Adler  <darin@apple.com>
1690
1691         - fix mistake from fix for 8952 that was breaking layout tests
1692
1693         * editing/SelectionController.cpp: (WebCore::SelectionController::nodeWillBeRemoved):
1694         Justin was right! I changed the behavior of the function by accident. Changed it
1695         back so that it doesn't blow away the selection on the DOM side in the case where
1696         it didn't before.
1697
1698         I will write 100 times on the blackboard: "When Justin says something about editing,
1699         assume he is right."
1700
1701 2006-07-15  Darin Adler  <darin@apple.com>
1702
1703         - try to fix Windows build
1704
1705         * WebCore.vcproj/WebCore/WebCore.vcproj: Added StreamingTextDecoderICU.cpp/h.
1706
1707 2006-07-15  Darin Adler  <darin@apple.com>
1708
1709         Reviewed by John Sullivan.
1710
1711         - fix http://bugzilla.opendarwin.org/show_bug.cgi?id=8952
1712           <rdar://problem/4575185>
1713           REGRESSION: crash on drag of highlighted Google custom home page modules
1714
1715         Test: fast/dynamic/move-node-with-selection.html
1716
1717         * editing/SelectionController.cpp: (WebCore::SelectionController::nodeWillBeRemoved):
1718         Call updateRendering before calling clearSelection(), since it's important to do any
1719         work beforehand, and there are calls inside clearSelection that will indirectly do an
1720         updateRendering. Also change code to make fewer assumptions about object lifetime.
1721
1722 2006-07-15  Darin Adler  <darin@apple.com>
1723
1724         Reviewed by John Sullivan.
1725
1726         - fix http://bugzilla.opendarwin.org/show_bug.cgi?id=8587
1727           <rdar://problem/4631844>
1728           REGRESSION: {display:list-item} on items outside an ol or ul element no longer causes incremental numbering
1729
1730         Test: fast/lists/numeric-markers-outside-list.html
1731
1732         * rendering/RenderListItem.cpp: (WebCore::previousListItem):
1733         Look for list items, even when we're outside any list.
1734
1735 2006-07-15  Adele Peterson  <adele@apple.com>
1736
1737         Reviewed by John.
1738
1739         - Fix for <rdar://problem/4593970> REGRESSION (NativeTextField): autofill menu disappears after typing a 2nd character
1740
1741         This was caused by composite editing commands, (like typing or pasting) that include a DeleteCommand, being
1742         interpreted by the form delegate as an actual delete.  This fix doesn't notify the form delegate if the deletion is
1743         part of an editing command to replace the selected text.
1744
1745         * editing/DeleteSelectionCommand.h: Added m_replace to keep track of whether this deletion is
1746           part of a composite command to replace the text being deleted.
1747         * editing/DeleteSelectionCommand.cpp:
1748         (WebCore::DeleteSelectionCommand::DeleteSelectionCommand): Initialize m_replace.
1749         (WebCore::DeleteSelectionCommand::doApply): Only notify the form delegate of the deletion if the text is not being replaced.
1750         * editing/CompositeEditCommand.h:
1751         * editing/CompositeEditCommand.cpp: (WebCore::CompositeEditCommand::deleteSelection):
1752           Added replace argument to construct DeleteSelectionCommand.
1753
1754         * editing/InsertTextCommand.cpp: (WebCore::InsertTextCommand::input): Call deleteSelection with replace argument.
1755         * editing/ReplaceSelectionCommand.cpp: (WebCore::ReplaceSelectionCommand::doApply): ditto.
1756
1757 2006-07-15  Geoffrey Garen  <ggaren@apple.com>
1758
1759         Reviewed by Maciej, tweaked to match Darin's patch.
1760
1761         - Fixed <rdar://problem/4631837> REGRESSION: Reproducible crash on
1762         FCKeditor demo (9911)
1763
1764         * WebCore.xcodeproj/project.pbxproj:
1765         * bindings/js/kjs_html.cpp:
1766         (KJS::JSHTMLDocument::putValueProperty): Check for a null body element before
1767         forwarding the put request to it. I confirmed that no other parts of the
1768         file use 'body' or 'bodyElement' without checking for null.
1769
1770         Also, use 'body' rather than 'bodyElement' because dir can bet set on
1771         any element, not just HTMLBodyElement.
1772
1773 2006-07-15  Darin Adler  <darin@apple.com>
1774
1775         Reviewed by Adele.
1776
1777         - fix http://bugzilla.opendarwin.org/show_bug.cgi?id=8871
1778           <rdar://problem/4575417>
1779           REGRESSION: Pressing Enter/Return in a text input removes the selected text
1780
1781         - fix http://bugzilla.opendarwin.org/show_bug.cgi?id=9743
1782           <rdar://problem/4614228>
1783           REGRESSION: crash dispatching JavaScript-created keyboard event to input element
1784
1785         * bridge/mac/FrameMac.mm: (WebCore::FrameMac::doTextFieldCommandFromEvent):
1786         Add a null check to fix the crash, and a FIXME explaining why this is not necessarily
1787         enough for the future.
1788
1789         * html/HTMLInputElement.cpp: (WebCore::HTMLInputElement::defaultEventHandler):
1790         Consume Enter key keypress events even if the element is not in a form.
1791
1792         * platform/mac/KeyEventMac.mm: (WebCore::keyIdentifierForKeyEvent): Added \n to the
1793         characters that turn into "Enter". Actual keyboard events always are \r or \003 on
1794         the Macintosh, but in layout tests we can use \n, and everything other than the
1795         code path here works, so worth fixing.
1796
1797 2006-07-15  Darin Adler  <darin@apple.com>
1798
1799         Reviewed by John Sullivan.
1800
1801         - fix http://bugzilla.opendarwin.org/show_bug.cgi?id=9883
1802           <rdar://problem/4631821>
1803           REGRESSION: NativePopups don't work correctly in some forum software
1804
1805         Test: fast/forms/select-selected.html
1806
1807         * rendering/RenderMenuList.h:
1808         * rendering/RenderMenuList.cpp:
1809         (WebCore::RenderMenuList::updateFromElement): Added code to map the selected option
1810         index to a list index before using it to index into the list.
1811         (WebCore::RenderMenuList::showPopup): Added code to map the selection option index
1812         to a list index before passing it to the menu renderer.
1813         (WebCore::RenderMenuList::valueChanged): Added code to map the list index back to a
1814         option index before calling setSelectedIndex (that function takes an option index).
1815
1816 2006-07-15  Darin Adler  <darin@apple.com>
1817
1818         Reviewed by John Sullivan.
1819
1820         - fix http://bugzilla.opendarwin.org/show_bug.cgi?id=9928
1821           REGRESSION: Text Encoding menu inoperative (after gcc protocol build fix)
1822
1823         * bindings/objc/DOM.mm:
1824         (-[DOMNode addEventListener:::]): Moved into DOMEventTarget category.
1825         (-[DOMNode removeEventListener:::]): Ditto.
1826         (-[DOMNode dispatchEvent:]): Ditto.
1827
1828         * WebCore.xcodeproj/project.pbxproj: Allow Xcode to do its thing.
1829
1830 2006-07-15  Maciej Stachowiak  <mjs@apple.com>
1831
1832         Reviewed by Anders.
1833
1834         <rdar://problem/4632144> REGRESSION: table column tests failing as a result of very recent fix
1835
1836         * rendering/RenderTable.cpp:
1837         (WebCore::RenderTable::addChild): columns and colgroups are valid children of a table too, not
1838         just table sections!
1839
1840 2006-07-15  Anders Carlsson  <acarlsson@apple.com>
1841
1842         Reviewed by Darin.
1843
1844         http://bugzilla.opendarwin.org/show_bug.cgi?id=9866
1845         <rdar://problem/4631561>
1846         REGRESSION: Repro crash from mangleme using iframe, only from server.
1847
1848         * html/HTMLIFrameElement.cpp:
1849         (WebCore::HTMLIFrameElement::attach):
1850         Add null check for contentFrame() since content frames won't be created for invalid URLs.
1851
1852 2006-07-14  Alexey Proskuryakov  <ap@nypop.com>
1853
1854         Reviewed by Darin.
1855
1856         Fix http://bugzilla.opendarwin.org/show_bug.cgi?id=4195
1857         REGRESSION: KOI8-U encoding no longer supported.
1858
1859         Tests:
1860         * http/tests/misc/BOM-override.pl
1861         * http/tests/misc/BOM-override-script.html
1862         * fast/encoding/charset-koi8-u.html
1863         * fast/encoding/charset-x-nextstep.html
1864
1865         Restored a TEC code path for encodings that are not supported by ICU (but which currently
1866         passes all layout tests even by itself with ICU disabled). A lot of refactoring is
1867         still needed - most importantly, round-tripping encoding names via CFStringEncoding
1868         makes little sense now.
1869
1870         * WebCore.exp:
1871         * bridge/mac/WebCoreTextDecoder.h: Removed.
1872         * bridge/mac/WebCoreTextDecoder.mm: Removed.
1873         WebCoreTextDecoder was not used anywhere since WebTextView was moved into WebCore.
1874
1875         * loader/Decoder.cpp:
1876         (Decoder::Decoder):
1877         (Decoder::setEncodingName):
1878         (Decoder::decode):
1879         Use StreamingTextDecoder::create().
1880
1881         * platform/StreamingTextDecoder.cpp:
1882         (WebCore::StreamingTextDecoder::create):
1883         (WebCore::StreamingTextDecoder::~StreamingTextDecoder):
1884         * platform/StreamingTextDecoder.h:
1885         StreamingTextDecoder is just an abstract interface to implementations now.
1886
1887         * platform/StreamingTextDecoderICU.cpp: Added.
1888         (WebCore::StreamingTextDecoderICU::StreamingTextDecoderICU):
1889         (WebCore::StreamingTextDecoderICU::~StreamingTextDecoderICU):
1890         (WebCore::StreamingTextDecoderICU::releaseICUConverter):
1891         (WebCore::StreamingTextDecoderICU::textEncodingSupported):
1892         (WebCore::StreamingTextDecoderICU::convertUTF16):
1893         (WebCore::StreamingTextDecoderICU::convertIfASCII):
1894         (WebCore::StreamingTextDecoderICU::createICUConverter):
1895         (WebCore::StreamingTextDecoderICU::appendOmittingBOM):
1896         (WebCore::StreamingTextDecoderICU::convertUsingICU):
1897         (WebCore::StreamingTextDecoderICU::convert):
1898         (WebCore::StreamingTextDecoderICU::toUnicode):
1899         (WebCore::StreamingTextDecoderICU::fromUnicode):
1900         * platform/StreamingTextDecoderICU.h: Added.
1901         Renamed from StreamingTextDecoder; added a way to tell whether the encoding is actually
1902         supported by the decoder; minor cleanup.
1903
1904         * platform/TextEncoding.cpp:
1905         (WebCore::TextEncoding::effectiveEncoding): Moved from StreamingTextDecoder.
1906         (WebCore::TextEncoding::toUnicode): Use StreamingTextDecoder::create().
1907         (WebCore::TextEncoding::fromUnicode): Moved to StreamingTextDecoderICU.
1908
1909         * platform/TextEncoding.h: Changed __APPLE__ to PLATFORM(MAC); added effectiveEncoding().
1910
1911         * platform/mac/StreamingTextDecoderMac.cpp: Added.
1912         (WebCore::StreamingTextDecoderMac::StreamingTextDecoderMac):
1913         (WebCore::StreamingTextDecoderMac::~StreamingTextDecoderMac):
1914         (WebCore::StreamingTextDecoderMac::releaseTECConverter):
1915         (WebCore::StreamingTextDecoderMac::textEncodingSupported):
1916         (WebCore::StreamingTextDecoderMac::convertUTF16):
1917         (WebCore::StreamingTextDecoderMac::convertIfASCII):
1918         (WebCore::StreamingTextDecoderMac::createTECConverter):
1919         (WebCore::StreamingTextDecoderMac::appendOmittingBOM):
1920         (WebCore::StreamingTextDecoderMac::convertOneChunkUsingTEC):
1921         (WebCore::StreamingTextDecoderMac::convertUsingTEC):
1922         (WebCore::StreamingTextDecoderMac::convert):
1923         (WebCore::StreamingTextDecoderMac::toUnicode):
1924         (WebCore::StreamingTextDecoderMac::fromUnicode):
1925         * platform/mac/StreamingTextDecoderMac.h: Added.
1926         (WebCore::StreamingTextDecoderMac::convert):
1927         This is a TEC+CFString code path for decoding, basically restored from a year-old revision.
1928
1929         * platform/mac/TextEncodingMac.cpp: Removed. Code moved to StreamingTextDecoderMac.
1930
1931         * WebCore.xcodeproj/project.pbxproj:
1932
1933 === Safari-521.17 ===
1934
1935 2006-07-14  Timothy Hatcher  <timothy@apple.com>
1936
1937         <rdar://problem/4623957> SWB: gcc-5412 (new?) objc warning causes WebCore project failure
1938
1939         Made a DOMNode category in DOMEvents.h that lets DOMNode conform to the
1940         DOMEventTarget protocol that works with GCC 5412.
1941
1942         Removed forward declarations of WebCoreWidgetHolder and imported WebCoreWidgetHolder.h.
1943
1944         Removed all <Cocoa/Cocoa.h> and <Foundation/Foundation.h> imports in other headers,
1945         we import <Cocoa/Cocoa.h> in the prefix header for ObjC.
1946
1947         * WebCore.xcodeproj/project.pbxproj:
1948         * bindings/objc/DOMCore.h:
1949         * bindings/objc/DOMEvents.h:
1950         * bridge/mac/FormDataMac.h:
1951         * bridge/mac/WebCoreAXObject.h:
1952         * bridge/mac/WebCoreCache.h:
1953         * bridge/mac/WebCoreEncodings.h:
1954         * bridge/mac/WebCoreFrameBridge.h:
1955         * bridge/mac/WebCoreFrameNamespaces.h:
1956         * bridge/mac/WebCoreJavaScript.h:
1957         * bridge/mac/WebCorePageBridge.h:
1958         * bridge/mac/WebCorePageState.h:
1959         * bridge/mac/WebCoreResourceLoader.h:
1960         * bridge/mac/WebCoreScriptDebugger.h:
1961         * bridge/mac/WebCoreSettings.h:
1962         * bridge/mac/WebCoreStringTruncator.h:
1963         * bridge/mac/WebCoreStringTruncator.mm:
1964         * bridge/mac/WebCoreTextDecoder.h:
1965         * bridge/mac/WebDashboardRegion.h:
1966         * kcanvas/device/quartz/KRenderingDeviceQuartz.mm:
1967         * platform/mac/CookieJar.mm:
1968         * platform/mac/DeprecatedStringMac.mm:
1969         * platform/mac/FontDataMac.mm:
1970         * platform/mac/FoundationExtras.h:
1971         * platform/mac/TextBoundaries.mm:
1972         * platform/mac/WebCoreHistory.h:
1973         * platform/mac/WebCoreHistory.m:
1974         * platform/mac/WebCoreKeyGenerator.h:
1975         * platform/mac/WebCoreTextArea.h:
1976         * platform/mac/WebCoreTextField.h:
1977         * platform/mac/WebCoreTextRenderer.h:
1978         * platform/mac/WebCoreView.h:
1979         * platform/mac/WebCoreWidgetHolder.h:
1980
1981 2006-07-14  Geoffrey Garen  <ggaren@apple.com>
1982
1983         Reviewed by Maciej.
1984
1985         - Updated ScriptInterpreter to work with Interpreter ref-counting in
1986         JavaScriptCore.
1987
1988         (KJS::ScriptInterpreter::~ScriptInterpreter): Now protected to catch
1989         manual delete.
1990
1991 2006-07-14  Anders Carlsson  <acarlsson@apple.com>
1992
1993         Reviewed by Adele and Justin.
1994
1995         http://bugzilla.opendarwin.org/show_bug.cgi?id=9658
1996         <rdar://problem/4613948>
1997         REGRESSION: Check Spelling does not work in textarea elements
1998
1999         * bridge/mac/FrameMac.mm:
2000         (WebCore::FrameMac::advanceToNextMisspelling):
2001         Don't use setStartBefore or setEndAfter on the search range because for shadow trees, there
2002         is no notion of before and after. Instead, use setStart and setEnd and pass in the start and end
2003         indices respectively.
2004
2005         * dom/Range.cpp:
2006         (WebCore::Range::checkNodeBA):
2007         Allow range operations on shadow trees.
2008
2009         * manual-tests/form-element-spelling.html: Added.
2010
2011 2006-07-14  Beth Dakin  <bdakin@apple.com>
2012
2013         Reviewed by Hyatt.
2014
2015         Fix for <rdar://problem/4621660> REGRESSION: Safari crashing in
2016         WebCore::FrameView::updateOverflowStatus
2017
2018         m_viewportRenderer is never initialized for framesets, and it
2019         shouldn't be. So we just need to nil-check for it in
2020         updateOverflowStatus() and return early.
2021
2022         * page/FrameView.cpp:
2023         (WebCore::FrameView::updateOverflowStatus): Nil check.
2024
2025 2006-07-14  Adele Peterson  <adele@apple.com>
2026
2027         RS by Darin.
2028
2029         Backing out fix for <rdar://problem/4604703>
2030         REGRESSION (NativeTextField): Focus is not removed from password field after I ctrl-click into a different field
2031
2032         Darin had a better fix in WebKit for this.
2033
2034         * bridge/mac/FrameMac.mm: (WebCore::FrameMac::sendContextMenuEvent):
2035
2036 2006-07-14  Adele Peterson  <adele@apple.com>
2037
2038         Reviewed by Darin.
2039
2040         - Fix for <rdar://problem/4614054> REGRESSION: Safari submits forms when the Return key is hit to complete inline inputs
2041
2042         * page/Frame.h: (WebCore::Frame::inputManagerHasMarkedText): Added.
2043         * bridge/mac/FrameMac.h: ditto.
2044         * bridge/mac/FrameMac.mm: (WebCore::FrameMac::inputManagerHasMarkedText): Added. Asks the input manager if there's marked text.
2045         * html/HTMLInputElement.cpp: (WebCore::HTMLInputElement::defaultEventHandler): For text fields, don't submit the form on Enter
2046         if the input manager says there's marked text.  I added this code for all text field paths.  For widgets, WebCoreTextField.mm
2047         has code to deal with this case.  But as we convert search, password, and isindex, they will need to do this check too.
2048
2049 2006-07-14  Mitz Pettel  <opendarwin.org@mitzpettel.com>
2050
2051         Reviewed by Darin.
2052
2053         - fix http://bugzilla.opendarwin.org/show_bug.cgi?id=9907
2054           REGRESSION (r15418): editing/pasteboard/paste-table-001 fails in pixel  mode
2055
2056         * rendering/RenderText.cpp:
2057         (WebCore::RenderText::caretRect): Fixed the calculation of the max/min allowed caret
2058         position.
2059
2060 2006-07-14  Mitz Pettel  <opendarwin.org@mitzpettel.com>
2061
2062         Reviewed by Darin.
2063
2064         - fix http://bugzilla.opendarwin.org/show_bug.cgi?id=9905
2065           REGRESSION (r15404-r15415): Repro crash when pressing delete in an empty editable div
2066
2067         Test: editing/deleting/delete-at-start-or-end.html
2068
2069         * editing/TypingCommand.cpp:
2070         (WebCore::TypingCommand::deleteKeyPressed): Added null check.
2071         (WebCore::TypingCommand::forwardDeleteKeyPressed): Ditto.
2072
2073 2006-07-14  David Kilzer  <ddkilzer@kilzer.net>
2074
2075         Reviewed by Geoffrey.
2076
2077         - fix http://bugzilla.opendarwin.org/show_bug.cgi?id=9903
2078           Simplify logic in JSHTMLOptionsCollection::setLength() by using no-arg getNumber()
2079
2080         No test cases since there is no change in functionality.
2081
2082         * bindings/js/JSHTMLOptionsCollectionCustom.cpp:
2083         (WebCore::JSHTMLOptionsCollection::setLength):  Simplified logic by using the
2084         no-argument getNumber() method.
2085
2086 2006-07-14  Geoffrey Garen  <ggaren@apple.com>
2087
2088         Reviewed by Adele.
2089
2090         - Build fix: don't need BLOCK_OBJC_EXCEPTIONS because we already have
2091         one surrounding this function, and the nested one makes the compiler
2092         think our local variable is volatile (seems like a compiler bug to me).
2093
2094         * bridge/mac/FrameMac.mm:
2095         (WebCore::FrameMac::sendContextMenuEvent):
2096
2097 2006-07-14  Timothy Hatcher  <timothy@apple.com>
2098
2099         Reviewed by Maciej.
2100
2101         Make JavaScriptCore a public framework. Adjusted the paths.
2102
2103         * WebCore.xcodeproj/project.pbxproj:
2104
2105 2006-07-14  Maciej Stachowiak  <mjs@apple.com>
2106
2107         Reviewed by Hyatt.
2108
2109         - fixed <rdar://problem/4567031> REGRESSION: Crash at WebCore::RenderBlock::createLineBoxes (seems to be a security hole?)
2110
2111         I also fixed all the similar crash / assertion failure cases I could think of.
2112
2113         * dom/Node.cpp:
2114         (WebCore::Node::nextRendererWithSameParent): Helper function for some of the above.
2115         * dom/Node.h:
2116         * rendering/RenderTable.cpp:
2117         (WebCore::RenderTable::addChild): Don't allow inserting forms when not in
2118         an HTML document, since we don't need that quirk and because parsing won't
2119         do certain render tree fixups. Also watch out for case when inserting before
2120         the renderer of a misnested child.
2121         * rendering/RenderTableRow.cpp:
2122         (WebCore::RenderTableRow::addChild): ditto
2123         * rendering/RenderTableSection.cpp:
2124         (WebCore::RenderTableSection::addChild): ditto
2125
2126 2006-07-13  Adele Peterson  <adele@apple.com>
2127
2128         Reviewed by Maciej.
2129
2130         - Fix for <rdar://problem/4604703>
2131         REGRESSION (NativeTextField): Focus is not removed from password field after I ctrl-click into a different field
2132
2133         Test:
2134         * manual-tests/password-ctrl-click-lose-focus.html: Added.
2135
2136         * bridge/mac/FrameMac.mm: (WebCore::FrameMac::sendContextMenuEvent):
2137         If we're about to set a selection in the current view, make sure its the first responder.
2138         In this case, this will cause the password field to resign first responder at the right time.
2139
2140 2006-07-13  Mitz Pettel  <opendarwin.org@mitzpettel.com>
2141
2142         Reviewed by Hyatt.
2143
2144         - fix http://bugzilla.opendarwin.org/show_bug.cgi?id=9670
2145           REGRESSION: RTL white-space:pre-wrap text is offset to the right
2146
2147         Test: fast/text/international/rtl-white-space-pre-wrap.html
2148
2149         * rendering/RenderText.cpp:
2150         (WebCore::RenderText::caretRect): Fixed LTR case and added the RTL case
2151         of clipping the caret position to the text box when the caret is after
2152         the trailing space of an autowrapped line.
2153         * rendering/bidi.cpp:
2154         (WebCore::RenderBlock::computeHorizontalPositionsForLine): Changed to truncate-
2155         to-fit the logically last text box if it contains the trailing spaces of an
2156         autowrapped line.
2157         (WebCore::RenderBlock::bidiReorderLine): Remember the logically last text run.
2158         In the case of autowrapped text with white space that overflows beyond the line,
2159         the last text run is the one containing the overflowing white space.
2160         (WebCore::RenderBlock::findNextLineBreak): Split overflowing white space on a
2161         line that autowraps only after white space into a separate text run.
2162
2163 2006-07-13  David Harrison  <harrison@apple.com>
2164
2165         Reviewed by Dave Hyatt.
2166
2167         <rdar://problem/4624203> -webkit-highlight should be behind images
2168
2169         * rendering/RenderImage.cpp:
2170         (WebCore::RenderImage::paint):
2171         * rendering/RenderListMarker.cpp:
2172         (WebCore::RenderListMarker::paint):
2173         * rendering/RenderWidget.cpp:
2174         (WebCore::RenderWidget::paint):
2175         Call custom highlighter before painting the image, marker, or widget.
2176
2177 2006-07-13  David Harrison  <harrison@apple.com>
2178
2179         Reviewed by Justin and Levi.
2180
2181         <rdar://problem/4620743> REGRESSION: Option-Delete doesn't delete words during typing
2182
2183         * Tests:
2184         editing/deleting/delete-by-word-001.html
2185         editing/deleting/delete-by-word-002.html
2186
2187         * editing/TypingCommand.cpp:
2188         (WebCore::TypingCommand::deleteKeyPressed):
2189         (WebCore::TypingCommand::forwardDeleteKeyPressed):
2190         (WebCore::TypingCommand::doApply):
2191         * editing/TypingCommand.h:
2192         Delete and forward delete to use specified granularity.
2193         Undo of delete and forward delete to select what had been deleted (non-char granularity only).
2194
2195         * page/Frame.cpp:
2196         (WebCore::Frame::setSelection):
2197         Close typing and end style even if selection is not changing.
2198         * page/Frame.h:
2199         Remove unused setSelection parameter keepTypingStyle.
2200
2201 2006-07-12  Levi Weintraub  <lweintraub@apple.com>
2202
2203         Reviewed by justin
2204
2205         <rdar://problem/4622763> Deleting from beginning of paragraph following a table deletes rather than selects the table
2206
2207         * editing/DeleteSelectionCommand.cpp:
2208         (WebCore::DeleteSelectionCommand::initializeStartEnd): Added selection expansion for HRs.
2209         (WebCore::DeleteSelectionCommand::initializePositionData): Now uses enclosingBlockFlowOrTableElement
2210         instead of enclosingBlockFlowElement.
2211         (WebCore::DeleteSelectionCommand::removeNode): Use to identify that we need a placeholder
2212         when the start or end block is removed.
2213         (WebCore::DeleteSelectionCommand::handleGeneralDelete): Added check for canHaveChildrenForEditing
2214         to keep things like HRs from being given children.
2215         (WebCore::DeleteSelectionCommand::doApply): Switched to use member variable for needPlaceholder.
2216         * editing/DeleteSelectionCommand.h: Made needPlaceholder a member variable.
2217         * editing/TypingCommand.cpp:
2218         (WebCore::TypingCommand::deleteKeyPressed): Fixed selection bug for tables and open typing commands.
2219         (WebCore::TypingCommand::forwardDeleteKeyPressed): Ditto.
2220         * editing/htmlediting.cpp:
2221         (WebCore::editingIgnoresContent): Added check for HRs, since it's not considered a widget.
2222         * editing/visible_units.cpp:
2223         (WebCore::startOfParagraph): Fix for HRs and tables.
2224         (WebCore::endOfParagraph): Ditto.
2225
2226 2006-07-13  Alexey Proskuryakov  <ap@nypop.com>
2227
2228         Reviewed by Darin.
2229
2230         - fix http://bugzilla.opendarwin.org/show_bug.cgi?id=9880
2231           Memory leaks running DOM-Hanoi
2232
2233         No change in behavior, thus no test included.
2234
2235         * bridge/mac/FrameMac.mm:
2236         (WebCore::FrameMac::setStatusBarText): Use a local autorelease pool to release the temporaries -
2237         the test runs non-stop, and the enclosing pool doesn't get a chance to be drained.
2238
2239 2006-07-12  Mitz Pettel  <opendarwin.org@mitzpettel.com>
2240
2241         Reviewed by Adele.
2242
2243         - fix http://bugzilla.opendarwin.org/show_bug.cgi?id=9862
2244           REGRESSION: GMail: Crash in RenderView::repaintViewRectangle when spoofing as FF
2245         - see also <rdar://problem/4622407>
2246
2247         Test: fast/frames/repaint-display-none-crash.html
2248
2249         * rendering/RenderView.cpp:
2250         (WebCore::RenderView::repaintViewRectangle): Added null checking of the owner element's
2251         renderer, which can be null if the iframe is set to display:none.
2252
2253 2006-07-12  Justin Garcia  <justin.garcia@apple.com>
2254
2255         Reviewed by levi
2256
2257         <rdar://problem/4509393>
2258         selected DOM range starts with <object>, 0 offset but selection should include the <object>
2259
2260         * editing/ReplaceSelectionCommand.cpp:
2261         (WebCore::ReplaceSelectionCommand::shouldMergeStart):
2262         Don't pull content out of a table cell.
2263         * editing/htmlediting.cpp:
2264         (WebCore::editingIgnoresContent): Added <select> nodes.
2265         (WebCore::rangeCompliantEquivalent): Convert [node, 0] positions to positionBeforeNode(node)
2266         for more types of nodes.
2267         * rendering/RenderContainer.cpp:
2268         (WebCore::RenderContainer::positionForCoordinates):
2269         Fix a crasher when right clicking on an anonymous table.
2270
2271 2006-07-12  John Sullivan  <sullivan@apple.com>
2272
2273         Reviewed by Brady Eidson
2274
2275         - fixed <rdar://problem/4611164> REGRESSION: Crash occurs when undoing a series of
2276         misspelled words (WebCore::RenderObject::repaint(bool)
2277
2278         * dom/Document.cpp:
2279         (WebCore::Document::removeMarkers):
2280         put (it - markers.begin()) in a local variable before altering markers, in every case where this was
2281         happening. One of the cases like this was fixed a while back, but other cases were
2282         either missed at that time or crept in since.
2283
2284 2006-07-13  Anders Carlsson  <acarlsson@apple.com>
2285
2286         Reviewed by Justin.
2287
2288         http://bugzilla.opendarwin.org/show_bug.cgi?id=9663
2289         REGRESSION (r14948-r14956): Selection in text field remains highlighted when the text field loses focus
2290
2291         * dom/Document.cpp:
2292         (WebCore::Document::updateSelection):
2293         Don't return early if the selection is empty.
2294
2295 2006-07-12  Anders Carlsson  <acarlsson@apple.com>
2296
2297         Reviewed by Adele.
2298
2299         <rdar://problem/4614656> REGRESSION: onpaste() handlers don't run for textarea elements
2300
2301         * bridge/mac/FrameMac.mm:
2302         (WebCore::FrameMac::dispatchCPPEvent):
2303         If the element is a shadow node, dispatch the event to its real parent.
2304
2305         * manual-tests/textarea-onpaste.html: Added.
2306
2307 2006-07-12  Beth Dakin  <bdakin@apple.com>
2308
2309         Reviewed by Anders.
2310
2311         Fix for layout test regressions after my check-in last night. Just
2312         a silly mistake where I should have asked if we were NOT printing
2313         instead of if we were in the listbox code.
2314
2315         * WebCore.xcodeproj/project.pbxproj: Project file wars. Back to
2316         XCode 2.3
2317         * platform/mac/ListBoxMac.mm:
2318         (itemTextRenderer): Inverted check.
2319         (groupLabelTextRenderer): Inverted check.
2320
2321 2006-07-12  Darin Adler  <darin@apple.com>
2322
2323         Reviewed by Maciej.
2324
2325         - fix http://bugzilla.opendarwin.org/show_bug.cgi?id=9806
2326           <rdar://problem/4622622>
2327           REGRESSION: Large rowspan causes WebKit to call abort()
2328
2329         Test: fast/table/large-rowspan-crash.html
2330
2331         * rendering/RenderTableSection.cpp:
2332         (WebCore::RenderTableSection::ensureRows):
2333         Use sizeof(RowStruct) instead of sizeof(int).
2334         Clearly we'll need something better to solve this completely.
2335         I expect another smaller, but still huge, value will still cause a problem.
2336
2337 2006-07-12  Anders Carlsson  <acarlsson@apple.com>
2338
2339         Reviewed by Maciej.
2340
2341         <rdar://problem/4586665> REGRESSION: autorestore.apple.com: Crashes Safari in WebCore::Widget::client() const
2342
2343         * bridge/mac/FrameMac.h:
2344         Add focusCallResultedInViewBeingCreated argument.
2345
2346         * bridge/mac/FrameMac.mm:
2347         (WebCore::FrameMac::nextKeyViewInFrame):
2348         If the call to focus() caused the node to get a native widget, set focusCallResultedInViewBeingCreated to true.
2349
2350         (WebCore::FrameMac::nextKeyViewInFrameHierarchy):
2351         Don't reset the focus node if focusCallResultedInViewBeingCreated is true. Also, add magic to prevent setting
2352         a text field as the first responder if its field editor already is the current first responder.
2353
2354         * page/FrameView.cpp:
2355         (WebCore::FrameView::handleMousePressEvent):
2356         In some cases, get the event target node again after dispatching the mouse event.
2357
2358 2006-07-12  Beth Dakin  <bdakin@apple.com>
2359
2360         Reviewed by Adele.
2361
2362         Fix for <rdar://problem/4615765> Input[type='search' fields
2363         initially render too large in Widgets
2364
2365         Fundamentally, the problem here is that we were miscalculating
2366         things because [NSGraphicsContext currentContextDrawingToScreen]
2367         was evaluating to false when widgets first load. We only ever used
2368         this check, however, to determine if we were printing or not, so it
2369         should not have evaluated to false for loading in Dashbaord.
2370         Instead, if we query the RenderView about whether or not we are
2371         printing, we will get the right answer.
2372
2373         No test case added since this appears only to affect Dashboard.
2374
2375         * bridge/mac/WebCoreFrameBridge.mm:
2376         (-[WebCoreFrameBridge drawRect:]): Ask the RenderView if we are
2377         printing instead. This is not part of the actual bug fix, but it
2378         seems wise to get rid of all calls to [NSGraphicsContext
2379         currentContextDrawingToScreen] when we are just trying to determine
2380         if we are printing or not.
2381         * platform/mac/ListBoxMac.mm:
2382         (itemTextRenderer):This function now takes a boolean, isPrinting.
2383         (groupLabelTextRenderer): Same as above.
2384         (ListBox::sizeForNumberOfLines): Ask the RenderView if we are
2385         printing.
2386         (-[WebCoreTableView drawRow:clipRect:]): Same as above.
2387         * platform/mac/PopUpButtonMac.mm:
2388         (PopUpButton::sizeHint): Same as above.
2389         * platform/mac/TextFieldMac.mm:
2390         (-[NSSearchFieldCell _addStringToRecentSearches:]):
2391         * rendering/RenderLineEdit.cpp: Same as above.
2392
2393 2006-07-12  Adele Peterson  <adele@apple.com>
2394
2395         Reviewed by Maciej.
2396
2397         Fix for: http://bugzilla.opendarwin.org/show_bug.cgi?id=9813
2398         OPTION text can paint over select element's scrollbar
2399
2400         Test: fast/forms/option-text-clip.html
2401
2402         * rendering/RenderMenuList.cpp: (WebCore::RenderMenuList::paintObject):
2403         When calculating the clip to apply to the button text, also consider padding.
2404         In the future, we may have separate renderers for the text part, and the arrow part
2405         of the control, and then the separation should be natural.  For now, we use padding.
2406
2407 2006-07-11  David Kilzer  <ddkilzer@kilzer.net>
2408
2409         Windows build fix.  Reviewed by NOBODY.
2410
2411         * WebCore.vcproj/WebCore/WebCore.vcproj:  Added JSHTMLOptionsCollection.cpp/h
2412         and JSHTMLOptionsCollectionCuston.cpp.  VC++ Express realphabetized the file list.
2413
2414 2006-07-11  Darin Adler  <darin@apple.com>
2415
2416         Reviewed by Adele.
2417
2418         - fix http://bugzilla.opendarwin.org/show_bug.cgi?id=9809
2419           <rdar://problem/4619515>
2420           focus ring fails to appear on select element after choosing item from popup
2421
2422         - includes http://bugzilla.opendarwin.org/show_bug.cgi?id=9853
2423           improvements to select element, including some storage leak fixes
2424
2425         * html/HTMLOptionElement.cpp: (WebCore::HTMLOptionElement::index): Use a const
2426         reference for the list items, so we don't have to copy a vector.
2427         * html/HTMLSelectElement.cpp:
2428         (WebCore::HTMLSelectElement::selectedIndex): Ditto.
2429         (WebCore::HTMLSelectElement::setSelectedIndex): Ditto.
2430         (WebCore::HTMLSelectElement::length): Ditto.
2431         (WebCore::HTMLSelectElement::remove): Ditto.
2432         (WebCore::HTMLSelectElement::value): Ditto.
2433         (WebCore::HTMLSelectElement::setValue): Ditto.
2434         (WebCore::HTMLSelectElement::stateValue): Ditto.
2435         (WebCore::HTMLSelectElement::restoreState): Ditto.
2436         (WebCore::HTMLSelectElement::appendFormData): Ditto.
2437         (WebCore::HTMLSelectElement::optionToListIndex): Ditto.
2438         (WebCore::HTMLSelectElement::listToOptionIndex): Ditto.
2439         (WebCore::HTMLSelectElement::recalcListItems): Made const, with the appropriate
2440         fields mutable.
2441         (WebCore::HTMLSelectElement::reset): Use a const reference for
2442         the list items, so we don't have to copy the vector. Remove the call to
2443         setSelectionChanged for the RenderMenuList case.
2444         (WebCore::HTMLSelectElement::notifyOptionSelected): Ditto, on both counts.
2445         (WebCore::HTMLSelectElement::defaultEventHandler): Call focus() before showing
2446         the pop-up.
2447         * html/HTMLSelectElement.h: The RenderMenuList class is no longer a friend.
2448         Changed the listItems function to return a const reference to the vector so
2449         it no longer copies the vector. Removed the const_cast to the call to
2450         recalcListItems and changed it to a const member function. Made m_recalcListItems
2451         mutable.
2452         * rendering/DeprecatedRenderSelect.cpp:
2453         (WebCore::DeprecatedRenderSelect::updateFromElement): Removed an unnecessary call
2454         to recalcListItems, which is called automatically. Use a const reference for the
2455         list items so we don't have to copy a vector.
2456         (WebCore::DeprecatedRenderSelect::layout): Ditto.
2457         (WebCore::DeprecatedRenderSelect::selectionChanged): Ditto.
2458         (WebCore::DeprecatedRenderSelect::updateSelection): Ditto.
2459         * rendering/RenderMenuList.cpp:
2460         (WebCore::RenderMenuList::RenderMenuList): Updated for renamed data members.
2461         (WebCore::RenderMenuList::createInnerBlock): Ditto.
2462         (WebCore::RenderMenuList::addChild): Ditto.
2463         (WebCore::RenderMenuList::removeChild): Ditto.
2464         (WebCore::RenderMenuList::setStyle): Ditto. Also removed code to set the style
2465         on the pop-up menu, because it's created with the correct style and destroyed
2466         before it a style change could occur.
2467         (WebCore::RenderMenuList::updateFromElement): Rearranged code to compute the
2468         maximum width in a simpler fashion, and to not bother trying to maintain
2469         the "selected" flags on the elements, since the HTMLSelectElement class
2470         takes care of that. Store the width as an int. Call setText to set the text
2471         based on the selected element's option text.
2472         (WebCore::RenderMenuList::paintObject): Don't check m_inner when setting
2473         up the clip -- always set up the clip.
2474         (WebCore::RenderMenuList::calcMinMaxWidth): Use m_optionsWidth directly
2475         instead of calling ceilf on m_longestWidth.
2476         (WebCore::RenderMenuList::showPopup): Don't use m_popupMenu to store the
2477         menu -- instead keep the pointer in a local variable. Get the selected
2478         index from the HTMLSelectElement.
2479         (WebCore::RenderMenuList::valueChanged): Call HTMLSelectElement::setSelectedIndex
2480         to do most of the work.
2481         * rendering/RenderMenuList.h: Renamed m_inner to m_innerBlock. Removed
2482         m_popupMenu, m_size, m_selectionChanged, and m_selectedIndex. Renamed
2483         m_longestWidth to m_optionsWidth and changed it to be an int. Removed
2484         unneeded override of removeLeftoverAnonymousBoxes function. Removed
2485         unneeded selectionChanged, setSelectionChanged, updateSelection, and
2486         hasPopupMenu functions. Removed extra includes.
2487         * rendering/RenderPopupMenu.cpp: (WebCore::RenderPopupMenu::populate):
2488         Change to iterate the list items instead of iterating all children
2489         of the select node.
2490         * rendering/RenderPopupMenu.h: Renamed getRenderMenuList to menuList.
2491         * rendering/RenderPopupMenuMac.mm:
2492         (WebCore::RenderPopupMenuMac::populate): Moved code to clear and create
2493         the pop-up here from the caller. Removed an extra retain that would cause
2494         the NSPopUpButtonCell to leak.
2495         (WebCore::RenderPopupMenuMac::showPopup): Removed unnecessary code to
2496         create the pop-up, which is now in populate, and also the call to the
2497         clear function, for the same reason. Reorganized code to make it a bit
2498         more readable. Removed an unnecessary if to check if frame is nil.
2499         Used a RefPtr to make sure we don't make a call on a frame after it's
2500         deleted. As part of the reorganization fixed a problem where we'd retain
2501         the event and then return early without releasing it in one case.
2502         (WebCore::RenderPopupMenuMac::addSeparator): Tweaked a little.
2503         (WebCore::RenderPopupMenuMac::addGroupLabel): Grouped all the code to
2504         manage the NSMenu at the bottom of the function.
2505         (WebCore::RenderPopupMenuMac::addOption): Ditto.
2506
2507 2006-07-11  Justin Garcia  <justin.garcia@apple.com>
2508
2509         Reviewed by levi
2510
2511         <rdar://problem/4620686>
2512         REGRESSION: Mispelling markers are still displayed when using the delete key to place caret into a misspelled word
2513
2514         * bridge/mac/FrameMac.mm:
2515         (WebCore::FrameMac::respondToChangedSelection): Remove markers from newAdjacentWords
2516         even if oldAdjacentWords is equal to newAdjacentWords.  This happens during a deletion.
2517
2518
2519 2006-07-11  Tim Omernick  <timo@apple.com>
2520
2521         Reviewed by NOBODY (build fix)
2522
2523         * bindings/objc/DOMPrivate.h:
2524         Don't @class NPObject; that breaks files that actually #import <JavaScriptCore/npruntime.h>, since the NPObject
2525         there is of a different type.
2526         Also clarified a highly misleading comment from my last commit.  I meant to clarify the comment
2527         before landing.
2528
2529 2006-07-12  Anders Carlsson  <acarlsson@apple.com>
2530
2531         Reviewed by Geoff.
2532
2533         http://bugzilla.opendarwin.org/show_bug.cgi?id=9635
2534         REGRESSION: Crash when adding to cart at <http://www.yemeksepeti.com/>
2535
2536         * html/HTMLTokenizer.cpp:
2537         (WebCore::HTMLTokenizer::scriptHandler):
2538         Don't load external scripts if the parser is stopped.
2539
2540         * manual-tests/open-close-tokenizer-crash.html: Added.
2541         * manual-tests/resources/empty-file.js: Added.
2542         * manual-tests/resources/open-close-tokenizer-crash.html: Added.
2543         Add manual test.
2544
2545 2006-07-11  John Sullivan  <sullivan@apple.com>
2546
2547         Reviewed by Kevin, Timo O, Brady, Darin, and Hyatt (whee!)
2548
2549         - added support for creating a selection image with white text
2550
2551         * rendering/RenderObject.h:
2552         (WebCore::):
2553         add PaintRestriction enum
2554         (WebCore::RenderObject::PaintInfo::PaintInfo):
2555         add forceWhiteText boolean to PaintInfo struct
2556
2557         * page/FramePrivate.h:
2558         (WebCore::FramePrivate::FramePrivate):
2559         replaced m_selectionOnly bool with m_paintRestriction, which also handles forcing the
2560         text to white
2561
2562         * page/Frame.cpp:
2563         (WebCore::Frame::paint):
2564         updated to use & pass m_paintRestriction where it used to use & pass just m_selectionOnly
2565
2566         * bridge/mac/FrameMac.h:
2567         * bridge/mac/FrameMac.mm:
2568         (WebCore::FrameMac::selectionImage):
2569         now takes forceWhiteText parameter, stored in FramePrivate as part of PaintRestriction
2570
2571         * bridge/mac/WebCoreFrameBridge.h:
2572         * bridge/mac/WebCoreFrameBridge.mm:
2573         (-[WebCoreFrameBridge selectionImageForcingWhiteText:]):
2574         renamed from selectionImage, now takes forceWhiteText parameter, which is passed
2575         down into Frame
2576
2577         * rendering/RenderLayer.h:
2578         * rendering/RenderLayer.cpp:
2579         (WebCore::RenderLayer::paint):
2580         use a PaintRestriction value where we used to use a selectionOnly bool
2581         (WebCore::RenderLayer::paintLayer):
2582         ditto; also stores forceWhiteText in PaintInfo struct
2583
2584         * kcanvas/KCanvasResources.cpp:
2585         (WebCore::KCanvasMarker::draw):
2586         updated for changed signature of PaintInfo constructor
2587
2588         * ksvg2/svg/SVGMaskElement.cpp:
2589         (WebCore::SVGMaskElement::drawMaskerContent):
2590         ditto
2591
2592         * ksvg2/svg/SVGPatternElement.cpp:
2593         (WebCore::SVGPatternElement::drawPatternContentIntoTile):
2594         ditto
2595
2596         * rendering/InlineTextBox.cpp:
2597         (WebCore::InlineTextBox::paint):
2598         paint with white text color if forceWhiteText is set in PaintInfo struct
2599
2600 2006-07-11  Tim Omernick  <timo@apple.com>
2601
2602         Reviewed by Anders.
2603
2604         <http://bugzilla.opendarwin.org/show_bug.cgi?id=9843>:
2605         Give Netscape plug-ins access to their own DOM element
2606
2607         * html/HTMLPlugInElement.h:
2608         Just import JSC headers on Mac.  The NPObject and Bindings::Instance stuff is only used on Mac anyway.
2609         Changed __APPLE__ to PLATFORM(MAC).
2610         Added m_NPObject ivar.
2611         * html/HTMLPlugInElement.cpp:
2612         (WebCore::HTMLPlugInElement::HTMLPlugInElement):
2613         Initialize the NPObject.
2614         (WebCore::HTMLPlugInElement::~HTMLPlugInElement):
2615         Release the NPObject when the element is destroyed.
2616         (WebCore::HTMLPlugInElement::createNPObject):
2617         Creates an NPObject for the element.
2618         (WebCore::HTMLPlugInElement::getNPObject):
2619         Returns the NPObject for the element, creating and caching it if necessary.
2620
2621         * html/HTMLAppletElement.h:
2622         Changed __APPLE__ to PLATFORM(MAC).
2623         * html/HTMLAppletElement.cpp:
2624         ditto
2625         * html/HTMLEmbedElement.h:
2626         ditto
2627         * html/HTMLEmbedElement.cpp:
2628         ditto
2629         * html/HTMLObjectElement.h:
2630         ditto
2631         * html/HTMLObjectElement.cpp:
2632         ditto
2633
2634         * bindings/objc/DOMPrivate.h:
2635         Added -[DOMElement _NPObject].
2636
2637         * bindings/objc/DOM.mm:
2638         (-[DOMElement _NPObject]):
2639         Returns the NPObject for the element.  Since this is only needed by the Netscape plug-in API, you can only get
2640         the NPObject for applet, embed, and object elements.
2641
2642 2006-07-11  Adele Peterson  <adele@apple.com>
2643
2644         Reviewed by Hyatt.
2645
2646         - Fix for http://bugzilla.opendarwin.org/show_bug.cgi?id=9861
2647         REGRESSION: Can't dynamically change list box to popup
2648
2649         and updated fix for http://bugzilla.opendarwin.org/show_bug.cgi?id=9859
2650         REGRESSION: Can't dynamically change popup to list box
2651
2652         The original fix for 9859 was in the renderer, and I think we can catch the change
2653         earlier in the element to detach and reattach.
2654
2655         I also did some cleanup to remove the PopupButton code path from DeprecatedRenderSelect
2656
2657         Test: fast/forms/select-change-listbox-to-popup.html
2658
2659         * html/HTMLSelectElement.cpp:
2660         (WebCore::HTMLSelectElement::parseMappedAttribute): If the new attribute value is going to require us to change from listbox to popup or vice versa,
2661         and we're already attached, then detach and reattach to create the correct renderer.  If we're not attached, then we'll create the correct renderer
2662         when we attach.
2663         (WebCore::HTMLSelectElement::recalcStyle): Updated shouldUseMenuList since it no longer takes a RenderStyle.
2664         (WebCore::HTMLSelectElement::isKeyboardFocusable): ditto.
2665         (WebCore::HTMLSelectElement::isMouseFocusable): ditto.
2666         (WebCore::HTMLSelectElement::createRenderer): ditto.
2667         (WebCore::HTMLSelectElement::setRecalcListItems): ditto.
2668         (WebCore::HTMLSelectElement::reset): ditto.
2669         (WebCore::HTMLSelectElement::notifyOptionSelected): ditto.
2670         (WebCore::HTMLSelectElement::defaultEventHandler): ditto.
2671
2672         * html/HTMLSelectElement.h: (WebCore::HTMLSelectElement::shouldUseMenuList):
2673         This method no longer takes a RenderStyle.  That was leftover from when we used to use the appearance
2674         to determine whether or not to use the new menu list implementation.
2675
2676         * rendering/DeprecatedRenderSelect.h: Removed PopupButton code path.
2677         * rendering/DeprecatedRenderSelect.cpp:
2678         (WebCore::DeprecatedRenderSelect::DeprecatedRenderSelect):
2679         (WebCore::DeprecatedRenderSelect::setWidgetWritingDirection):
2680         (WebCore::DeprecatedRenderSelect::updateFromElement):
2681         (WebCore::DeprecatedRenderSelect::baselinePosition):
2682         (WebCore::DeprecatedRenderSelect::layout):
2683         (WebCore::DeprecatedRenderSelect::updateSelection):
2684
2685         * rendering/RenderMenuList.cpp: (WebCore::RenderMenuList::updateFromElement):
2686         Backed out previous fix since this is now done in HTMLSelectElement.
2687
2688 2006-07-11  Adele Peterson  <adele@apple.com>
2689
2690         Reviewed by Hyatt.
2691
2692         Fix for http://bugzilla.opendarwin.org/show_bug.cgi?id=9859
2693         REGRESSION: Can't dynamically change popup to list box
2694
2695         Test: fast/forms/select-change-popup-to-listbox.html
2696
2697         * rendering/RenderMenuList.cpp: (WebCore::RenderMenuList::updateFromElement):
2698         If the select element has the multiple property set, or has a size > 1, then it
2699         really should be a list box, so we detach and attach the element so it creates the
2700         correct type of renderer.
2701
2702 2006-07-11  John Sullivan  <sullivan@apple.com>
2703
2704         Reviewed by Beth Dakin and Tim Omernick
2705
2706         - fixed <rdar://problem/4622794> HiDPI: dragging the selection in Safari can show pixel cracks
2707         at non-integral scale factors
2708
2709         * bridge/mac/FrameMac.mm:
2710         (WebCore::FrameMac::imageFromRect):
2711         round image rect in window coordinate space
2712
2713 2006-07-10  Darin Adler  <darin@apple.com>
2714
2715         Reviewed by Geoff.
2716
2717         - fix http://bugzilla.opendarwin.org/show_bug.cgi?id=9811
2718           rdar://problem/4619572
2719           OPTION text contained in OPTGROUP appears at the wrong size
2720
2721         * dom/Node.h:
2722         * dom/Node.cpp: (WebCore::Node::renderStyle): Made a const member function.
2723         The derived classes were overriding with const member functions, and I could
2724         change either this or the derived -- decided to do this.
2725
2726 2006-07-10  Darin Adler  <darin@apple.com>
2727
2728         Reviewed by Geoff.
2729
2730         - fix http://bugzilla.opendarwin.org/show_bug.cgi?id=9810
2731           rdar://problem/4619534
2732           OPTGROUP label text is using the default font instead of Lucida Grande
2733
2734         * rendering/RenderThemeMac.mm: (WebCore::RenderThemeMac::setFontFromControlSize):
2735         Code wants the family name, but was calling fontName, which is not the same thing.
2736         Also removed the unneeded conversion to DeprecatedString.
2737
2738 === Safari-521.16 ===
2739
2740 2006-07-10  David Kilzer  <ddkilzer@kilzer.net>
2741
2742         Reviewed by Darin.
2743
2744         - fix http://bugzilla.opendarwin.org/show_bug.cgi?id=9179
2745           Implement select.options.add() method
2746
2747         Tests:
2748         - fast/dom/select-selectedIndex-multiple.html
2749         - fast/dom/select-selectedIndex.html
2750         - fast/js/select-options-add.html
2751
2752         * DerivedSources.make: Added JSHTMLOptionsCollection.h.
2753         * ForwardingHeaders/kjs/operations.h: Added.
2754         * WebCore.xcodeproj/project.pbxproj: Added new source files.
2755         * bindings/js/JSHTMLOptionsCollectionCustom.cpp: Added.
2756         (WebCore::JSHTMLOptionsCollection::length):
2757         (WebCore::JSHTMLOptionsCollection::setLength):
2758         (WebCore::JSHTMLOptionsCollection::indexSetter):
2759         * bindings/js/kjs_html.cpp: Removed JSHTMLOptionsCollection implementation.
2760         Renamed classes and methods for consistency.
2761         (KJS::JSHTMLElement::selectGetter):
2762         (KJS::JSHTMLElement::put):
2763         (KJS::JSHTMLElement::selectSetter):
2764         (KJS::JSHTMLCollection::JSHTMLCollection):
2765         (KJS::JSHTMLCollectionProtoFunc::callAsFunction):
2766         (KJS::getHTMLOptionsCollection):
2767         * bindings/js/kjs_html.h: Ditto.
2768         * bindings/scripts/CodeGeneratorJS.pm: Added support for HasCustomIndexSetter class attribute.
2769         Added support for Optional parameter attribute, which makes generated code assume overloaded
2770         implementation methods are available for a JavaScript function with optional arguments.  Changed
2771         local 'impl' variables to 'imp' so that impl() methods could be called without class designation.
2772         * html/HTMLOptionElement.idl: Added GenerateNativeConverter attribute.
2773         * html/HTMLOptionsCollection.cpp: Added methods used by generated JSHTMLOptionsCollection class.
2774         (WebCore::HTMLOptionsCollection::HTMLOptionsCollection):
2775         (WebCore::HTMLOptionsCollection::add):
2776         (WebCore::HTMLOptionsCollection::selectedIndex):
2777         (WebCore::HTMLOptionsCollection::setSelectedIndex):
2778         (WebCore::HTMLOptionsCollection::setLength):
2779         * html/HTMLOptionsCollection.h: Ditto.
2780         * html/HTMLOptionsCollection.idl: Added.
2781
2782 2006-07-10  David Harrison  <harrison@apple.com>
2783
2784         Reviewed by Dave Hyatt.
2785
2786         <rdar://problem/4602408> -webkit-highlight needs to support images and elements other than text
2787
2788         * rendering/RenderBox.cpp:
2789         (WebCore::RenderBox::paintCustomHighlight):
2790         * rendering/RenderBox.h:
2791         Utility for subclasses.
2792
2793         * rendering/RenderImage.cpp:
2794         (WebCore::RenderImage::paint):
2795         Custom highlight in front of images.
2796
2797         * rendering/RenderListMarker.cpp:
2798         (WebCore::RenderListMarker::paint):
2799         Custom highlight in front of list markers.
2800
2801         * rendering/RenderWidget.cpp:
2802         (WebCore::RenderWidget::paint):
2803         Custom highlight in front of objects.
2804
2805 2006-07-10  David Hyatt  <hyatt@apple.com>
2806
2807         <rdar://problem/4620557> REGRESSION: In a mail message, text isn't painted
2808         correctly after choosing Redo Typing
2809
2810         Better fix for repainting issue with positioned objects when height
2811         changes.  Optimize when only a positioned child changes so that we don't
2812         incorrectly do normal flow layout.
2813
2814         Reviewed by darin
2815
2816         * WebCore.xcodeproj/project.pbxproj:
2817         * rendering/RenderBlock.cpp:
2818         (WebCore::RenderBlock::layoutBlock):
2819         * rendering/RenderFlexibleBox.cpp:
2820         (WebCore::RenderFlexibleBox::layoutBlock):
2821         * rendering/RenderTable.cpp:
2822         (WebCore::RenderTable::layout):
2823         * rendering/RenderView.cpp:
2824         (WebCore::RenderView::layout):
2825         * rendering/bidi.cpp:
2826         (WebCore::RenderBlock::layoutInlineChildren):
2827
2828 2006-07-10  Adele Peterson  <adele@apple.com>
2829
2830         Reviewed by Beth.
2831
2832         - Fix for <rdar://problem/4621442> REGRESSION (NativePopup) : Popup menus aren't drawn correctly on page at orbitz.com; cheaptickets.com
2833
2834         Test: fast/forms/select-style.html
2835
2836         * rendering/RenderThemeMac.mm: (WebCore::RenderThemeMac::isControlStyled):
2837         Don't let popups be style-able for now.  We'll allow it later, when we
2838         have a good way to handle border & backgrounds and still having it look like a control.
2839
2840 2006-07-10  Justin Garcia  <justin.garcia@apple.com>
2841
2842         Reviewed by harrison
2843
2844         <rdar://problem/4619260>
2845         9A213 + Safari 521.15.1 Crash with To Do's alarm icon (WebCore::maxDeepOffset(WebCore::Node const*) )
2846         <rdar://problem/4619841>
2847         REGRESSION: Dragging selection over input field results in a crash (WebCore::maxDeepOffset(WebCore::Node const*)
2848
2849         * WebCore.xcodeproj/project.pbxproj:
2850         * editing/Selection.cpp:
2851         (WebCore::Selection::adjustForEditableContent): Iterate using next/previousVisuallyDistinctCandidate
2852         instead of traverseNext/Previous node because we were skipping positions.  Jump out of shadow trees.
2853         Migrate to isEditablePosition instead of isContentEditable.
2854
2855
2856 2006-07-10  Beth Dakin  <bdakin@apple.com>
2857
2858         Reviewed by Adele.
2859
2860         Shuffling comments around from my last check-in to make things read
2861         more clearly.
2862
2863         * page/DOMWindow.h:
2864         * page/DOMWindow.idl:
2865
2866 2006-07-10  Beth Dakin  <bdakin@apple.com>
2867
2868         Reviewed by Darin.
2869
2870         Fix for <rdar://problem/4621095> The should be a way to
2871         access the user interface scale factor through the DOM
2872
2873         window.devicePixelRatio returns the user interface scale factor.
2874
2875         * page/DOMWindow.cpp:
2876         (WebCore::DOMWindow::devicePixelRatio):
2877         * page/DOMWindow.h:
2878         * page/DOMWindow.idl:
2879
2880 2006-07-10  Brady Eidson  <beidson@apple.com>
2881
2882         Reviewed by Alexey
2883
2884         Resolved the console error messages people got from the new DB even if they didn't have it enabled
2885
2886         * icon/IconDatabase.cpp:
2887         (WebCore::IconDatabase::pruneUnreferencedIcons):
2888         (WebCore::IconDatabase::pruneUnretainedIcons):
2889         Added quick check to bail if the DB isn't open
2890
2891 2006-07-10  David Harrison  <harrison@apple.com>
2892
2893         Reviewed by Tim Hatcher
2894
2895         Partial fix for...
2896         <rdar://problem/4619260> 9A213 + Safari 521.15.1 Crash with To Do's alarm icon (WebCore::maxDeepOffset(WebCore::Node const*) )
2897
2898         * editing/Selection.cpp:
2899         (WebCore::Selection::adjustForEditableContent):
2900         Allow for node being 0.  This is a bandaid fix to prevent the crash.  It will still assert in a debug build.  See bug for more info.
2901
2902 2006-07-10  Mitz Pettel  <opendarwin.org@mitzpettel.com>
2903
2904         Reviewed by Darin.
2905
2906         - fix http://bugzilla.opendarwin.org/show_bug.cgi?id=9808
2907           REGRESSION: Incorrect layout (and ERROR) when forcing ATSU For All Text
2908
2909         * manual-tests/ATSU-bad-layout.html: Added.
2910         * platform/mac/FontMac.mm:
2911         (WebCore::Font::floatWidthForComplexText): Return 0 immediately for empty runs.
2912
2913 2006-07-10  Darin Adler  <darin@apple.com>
2914
2915         - try to fix the no-SVG build
2916
2917         * ksvg2/misc/SVGImageLoader.cpp: Move namespace brace inside the #if.
2918
2919 2006-07-10  Darin Adler  <darin@apple.com>
2920
2921         Reviewed by Anders.
2922
2923         - fix http://bugzilla.opendarwin.org/show_bug.cgi?id=9833
2924           REGRESSION: Reproducible crash: RenderMenuList.cpp:58: failed assertion `!m_first'
2925
2926         * manual-tests/empty-title-popup.html: Added.
2927
2928         * rendering/RenderMenuList.h: Add createInnerBlock.
2929         * rendering/RenderMenuList.cpp:
2930         (WebCore::RenderMenuList::createInnerBlock): Factored out of addChild.
2931         (WebCore::RenderMenuList::addChild): Call createInnerBlock.
2932         (WebCore::RenderMenuList::setText): Changed parameter type.
2933         (WebCore::RenderMenuList::showPopup): Call createInnerBlock before calling
2934         the parent class's addChild.
2935
2936 2006-07-10  Rob Buis  <buis@kde.org>
2937
2938         Reviewed by Maciej via IRC.
2939
2940         Make sure the attributes are calculated against viewport width value,
2941         as defined in the spec.  http://paste.lisp.org/display/22342
2942
2943         * ksvg2/svg/SVGEllipseElement.cpp:
2944         (WebCore::SVGEllipseElement::rx): Changed LM_HEIGHT to LM_WIDTH.
2945         * ksvg2/svg/SVGLineElement.cpp:
2946         (SVGLineElement::x2): Ditto.
2947
2948 2006-07-10  Beth Dakin  <bdakin@apple.com>
2949
2950         Reviewed by Maciej.
2951
2952         Fix for <rdar://problem/4610314> Support the CSS3 content property
2953         for images
2954
2955         This patch provides initial support for the CSS3 content property.
2956         It lacks support for any content other than images, and only allows
2957         content for elements that seemed safe for now.
2958
2959         * css/cssstyleselector.cpp:
2960         (WebCore::CSSStyleSelector::applyProperty): Remove restrictions for
2961         just :before and :after
2962         * html/HTMLBRElement.cpp:
2963         (WebCore::HTMLBRElement::createRenderer):
2964         * html/HTMLFrameSetElement.cpp:
2965         (WebCore::HTMLFrameSetElement::createRenderer):
2966         * html/HTMLImageElement.cpp:
2967         (WebCore::HTMLImageElement::createRenderer):
2968         * html/HTMLLegendElement.cpp:
2969         (WebCore::HTMLLegendElement::createRenderer):
2970         * rendering/RenderImage.cpp:
2971         (WebCore::RenderImage::RenderImage):
2972         (WebCore::RenderImage::setCachedImage): Don't allow the image to be
2973         set this way if it was already set with the content property.
2974         * rendering/RenderImage.h:
2975         (WebCore::RenderImage::setIsAnonymousImage): Anonymous images are
2976         those set through the content property.
2977         (WebCore::RenderImage::isAnonymousImage):
2978         * rendering/RenderObject.cpp:
2979         (WebCore::RenderObject::createObject): If there is content
2980         specified, create a RenderImage with the content and initialize the
2981         cached image to whatever was specified in the CSS.
2982
2983 2006-07-10  Brady Eidson  <beidson@apple.com>
2984
2985         Reviewed by NOBODY (Build Fix!)
2986
2987         * icon/SiteIcon.cpp:
2988         (SiteIcon::getImage):
2989         Release build made an "unused variable" warning, which is, of course, an error.  fixed that.
2990
2991 2006-07-10  Brady Eidson  <beidson@apple.com>
2992
2993         Reviewed by Maciej
2994
2995         SQLite Icon DB now fully replicates functionality of the old DB, including pruning to keep unused
2996         information out to keep down disk usage.  For now, it is still off by default, as the critical
2997         feature of importing old icons into the new DB is still unrealized.
2998         If you'd like to test, add #define ICONDEBUG to your WebKitPrefix.h
2999
3000         * icon/IconDatabase.cpp:
3001         (WebCore::IconDatabase::IconDatabase):
3002         (WebCore::IconDatabase::open):
3003           -Sets up a timer for initial pruning
3004         (WebCore::IconDatabase::iconForPageURL):
3005           -Switched away from "dirty blob hack" as the problem is provisionally solved
3006         (WebCore::IconDatabase::retainIconForURL):
3007         (WebCore::IconDatabase::releaseIconForURL):
3008           -Same as found in WebKit
3009         (WebCore::IconDatabase::totalRetainCountForIconURL):
3010           -Adds up all retainers of icon
3011         (WebCore::IconDatabase::forgetIconForIconURLFromDatabase):
3012           -Wipes an Icon table entry out
3013         (WebCore::IconDatabase::establishTemporaryIconIDForIconURL):
3014         (WebCore::IconDatabase::establishTemporaryIconIDForEscapedIconURL):
3015         (WebCore::IconDatabase::establishIconIDForIconURL):
3016         (WebCore::IconDatabase::establishIconIDForEscapedIconURL):
3017           -Added ability to get an iconID without creating a new one if a lookup is all
3018            that is genuinely wanted
3019         (WebCore::IconDatabase::pruneUnreferencedIcons):
3020         (WebCore::IconDatabase::pruneUnretainedIcons):
3021           -Called on a timer on DB startup to clean it out
3022         * icon/IconDatabase.h:
3023         * icon/SiteIcon.cpp:
3024         (SiteIcon::SiteIcon):
3025         (SiteIcon::~SiteIcon):
3026           -Deletes the image on deletion
3027         (SiteIcon::getImage):
3028           -Cut down on no-longer-necessary debug info
3029
3030 2006-07-10  Anders Carlsson  <acarlsson@apple.com>
3031
3032         Reviewed by Adele.
3033
3034         http://bugzilla.opendarwin.org/show_bug.cgi?id=9245
3035         Quirksmode: Incorrect handling of disabled BUTTON elements
3036
3037         * html/HTMLButtonElement.cpp:
3038         (WebCore::HTMLButtonElement::defaultEventHandler):
3039         Don't call prepareSubmit if the element is disabled.
3040
3041         * html/HTMLInputElement.cpp:
3042         (WebCore::HTMLInputElement::defaultEventHandler):
3043         Don't call prepareSubmit if the element is disabled.
3044
3045 2006-07-10  Darin Adler  <darin@apple.com>
3046
3047         - try to fix Windows build
3048
3049         * platform/win/TemporaryLinkStubs.cpp: Put a few functions in the WebCore namespace.
3050         * platform/gdk/TemporaryLinkStubs.cpp: Ditto.
3051
3052 2006-07-10  Darin Adler  <darin@apple.com>
3053
3054         Reviewed by Geoff.
3055
3056         - fix failures seen in layout tests
3057
3058         * bindings/js/kjs_dom.cpp: (KJS::toJS): Restore one of the few WebCore:: prefixes
3059         that really needed to be there -- because we have two classes named JSHTMLDocument
3060         at the moment! (One in KJS and one in WebCore namespace.)
3061
3062 2006-07-10  Darin Adler  <darin@apple.com>
3063
3064         - try to fix Windows build
3065
3066         * rendering/RenderLayer.h: Correct a forward declaration of ScrollBar.
3067
3068 2006-07-09  Adele Peterson  <adele@apple.com>
3069
3070         Reviewed by Maciej.
3071
3072         Fix for <rdar://problem/4463835> Switch to use new popup menu implementation for <select>
3073
3074         * css/html4.css: Added style for new selects, and for list boxes.
3075         * html/HTMLSelectElement.h: (WebCore::HTMLSelectElement::shouldUseMenuList): Removed check for appearance so new popups can be styled.
3076
3077         * rendering/RenderMenuList.cpp:
3078         (WebCore::RenderMenuList::setStyle): Removed padding.  This is now set by the theme.
3079         (WebCore::RenderMenuList::calcMinMaxWidth): Now takes minimum text size into account.
3080         * rendering/RenderMenuList.h: Removed baselinePosition since we're letting the theme calculate that.
3081
3082         * rendering/RenderTheme.cpp: (WebCore::RenderTheme::isControlContainer): Added menuList as a ControlContainer,
3083         so RenderThemeMac will compute its baseline.
3084         * rendering/RenderTheme.h: (WebCore::RenderTheme::minimumTextSize): Added.
3085         * rendering/RenderThemeMac.h: Added minimumTextSize, setPopupPaddingFromControlSize, popupButtonSizes, popupButtonMargins, popupButtonPadding.
3086         Removed sizeOfArrowControl.
3087         * rendering/RenderThemeMac.mm:
3088         (WebCore::): Added enum for padding.
3089         (WebCore::RenderThemeMac::adjustRepaintRect): Now inflates rect to account for control shadow.
3090         (WebCore::RenderThemeMac::baselinePosition): Now calculates baseline for menulist.
3091         (WebCore::RenderThemeMac::popupButtonMargins): Tweaked values.
3092         (WebCore::RenderThemeMac::popupButtonSizes): Added.
3093         (WebCore::RenderThemeMac::popupButtonPadding): Added, so different padding can be set for different control sizes.
3094         (WebCore::RenderThemeMac::setPopupPaddingFromControlSize): Added.
3095         (WebCore::RenderThemeMac::paintMenuList): Now inflates rect to account for control shadow.
3096         (WebCore::RenderThemeMac::adjustMenuListStyle): Resets border, sets padding, sets control size.
3097         (WebCore::RenderThemeMac::setPopupButtonCellState): Removed unnecessary class name.
3098         (WebCore::RenderThemeMac::minimumTextSize): Added.
3099
3100 2006-07-09  Darin Adler  <darin@apple.com>
3101
3102         - add back "unused code" for plain text fields, still used by <isindex>!
3103
3104         * platform/TextField.h:
3105         * platform/mac/TextFieldMac.mm:
3106         * platform/mac/WebCoreTextField.h:
3107         * platform/mac/WebCoreTextField.mm:
3108         * rendering/RenderLineEdit.cpp:
3109
3110 2006-07-09  Darin Adler  <darin@apple.com>
3111
3112         Reviewed by Tim Hatcher.
3113
3114         - put more into the WebCore namespace
3115         - removed unused code for plain text fields
3116
3117         Put more classes, class templates, enums, functions, and constants into
3118         the WebCore namespace, including all the headers in the platform directory.
3119
3120         Removed lots of unneeded "WebCore::" qualifiers and "using" directives.
3121
3122         Added some "WebCore::" qualifiers in a few places.
3123
3124         Removed "uses WebCore::" in headers that we said we'd remove "when
3125         everything is in the WebCore namespace".
3126
3127         * bindings/js/JSDOMParser.cpp:
3128         * bindings/js/JSHTMLElementWrapperFactory.cpp:
3129         * bindings/js/JSXSLTProcessor.cpp:
3130         * bindings/js/kjs_binding.cpp:
3131         * bindings/js/kjs_css.cpp:
3132         * bindings/js/kjs_dom.cpp:
3133         * bindings/js/kjs_dom.h:
3134         * bindings/js/kjs_events.cpp:
3135         * bindings/js/kjs_proxy.h:
3136         * bindings/js/kjs_window.h:
3137         * bindings/objc/DOM.mm:
3138         * bridge/History.h:
3139         * bridge/JavaAppletWidget.h:
3140         * bridge/mac/FormDataMac.h:
3141         * bridge/mac/FormDataMac.mm:
3142         * bridge/mac/WebCoreAXObject.mm:
3143         * bridge/mac/WebCoreEncodings.mm:
3144         * bridge/mac/WebCoreFrameBridge.h:
3145         * bridge/mac/WebCoreIconDatabaseBridge.mm:
3146         * bridge/mac/WebCorePageState.h:
3147         * bridge/mac/WebCoreScriptDebugger.mm:
3148         * bridge/mac/WebCoreSettings.h:
3149         * css/CSSGrammar.y:
3150         * css/CSSPageRule.cpp:
3151         * css/CSSPageRule.h:
3152         * css/MediaList.cpp:
3153         * css/cssstyleselector.h:
3154         * dom/Clipboard.h:
3155         * dom/Document.cpp:
3156         * dom/Document.h:
3157         * dom/EventTargetNode.h:
3158         * dom/Node.cpp:
3159         * dom/Node.h:
3160         * dom/Position.h:
3161         * dom/Range.cpp:
3162         * dom/XMLTokenizer.cpp:
3163         * editing/AppendNodeCommand.h:
3164         * editing/ApplyStyleCommand.cpp:
3165         * editing/BreakBlockquoteCommand.cpp:
3166         * editing/BreakBlockquoteCommand.h:
3167         * editing/CompositeEditCommand.cpp:
3168         * editing/CompositeEditCommand.h:
3169         * editing/CreateLinkCommand.h:
3170         * editing/DeleteFromTextNodeCommand.h:
3171         * editing/DeleteSelectionCommand.h:
3172         * editing/FormatBlockCommand.h:
3173         * editing/HTMLInterchange.cpp:
3174         * editing/HTMLInterchange.h:
3175         * editing/IndentOutdentCommand.h:
3176         * editing/InsertIntoTextNodeCommand.h:
3177         * editing/InsertLineBreakCommand.h:
3178         * editing/InsertListCommand.h:
3179         * editing/InsertNodeBeforeCommand.h:
3180         * editing/InsertTextCommand.cpp:
3181         * editing/InsertTextCommand.h:
3182         * editing/JSEditor.cpp:
3183         * editing/JSEditor.h:
3184         * editing/JoinTextNodesCommand.h:
3185         * editing/MergeIdenticalElementsCommand.cpp:
3186         * editing/MergeIdenticalElementsCommand.h:
3187         * editing/ModifySelectionListLevel.h:
3188         * editing/MoveSelectionCommand.h:
3189         * editing/RebalanceWhitespaceCommand.h:
3190         * editing/RemoveNodeAttributeCommand.h:
3191         * editing/RemoveNodeCommand.h:
3192         * editing/RemoveNodePreservingChildrenCommand.h:
3193         * editing/ReplaceSelectionCommand.h:
3194         * editing/SetNodeAttributeCommand.h:
3195         * editing/SplitElementCommand.cpp:
3196         * editing/SplitElementCommand.h:
3197         * editing/SplitTextNodeCommand.cpp:
3198         * editing/SplitTextNodeCommand.h:
3199         * editing/SplitTextNodeContainingElementCommand.h:
3200         * editing/TypingCommand.cpp:
3201         * editing/TypingCommand.h:
3202         * editing/UnlinkCommand.h:
3203         * editing/VisiblePosition.cpp:
3204         * editing/WrapContentsInDummySpanCommand.cpp:
3205         * editing/WrapContentsInDummySpanCommand.h:
3206         * editing/markup.cpp:
3207         * editing/markup.h:
3208         * html/HTMLAnchorElement.cpp:
3209         * html/HTMLBodyElement.cpp:
3210         * html/HTMLButtonElement.h:
3211         * html/HTMLCollection.h:
3212         * html/HTMLDocument.h:
3213         * html/HTMLEmbedElement.cpp:
3214         * html/HTMLFormElement.cpp:
3215         * html/HTMLFormElement.h:
3216         * html/HTMLFrameElement.cpp:
3217         * html/HTMLFrameSetElement.cpp:
3218         * html/HTMLKeygenElement.cpp:
3219         * html/HTMLLabelElement.cpp:
3220         * html/HTMLObjectElement.cpp:
3221         * html/HTMLObjectElement.h:
3222         * html/HTMLTableElement.cpp:
3223         * html/HTMLTablePartElement.cpp:
3224         * icon/IconDatabase.h:
3225         * kcanvas/KCanvasPath.h:
3226         * kcanvas/KCanvasResources.cpp:
3227         * kcanvas/KCanvasResources.h:
3228         * kcanvas/KCanvasTreeDebug.cpp:
3229         * kcanvas/KCanvasTreeDebug.h:
3230         * kcanvas/RenderSVGContainer.cpp:
3231         * kcanvas/RenderSVGText.h:
3232         * kcanvas/device/KRenderingPaintServer.h:
3233         * kcanvas/device/KRenderingPaintServerGradient.h:
3234         * kcanvas/device/KRenderingPaintServerSolid.cpp:
3235         * ksvg2/misc/SVGImageLoader.cpp:
3236         * ksvg2/svg/SVGElement.cpp:
3237         * ksvg2/svg/SVGElement.h:
3238         * ksvg2/svg/SVGForeignObjectElement.cpp:
3239         * ksvg2/svg/SVGForeignObjectElement.h:
3240         * ksvg2/svg/SVGImageElement.cpp:
3241         * ksvg2/svg/SVGPoint.h:
3242         * ksvg2/svg/SVGStylable.h:
3243         * ksvg2/svg/SVGTRefElement.cpp:
3244         * ksvg2/svg/SVGTRefElement.h:
3245         * ksvg2/svg/SVGTSpanElement.cpp:
3246         * ksvg2/svg/SVGTSpanElement.h:
3247         * ksvg2/svg/SVGTextElement.h:
3248         * ksvg2/svg/svgpathparser.cpp:
3249         * ksvg2/svg/svgpathparser.h:
3250         * loader/Cache.h:
3251         * loader/DocLoader.h:
3252         * loader/LoaderFunctions.h:
3253         * loader/mac/LoaderFunctionsMac.mm:
3254         * page/Frame.h:
3255         * page/FrameView.h:
3256         * page/Page.h:
3257         * page/Settings.h:
3258         * platform/AffineTransform.cpp:
3259         * platform/AffineTransform.h:
3260         * platform/Arena.cpp:
3261         * platform/Arena.h:
3262         * platform/CookieJar.h:
3263         * platform/DeprecatedCString.cpp:
3264         * platform/DeprecatedCString.h:
3265         * platform/DeprecatedPtrList.h:
3266         * platform/DeprecatedPtrListImpl.cpp:
3267         * platform/DeprecatedPtrListImpl.h:
3268         * platform/DeprecatedPtrQueue.h:
3269         * platform/DeprecatedString.cpp:
3270         * platform/DeprecatedString.h:
3271         * platform/DeprecatedStringList.cpp:
3272         * platform/DeprecatedStringList.h:
3273         * platform/DeprecatedValueList.h:
3274         * platform/DeprecatedValueListImpl.cpp:
3275         * platform/DeprecatedValueListImpl.h:
3276         * platform/FileButton.h:
3277         * platform/FloatPoint.h:
3278         * platform/FloatRect.h:
3279         * platform/FloatSize.h:
3280         * platform/FontData.h:
3281         * platform/FontFallbackList.h:
3282         * platform/GraphicsContext.h:
3283         * platform/IntPoint.h:
3284         * platform/IntRect.h:
3285         * platform/IntSize.h:
3286         * platform/KURL.cpp:
3287         * platform/KURL.h:
3288         * platform/ListBox.h:
3289         * platform/Logging.cpp:
3290         * platform/Logging.h:
3291         * platform/Pen.h:
3292         * platform/PopUpButton.h:
3293         * platform/RegularExpression.cpp:
3294         * platform/RegularExpression.h:
3295         * platform/SSLKeyGenerator.h:
3296         * platform/ScrollBar.h:
3297         * platform/SegmentedString.h:
3298         * platform/Shared.h:
3299         * platform/Slider.h:
3300         * platform/StringImpl.h:
3301         * platform/TextBox.h:
3302         * platform/TextField.h:
3303         * platform/TextStream.cpp:
3304         * platform/TextStream.h:
3305         * platform/TransferJob.h:
3306         * platform/TransferJobClient.h:
3307         * platform/Widget.h:
3308         * platform/cg/AffineTransformCG.cpp:
3309         * platform/mac/ClipboardMac.h:
3310         * platform/mac/ClipboardMac.mm:
3311         * platform/mac/DeprecatedStringListMac.mm:
3312         * platform/mac/FontCacheMac.mm:
3313         * platform/mac/KURLMac.mm:
3314         * platform/mac/ListBoxMac.mm:
3315         * platform/mac/PopUpButtonMac.mm:
3316         * platform/mac/SSLKeyGeneratorMac.mm:
3317         * platform/mac/TextBoxMac.mm:
3318         * platform/mac/TextFieldMac.mm:
3319         * platform/mac/WebCoreTextArea.h:
3320         * platform/mac/WebCoreTextArea.mm:
3321         * platform/mac/WebCoreTextField.h:
3322         * platform/mac/WebCoreTextField.mm:
3323         * rendering/DeprecatedRenderSelect.h:
3324         * rendering/RenderArena.cpp:
3325         * rendering/RenderArena.h:
3326         * rendering/RenderBlock.cpp:
3327         * rendering/RenderBox.cpp:
3328         * rendering/RenderContainer.cpp:
3329         * rendering/RenderFlexibleBox.cpp:
3330         * rendering/RenderFlow.cpp:
3331         * rendering/RenderLayer.cpp:
3332         * rendering/RenderLineEdit.cpp:
3333         * rendering/RenderObject.h:
3334         * rendering/RenderStyle.cpp:
3335         * rendering/RenderStyle.h:
3336         * rendering/RenderText.cpp:
3337         * rendering/RenderTextFragment.cpp:
3338         * rendering/RenderTheme.h:
3339         * rendering/RenderTreeAsText.cpp:
3340         * rendering/RenderTreeAsText.h:
3341         * rendering/RenderView.h:
3342         * rendering/bidi.h:
3343         * xml/XSLStyleSheet.cpp:
3344         * xml/XSLTProcessor.cpp:
3345
3346 2006-07-09  Geoffrey Garen  <ggaren@apple.com>
3347
3348         Reviewed by Beth.
3349
3350         * bindings/js/kjs_window.cpp:
3351         (KJS::Window::clear): Garbage collect after reinitalizing the global
3352         object, not before, since the reinitialization tends to create garbage.
3353
3354 2006-07-09  Timothy Hatcher  <timothy@apple.com>
3355
3356         Reviewed by Darin.
3357
3358         Bug 9820: Move new DOM API that has been through API review to public headers
3359         http://bugzilla.opendarwin.org/show_bug.cgi?id=9820
3360
3361         * bindings/objc/DOM.mm:
3362         (-[DOMElement focus]):
3363         (-[DOMElement blur]):
3364         (-[DOMRange text]):
3365         (-[DOMRange _text]):
3366         * bindings/objc/DOMCSS.mm:
3367         (-[DOMRGBColor color]):
3368         (-[DOMRGBColor _color]):
3369         * bindings/objc/DOMCore.h:
3370         * bindings/objc/DOMEvents.h:
3371         * bindings/objc/DOMExtensions.h:
3372         * bindings/objc/DOMHTML.mm:
3373         (-[DOMHTMLElement titleDisplayString]):
3374         (-[DOMHTMLDocument createDocumentFragmentWithMarkupString:baseURL:]):
3375         (-[DOMHTMLDocument createDocumentFragmentWithText:]):
3376         (-[DOMHTMLDocument _createDocumentFragmentWithMarkupString:baseURLString:]):
3377         (-[DOMHTMLDocument _createDocumentFragmentWithText:]):
3378         * bindings/objc/DOMPrivate.h:
3379         * bindings/objc/DOMXPath.h:
3380         * bindings/objc/DOMXPath.mm:
3381
3382 2006-07-09  Beth Dakin  <bdakin@apple.com>
3383
3384         Reviewed by Maciej.
3385
3386         Fix for http://bugzilla.opendarwin.org/show_bug.cgi?id=9749 A
3387         nested content editable div causes unnecessary screen redraws.
3388
3389         We were repainting too much because we were deciding that we needed
3390         a full layout because we didn't have a first line box. Our code
3391         used to hold a lot of assumptions that positioned elements would
3392         always have line boxes, but I fixed a few bugs a while back that
3393         appeared because this assumption isn't true. This assumption was
3394         also causing the extra redraws because the check meant that we
3395         would decide to do a full layout for any positioned element that
3396         did not have any line boxes in the first place. Maciej and I talked
3397         about this, and we couldn't figure out why the check for line boxes
3398         was part of determining if we need a full layout in the first
3399         place. After hunting around the code, we decided to assume that
3400         this check is antiquated. Removing it fixes the bug and doesn't
3401         seem to break any layout tests. So, fingers crossed!
3402
3403         Will add layout test so soon.
3404
3405         * rendering/bidi.cpp:
3406         (WebCore::RenderBlock::layoutInlineChildren): Remove firstLineBox
3407         check.
3408
3409 2006-07-09  David Harrison  <harrison@apple.com>
3410
3411         Reviewed by John Sullivan.
3412
3413         <rdar://problem/4598149> Deleting the contents of a DOMRange and then inserting a node into it crashes in WebCore::Font::canUseGlyphCache
3414
3415         Test: fast/dom/delete-contents.html
3416
3417         * dom/Range.cpp:
3418         (WebCore::Range::processContents):
3419
3420 2006-07-09  Darin Adler  <darin@apple.com>
3421
3422         - another round of renaming; and KWQ is dead
3423
3424         * WebCore.vcproj/WebCore/WebCore.vcproj: Remove paths for kwq and khtml directories.
3425         Move files to their new locations. Did rename.
3426         * WebCore.xcodeproj/project.pbxproj: Move files to their new locations. Did rename.
3427         * WebCoreSources.bkl: Did rename.
3428         * webcore-base.bkl: Removed kwq.
3429
3430         * kwq/AccessibilityObjectCache.h: Removed.
3431         * kwq/AccessibilityObjectCache.mm: Removed.
3432         * kwq/FormDataMac.h: Removed.
3433         * kwq/FormDataMac.mm: Removed.
3434         * kwq/WebCoreAXObject.h: Removed.
3435         * kwq/WebCoreAXObject.mm: Removed.
3436         * kwq/WebCoreEditCommand.h: Removed.
3437         * kwq/WebCoreEditCommand.mm: Removed.
3438         * kwq/WebCorePageState.h: Removed.
3439         * kwq/WebCorePageState.mm: Removed.
3440         * kwq/WebCoreResourceLoaderImp.h: Removed.
3441         * kwq/WebCoreResourceLoaderImp.mm: Removed.
3442
3443         * bridge/AXObjectCache.h: Added.
3444         * bridge/mac/AXObjectCacheMac.mm: Added.
3445         * bridge/mac/WebCoreAXObject.h: Added.
3446         * bridge/mac/WebCoreAXObject.mm: Added.
3447
3448         * loader/CachedObject.cpp: Removed.
3449         * loader/CachedObject.h: Removed.
3450         * loader/CachedObjectClient.h: Removed.
3451         * loader/CachedObjectClientWalker.cpp: Removed.
3452         * loader/CachedObjectClientWalker.h: Removed.
3453
3454         * loader/CachedResource.cpp: Added.
3455         * loader/CachedResource.h: Added.
3456         * loader/CachedResourceClient.h: Added.
3457         * loader/CachedResourceClientWalker.cpp: Added.
3458         * loader/CachedResourceClientWalker.h: Added.
3459
3460         * bridge/mac/FrameMac.mm:
3461         * bridge/mac/WebCoreFrameBridge.mm:
3462         * css/CSSImageValue.h:
3463         * css/CSSImportRule.h:
3464         * dom/Document.cpp:
3465         * dom/Document.h:
3466         * dom/Notation.h:
3467         * dom/ProcessingInstruction.h:
3468         * dom/XMLTokenizer.cpp:
3469         * html/CanvasPattern.h:
3470         * html/HTMLDocument.h:
3471         * html/HTMLImageLoader.cpp:
3472         * html/HTMLImageLoader.h:
3473         * html/HTMLLinkElement.h:
3474         * html/HTMLScriptElement.cpp:
3475         * html/HTMLScriptElement.h:
3476         * html/HTMLTokenizer.cpp:
3477         * html/HTMLTokenizer.h:
3478         * kcanvas/KCanvasFilters.h:
3479         * ksvg2/svg/SVGCursorElement.cpp:
3480         * ksvg2/svg/SVGCursorElement.h:
3481         * ksvg2/svg/SVGFEImageElement.cpp:
3482         * ksvg2/svg/SVGFEImageElement.h:
3483         * loader/Cache.cpp:
3484         * loader/Cache.h:
3485         * loader/CachedCSSStyleSheet.cpp:
3486         * loader/CachedCSSStyleSheet.h:
3487         * loader/CachedImage.cpp:
3488         * loader/CachedImage.h:
3489         * loader/CachedScript.cpp:
3490         * loader/CachedScript.h:
3491         * loader/CachedXBLDocument.cpp:
3492         * loader/CachedXBLDocument.h:
3493         * loader/CachedXSLStyleSheet.cpp:
3494         * loader/CachedXSLStyleSheet.h:
3495         * loader/DocLoader.cpp:
3496         * loader/DocLoader.h:
3497         * loader/LoaderFunctions.h:
3498         * loader/Request.cpp:
3499         * loader/Request.h:
3500         * loader/loader.cpp:
3501         * loader/loader.h:
3502         * loader/mac/LoaderFunctionsMac.mm:
3503         * page/Frame.cpp:
3504         * page/FrameView.cpp:
3505         * platform/gdk/TemporaryLinkStubs.cpp:
3506         * platform/mac/ClipboardMac.h:
3507         * platform/win/TemporaryLinkStubs.cpp:
3508         * rendering/RenderBox.h: