WebCore/ChangeLog-2005-08-23

   1 === Safari-188 ===
   2
   3 2005-02-21  Darin Adler  <darin@apple.com>
   4
   5         - fixed Panther deployment build
   6
   7         * kwq/KWQTextEdit.mm: (QTextEdit::setScrollBarModes): Put more stuff inside #if.
   8
   9 2005-02-21  David Harrison  <harrison@apple.com>
  10
  11         Reviewed by Darin.
  12
  13         <rdar://problem/3943090> REGRESSION (Mail): Spelling underline incompletely erased following certain steps
  14
  15         * khtml/rendering/render_text.cpp:
  16         (InlineTextBox::paintMarker):
  17         Make sure underline is placed within the text bounds.
  18
  19         * kwq/KWQPainter.h:
  20         * kwq/KWQPainter.mm:
  21         (QPainter::misspellingLineThickness):
  22         * kwq/WebCoreTextRenderer.h:
  23         Add misspellingLineThickness for use by InlineTextBox::paintMarker.
  24
  25 2005-02-21  Darin Adler  <darin@apple.com>
  26
  27         Reviewed by John.
  28
  29         - fixed <rdar://problem/4012978> -[DOMRange markupString] crashes when range contains only a text node with a single space
  30
  31         * khtml/editing/markup.cpp: (khtml::createMarkup): Added updateLayout calls, and added a missing
  32         nil check.
  33
  34 2005-02-21  Darin Adler  <darin@apple.com>
  35
  36         Reviewed by John.
  37
  38         - fixed <rdar://problem/4005435> Safari hung while pasting text into a <textarea> (Panther-only)
  39
  40         * kwq/KWQTextEdit.mm: (QTextEdit::setScrollBarModes): Don't setAutohidesScrollers:YES on Panther.
  41
  42 2005-02-21  Ken Kocienda  <kocienda@apple.com>
  43
  44         Reviewed by John
  45
  46         Fix for this bug:
  47
  48         <rdar://problem/4015499> REGRESSION (186-187): pasted quoted text starting with a blank line increases quote level of pasted text when pasted
  49
  50         * khtml/editing/markup.cpp:
  51         (khtml::markup): Changed over to ASSERT instead of assert.
  52         (khtml::createMarkup): The issue was that the code to add parents all the way back to the common ancestor
  53         block did not check for blocks whose markup had already been added, and could result in adding markup for
  54         nodes twice (hence the additional and erroneous quote level). Now there is a new check that will
  55         only add markup for those nodes before the start of the selection range. This fixes the bug.
  56         (khtml::createFragmentFromMarkup): Changed over to ASSERT instead of assert.
  57         (khtml::createFragmentFromText): Ditto.
  58
  59 2005-02-21  Ken Kocienda  <kocienda@apple.com>
  60
  61         Reviewed by John
  62
  63         Fix for this bug:
  64
  65         <rdar://problem/4015494> REGRESSION (186-187) <cr> removed if quoted word is copy/pasted on line immediately above quoted text
  66
  67         * khtml/editing/htmlediting.cpp:
  68         (khtml::ReplaceSelectionCommand::doApply): Relatively new check designed to remove a <br> element when
  69         that element was on a line by itself did not do an adequate check for this condition, causing the
  70         failure described in the bug. Fixed.
  71
  72 2005-02-19  Ken Kocienda  <kocienda@apple.com>
  73
  74         Reviewed by Maciej
  75
  76         * khtml/editing/htmlediting.cpp:
  77         (khtml::matchNearestBlockquoteColorString): New function which abstracts away the CSS property we use
  78         for this pseudo-color.
  79         (khtml::ReplacementFragment::removeBlockquoteColorsIfNeeded): Set the new nearestBlockquoteColorString
  80         psuedo-color here, rather than removing colors as was done before.
  81         (khtml::ReplaceSelectionCommand::applyStyleToInsertedNodes): Check for the pseudo-color in the desired
  82         style, and determine the real color based on the nearest blockquote (or none) to the node.
  83         (khtml::nearestMailBlockquote): Name change from closestMailBlockquote to match new property better.
  84
  85         All the rest of this change is the mechanical coding you need to do to add a new CSS property, in this
  86         case, -khtml-match-nearest-mail-blockquote-color.
  87
  88         * khtml/css/css_computedstyle.cpp:
  89         (DOM::CSSComputedStyleDeclarationImpl::getPropertyCSSValue)
  90         * khtml/css/cssparser.cpp:
  91         (CSSParser::parseValue):
  92         * khtml/css/cssproperties.c:
  93         (hash_prop):
  94         (findProp):
  95         * khtml/css/cssproperties.h:
  96         * khtml/css/cssproperties.in:
  97         * khtml/css/cssstyleselector.cpp:
  98         (khtml::CSSStyleSelector::applyProperty)
  99         * khtml/css/cssvalues.c:
 100         (findValue)
 101         * khtml/css/cssvalues.h:
 102         * khtml/css/cssvalues.in:
 103         * khtml/rendering/render_style.h:
 104         (khtml::RenderStyle::matchNearestMailBlockquoteColor)
 105         (khtml::RenderStyle::setMatchNearestMailBlockquoteColor)
 106         (khtml::RenderStyle::initialMatchNearestMailBlockquoteColor)
 107
 108 2005-02-19  Ken Kocienda  <kocienda@apple.com>
 109
 110         Reviewed by Chris
 111
 112         Fix for these bugs:
 113
 114         <rdar://problem/4014228> REGRESSION (186-187) extra, uneditable lines inserted above and below a line of pasted quoted text
 115         <rdar://problem/4014393> REGRESSION (186-187) pasted quoted text gets extra <cr>s when pasted at top of document
 116
 117         * khtml/editing/htmlediting.cpp:
 118         (khtml::ReplacementFragment::ReplacementFragment): Part of a general refactoring of how
 119         the fragment is inserted into the document, rendered, and then tested for certain
 120         important pieces of information that are required for pasting.
 121         (khtml::ReplacementFragment::insertFragmentForTestRendering): New helper. Handles inserting
 122         the fragment nodes into the document.
 123         (khtml::ReplacementFragment::restoreTestRenderingNodesToFragment): Removes nodes from the
 124         document, and restores them to the fragment.
 125         (khtml::ReplacementFragment::computeStylesUsingTestRendering): Factored out code that
 126         did this before into its own function.
 127         (khtml::ReplacementFragment::removeUnrenderedNodesUsingTestRendering): Ditto.
 128         (khtml::ReplacementFragment::countRenderedBlocks): This is a real improvement, as it
 129         eliminates a major use of the isProbablyBlock() function. Now, the blocks that are
 130         counted are real, rendered blocks.
 131         (khtml::ReplacementFragment::removeStyleNodes): Made this function retain margin-zeroing
 132         CSS properties on paragraphs. This does two things: 1) It helps us to maintain good behavior
 133         in the short term while there are still versions of Mail out there that use <p> elements
 134         instead of <div> elements for new paragraphs; and 2) It will help to maintain the compatibility
 135         with other mail clients that use <p> elements for their paragraphs but render them themselves
 136         with no margins as the result of quirks.
 137         (khtml::ReplaceSelectionCommand::doApply): Do some work to fix up and improve the handling
 138         of blank lines, be they <p> elements or <br> elements, that can be removed after pasting. This,
 139         coupled with the refactoring, fixes 4014393.
 140         * khtml/editing/htmlediting.h: Updated for new functions.
 141         * layout-tests/editing/pasteboard/paste-text-010-expected.txt: Updated results, actually improved with this change.
 142         * layout-tests/editing/pasteboard/paste-text-011-expected.txt: Ditto.
 143
 144 2005-02-19  Kevin Decker  <kdecker@apple.com>
 145
 146         Reviewed by Chris.
 147
 148         Fixed <rdar://problem/4010765> Flash player can be used to arbitrarily open popup windows without user permission
 149
 150         Our window.open() policy is to refuse the <script>window.open(...)</script> case and allow the inline the <a href="javascript:window.open('foo')> case.  Clever advertisers at some point realized that by executing their Javascript through the Flash plugin, Safari would always treat their code as the inline case, and thus, they were able to work around our popup blocker.
 151
 152         * kwq/WebCoreBridge.h:
 153         * kwq/WebCoreBridge.mm:
 154         (-[WebCoreBridge stringByEvaluatingJavaScriptFromString:]): Changed this to invoke the new stringByEvaluatingJavaScriptFromString:forceUserGesture method below.
 155         (-[WebCoreBridge stringByEvaluatingJavaScriptFromString:forceUserGesture:]): Let the WebKit make the determination if this was a user originated gesture or not; we must no longer assume this is always the case.
 156
 157 2005-02-18  Chris Blumenberg  <cblu@apple.com>
 158
 159         Fixed: <rdar://problem/3951196> REGRESSION (Mail): too many levels of reply quotes after certain steps
 160
 161         Reviewed by kocienda.
 162
 163         * khtml/editing/htmlediting.cpp:
 164         (khtml::InsertParagraphSeparatorInQuotedContentCommand::doApply):
 165         - If we find a new start node, update topBlockquote so we don't use too many block quotes for the contents following the new line.
 166         - Build up the list of ancestors after we've determined the actual topBlockquote.
 167         - Don't insert an extra new line if there is a new start node.
 168
 169 2005-02-19  Chris Blumenberg  <cblu@apple.com>
 170
 171         Fixed: <rdar://problem/3978461> smart paste is broken
 172
 173         Reviewed by kevin.
 174
 175         * khtml/editing/htmlediting.cpp:
 176         (khtml::ReplaceSelectionCommand::doApply): properly check for leading and trailing whitespace. These checks were incorrectly reversed. Also check if we're pasting at the beginning or end of a line. We should not insert spaces in either case.
 177
 178 2005-02-18  Adele Amchan  <adele@apple.com>
 179
 180         Reviewed by Darin.
 181
 182         Fix for <rdar://problem/3975568> REGRESSION(125-180)Australian Open pages have drawing problem
 183
 184         This patch fixes two problems caused by our added support for custom tags.  First, the layout problem at the sites mentioned in the bug
 185         was caused by custom tags within tables.  In checkChild, we needed to treat these tags as spans so they get placed correctly in the DOM tree.
 186         Also, we were indexing the tagPriority and endTag arrays with id values from the custom tags that were greater than the size of the array.  So now
 187         we have functions to check for the custom tags, and again, treat them as spans.  To avoid confusion, we changed the names of the arrays to
 188         endTagArray and tagPriorityArray.
 189
 190         * khtml/html/dtd.h: changed all uses of the endTag array to endTagArray for our new wrapper functions
 191         (DOM::tagPriority): added function to check array bounds and to treat custom tags as spans
 192         (DOM::endTagRequirement): added function to check array bounds and to treat custom tags as spans
 193         * khtml/html/dtd.cpp: changed name of endTag and tagPriority arrays to endTagArray and tagPriorityArray
 194         (DOM::checkChild): treat custom tags as spans during this check
 195         * khtml/html/html_elementimpl.cpp: changed all uses of the endTag array to the endTagRequirement function
 196         (HTMLElementImpl::createContextualFragment):
 197         (HTMLElementImpl::setInnerText):
 198         (HTMLElementImpl::setOuterText):
 199         (HTMLElementImpl::toString):
 200         * khtml/html/htmlparser.cpp: changed all uses of the endTag array to the endTagRequirement function and all uses of the tagPriority array to the tagPriority function.
 201         (KHTMLParser::parseToken):
 202         (KHTMLParser::insertNode):
 203         * khtml/html/htmltokenizer.cpp: (khtml::HTMLTokenizer::parseTag): changed all uses of the endTag array to the endTagRequirement function
 204        * khtml/editing/markup.cpp: changed all uses of the endTag array to the endTagRequirement function
 205         (khtml::endMarkup):
 206         (khtml::markup):
 207
 208 2005-02-18  Jens Alfke  <jens@apple.com>
 209
 210         Reviewed by Hyatt.
 211
 212         Fixed build: Whoops, setNeedsLayout's parameter is NOT optional.
 213
 214         * khtml/html/html_objectimpl.cpp:
 215         (HTMLAppletElementImpl::setAllParamsAvailable):
 216
 217 2005-02-18  Jens Alfke  <jens@apple.com>
 218
 219         Reviewed by cblu, hyatt.
 220
 221         Fixes <rdar://problem/3603191> "REGRESSION: Applets not receiving all of the Applet Parameters in Java 1.4.1/1.4.2"
 222         Defer instantiation of Java applet plugin until after all of the <applet> tag's nested <param> tags have been parsed, otherwise the list of parameters passed to the applet is incomplete. The regression was introduced (says Dave) when the parser's close-tag notifications were removed in the name of performance.
 223
 224         * khtml/html/html_objectimpl.cpp:
 225         (HTMLAppletElementImpl::HTMLAppletElementImpl):
 226         (HTMLAppletElementImpl::getAppletInstance):
 227         (HTMLAppletElementImpl::setAllParamsAvailable):
 228         (HTMLAppletElementImpl::allParamsAvailable):
 229         * khtml/html/html_objectimpl.h:
 230         * khtml/html/htmlparser.cpp:
 231         (KHTMLParser::processCloseTag):
 232         * khtml/rendering/render_applet.cpp:
 233         (RenderApplet::createWidgetIfNecessary):
 234
 235 2005-02-18  Richard Williamson   <rjw@apple.com>
 236
 237         Fixed <rdar://problem/4006161> Tiger8A380: Widgets leak dashboard regions
 238
 239         We were leaking the DashboardRegionImpls.
 240
 241         Reviewed by Kevin.
 242
 243         * khtml/css/css_valueimpl.cpp:
 244         (DOM::CSSPrimitiveValueImpl::cleanup):
 245
 246 2005-02-18  Chris Blumenberg  <cblu@apple.com>
 247
 248         Fixed: <rdar://problem/3945271> REGRESSION (Mail): pasted plain text should pick up typing style instead of being unstyled
 249
 250         Reviewed by kocienda.
 251
 252         * khtml/editing/htmlediting.cpp:
 253         (khtml::ApplyStyleCommand::addInlineStyleIfNeeded): don't set class on element returned by createStyleSpanElement since that's already done
 254         (khtml::ReplacementFragment::ReplacementFragment): take matchStyle param, don't call computeStylesAndRemoveUnrendered() if !matchStyle
 255         (khtml::ReplaceSelectionCommand::ReplaceSelectionCommand): take matchStyle param
 256         (khtml::ReplaceSelectionCommand::doApply): if m_matchStyle, use selection.start() as the insertion position and don't call applyStyleToInsertedNodes()
 257         * khtml/editing/htmlediting.h:
 258         * kwq/WebCoreBridge.h:
 259         * kwq/WebCoreBridge.mm:
 260         (-[WebCoreBridge replaceSelectionWithFragment:selectReplacement:smartReplace:matchStyle:]): take matchStyle param and pass it
 261         (-[WebCoreBridge replaceSelectionWithNode:selectReplacement:smartReplace:]): pass NO for matchStyle
 262         (-[WebCoreBridge replaceSelectionWithMarkupString:baseURLString:selectReplacement:smartReplace:]): pass NO for matchStyle
 263         (-[WebCoreBridge replaceSelectionWithText:selectReplacement:smartReplace:]): pass YES for matchStyle
 264
 265 2005-02-18  Ken Kocienda  <kocienda@apple.com>
 266
 267         Reviewed by Chris
 268
 269         Fix for these bugs:
 270
 271         <rdar://problem/4013025> Copy/Paste of quoted word results in removal of any following <cr>
 272         <rdar://problem/4013100> Copy/Paste quoted text and then decrease quote level does not change text color
 273
 274         For the most part, these bugs were caused by errors and lack of foresight on my part when
 275         I added the better paste code. Chalk these fixes up to the result of bake time.
 276
 277         * khtml/editing/htmlediting.cpp:
 278         (khtml::ReplacementFragment::ReplacementFragment): Need to move count of number of blocks in
 279         fragment after the call to remove unrendered nodes. Meant to do this before, but forgot to.
 280         (khtml::ReplacementFragment::removeStyleNodes): Need to remove inline styles from elements!
 281         Terrible omission now fixed.
 282         (khtml::ReplacementFragment::removeBlockquoteColorsIfNeeded): Remove blockquote colors for now.
 283         Code has a more extensive comment in it now to explain the difficulty, and the need for more
 284         study and changes.
 285         (khtml::ReplaceSelectionCommand::doApply): Need to call applyStyleToInsertedNodes() in the
 286         m_fragment.hasInterchangeNewline() case. This was just missed before.
 287         * layout-tests/editing/pasteboard/paste-text-011-expected.txt: Updated results, subtly different, but OK.
 288         * layout-tests/editing/pasteboard/paste-text-017-expected.txt: Updated for <p> to <div> change in test content.
 289         * layout-tests/editing/pasteboard/paste-text-017.html: Needed to change <p> to <div> to
 290         make this test go with the new design of using <div> tags for default paragraphs.
 291
 292 2005-02-18  David Hyatt  <hyatt@apple.com>
 293
 294         Fix for 3974263 (and possibly others).  Don't let fixed tables use maxint as their maxwidth when some
 295         cells have percentage values.
 296
 297         Reviewed by kocienda
 298
 299         * khtml/misc/arena.cpp:
 300         (ArenaAllocate):
 301         * khtml/rendering/table_layout.cpp:
 302         (FixedTableLayout::calcWidthArray):
 303         (FixedTableLayout::calcMinMaxWidth):
 304         (AutoTableLayout::layout):
 305
 306 2005-02-17  Darin Adler  <darin@apple.com>
 307
 308         Reviewed by Kevin.
 309
 310         - fixed <rdar://problem/3998627> WebKit crashes when deleting text in a modified div
 311
 312         * khtml/khtml_part.cpp: (KHTMLPart::selectionComputedStyle): Add a ref and deref to keep
 313         the element alive until it's added to the DOM tree.
 314         * kwq/KWQKHTMLPart.mm: (KWQKHTMLPart::styleForSelectionStart): Ditto.
 315
 316 2005-02-17  Richard Williamson   <rjw@apple.com>
 317
 318         Fixed <rdar://problem/3999467> when Osaka-Mono is specified as fixed width font, Osaka used instead
 319
 320         Fixed w/o introducing a performance regression.  Add early
 321         check for Osaka-Mono to avoid expensive call into WebKit.
 322
 323         Reviewed by Vicki.
 324
 325         * kwq/KWQFont.mm:
 326         (QFont::isFixedPitch):
 327
 328 2005-02-17  Darin Adler  <darin@apple.com>
 329
 330         Reviewed by Kevin.
 331
 332         - fixed <rdar://problem/4011210> REGRESSION (180-181): Maxlength property for INPUT object not working
 333
 334         * kwq/KWQTextField.mm:
 335         (-[KWQTextField textView:shouldChangeTextInRange:replacementString:]): If controller returns
 336         YES, go on to call super, since super does the text formatter handling, which we need.
 337         (-[KWQSecureTextField textView:shouldChangeTextInRange:replacementString:]): Ditto.
 338         (-[KWQSearchField textView:shouldChangeTextInRange:replacementString:]): Ditto.
 339
 340 === Safari-187 ===
 341
 342 2005-02-17  Ken Kocienda  <kocienda@apple.com>
 343
 344         Reviewed by Hyatt
 345
 346         Fix for this bug:
 347
 348         <rdar://problem/4012058> Copy from quoted text and paste results in blue text
 349
 350         The relatively-new paste code tries hard to retain style of the content from the
 351         source location. However, in the case of quoted material in mail messages, we do
 352         not want to carry the quoting color along. This fixes the problem by factoring
 353         out such colors.
 354
 355         * khtml/editing/htmlediting.cpp:
 356         (khtml::ReplacementFragment::computeStylesAndRemoveUnrendered): Now calls removeBlockquoteColorsIfNeeded()
 357         to fix bug.
 358         (khtml::ReplacementFragment::removeBlockquoteColorsIfNeeded): New function. Factors out colors that are
 359         the result of text being quoted.
 360         (khtml::isNodeRendered): Function moved in file. No other change.
 361         (khtml::isProbablyBlock): Function moved in file. No other change.
 362         (khtml::closestMailBlockquote): New function. Helps fix bug.
 363         (khtml::isMailBlockquote): Function moved to be free-floating instead of being a
 364         member of CompositeEditCommand. No other change.
 365         * khtml/editing/htmlediting.h: Moved around some functions. Added removeBlockquoteColorsIfNeeded().
 366         * khtml/editing/markup.cpp: Remove redundant static implementation of isMailBlockquote.
 367
 368 2005-02-17  Richard Williamson   <rjw@apple.com>
 369
 370         Fixed <rdar://problem/4008163> dynamic support for -apple-dashboard-region is flakey
 371
 372         Style operator== wasn't including regions.
 373
 374         Reviewed by Darin.
 375
 376         * khtml/rendering/render_style.cpp:
 377         (StyleCSS3NonInheritedData::operator==):
 378
 379 2005-02-17  Adele Amchan  <adele@apple.com>
 380
 381         Reviewed by Chris.
 382
 383         fix for <rdar://problem/4010028> 8A383: Safari v185 crash loading united.com multi city fare finder page.
 384
 385         * khtml/dom/dom_string.cpp: (DOM::DOMString::operator += ): prevent nil dereference when DOMString being added is nil
 386
 387 2005-02-17  Jens Alfke  <jens@apple.com>
 388
 389         Reviewed by Hyatt.
 390
 391         Fix for rdar://3963151 "Mail only pasted ~950 of 1407 text lines into my message!"
 392         Force tokenizer to run synchronously while parsing document fragments, so it doesn't stop halfway through and cause truncated content.
 393
 394         * khtml/html/html_elementimpl.cpp:
 395         (HTMLElementImpl::createContextualFragment):
 396         * khtml/html/htmltokenizer.cpp:
 397         (khtml::HTMLTokenizer::reset):
 398         (khtml::HTMLTokenizer::begin):
 399         (khtml::HTMLTokenizer::setForceSynchronous):
 400         (khtml::HTMLTokenizer::continueProcessing):
 401         * khtml/html/htmltokenizer.h:
 402
 403 2005-02-17  Ken Kocienda  <kocienda@apple.com>
 404
 405         Reviewed by Hyatt
 406
 407         Fix for this bug:
 408
 409         <rdar://problem/3982183> Mail's HTML paragraphs appear with unintended margins in Entourage and Eudora
 410
 411         * khtml/editing/htmlediting.cpp:
 412         (khtml::createDefaultParagraphElement): Change element we make from <p> to <div>.
 413         This fixes the problem, as these other mailers show <div> elements without margins.
 414
 415         Some test files needed to change to preserve the ability to "eyeball" the results.
 416         Many others changed just because their <p> elements changed to <div> elements.
 417
 418         * layout-tests/editing/deleting/delete-3928305-fix-expected.txt
 419         * layout-tests/editing/deleting/delete-at-paragraph-boundaries-002-expected.txt
 420         * layout-tests/editing/inserting/insert-div-013-expected.txt
 421         * layout-tests/editing/inserting/insert-div-013.html
 422         * layout-tests/editing/inserting/insert-div-014-expected.txt
 423         * layout-tests/editing/inserting/insert-div-014.html
 424         * layout-tests/editing/inserting/insert-div-018-expected.txt
 425         * layout-tests/editing/inserting/insert-div-018.html
 426         * layout-tests/editing/inserting/insert-div-019-expected.txt
 427         * layout-tests/editing/inserting/insert-div-019.html
 428         * layout-tests/editing/inserting/insert-div-020-expected.txt
 429         * layout-tests/editing/inserting/insert-div-020.html
 430         * layout-tests/editing/inserting/insert-div-021-expected.txt
 431         * layout-tests/editing/inserting/insert-div-021.html
 432         * layout-tests/editing/style/remove-underline-across-paragraph-expected.txt
 433         * layout-tests/editing/style/remove-underline-across-paragraph-in-bold-expected.txt
 434         * layout-tests/editing/style/remove-underline-after-paragraph-expected.txt
 435         * layout-tests/editing/style/remove-underline-after-paragraph-in-bold-expected.txt
 436
 437 2005-02-17  Ken Kocienda  <kocienda@apple.com>
 438
 439         Reviewed by Hyatt
 440
 441         Fix for this bug:
 442
 443         <rdar://problem/3996737> REGRESSION (Mail): Copy/paste in Mail inserts returns
 444
 445         The root cause of the problem is that a couple of pieces of code in AppKit and Mail
 446         insert newlines into markup as it is generated, and the paste code in WebCore was
 447         not smart about nodes that do not render (as is the case with these added newlines).
 448         So, the solution is to remove these unrendered nodes in the paste code in a pre-pass.
 449         Fortunately, my recent addition of such a pass to the paste code to handle styles
 450         gave me a convenient place to put this new logic.
 451
 452         * khtml/editing/htmlediting.cpp:
 453         (khtml::ReplacementFragment::ReplacementFragment): Use new name for function below.
 454         (khtml::ReplacementFragment::computeStylesAndRemoveUnrendered): Name change from computeStylesForNodes().
 455         Now does the additional work of removing unrendered nodes, as mentioned above.
 456         (khtml::isNodeRendered): New helper function.
 457         * khtml/editing/htmlediting.h: Updated as necessary.
 458
 459 2005-02-17  Ken Kocienda  <kocienda@apple.com>
 460
 461         Reviewed by Hyatt
 462
 463         Fix for this bug:
 464
 465         <rdar://problem/3998892> REGRESSION (Mail): bolding a selection from end of line changes unselected text on starting line.
 466
 467         The styling code did not move to the next node when the starting position was
 468         at the last offset of a node. Instead, it styled it. Clearly wrong. Solution
 469         is to borrow a check from the delete algorithm. Though I call caretMaxOffset(),
 470         which I consider deprecated, it is still the simplest and most strightforward
 471         way to ask the right question for this kind of problem.
 472
 473         * khtml/editing/htmlediting.cpp:
 474         (khtml::ApplyStyleCommand::applyInlineStyle): Fixed, as described above.
 475         * layout-tests/editing/style/style-3998892-fix-expected.txt: Added.
 476         * layout-tests/editing/style/style-3998892-fix.html: Added.
 477
 478 2005-02-16  Chris Blumenberg  <cblu@apple.com>
 479
 480         Fixed: <rdar://problem/3954842> Forward/reply to an HTML email can result in nothing (in cases with <link> tags for CSS)
 481
 482         Reviewed by john.
 483
 484         * khtml/xml/dom_nodeimpl.cpp:
 485         (NodeBaseImpl::addChild): don't call insertedIntoDocument on the added child if "this" itself is not in the document.
 486
 487 2005-02-16  John Sullivan  <sullivan@apple.com>
 488
 489         Written by Darin, reviewed by Maciej and me
 490
 491         - change required to make previous checkin work with English language RSS pages
 492
 493         * kwq/KWQKURL.h:
 494         new private method isHierarchical
 495         * kwq/KWQKURL.mm:
 496         (KURL::KURL):
 497         add hierarchical base URL check when determining whether the URL is absolute
 498         (KURL::isHierarchical):
 499         new method, returns true if this is a valid URL with a slash just past the scheme's trailing colon
 500
 501 2005-02-16  John Sullivan  <sullivan@apple.com>
 502
 503         Written by Darin, reviewed by me.
 504
 505         - WebCore part of fix for <rdar://problem/4007384>
 506         FILTER: Bookmark of RSS with Japanese search word & multiple RSS pages loses filter
 507
 508         * kwq/DOM.mm:
 509         (-[DOMElement _getURLAttribute:]):
 510         new SPI method, uses parseURL and completeURL to get valid URL from attribute value
 511
 512         * kwq/DOMPrivate.h:
 513         declare new SPI method
 514
 515 2005-02-15  Maciej Stachowiak  <mjs@apple.com>
 516
 517         Reviewed by Darin.
 518
 519         <rdar://problem/3942428> reproducible crash loading cbs.sportsline.com
 520
 521         This change reverts the fix for <rdar://problem/3805311>, and
 522         re-fixes it in a different (better) way. Instead of preventing
 523         programmatic open from setting the parsing flag, instead make sure
 524         that programmatic close resets it.
 525
 526         * khtml/khtml_part.cpp:
 527         (KHTMLPart::openURL):
 528         (KHTMLPart::didExplicitOpen):
 529         (KHTMLPart::closeURL):
 530         (KHTMLPart::begin):
 531         (KHTMLPart::end):
 532         (KHTMLPart::endIfNotLoading):
 533         (KHTMLPart::slotFinishedParsing):
 534         (KHTMLPart::checkEmitLoadEvent):
 535         * khtml/khtml_part.h:
 536         * khtml/khtmlpart_p.h:
 537         (KHTMLPartPrivate::KHTMLPartPrivate):
 538         * khtml/xml/dom_docimpl.cpp:
 539         (DocumentImpl::open):
 540         (DocumentImpl::implicitOpen):
 541         (DocumentImpl::close):
 542         (DocumentImpl::implicitClose):
 543         * khtml/xml/dom_docimpl.h:
 544
 545 2005-02-15  David Harrison  <harrison@apple.com>
 546
 547         Reviewed by Maciej.
 548
 549         <rdar://problem/3933665> smart delete seems to delete too much after expanding selection with arrow keys
 550
 551         Fixed by updating the selection granularity.  Only byWord granularity enables smart delete, but in this
 552         case the granularity changed from byWord to byCharacter.
 553
 554         * khtml/khtml_part.cpp:
 555         (KHTMLPart::setSelectionGranularity):
 556         * khtml/khtml_part.h:
 557         * kwq/WebCoreBridge.mm:
 558         (-[WebCoreBridge rangeByExpandingSelectionWithGranularity:]):
 559         (-[WebCoreBridge rangeByAlteringCurrentSelection:direction:granularity:]):
 560         (-[WebCoreBridge alterCurrentSelection:direction:granularity:]):
 561
 562 2005-02-15  David Harrison  <harrison@apple.com>
 563
 564         Reviewed by Darin.
 565
 566         (continued) <rdar://problem/4004305> REGRESSION (Mail): Command-right-arrow on wrapped text goes to end of previous line
 567
 568         This was supposed have been committed yesterday with the other part of the fix, but was not.
 569
 570         * khtml/editing/visible_units.cpp:
 571         (khtml::endOfWord):
 572
 573 2005-02-15  Ken Kocienda  <kocienda@apple.com>
 574
 575         Reviewed by John
 576
 577         Fix for this bug:
 578
 579         <rdar://problem/3951178> REGRESSION (Mail): blank line lost after pasting as quotation
 580
 581         Problem was that the blank line after the selection was getting deleted incorrectly since the
 582         paste code thought this was an unneeded placeholder rather than a placeholder outside of the
 583         selection.
 584
 585         * khtml/editing/htmlediting.cpp:
 586         (khtml::CompositeEditCommand::removeBlockPlaceholderIfNeeded): Now calls findBlockPlaceholder.
 587         (khtml::CompositeEditCommand::findBlockPlaceholder): Moved finding code formerly in
 588         removeBlockPlaceholderIfNeeded to this new helper.
 589         (khtml::ReplaceSelectionCommand::doApply): Do not delete placeholder up front. Call
 590         findBlockPlaceholder, and delete it later if needed in the already-existing cleanup step.
 591         * khtml/editing/htmlediting.h: Add new function.
 592
 593         New layout test.
 594
 595         * layout-tests/editing/pasteboard/paste-text-017-expected.txt: Added.
 596         * layout-tests/editing/pasteboard/paste-text-017.html: Added.
 597
 598         Result changed for the better.
 599
 600         * layout-tests/editing/pasteboard/paste-text-011-expected.txt
 601
 602 2005-02-14  David Harrison  <harrison@apple.com>
 603
 604         Reviewed by Darin.
 605
 606         <rdar://problem/4004305> REGRESSION (Mail): Command-right-arrow on wrapped text goes to end of previous line
 607
 608         Reviewed by Darin.
 609
 610         <rdar://problem/4004305> REGRESSION (Mail): Command-right-arrow on wrapped text goes to end of previous line
 611
 612         * khtml/editing/visible_text.cpp:
 613         (khtml::SimplifiedBackwardsTextIterator::advance):
 614         Add BR in for <rdar://problem/3917929> fix only if leaving a visible text node.
 615
 616         * khtml/editing/visible_units.cpp:
 617         (khtml::endOfWord):
 618         Do not move left over a paragraph boundary.
 619
 620 2005-02-14  Darin Adler  <darin@apple.com>
 621
 622         Reviewed by John.
 623
 624         - fixed <rdar://problem/3686434> Safari uses too much RAM on file upload, leading to malloc errors and crashes (HP printers)
 625
 626         * khtml/html/html_formimpl.cpp:
 627         (DOM::FormDataList::begin): Updated to use a list of FormDataListItem instead of QCString.
 628         (DOM::FormDataList::end): Ditto.
 629         (DOM::HTMLFormElementImpl::formData): Same here, and also use appendFile rather than appendData
 630         when we encounter a path name rather than data.
 631         (DOM::HTMLInputElementImpl::appendFormData): Use appendFile here rather than reading the file
 632         in; the reading now happens inside WebKit.
 633         (DOM::FormDataList::appendString): Updated for FormDataListItem.
 634         (DOM::FormDataList::appendFile): Added.
 635
 636         * ForwardingHeaders/kfileitem.h: Emptied out the file; no longer includes KWQKFileItem.h.
 637         * ForwardingHeaders/netaccess.h: Emptied out the file; no longer includes KWQKIONetAccess.h.
 638         * kwq/KWQKFileItem.h: Removed.
 639         * kwq/KWQKFileItem.mm: Removed.
 640         * kwq/KWQKIONetAccess.h: Removed.
 641         * kwq/KWQKIONetAccess.mm: Removed.
 642         * WebCore.pbproj/project.pbxproj: Removed the 4 files above.
 643
 644         - small unrelated fix (not reviewed by John)
 645
 646         * khtml/ecma/kjs_window.cpp: (Window::isSafeScript): Put some logging code outside if statements so it works all the time.
 647
 648 2005-02-14  David Harrison  <harrison@apple.com>
 649
 650         Reviewed by Darin.
 651
 652         <rdar://problem/4004305> REGRESSION (Mail): Command-right-arrow on wrapped text goes to end of previous line
 653
 654         Fixed nextLinePosition to calculate affinity rather than take it as a parameter.  Propagated the parameter change out to related methods.
 655
 656         * khtml/editing/htmlediting.cpp:
 657         (khtml::DeleteSelectionCommand::initializePositionData):
 658         (khtml::InsertLineBreakCommand::doApply):
 659         (khtml::InsertParagraphSeparatorCommand::doApply):
 660         (khtml::InsertTextCommand::input):
 661         (khtml::ReplaceSelectionCommand::doApply):
 662         (khtml::ReplaceSelectionCommand::applyStyleToInsertedNodes):
 663         * khtml/editing/selection.cpp:
 664         (khtml::Selection::modifyExtendingRightForward):
 665         (khtml::Selection::modifyMovingRightForward):
 666         (khtml::Selection::modifyExtendingLeftBackward):
 667         (khtml::Selection::modifyMovingLeftBackward):
 668         (khtml::Selection::modify):
 669         (khtml::Selection::validate):
 670         * khtml/editing/visible_position.cpp:
 671         (khtml::visiblePositionsOnDifferentLines):
 672         * khtml/editing/visible_units.cpp:
 673         (khtml::rootBoxForLine):
 674         (khtml::startOfLine):
 675         (khtml::endOfLine):
 676         (khtml::inSameLine):
 677         (khtml::isStartOfLine):
 678         (khtml::isEndOfLine):
 679         (khtml::previousLinePosition):
 680         (khtml::nextLinePosition):
 681         (khtml::previousSentencePosition):
 682         (khtml::nextSentencePosition):
 683         (khtml::previousParagraphPosition):
 684         (khtml::nextParagraphPosition):
 685         * khtml/editing/visible_units.h:
 686         * khtml/khtml_events.cpp:
 687         (khtml::MouseEvent::offset):
 688         * khtml/khtml_part.cpp:
 689         (KHTMLPart::isPointInsideSelection):
 690         (KHTMLPart::selectClosestWordFromMouseEvent):
 691         (KHTMLPart::handleMousePressEventTripleClick):
 692         (KHTMLPart::handleMousePressEventSingleClick):
 693         (KHTMLPart::handleMouseMoveEventSelection):
 694         (KHTMLPart::khtmlMouseReleaseEvent):
 695         * khtml/rendering/render_block.cpp:
 696         (khtml::RenderBlock::positionForCoordinates):
 697         * khtml/rendering/render_block.h:
 698         * khtml/rendering/render_br.cpp:
 699         (RenderBR::positionForCoordinates):
 700         * khtml/rendering/render_br.h:
 701         * khtml/rendering/render_container.cpp:
 702         (RenderContainer::positionForCoordinates):
 703         * khtml/rendering/render_container.h:
 704         * khtml/rendering/render_inline.cpp:
 705         (RenderInline::positionForCoordinates):
 706         * khtml/rendering/render_inline.h:
 707         * khtml/rendering/render_object.cpp:
 708         (RenderObject::caretRect):
 709         (RenderObject::positionForCoordinates):
 710         * khtml/rendering/render_object.h:
 711         * khtml/rendering/render_replaced.cpp:
 712         (RenderReplaced::positionForCoordinates):
 713         * khtml/rendering/render_replaced.h:
 714         * khtml/rendering/render_text.cpp:
 715         (RenderText::positionForCoordinates):
 716         * khtml/rendering/render_text.h:
 717         * khtml/xml/dom_position.cpp:
 718         (DOM::Position::previousCharacterPosition):
 719         (DOM::Position::nextCharacterPosition):
 720         (DOM::Position::leadingWhitespacePosition):
 721         (DOM::Position::trailingWhitespacePosition):
 722         * khtml/xml/dom_position.h:
 723         * kwq/KWQAccObject.mm:
 724         (-[KWQAccObject value]):
 725         (-[KWQAccObject accessibilityAttributeValue:]):
 726         (-[KWQAccObject doAXLineForTextMarker:]):
 727         (-[KWQAccObject doAXTextMarkerRangeForLine:]):
 728         (-[KWQAccObject doAXTextMarkerForPosition:]):
 729         (-[KWQAccObject doAXLeftLineTextMarkerRangeForTextMarker:]):
 730         (-[KWQAccObject doAXRightLineTextMarkerRangeForTextMarker:]):
 731         (-[KWQAccObject doAXNextLineEndTextMarkerForTextMarker:]):
 732         (-[KWQAccObject doAXPreviousLineStartTextMarkerForTextMarker:]):
 733         * kwq/KWQKHTMLPart.mm:
 734         * kwq/WebCoreBridge.mm:
 735         (-[WebCoreBridge _visiblePositionForPoint:]):
 736
 737 05-02-07  Maciej Stachowiak  <mjs@apple.com>
 738
 739         Reviewed by Ken and John.
 740
 741         Re-fixed a specific case of the following:
 742
 743         <rdar://problem/3790449> REGRESSION (Mail): underline behavior is flaky because of how CSS handles text-decoration
 744
 745         Oddly, Cmd-B, Cmd-U, type some text, return, Cmd-U, Cmd-B, type
 746         some text, worked fine. But hitting the second Cmd-B before the
 747         second Cmd-U still failed to remove underlining. The reason for
 748         this is that our code to compute the style of the current position
 749         did not work when you had a typing style and were positioned right
 750         at a <br> element. For various reasons, this did not show up for
 751         bold and italic, since those are handled through the font manager.
 752
 753         The following change fixes this - for elements that can't have
 754         children, we add the dummy span after the element of interest,
 755         rather than as a child of it.
 756
 757         * khtml/khtml_part.cpp:
 758         (KHTMLPart::selectionComputedStyle):
 759
 760 2005-02-11  David Harrison  <harrison@apple.com>
 761
 762         Reviewed by Darin.
 763
 764         <rdar://problem/3978980> Double Clicking on a line in Mail selected the entire body
 765
 766         * khtml/editing/visible_units.cpp:
 767         (khtml::startOfWord):
 768         (khtml::endOfWord):
 769         Pay attention to being at the end of a paragraph.
 770
 771         (khtml::previousLinePosition):
 772         (khtml::nextLinePosition):
 773         (khtml::endOfParagraph):
 774         Use DOWNSTREAM per recent affinity changes.
 775
 776 2005-02-11  Richard Williamson   <rjw@apple.com>
 777
 778         Fixed <rdar://problem/3985118> DOM objects not being marshaled on JS->native calls
 779
 780         Re-factored how 'native' wrappers for JS objects are created.  The interpreter now
 781         creates these wrappers.  The WebCore subclass of the interpreter now overrides
 782         createLanguageInstanceForValue() and creates a DOM ObjC wrapper for DOM objects.
 783
 784         * WebCore.pbproj/project.pbxproj:
 785         * khtml/ecma/kjs_binding.cpp:
 786         (ScriptInterpreter::createLanguageInstanceForValue):
 787         * khtml/ecma/kjs_binding.h:
 788         * kwq/DOMUtility.mm: Added.
 789         (KJS::ScriptInterpreter::createObjcInstanceForValue):
 790         * kwq/KWQKHTMLPart.mm:
 791         (KWQKHTMLPart::getAppletInstanceForView):
 792         (getInstanceForView):
 793         (KWQKHTMLPart::getEmbedInstanceForView):
 794         (KWQKHTMLPart::getObjectInstanceForView):
 795
 796 2005-02-11  Chris Blumenberg  <cblu@apple.com>
 797
 798         Fixed: <rdar://problem/3937352> Quote level not maintained when copied and pasted within a Mail message
 799
 800         Reviewed by harrison.
 801
 802         * khtml/editing/markup.cpp:
 803         (khtml::isMailBlockquote): new
 804         (khtml::createMarkup): retain the Mail quote level by including all ancestor mail block quotes
 805         * khtml/rendering/render_block.cpp:
 806         (khtml::RenderBlock::positionForCoordinates): default to downstream for the affinity since the affinity is only upstream when the cursor is clicked to the right of a wrapped line
 807         * khtml/rendering/render_text.cpp:
 808         (RenderText::positionForCoordinates): ditto
 809
 810 2005-02-11  Adele Amchan  <adele@apple.com>
 811
 812         Reviewed by Darin.
 813
 814         fix for <rdar://problem/4004004> no need to add body element for xml documents
 815
 816         In the fix for <rdar://problem/3758785> we decided to add a body element when closing a document
 817         to ensure that the onload handler would fire.  This is unnecessary for xml documents,
 818         so now we also check to see if we're dealing with an html document before adding the body element.
 819
 820         * khtml/xml/dom_docimpl.cpp: (DocumentImpl::close):
 821
 822 2005-02-11  Adele Amchan  <adele@apple.com>
 823
 824         Reviewed by Darin.
 825
 826         * layout-tests/apple-only/base/www.excite.com/index-expected.txt: updates tests because of custom tag change.
 827         * layout-tests/apple-only/base/www.msnbc.com/index-expected.txt:
 828         * layout-tests/apple-only/base/www.time.com/index-expected.txt:
 829         * layout-tests/apple-only/base/www.travelocity.com/index-expected.txt:
 830         * layout-tests/apple-only/base/www.zdnet.com/index-expected.txt:
 831         * layout-tests/fast/overflow/003-expected.txt:
 832
 833         * layout-tests/fast/js/window-object-cross-frame-calls-expected.txt: updated test because of added Body (see rdar://problem/3758785)
 834
 835 === Safari-186 ===
 836
 837 2005-02-11  Darin Adler  <darin@apple.com>
 838
 839         Reviewed by John.
 840
 841         - fixed <rdar://problem/3915449> paths are relative to the old src URL after document.open, which is supposed to clear the document, including the URL
 842
 843         * khtml/xml/dom_docimpl.h: Added a new openInternal function for use by KHTMLPart.
 844         * khtml/xml/dom_docimpl.cpp:
 845         (DocumentImpl::open): Changed to do everything we did before, but also clear the URL and set the
 846         base URL based on the enclosing document. This is the basic JavaScript/DOM operation of opening a
 847         document, which is supposed to clear the document, including the URL. In the long run we might want
 848         to do even more document "resetting and clearing" in here, but this URL clearing is what's needed
 849         now to fix the most important problem.
 850         (DocumentImpl::openInternal): Moved the old open code in here, except for the "parsing" check, which
 851         is unnecessary and inappropriate in the one place we call this.
 852
 853         * khtml/khtml_part.cpp: (KHTMLPart::begin): Call openInternal instead of open.
 854
 855 2005-02-10  Ken Kocienda  <kocienda@apple.com>
 856
 857         Reviewed by Hyatt
 858
 859         Fix for this bug:
 860
 861         <rdar://problem/3992092> 8A374: Mail crash while pasting - RemoveNodeCommand
 862
 863         * khtml/editing/htmlediting.cpp:
 864         (khtml::ReplaceSelectionCommand::doApply): Code tried to remove a node that was no longer in
 865         the document. We try to do a good job of detecting all these cases, and generally do. This
 866         one was missed. Fixed.
 867
 868 2005-02-10  Darin Adler  <darin@apple.com>
 869
 870         Reviewed by Harrison.
 871
 872         - added support needed to fix <rdar://problem/3991225> Format->Style->Underline menu item does not get checked when selected text is underlined
 873
 874         * kwq/WebCoreBridge.h: Added selectionHasStyle: method.
 875         * kwq/WebCoreBridge.mm: (-[WebCoreBridge selectionHasStyle:]): Added. Calls selectionHasStyle on the part.
 876
 877 2005-02-10  Darin Adler  <darin@apple.com>
 878
 879         Reviewed by Harrison.
 880
 881         - fixed <rdar://problem/3990484> cursor changes to I-beam when moving over text in widgets, even when selection off via CSS
 882
 883         * khtml/khtmlview.cpp: (KHTMLView::viewportMouseMoveEvent): Changed code to check canSelect when deciding
 884         whether to show an I-beam cursor.
 885
 886         * khtml/rendering/render_object.h: Added canSelect.
 887         * khtml/rendering/render_object.cpp:
 888         (selectStartNode): Added. Helper function with the guts of shouldSelect.
 889         (RenderObject::canSelect): Added. Like shouldSelect, but does not call the "start selecting" event handler.
 890         (RenderObject::shouldSelect): Refactored to use selectStartNode.
 891
 892         * khtml/css/cssvalues.c: Regnerated with newer gperf.
 893
 894 2005-02-10  David Hyatt  <hyatt@apple.com>
 895
 896         Fix for 3867759, .mac regression where scrollers don't show up.  Make sure you can set the .width and
 897         .height properties on the Image object.
 898
 899         Reviewed by John Sullivan
 900
 901         * khtml/ecma/kjs_html.cpp:
 902         (ImageConstructorImp::construct):
 903         (Image::getValueProperty):
 904         (Image::putValue):
 905         (Image::Image):
 906         * khtml/ecma/kjs_html.h:
 907
 908 2005-02-10  Ken Kocienda  <kocienda@apple.com>
 909
 910         Reviewed by Chris
 911
 912         Fix for this bug:
 913
 914         <rdar://problem/3965158> Drag-n-drop within a rich text message sometimes changes the color of the dragged text
 915
 916         This change fixes the bug....and much more. Now, for the first time, the paste code can do "smart merging"
 917         or "smooshing" of styles during its operation. Since this new code is actively, rather than passively
 918         working with styles, it fixes the bug, and lays the groundwork for similar work we need to do to
 919         preserve quote levels in Mail.
 920
 921         * khtml/css/css_valueimpl.cpp:
 922         (DOM::CSSMutableStyleDeclarationImpl::clear): New method.
 923         (DOM::CSSMutableStyleDeclarationImpl::removeBlockProperties): Ditto.
 924         (DOM::CSSMutableStyleDeclarationImpl::removePropertiesInSet): Ditto.
 925         (DOM::operator==): Add operator for CSSProperty.
 926         * khtml/css/css_valueimpl.h: Declare new functions.
 927         * khtml/editing/htmlediting.cpp:
 928         (khtml::isEmptyStyleSpan): Improved the test in this function, rolling together the old implementation
 929         with some code that did this work inline elsewhere. Sum of the parts is better than either test was by itself.
 930         (khtml::isStyleSpan): Check for ID_SPAN.
 931         (khtml::ApplyStyleCommand::removeCSSStyle): Call isEmptyStyleSpan. This was the place with an inline implementation before.
 932         (khtml::ReplacementFragment::ReplacementFragment): Now takes a DocumentImpl argument. No longer does a "default style"
 933         check, but rather calls functions which do a similar check to that, and much more.
 934         (khtml::ReplacementFragment::~ReplacementFragment): Deref document, and computed styles.
 935         (khtml::ReplacementFragment::styleForNode): New helper. Looks up and returns computed style for a node.
 936         (khtml::ReplacementFragment::removeNodePreservingChildren): New helper.
 937         (khtml::ReplacementFragment::computeStylesForNodes): New function which computes the "desired" style for
 938         every node in the fragment. This information is used later after paste is done as a reference for testing
 939         what styles need to be added, and which can be removed as redundant, from all the nodes inserted by the
 940         replacement code.
 941         (khtml::ReplacementFragment::removeStyleNodes): Clears out all style nodes from the fragment. They are
 942         no longer needed after the call to computeStylesForNodes(),
 943         (khtml::ReplaceSelectionCommand::ReplaceSelectionCommand): Add a document to the call to initialize the
 944         command's ReplacementFragment.
 945         (khtml::ReplaceSelectionCommand::doApply): Call applyStyleToInsertedNodes() after inserting nodes to make
 946         styles come out right.
 947         (khtml::ReplaceSelectionCommand::applyStyleToInsertedNodes): This is the "style smooshing" function. It
 948         computes the styles that need to be added to each node inserted, comparing the style it gets from just
 949         being inserted into its correct destination with the computed "desired style" done in the
 950         ReplacementFragment constructor. It then adds in all the necessary styles, and will also remove redundant styles.
 951         * khtml/editing/htmlediting.h: Update declarations and member variables as needed.
 952         * khtml/editing/markup.cpp:
 953         (khtml::startMarkup): Add additional style annotations to the markup we generate, so that paste code can preserve it.
 954         (khtml::markup): Ditto.
 955         (khtml::createMarkup): Ditto.
 956
 957         These test results are subtly better with this change. They no longer have an unneeded empty span.
 958         Visually the same as before.
 959
 960         * layout-tests/editing/style/remove-underline-across-paragraph-expected.txt
 961         * layout-tests/editing/style/remove-underline-across-paragraph-in-bold-expected.txt
 962         * layout-tests/editing/style/remove-underline-expected.txt
 963         * layout-tests/editing/style/remove-underline-from-stylesheet-expected.txt
 964
 965         New tests:
 966
 967         * layout-tests/editing/style/smoosh-styles-001-expected.txt
 968         * layout-tests/editing/style/smoosh-styles-002-expected.txt
 969         * layout-tests/editing/style/smoosh-styles-001.html
 970         * layout-tests/editing/style/smoosh-styles-002.html
 971
 972 2005-02-10  Darin Adler  <darin@apple.com>
 973
 974         Reviewed by Adele.
 975
 976         - fixed <rdar://problem/3974988> WebHTMLView drops scroll wheel events when deltas are 0
 977
 978         * kwq/WebCoreScrollView.m: (-[WebCoreScrollView autoforwardsScrollWheelEvents]): Added, for Tiger only.
 979         For Tiger, removed the override of scrollWheel:, which doesn't work quite right because of how the
 980         delta attributes return all 0 for newfangled scrolling events from the new trackpads.
 981
 982 2005-02-10  David Harrison  <harrison@apple.com>
 983
 984         Reviewed by Darin.
 985
 986         <rdar://problem/3991532> REGRESSION (Mail): Triple-click on trailing blank line should select previous line but doesn't
 987
 988         * khtml/css/cssvalues.c:
 989         (hash_val):
 990         (findValue):
 991         * khtml/editing/selection.cpp:
 992         (khtml::Selection::validate):
 993         Do moral equiavalent of LeftWordIfOnBoundary for PARAGRAPH.
 994
 995 2005-02-10  David Harrison  <harrison@apple.com>
 996
 997         Reviewed by Darin.
 998
 999         <rdar://problem/3991848> Double-click on first character selects wrong item
1000
1001         * khtml/editing/selection.cpp:
1002         (khtml::Selection::validate):
1003         Honor the fact that clicking on a character positions the cursor on the left side of the character.
1004
1005 2005-02-10  David Hyatt  <hyatt@apple.com>
1006
1007         Fix for 3980778, repro crash in RootInlineBox::paint.  Make sure that when merging blocks we dirty them for layout
1008         up front. That way as we do the merge, we dont attempt to examine the lines in dirtyLinesForChangedChild.
1009
1010         Reviewed by mjs
1011
1012         * khtml/rendering/render_block.cpp:
1013         (khtml::RenderBlock::removeChild):
1014
1015 2005-02-10  David Hyatt  <hyatt@apple.com>
1016
1017         Fix for 3987010, fix a bug that caused the self-collapsing-block-cleared-float bit to never get set correctly.
1018
1019         Reviewed by mjs
1020
1021         * khtml/rendering/render_block.h:
1022         (khtml::RenderBlock::MarginInfo::setSelfCollapsingBlockClearedFloat):
1023
1024 2005-02-10  David Hyatt  <hyatt@apple.com>
1025
1026         Fix  for 3992440, text not wrapping correctly on half moon bay web site.  Refine an incorrect optimization I
1027         added to be less restrictive.
1028
1029         Reviewed by mjs
1030
1031         * khtml/rendering/render_block.cpp:
1032         (khtml::RenderBlock::layoutBlockChildren):
1033
1034 2005-02-08  Maciej Stachowiak  <mjs@apple.com>
1035
1036         Reviewed by Darin.
1037
1038         <rdar://problem/3977973> pages on ebay leak referenced JavaScript objects -- over time browsing becomes super-slow
1039
1040         I fixed this by removing all event listeners for a document, it's
1041         children, and any disconnected nodes that used to be in the
1042         document at document detach time. Mozilla temporarily disables
1043         event listeners on such nodes, but re-enables them if you
1044         re-parant a node into a new document. However, in WebCore, you
1045         can't re-parent a node into another document, so there is no
1046         observable change in behavior.
1047
1048         We have to do this to break the possible reference cycles between
1049         event listeners and the dom nodes they are attached to (e.g. via
1050         scope chain, as in this case).
1051
1052         * khtml/xml/dom_docimpl.cpp:
1053         (DocumentImpl::detach):
1054         (DocumentImpl::removeAllEventListenersFromAllNodesx):
1055         (DocumentImpl::registerDisconnectedNodeWithEventListeners):
1056         (DocumentImpl::unregisterDisconnectedNodeWithEventListeners):
1057         (DocumentImpl::removeAllDisconnectedNodeEventListeners):
1058         * khtml/xml/dom_docimpl.h:
1059         * khtml/xml/dom_nodeimpl.cpp:
1060         (NodeImpl::~NodeImpl):
1061         (NodeImpl::addEventListener):
1062         (NodeImpl::removeEventListener):
1063         (NodeImpl::removeAllEventListeners):
1064         (NodeImpl::removeHTMLEventListener):
1065         (NodeImpl::insertedIntoDocument):
1066         (NodeImpl::removedFromDocument):
1067         * khtml/xml/dom_nodeimpl.h:
1068
1069 2005-02-09  Chris Blumenberg  <cblu@apple.com>
1070
1071         Fixed: <rdar://problem/3999213> Sometimes 2 Windows Media Player plugin instances are loaded
1072
1073         Reviewed by hyatt.
1074
1075         * khtml/html/html_objectimpl.cpp:
1076         (HTMLObjectElementImpl::attach): Set needWidgetUpdate to false before calling updateWidget because updateWidget may cause this method or another method (which also calls updateWidget) to be called.
1077         (HTMLObjectElementImpl::recalcStyle): ditto
1078
1079 2005-02-09  David Harrison  <harrison@apple.com>
1080
1081         Reviewed by Darin.
1082
1083         <rdar://problem/3937447> Mail-689: Arrow key navigation in new message body stops working when a line ends with a bold word
1084
1085         Added affinity to VisiblePosition.  Changed Selection code to use affinity more.
1086
1087         (Partial) <rdar://problem/3982096> editing/pasteboard/paste-text-007 is failing
1088
1089         Changed ReplaceSelectionCommand to also pay attention to the top children of the incoming fragment, not just the very last node inserted, when deciding whether to insert a paragraph for the Apple interchange newline.
1090
1091         * khtml/ecma/kjs_window.cpp:
1092         (SelectionFunc::tryCall):
1093         * khtml/editing/htmlediting.cpp:
1094         (khtml::EditCommandPtr::setStartingSelection):
1095         (khtml::EditCommandPtr::setEndingSelection):
1096         (khtml::EditCommand::setStartingSelection):
1097         (khtml::EditCommand::setEndingSelection):
1098         (khtml::CompositeEditCommand::deleteInsignificantTextDownstream):
1099         (khtml::CompositeEditCommand::moveParagraphContentsToNewBlockIfNecessary):
1100         (khtml::ApplyStyleCommand::splitTextAtStartIfNeeded):
1101         (khtml::ApplyStyleCommand::splitTextAtEndIfNeeded):
1102         (khtml::ApplyStyleCommand::splitTextElementAtStartIfNeeded):
1103         (khtml::ApplyStyleCommand::splitTextElementAtEndIfNeeded):
1104         (khtml::ApplyStyleCommand::mergeStartWithPreviousIfIdentical):
1105         (khtml::ApplyStyleCommand::mergeEndWithNextIfIdentical):
1106         (khtml::ApplyStyleCommand::joinChildTextNodes):
1107         (khtml::DeleteSelectionCommand::initializePositionData):
1108         (khtml::DeleteSelectionCommand::handleGeneralDelete):
1109         (khtml::DeleteSelectionCommand::calculateTypingStyleAfterDelete):
1110         (khtml::DeleteSelectionCommand::doApply):
1111         (khtml::InsertLineBreakCommand::doApply):
1112         (khtml::InsertParagraphSeparatorCommand::doApply):
1113         (khtml::InsertParagraphSeparatorInQuotedContentCommand::doApply):
1114         (khtml::InsertTextCommand::input):
1115         (khtml::MoveSelectionCommand::doApply):
1116         (khtml::ReplaceSelectionCommand::ReplaceSelectionCommand):
1117         (khtml::ReplaceSelectionCommand::~ReplaceSelectionCommand):
1118         (khtml::ReplaceSelectionCommand::doApply):
1119         (khtml::ReplaceSelectionCommand::completeHTMLReplacement):
1120         (khtml::ReplaceSelectionCommand::updateNodesInserted):
1121         (khtml::TypingCommand::deleteKeyPressed):
1122         (khtml::TypingCommand::forwardDeleteKeyPressed):
1123         (khtml::TypingCommand::markMisspellingsAfterTyping):
1124         * khtml/editing/htmlediting.h:
1125         * khtml/editing/selection.cpp:
1126         (khtml::Selection::Selection):
1127         (khtml::Selection::init):
1128         (khtml::Selection::moveTo):
1129         (khtml::Selection::modifyExtendingRightForward):
1130         (khtml::Selection::modifyMovingRightForward):
1131         (khtml::Selection::modifyExtendingLeftBackward):
1132         (khtml::Selection::modifyMovingLeftBackward):
1133         (khtml::Selection::modify):
1134         (khtml::Selection::xPosForVerticalArrowNavigation):
1135         (khtml::Selection::clear):
1136         (khtml::Selection::setBase):
1137         (khtml::Selection::setExtent):
1138         (khtml::Selection::setBaseAndExtent):
1139         (khtml::Selection::layout):
1140         (khtml::Selection::validate):
1141         * khtml/editing/selection.h:
1142         (khtml::Selection::startAffinity):
1143         (khtml::Selection::endAffinity):
1144         (khtml::Selection::baseAffinity):
1145         (khtml::Selection::extentAffinity):
1146         (khtml::operator==):
1147         * khtml/editing/text_affinity.h:
1148         (khtml::):
1149         * khtml/editing/visible_position.cpp:
1150         (khtml::VisiblePosition::VisiblePosition):
1151         (khtml::VisiblePosition::init):
1152         (khtml::VisiblePosition::initUpstream):
1153         (khtml::VisiblePosition::initDownstream):
1154         (khtml::VisiblePosition::next):
1155         (khtml::VisiblePosition::previous):
1156         (khtml::startVisiblePosition):
1157         (khtml::endVisiblePosition):
1158         * khtml/editing/visible_position.h:
1159         (khtml::VisiblePosition::):
1160         (khtml::VisiblePosition::VisiblePosition):
1161         (khtml::VisiblePosition::affinity):
1162         (khtml::VisiblePosition::setAffinity):
1163         (khtml::operator==):
1164         * khtml/editing/visible_units.cpp:
1165         (khtml::previousBoundary):
1166         (khtml::nextBoundary):
1167         (khtml::startOfLine):
1168         (khtml::endOfLine):
1169         (khtml::previousLinePosition):
1170         (khtml::nextLinePosition):
1171         (khtml::startOfParagraph):
1172         (khtml::endOfParagraph):
1173         (khtml::previousParagraphPosition):
1174         (khtml::nextParagraphPosition):
1175         (khtml::startOfBlock):
1176         (khtml::endOfBlock):
1177         (khtml::startOfDocument):
1178         (khtml::endOfDocument):
1179         * khtml/editing/visible_units.h:
1180         * khtml/khtml_part.cpp:
1181         (KHTMLPart::findTextNext):
1182         (KHTMLPart::selectClosestWordFromMouseEvent):
1183         (KHTMLPart::handleMousePressEventTripleClick):
1184         (KHTMLPart::handleMousePressEventSingleClick):
1185         (KHTMLPart::handleMouseMoveEventSelection):
1186         (KHTMLPart::khtmlMouseReleaseEvent):
1187         (KHTMLPart::selectAll):
1188         (KHTMLPart::computeAndSetTypingStyle):
1189         (KHTMLPart::selectionComputedStyle):
1190         * khtml/rendering/render_br.cpp:
1191         (RenderBR::positionForCoordinates):
1192         * khtml/xml/dom_docimpl.cpp:
1193         (DocumentImpl::updateSelection):
1194         * khtml/xml/dom_nodeimpl.cpp:
1195         (NodeBaseImpl::setFocus):
1196         * khtml/xml/dom_position.cpp:
1197         (DOM::Position::previousCharacterPosition):
1198         (DOM::Position::nextCharacterPosition):
1199         * khtml/xml/dom_position.h:
1200         * kwq/KWQAccObject.mm:
1201         (-[KWQAccObject value]):
1202         (-[KWQAccObject visiblePositionForStartOfTextMarkerRange:]):
1203         (-[KWQAccObject visiblePositionForEndOfTextMarkerRange:]):
1204         (-[KWQAccObject accessibilityAttributeValue:]):
1205         (-[KWQAccObject doAXLineForTextMarker:]):
1206         (-[KWQAccObject doAXTextMarkerRangeForLine:]):
1207         (-[KWQAccObject doAXTextMarkerForPosition:]):
1208         (-[KWQAccObject doAXLeftLineTextMarkerRangeForTextMarker:]):
1209         (-[KWQAccObject doAXRightLineTextMarkerRangeForTextMarker:]):
1210         (-[KWQAccObject doAXNextWordEndTextMarkerForTextMarker:]):
1211         (-[KWQAccObject doAXPreviousWordStartTextMarkerForTextMarker:]):
1212         (-[KWQAccObject doAXNextLineEndTextMarkerForTextMarker:]):
1213         (-[KWQAccObject doAXPreviousLineStartTextMarkerForTextMarker:]):
1214         (-[KWQAccObject doSetAXSelectedTextMarkerRange:]):
1215         * kwq/KWQAccObjectCache.mm:
1216         (KWQAccObjectCache::textMarkerForVisiblePosition):
1217         (KWQAccObjectCache::visiblePositionForTextMarker):
1218         * kwq/KWQKHTMLPart.mm:
1219         (KWQKHTMLPart::findString):
1220         (KWQKHTMLPart::advanceToNextMisspelling):
1221         (KWQKHTMLPart::styleForSelectionStart):
1222         (KWQKHTMLPart::baseWritingDirectionForSelectionStart):
1223         (KWQKHTMLPart::setSelectionFromNone):
1224         (KWQKHTMLPart::respondToChangedSelection):
1225         * kwq/WebCoreBridge.mm:
1226         (-[WebCoreBridge setSelectedDOMRange:affinity:]):
1227         (-[WebCoreBridge selectionAffinity]):
1228         (-[WebCoreBridge setMarkDOMRange:]):
1229         (-[WebCoreBridge _visiblePositionForPoint:]):
1230         (-[WebCoreBridge moveDragCaretToPoint:]):
1231         (-[WebCoreBridge editableDOMRangeForPoint:]):
1232         (-[WebCoreBridge ensureSelectionVisible]):
1233         (-[WebCoreBridge rangeOfCharactersAroundCaret]):
1234
1235 2005-02-09  Chris Blumenberg  <cblu@apple.com>
1236
1237         Fixed: <rdar://problem/3985211> Seed: Mail: Drag-and-drop destination indicator / insertion point disappears
1238
1239         Reviewed by kocienda.
1240
1241         * khtml/rendering/render_block.cpp:
1242         (khtml::RenderBlock::paintCaret): new
1243         (khtml::RenderBlock::paintObject): call paintCaret for the cursor caret and the drag caret
1244         * khtml/rendering/render_block.h:
1245
1246 === Safari-185 ===
1247
1248 2005-02-08  Ken Kocienda  <kocienda@apple.com>
1249
1250         Reviewed by John
1251
1252         Fix for this bug:
1253
1254         <rdar://problem/3996344> Entire document content is deleted when only the first paragraph was supposed to be
1255
1256         Problem stems from my attempt to fix this bug: <rdar://problem/3986155> Insertion point goes
1257         to beginning of doc after deleting.
1258
1259         I added an "optimization" that proved to do more harm than good. So, basically, the fix is to remove code.
1260
1261         * khtml/editing/htmlediting.cpp:
1262         (khtml::DeleteSelectionCommand::handleGeneralDelete)
1263
1264 2005-02-07  Darin Adler  <darin@apple.com>
1265
1266         Reviewed by Kevin.
1267
1268         - fixed <rdar://problem/3890352> REGRESSION (125-170, Panther-only): pasted textarea contents disappear (exception in NSText)
1269
1270         * kwq/KWQTextArea.mm: (-[KWQTextArea tile]): Don't resize the text view immediately here.
1271         Use performWithDelay to defer it to avoid the re-entrancy that bothers the Panther version
1272         of NSTextView. On Tiger, leave the code alone.
1273
1274 2005-02-07  Adele Amchan  <adele@apple.com>
1275
1276         Reviewed by Darin.
1277
1278         Fix for <rdar://problem/3993628> REGRESSION (180-181): Logitelnet bank's website doesn't work
1279
1280         This fixes an oversight in the fix for <rdar://problem/3964286>.
1281         We're now making sure there is a document before calling checkCompleted in slotLoaderRequestDone.
1282         That function is called even when other parts finish loads of their subresources, so we could end
1283         up calling checkCompleted before we had a document. We also looked over all the other calls to
1284         checkCompleted to convince ourselves no other call sites had a similar issue.
1285
1286         * khtml/khtml_part.cpp: (KHTMLPart::slotLoaderRequestDone): Check if doc is nil before calling
1287         checkCompleted.
1288
1289 2005-02-07  Ken Kocienda  <kocienda@apple.com>
1290
1291         Reviewed by Darin
1292
1293         Fix for this bug:
1294
1295         <rdar://problem/3953302> Replacing quoted text ends up with blue-colored text that is not quoted
1296
1297         * khtml/editing/htmlediting.cpp:
1298         (khtml::DeleteSelectionCommand::doApply): Removed special case delete code for select all. It was
1299         causing as many bugs as it fixed. What's more, the design concept is a bit shaky. By removing
1300         this special case, the bug goes away.
1301         * khtml/editing/htmlediting.h: Remove declaration.
1302
1303         Test results using selectAll() updated to reflect changed behavior.
1304
1305         * layout-tests/editing/deleting/delete-3775172-fix-expected.txt:
1306         * layout-tests/editing/deleting/delete-select-all-001-expected.txt:
1307         * layout-tests/editing/deleting/delete-select-all-003-expected.txt:
1308         * layout-tests/editing/inserting/insert-3654864-fix-expected.txt:
1309         * layout-tests/editing/inserting/insert-3775316-fix-expected.txt:
1310         * layout-tests/editing/inserting/insert-3851164-fix-expected.txt:
1311         * layout-tests/editing/inserting/insert-3907422-fix-expected.txt:
1312         * layout-tests/editing/pasteboard/paste-text-015-expected.txt:
1313
1314 2005-02-06  Darin Adler  <darin@apple.com>
1315
1316         Reviewed by Ken.
1317
1318         - fixed <rdar://problem/3994164> setting a new style attribute with setAttribute("style") doesn't blow away the old style
1319
1320         * khtml/css/css_valueimpl.h: Renamed parseProperty to parseDeclaration, since it's used to parse
1321         an entire style declaration, not a single property.
1322         * khtml/css/css_valueimpl.cpp: (DOM::CSSMutableStyleDeclarationImpl::parseDeclaration): Renamed,
1323         and added code to clear m_values.
1324         * khtml/html/html_elementimpl.cpp: (HTMLElementImpl::parseHTMLAttribute): Call function by its
1325         new name, and remove some unnecessary comments.
1326
1327 2005-02-06  Darin Adler  <darin@apple.com>
1328
1329         Reviewed by Maciej.
1330
1331         - fixed <rdar://problem/3993722> need oncut and onpaste handlers implemented for <textarea> (needed for widgets)
1332
1333         * kwq/KWQTextArea.mm:
1334         (-[KWQTextAreaTextView dispatchHTMLEvent:]): Added.
1335         (-[KWQTextAreaTextView cut:]): Dispatch "before cut" and "cut" events.
1336         (-[KWQTextAreaTextView copy:]): Dispatch "before copy" and "copy" events.
1337         (-[KWQTextAreaTextView paste:]): Dispatch "before paste" and "paste" events.
1338         (-[KWQTextAreaTextView pasteAsPlainText:]): Ditto.
1339         (-[KWQTextAreaTextView pasteAsRichText:]): Ditto.
1340
1341 2005-02-06  Darin Adler  <darin@apple.com>
1342
1343         Reviewed by Maciej.
1344
1345         - fixed <rdar://problem/3943038> <input type=search> that is focused in onload handler doesn't have a visible editor
1346
1347         * khtml/xml/dom_docimpl.cpp: (DocumentImpl::setFocusNode): Always update layout before giving focus to a widget.
1348         This prevents the bad case where we give a widget focus before it has been positioned or sized, causing us to
1349         exercise edge cases AppKit doesn't handle well.
1350
1351         * kwq/KWQTextField.mm: (-[KWQTextFieldController initWithTextField:QLineEdit:]): Remove workaround code I added
1352         to fix bug 3943049. Updating the layout fixes the same problem in a cleaner, safer way, and solves another problem,
1353         with the way the clip view is set up.
1354
1355 2005-02-06  Darin Adler  <darin@apple.com>
1356
1357         Reviewed by Maciej.
1358
1359         - fixed <rdar://problem/3425232> textarea won't trigger onchange action when clicking to new field
1360
1361         * khtml/rendering/render_form.h: Added new m_dirty data member.
1362         * khtml/rendering/render_form.cpp:
1363         (RenderTextArea::RenderTextArea): Initialize m_dirty to false.
1364         (RenderTextArea::detach): Call value() on the DOM element to cause it to fetch the value from this object.
1365         (RenderTextArea::handleFocusOut): Ditto. Also send the onchange event based only on the m_dirty flag.
1366         (RenderTextArea::updateFromElement): Clear the m_dirty flag if the edited value is blown away by a value
1367         that comes from the DOM element. Remove code to clear the element's m_dirtyvalue flag; that's an unnecessary
1368         optimization, and ideally we'll stop having code in the render object that knows about that flag later.
1369         (RenderTextArea::slotTextChanged): Set the m_dirty flag.
1370
1371 2005-02-06  Darin Adler  <darin@apple.com>
1372
1373         Reviewed by Maciej.
1374
1375         - re-fixed <rdar://problem/3760910> please add support for custom tag names in HTML
1376
1377         * khtml/xml/dom_docimpl.cpp:
1378         (DocumentImpl::tagId): Use ID_LAST_TAG + 1 for the ID of the first assigned tag ID. Before we'd use ID_LAST_TAG,
1379         which resulted in the same number being used for ID_COMMENT and the first custom tag.
1380         (DocumentImpl::tagName): Use ID_LAST_TAG + 1, as above.
1381         (DocumentImpl::namespaceURI): Change a < ID_LAST_TAG to <= ID_LAST_TAG in the same spirit as above.
1382
1383 2005-02-06  Darin Adler  <darin@apple.com>
1384
1385         Reviewed by Maciej.
1386
1387         - fixed <rdar://problem/3986639> Crash occurs after choosing Undo Typing from the Edit menu
1388
1389         * kwq/KWQTextArea.mm: (-[KWQTextArea viewWillMoveToWindow:]): Remove actions from undo manager when the view is
1390         removed from the window. Unfortunately, to do this the code has to know that the target of NSTextView actions
1391         will be the text storage object, but given the NSText architecture, that's pretty clear.
1392
1393 2005-02-05  Chris Blumenberg  <cblu@apple.com>
1394
1395         Fixed: <rdar://problem/3991974> REGRESSION: www.jabra.com world location screen does not work
1396
1397         Reviewed by hyatt.
1398
1399         * kwq/KWQKJavaAppletWidget.mm:
1400         (KJavaAppletWidget::KJavaAppletWidget): fallback to the document's base URL if there is no baseURL attribute
1401         * kwq/WebCoreBridge.h:
1402         * kwq/WebCoreBridge.mm:
1403         (-[WebCoreBridge baseURL]): new
1404
1405 2005-02-04  David Harrison  <harrison@apple.com>
1406
1407         Reviewed by Darin.
1408
1409         Rolled out Ken's accidental checkin when committing <rdar://problem/3986155> fix yesterday.
1410
1411         * khtml/editing/selection.cpp:
1412         (khtml::Selection::validate):
1413
1414 2005-02-04  Adele Amchan  <adele@apple.com>
1415
1416         Reviewed by Darin.
1417
1418         Fix for <rdar://problem/3758785> Safari no longer works at BankOfAmerica online banking for military users
1419
1420         When closing the document, if a body doesn't exist we now create one.  This bug was preventing the onload handler from
1421         firing in cases where there wasn't a body.  In the BankOfAmerica case, the page was just a bit of script where the onload
1422         handler redirected to the correct page.
1423
1424         * khtml/xml/dom_docimpl.cpp: (DocumentImpl::close):
1425
1426 2005-02-03  Ken Kocienda  <kocienda@apple.com>
1427
1428         Reviewed by Chris
1429
1430         Fix for this bug:
1431
1432         <rdar://problem/3986155> Insertion point goes to beginning of doc after deleting
1433
1434         * khtml/editing/htmlediting.cpp:
1435         (khtml::DeleteSelectionCommand::setStartNode): New convenience to handle reference counting when setting.
1436         (khtml::DeleteSelectionCommand::handleGeneralDelete): This contains the crux of the bug fix. Improve tests
1437         that detect when a selected node needs to be retained, rather than deleted, to preserve the intent of the user.
1438         This has the side effect of causing the insertion point placement code to succeed rather than fail. Before
1439         this fix, the failure of the insertion point placement code caused the insertion point to jump to the start
1440         of the document, which is the symptom that can be perceived by users when editing.
1441         * khtml/editing/htmlediting.h: Add setStartNode declaration.
1442         * khtml/editing/visible_units.cpp:
1443         (khtml::startOfBlock): This function had a stubbed-in non-tested implementation. Implement and
1444         * layout-tests/editing/deleting/delete-at-paragraph-boundaries-001-expected.txt: Added.
1445         * layout-tests/editing/deleting/delete-at-paragraph-boundaries-001.html: Added.
1446         * layout-tests/editing/deleting/delete-at-paragraph-boundaries-002-expected.txt: Added.
1447         * layout-tests/editing/deleting/delete-at-paragraph-boundaries-002.html: Added.
1448         * layout-tests/editing/deleting/delete-at-paragraph-boundaries-003-expected.txt: Added.
1449         * layout-tests/editing/deleting/delete-at-paragraph-boundaries-003.html: Added.
1450         * layout-tests/editing/deleting/delete-at-paragraph-boundaries-004-expected.txt: Added.
1451         * layout-tests/editing/deleting/delete-at-paragraph-boundaries-004.html: Added.
1452         * layout-tests/editing/deleting/delete-at-paragraph-boundaries-005-expected.txt: Added.
1453         * layout-tests/editing/deleting/delete-at-paragraph-boundaries-005.html: Added.
1454         * layout-tests/editing/deleting/delete-at-paragraph-boundaries-006-expected.txt: Added.
1455         * layout-tests/editing/deleting/delete-at-paragraph-boundaries-006.html: Added.
1456
1457 === Safari-183 ===
1458
1459 2005-02-03  Richard Williamson   <rjw@apple.com>
1460
1461         Fixed <rdar://problem/3987419> Stocks and Weather leak what appears to XMLHTTPRequest results
1462
1463         XMLHTTPRequests were causing massive leaks.  _webcore_initWithHeaderString: did funky things with
1464         self replacement.  Re-wrote to use more traditional factory constructor avoiding self replacement.
1465
1466         Reviewed by David Harrison.
1467
1468         * kwq/KWQLoader.mm:
1469         (+[NSDictionary _webcore_dictionaryWithHeaderString:]):
1470         (KWQServeRequest):
1471         (KWQServeSynchronousRequest):
1472
1473 2005-02-03  Chris Blumenberg  <cblu@apple.com>
1474
1475         Fixed: <rdar://problem/3938763> Cannot view Windows Media Player videos (soundtrack is played twice with delay)
1476
1477         Reviewed by hyatt.
1478
1479         * khtml/html/html_objectimpl.cpp:
1480         (HTMLObjectElementImpl::attach): set needWidgetUpdate to false right after calling updateWidget because dispatchHTMLEvent can end up calling this method again
1481         (HTMLObjectElementImpl::recalcStyle): ditto
1482
1483 2005-02-02  John Sullivan  <sullivan@apple.com>
1484
1485         Reviewed by Chris.
1486
1487         - WebCore part of fix for <rdar://problem/3980651> REGRESSION (125-180): Huge number of pages printed from certain page,
1488         iFrame involved
1489
1490         This also fixes the problems with printing from GMail, yay!
1491
1492         * khtml/rendering/render_flow.cpp:
1493         (RenderFlow::paintLines):
1494         If the current line is taller than the entire page height (e.g. tall iFrame), don't try
1495         to avoid splitting it across pages.
1496
1497         * kwq/WebCoreBridge.mm:
1498         (-[WebCoreBridge computePageRectsWithPrintWidthScaleFactor:printHeight:]):
1499         Handle error cases in a more obvious manner; this will cause future problems like this
1500         to print a blank page and complain to the console on debug builds, rather than print a
1501         zillionty mostly-blank pages.
1502
1503 2005-02-02  Chris Blumenberg  <cblu@apple.com>
1504
1505         Fixed: <rdar://problem/3960304> can't load a particular applet (at www.escape.de) unless it's the first applet to be loaded
1506
1507         Reviewed by hyatt.
1508
1509         * khtml/rendering/render_frames.cpp:
1510         (RenderPartObject::updateWidget): when getting the MIME type from the PARAM tag, make sure the MIME type is the text from the left of the semi-colon if there is one. We do this elsewhere as well.
1511
1512 2005-02-02  Chris Blumenberg  <cblu@apple.com>
1513
1514         Fixed: <rdar://problem/3983628> control-click on WebView is not selecting the word under the cursor (Mail, non-editable WebView)
1515
1516         Reviewed by rjw.
1517
1518         * kwq/KWQKHTMLPart.mm:
1519         (KWQKHTMLPart::sendContextMenuEvent): check if SPI to always enable selecting closest word is enabled
1520         * kwq/WebCoreBridge.h:
1521
1522 2005-02-02  Ken Kocienda  <kocienda@apple.com>
1523
1524         Reviewed by me
1525
1526         * khtml/editing/markup.cpp:
1527         (khtml::createMarkup): Removed debug spam I committed earlier in error.
1528
1529 2005-02-02  Ken Kocienda  <kocienda@apple.com>
1530
1531         Reviewed by Darin
1532
1533         Fix for this bug:
1534
1535         <rdar://problem/3823109> WebKit should support -toggleBaseWritingDirection: (bidi editing support)
1536
1537         Since base writing direction is a paragraph-level property in AppKit, and we use the CSS direction
1538         property in WebCore, which can be applied to inline elements as well as blocks, a new notion has
1539         been added to the WebCore apply style logic. It is now possible to force all properties in a
1540         style declaration to be applied as though they were block properties.
1541
1542         * khtml/editing/htmlediting.cpp:
1543         (khtml::ApplyStyleCommand::ApplyStyleCommand): Add an enum and a flag to this command that controls whether to force all
1544         properties in the style to be applied as block styles. Default is "no", retaining previous behavior.
1545         (khtml::ApplyStyleCommand::doApply): Switch on new flag to apply styles as before, or force all preoperties to be applied
1546         as block styles.
1547         * khtml/editing/htmlediting.h:
1548         (khtml::ApplyStyleCommand::): Add an enum and a flag, as above.
1549         * khtml/khtml_part.cpp:
1550         (KHTMLPart::applyParagraphStyle): New method to force application of all style properties as block styles.
1551         * khtml/khtml_part.h: Update header declarations.
1552         * kwq/KWQKHTMLPart.h: Update header declarations.
1553         * kwq/KWQKHTMLPart.mm:
1554         (KWQKHTMLPart::baseWritingDirectionForSelectionStart): Accessor to help WebKit do a toggle operation on
1555         writing direction.
1556         * kwq/WebCoreBridge.h: Update header declarations.
1557         * kwq/WebCoreBridge.mm:
1558         (-[WebCoreBridge applyParagraphStyle:withUndoAction:]): New method to force application of all style properties as block styles.
1559         (-[WebCoreBridge baseWritingDirectionForSelectionStart]): Accessor to help WebKit do a toggle operation on
1560         writing direction.
1561
1562         Fix for this bug:
1563
1564         <rdar://problem/3985035> REGRESSION (Mail): Text copied from wrapped line contains extra character when pasted
1565
1566         * khtml/editing/markup.cpp:
1567         (khtml::renderedText): Add an enum and a flag to this command that controls whether to force all
1568         This is the result of an error in the code that computes the rendered text that is selected
1569         when copying. Since spaces collapse at the end of lines, and these spaces need to be copied
1570         when the selection spans line endings, code runs to compute this text. However, this code
1571         was also running incorrectly in cases where lines wrapped. I have now added the missing
1572         test to check that the selection does indeed extend to the end of the line.
1573
1574
1575 2005-02-02  Ken Kocienda  <kocienda@apple.com>
1576
1577         Reviewed by Darin
1578
1579         Fix for these bugs:
1580
1581         <rdar://problem/3984894> REGRESSION (Mail): Command-right-arrow in reply does the wrong thing (two ways)
1582         <rdar://problem/3985130> REGRESSION (Mail): command-right-arrow in pasted RTF selects only up to tab
1583
1584         * khtml/editing/visible_units.cpp:
1585         (khtml::endOfLine): Two separate problems in this new function recently added to take the place of
1586         selectionForLine. In the first, endOfLine did not stop at BR elements, and returned the position
1587         beyond them. In the second, the VisiblePosition constructor gives the wrong answer in certain cases.
1588         Darin is going to work on a solution for that problem. In the meantime, I can fix the symptom of this
1589         bug by going down to the last leaf child of the root line box; a one line change we will roll out when
1590         Darin's fix is in.
1591
1592 2005-02-01  Richard Williamson   <rjw@apple.com>
1593
1594         Fixed <rdar://problem/3985535> QT Plug-in JavaScript support now fails in <embed> tag only case
1595
1596         Reviewed by Maciej.
1597
1598         * khtml/ecma/kjs_html.cpp:
1599         (KJS::HTMLDocument::tryGet):
1600
1601 2005-02-01  John Sullivan  <sullivan@apple.com>
1602
1603         Written by Darin, reviewed and tested by me
1604
1605         - fixed <rdar://problem/3969684> Panther-only: extra blank lines between
1606         pasted content from Blot
1607
1608         * kwq/KWQKHTMLPart.mm:
1609         (KWQKHTMLPart::attributedString):
1610         don't emit paragraph breaks if the margins are tiny; matches what
1611         we do for plain-text conversion
1612
1613 === Safari-182 ===
1614
1615 2005-02-01  Ken Kocienda  <kocienda@apple.com>
1616
1617         Reviewed by John
1618
1619         Fix for this bug:
1620
1621         <rdar://problem/3985160> Deficiencies in pasting architecture blocking progress on other bugs
1622
1623         * khtml/editing/html_interchange.h: Move style span text used to mark element added to
1624         add style to this header.
1625         * khtml/editing/htmlediting.cpp:
1626         (khtml::styleSpanClassString): Change to use constant moved to html_interchange.h.
1627         (khtml::isStyleSpan): New helper function. Checks if this is a span we added to apply style.
1628         (khtml::CompositeEditCommand::insertNodeBefore): Added an assert to check that the node
1629         we are inserting before is not the body.
1630         (khtml::CompositeEditCommand::insertNodeAfter): Ditto, but check is for after.
1631         (khtml::ReplacementFragment::ReplacementFragment): Added code to process the "default style"
1632         that is added by the copy code.
1633         (khtml::ReplaceSelectionCommand::ReplaceSelectionCommand): first and last nodes inserted are
1634         now member variables instead of function locals. Initialize them here.
1635         (khtml::ReplaceSelectionCommand::~ReplaceSelectionCommand): Deref first and last nodes inserted
1636         if necessary.
1637         (khtml::ReplaceSelectionCommand::doApply): Change design to fix the bug. Major change is to
1638         separate out the code that inserts nodes into the tree so additional styling checks can
1639         be done in a centralized way. Also got rid of the notion of "merging into the end block." That
1640         concept was just wrong.
1641         (khtml::ReplaceSelectionCommand::completeHTMLReplacement): Tweak interface now that first and
1642         last nodes inserted are member variables.
1643         (khtml::ReplaceSelectionCommand::insertNodeAfterAndUpdateNodesInserted): New helper used
1644         by replace code to do the stated DOM operation and update state internal to the command.
1645         This will also be a catch point to handle the kinds of additional style checks needed to
1646         make paste work right.
1647         (khtml::ReplaceSelectionCommand::insertNodeAtAndUpdateNodesInserted): Ditto.
1648         (khtml::ReplaceSelectionCommand::insertNodeBeforeAndUpdateNodesInserted): Ditto.
1649         (khtml::ReplaceSelectionCommand::updateNodesInserted): Ditto.
1650         * khtml/editing/htmlediting.h: Update declarations as needed.
1651         * khtml/editing/markup.cpp:
1652         (khtml::createMarkup): Adds a "default style" span to the content written to the pasteboard.
1653         This will help us to fix some of the bugs blocked by the bug above.
1654         * khtml/xml/dom_nodeimpl.cpp:
1655         (NodeImpl::lastDescendent): New helper.
1656         * khtml/xml/dom_nodeimpl.h: Ditto.
1657         * khtml/xml/dom_position.cpp:
1658         (DOM::Position::upstream): Fixed a bug which would allow the upstream position returned to be
1659         in unrendered content.
1660         (DOM::Position::downstream): Ditto.
1661
1662         * layout-tests/editing/deleting/delete-3775172-fix-expected.txt: Ending positions tweaked due to
1663         changes in upstream() and downstream() functions.
1664         * layout-tests/editing/inserting/insert-3851164-fix-expected.txt: Ditto
1665         * layout-tests/editing/inserting/insert-3907422-fix-expected.txt: Ditto
1666         * layout-tests/editing/selection/extend-by-character-006-expected.txt: Ditto
1667
1668         * layout-tests/editing/pasteboard/paste-text-003-expected.txt: Changed what we expect, given
1669         new behavior of paste code.
1670
1671 2005-01-31  John Sullivan  <sullivan@apple.com>
1672
1673         Reviewed by Dave Hyatt.
1674
1675         - fixed <rdar://problem/3983097> Tabbing on RSS pages gets stuck in search field
1676         with "full keyboard navigation" off
1677
1678         * kwq/KWQSlider.h:
1679         * kwq/KWQSlider.mm:
1680         (QSlider::focusPolicy):
1681         Implemented this method for KWQSlider. It is needed for our widget subclasses that
1682         can become focused in some situations. I missed this when fixing 3949203 because
1683         it only comes into play when tab-to-controls is on, and because the other relevant
1684         methods are objective-C methods.
1685
1686 2005-01-31  David Harrison  <harrison@apple.com>
1687
1688         Reviewed by Darin.
1689
1690         <rdar://problem/3964164> AXWebArea needs to unregister on going invisible
1691
1692         * khtml/xml/dom_docimpl.cpp:
1693         (DocumentImpl::detach):
1694         Detach the AX UIElement for the AXWebArea.
1695
1696 2005-01-31  David Harrison  <harrison@apple.com>
1697
1698         Reviewed by Darin
1699
1700         <rdar://problem/3850876> AX: button exposed as only an AXImage at http://appleseed.apple.com/ site (<input type=image>)
1701
1702         * khtml/rendering/render_form.h:
1703         (khtml::RenderImageButton::isImageButton):
1704         New.
1705         * khtml/rendering/render_image.h:
1706         (khtml::RenderImage::isImageButton):
1707         New.
1708         * kwq/KWQAccObject.mm:
1709         (-[KWQAccObject isImageButton]):
1710         (-[KWQAccObject actionElement]):
1711         (-[KWQAccObject role]):
1712         (-[KWQAccObject accessibilityActionNames]):
1713         (-[KWQAccObject accessibilityPerformAction:]):
1714         Represent RenderImageButton elements as buttons instead of images.
1715
1716 2005-01-31  David Harrison  <harrison@apple.com>
1717
1718         Prevent nil dereference in debug code.
1719
1720         * khtml/xml/dom_nodeimpl.cpp:
1721         (NodeImpl::displayTree):
1722         Stop when parent is 0, too.
1723
1724 2005-01-31  John Sullivan  <sullivan@apple.com>
1725
1726         Reviewed by Dave Hyatt.
1727
1728         - fixed <rdar://problem/3949203> cannot tab to, within, or out of the RSS sidebar
1729
1730         * kwq/KWQSlider.mm:
1731         (-[KWQSlider becomeFirstResponder]):
1732         (-[KWQSlider resignFirstResponder]):
1733         (-[KWQSlider nextKeyView]):
1734         (-[KWQSlider previousKeyView]):
1735         (-[KWQSlider canBecomeKeyView]):
1736         (-[KWQSlider nextValidKeyView]):
1737         (-[KWQSlider previousValidKeyView]):
1738         Copied these methods from KWQButton. The lack of special handling for first-responder-ness
1739         here was causing the focus to get "stuck" on the RSS page's slider widget. Added FIXME
1740         about how it would be nice to share more of this code rather than replicating it in each
1741         KWQ widget subclass.
1742
1743 2005-01-31  Darin Adler  <darin@apple.com>
1744
1745         Reviewed by Harrison.
1746
1747         - fixed <rdar://problem/3980066> Double-click on single character moves insertion point to previous line
1748
1749         * khtml/khtml_part.cpp:
1750         (KHTMLPart::selectClosestWordFromMouseEvent): Set affinity too.
1751         (KHTMLPart::handleMousePressEventTripleClick): Ditto.
1752         (KHTMLPart::handleMouseMoveEventSelection): Ditto.
1753         (KHTMLPart::khtmlMouseReleaseEvent): Ditto.
1754
1755         * khtml/editing/selection.cpp:
1756         (khtml::Selection::modifyExtendingRightForward): Use endOfLine and endOfDocument.
1757         (khtml::Selection::modifyMovingRightForward): Ditto.
1758         (khtml::Selection::modifyExtendingLeftBackward): Use startOfLine and startOfDocument.
1759         (khtml::Selection::modifyMovingLeftBackward): Ditto.
1760         (khtml::Selection::validate): Rewrote the section that handles double-click. Two main fixes: 1) use isStartOfLine to
1761         check for another case where we want to select the word to the right, and 2) use isEndOfParagraph, which seems
1762         to work correctly in cases where isLastVisiblePositionInParagraph is giving the wrong answer. Also changed the line
1763         code to use startOfLine/endOfLine and the document code to use startOfDocument/endOfDocument.
1764
1765 2005-01-31  Darin Adler  <darin@apple.com>
1766
1767         Reviewed by Harrison.
1768
1769         - fixed <rdar://problem/3935275> unexpected quit scrolling over link; last.fm (works in IE and Firefox)
1770
1771         * kwq/WebCoreBridge.mm: (-[WebCoreBridge elementAtPoint:]): By using the correct document, fix the crash reported here.
1772         The page still crashes when you close the browser window, but it's not trivial to fix so I'll file a new bug after that.
1773
1774 2005-01-31  Darin Adler  <darin@apple.com>
1775
1776         Reviewed by Ken and Harrison.
1777
1778         - fixed <rdar://problem/3947901> REGRESSION (Mail): Pasting paragraph of rich text leaves insertion point before pasted text
1779         - fixed <rdar://problem/3949790> hitting return after underlined line results in too much or too little underlined
1780         - fixed <rdar://problem/3981759> nil-deref and crash when pasting just a paragraph break
1781         - fixed a couple problems I discovered while working with bug 3949790
1782
1783         * khtml/editing/htmlediting.cpp:
1784         (khtml::ApplyStyleCommand::applyInlineStyle): Pass StayInBlock to upstream. Without this, we end up going too far
1785         upstream in the test case in bug 3949790.
1786         (khtml::ApplyStyleCommand::removeInlineStyle): Pass StayInBlock to upstream and downstream. Same reason as above.
1787         (khtml::ReplaceSelectionCommand::doApply): Update endPos if inserting a new node and endPos is using that node's
1788         parent and an offset past the node being inserted. That change fixes a problem with the position of the insertion point
1789         after pasting into the top level of a document (from test cases in 3947901 and 3949790). When setting insertionPos, use
1790         code that works when lastNodeInserted is a block rather than a text node. That change fixes a problem where a newline is
1791         not added when pasting an entire paragraph into the end of a document (from test case in 3949790). Added nil check before
1792         checking if lastNodeInserted is a <br> element, which fixes the crash when pasting just a paragraph break.
1793
1794         * khtml/editing/visible_units.h: Filled out the set of calls to add some boolean checks for lines (needed for the
1795         bug fix), and calls for blocks (not yet implemented), and documents. The document checks may need refinement to
1796         properly handle documents with a mix of editable and non-editable content, but for now they just refactor code
1797         and make things a little clearer. Also removed the "include line break" parameter from endOfSentence.
1798         * khtml/editing/visible_units.cpp:
1799         (khtml::rootBoxForLine): Added.
1800         (khtml::startOfLine): Added. Algorithm taken from selectionForLine in selection.cpp.
1801         (khtml::endOfLine): Ditto.
1802         (khtml::inSameLine): Added.
1803         (khtml::isStartOfLine): Added.
1804         (khtml::isEndOfLine): Added.
1805         (khtml::endOfSentence): Removed "include line break" parameter.
1806         (khtml::inSameParagraph): Added a null check.
1807         (khtml::isStartOfParagraph): Ditto.
1808         (khtml::isEndOfParagraph): Ditto.
1809         (khtml::startOfBlock): Added.
1810         (khtml::endOfBlock): Added.
1811         (khtml::inSameBlock): Added.
1812         (khtml::isStartOfBlock): Added.
1813         (khtml::isEndOfBlock): Added.
1814         (khtml::startOfDocument): Added.
1815         (khtml::endOfDocument): Added.
1816         (khtml::inSameDocument): Added.
1817         (khtml::isStartOfDocument): Added.
1818         (khtml::isEndOfDocument): Added.
1819
1820 2005-01-30  Darin Adler  <darin@apple.com>
1821
1822         Reviewed by John.
1823
1824         - fixed <rdar://problem/3977000> form data set posted for <input type="image" ...> omits name/value pair
1825
1826         * khtml/html/html_formimpl.cpp:
1827         (DOM::HTMLInputElementImpl::appendFormData): Append a name/value pair for type IMAGE in addition to the X/Y
1828         coordinates.
1829
1830 2005-01-30  Darin Adler  <darin@apple.com>
1831
1832         Reviewed by John.
1833
1834         - fixed <rdar://problem/3974246> REGRESSION (125-180): popup menus are missing item text on Harmony Remote web site
1835
1836         * khtml/html/html_formimpl.cpp: (DOM::HTMLOptionElementImpl::text): Change this function to traverse the entire tree
1837         and gather all the text rather than just looking at immediate children.
1838
1839         * khtml/xml/dom_nodeimpl.h: Added const.
1840         * khtml/xml/dom_nodeimpl.cpp:
1841         (NodeImpl::traverseNextNode): Added const.
1842         (NodeImpl::traverseNextSibling): Ditto.
1843         (NodeImpl::traversePreviousNodePostOrder): Ditto.
1844         (NodeImpl::detach): Add a missing nil check.
1845
1846 2005-01-30  Darin Adler  <darin@apple.com>
1847
1848         Reviewed by John.
1849
1850         - fixed <rdar://problem/3864536> crash copying text from other apps and pasting in a sticky in dashboard
1851
1852         * khtml/editing/markup.cpp: (khtml::createFragmentFromText): Added missing ref/deref of newly created element.
1853
1854 2005-01-28  Darin Adler  <darin@apple.com>
1855
1856         * khtml/css/cssproperties.c: Regenerated with newer gperf.
1857
1858 2005-01-28  David Harrison  <harrison@apple.com>
1859
1860         Reviewed by John Sullivan.
1861
1862         <rdar://problem/3968144> AX need to be able to focus an AXLink by setting AXFocused to true
1863
1864         * kwq/KWQAccObject.mm:
1865         (-[KWQAccObject accessibilityPerformAction:]):
1866         (-[KWQAccObject accessibilityIsAttributeSettable:]):
1867         Add AXFocused for AXLinks.
1868
1869 2005-01-28  David Harrison  <harrison@apple.com>
1870
1871         Reviewed by Darin.
1872
1873         <rdar://problem/3584942> AX: Safari Accessibility parent-child mismatch
1874
1875         Use AppKit SPI _accessibilityParentForSubview to return KWQAccObject parent of AppKit AX object.
1876
1877         * kwq/KWQAccObject.mm:
1878         (-[KWQAccObject _accessibilityParentForSubview:]):
1879         New.
1880
1881 2005-01-27  Adele Amchan  <adele@apple.com>
1882
1883         fixed by Darin, reviewed by me.
1884
1885         <rdar://problem/3976314> REGRESSION (180-TOT): submitting password fields fail on Panther
1886
1887         * kwq/KWQTextField.mm: (-[KWQTextFieldController string]): on panther, the secure text field's editor
1888         does not contain the real string, so now we just call stringValue on the field in that case.
1889
1890 2005-01-26  Maciej Stachowiak  <mjs@apple.com>
1891
1892         Reviewed by Darin, Hyatt and Ken.
1893
1894         <rdar://problem/3790449> REGRESSION (Mail): underline behavior is flaky because of how CSS handles text-decoration
1895
1896         * khtml/css/css_computedstyle.cpp:
1897         (DOM::CSSComputedStyleDeclarationImpl::getPropertyCSSValue): Add support
1898         for -khtml-text-decorations-in-effect property. This works like text-docration
1899         but follows proper inline semantics, that is, if a parent is causing text
1900         decoration then -khtml-text-decorations-in-effect on this element will reflect that.
1901         * khtml/css/cssproperties.in: Add -khtml-text-decorations-in-effect property.
1902         * khtml/css/cssproperties.h: regenerated
1903         * khtml/css/cssvalues.c: regenerated
1904         * khtml/css/cssparser.cpp: regenerated
1905         * khtml/css/cssproperties.c: regenerated
1906
1907         * khtml/css/css_valueimpl.h: Made copyPropertiesInSet public.
1908
1909         * khtml/editing/htmlediting.cpp:
1910         (khtml::StyleChange::init): Handle -khtml-text-decorations-in-effect
1911         specially, translate to text-decoration
1912         (khtml::ApplyStyleCommand::applyInlineStyle): Fix a comment.
1913         (khtml::ApplyStyleCommand::removeInlineStyle): Add new way of pushing down
1914         text-decoration styles around the edges of the selected range.
1915         (khtml::ApplyStyleCommand::pushDownTextDecorationStyleAroundNode): Helper for this.
1916         (khtml::ApplyStyleCommand::pushDownTextDecorationStyleAtBoundaries): ditto
1917         (khtml::ApplyStyleCommand::nodeFullyUnselected): ditto
1918         (khtml::hasTextDecorationProperty): ditto
1919         (khtml::highestAncestorWithTextDecoration): ditto
1920         (khtml::ApplyStyleCommand::extractTextDecorationStyle): ditto
1921         (khtml::ApplyStyleCommand::extractAndNegateTextDecorationStyle): ditto
1922         (khtml::ApplyStyleCommand::applyTextDecorationStyle): ditto
1923         * khtml/editing/htmlediting.h: Prototype new methods.
1924         * khtml/editing/jsediting.cpp: Make underline execCommand use the new CSS property.
1925
1926         - as a necessary part of the fix, fixed the longstanding problem
1927         that mutating an element's inlineStyleDecl would not property
1928         update its style attribute. now it does.
1929
1930         * khtml/css/css_valueimpl.cpp:
1931         (DOM::CSSMutableStyleDeclarationImpl::setChanged): If this is an inline style
1932         declaration, tell the element it's style attribute needs updating.
1933         * khtml/html/html_elementimpl.cpp:
1934         (HTMLElementImpl::invalidateStyleAttribute): New method, just calls down
1935         to the dom element.
1936         (HTMLElementImpl::updateStyleAttribute): Regenerate style attribute from
1937         inline style declaration.
1938         * khtml/html/html_elementimpl.h: Prototype new stuff.
1939         * khtml/xml/dom_elementimpl.cpp:
1940         (ElementImpl::ElementImpl): Initialize new attribute.
1941         (ElementImpl::updateStyleAttributeIfNeeded): New method to check if the style
1942         is dirty and this is an html element, and if so call the html element to
1943         regenerate the style attribute.
1944         (ElementImpl::attributes): update style if needed
1945         (ElementImpl::getAttribute): ditto
1946         (ElementImpl::hasAttributes): ditto
1947         (ElementImpl::dump): ditto
1948         (XMLElementImpl::cloneNode): ditto
1949         * khtml/xml/dom_elementimpl.h:
1950         (DOM::ElementImpl::isStyleAttributeValid): new method to manage style attr
1951         dirty bit
1952         (DOM::ElementImpl::setStyleAttributeValid): ditto
1953
1954         - added new layout tests for underlining:
1955
1956         * layout-tests/editing/editing.js:
1957         * layout-tests/editing/style/remove-underline-across-paragraph-expected.txt: Added.
1958         * layout-tests/editing/style/remove-underline-across-paragraph-in-bold-expected.txt: Added.
1959         * layout-tests/editing/style/remove-underline-across-paragraph-in-bold.html: Added.
1960         * layout-tests/editing/style/remove-underline-across-paragraph.html: Added.
1961         * layout-tests/editing/style/remove-underline-after-paragraph-expected.txt: Added.
1962         * layout-tests/editing/style/remove-underline-after-paragraph-in-bold-expected.txt: Added.
1963         * layout-tests/editing/style/remove-underline-after-paragraph-in-bold.html: Added.
1964         * layout-tests/editing/style/remove-underline-after-paragraph.html: Added.
1965         * layout-tests/editing/style/remove-underline-expected.txt: Added.
1966         * layout-tests/editing/style/remove-underline-from-stylesheet-expected.txt: Added.
1967         * layout-tests/editing/style/remove-underline-from-stylesheet.html: Added.
1968         * layout-tests/editing/style/remove-underline-in-bold-expected.txt: Added.
1969         * layout-tests/editing/style/remove-underline-in-bold.html: Added.
1970         * layout-tests/editing/style/remove-underline.html: Added.
1971         * layout-tests/editing/style/underline-expected.txt: Added.
1972         * layout-tests/editing/style/underline.html: Added.
1973
1974 2005-01-27  David Hyatt  <hyatt@apple.com>
1975
1976         Fix for 3875199, search field needs to be able to  show a magnifying  glass without a dropdown menu.
1977
1978         Reviewed by kocienda
1979
1980         * khtml/html/html_formimpl.cpp:
1981         (DOM::HTMLInputElementImpl::HTMLInputElementImpl):
1982         (DOM::HTMLInputElementImpl::parseHTMLAttribute):
1983         * kwq/KWQLineEdit.mm:
1984         (QLineEdit::setMaxResults):
1985
1986 2005-01-27  Ken Kocienda  <kocienda@apple.com>
1987
1988         Reviewed by Maciej
1989
1990         Fix for this bug:
1991
1992         <rdar://problem/3973254> Deletions of ranges does not coalesce correctly with subsequent typing
1993
1994         * khtml/editing/htmlediting.cpp:
1995         (khtml::InsertTextCommand::deleteCharacter): Remove this dead code.
1996         (khtml::ReplaceSelectionCommand::editingAction): Moved this code to a better place in the file.
1997         It was in with unrelated functions.
1998         (khtml::TypingCommand::TypingCommand): Reorganize initialization list so it is easier to read.
1999         (khtml::TypingCommand::issueCommandForDeleteKey): Remove dead code. Roll in remaining code into deleteKeyPressed.
2000         (khtml::TypingCommand::deleteKeyPressed): Add in support for carrying along smart delete flag.
2001         (khtml::TypingCommand::forwardDeleteKeyPressed): New function. Makes forward delete work like
2002         "regular" delete in terms of coalescing typing.
2003         (khtml::TypingCommand::doApply): Add case for ForwardDeleteKey.
2004         (khtml::TypingCommand::preservesTypingStyle): Ditto.
2005         * khtml/editing/htmlediting.h:
2006         (khtml::TypingCommand::): Add ForwardDeleteKey constant. Remove a couple declarations for now-dead code.
2007         (khtml::TypingCommand::smartDelete): New accessor.
2008         (khtml::TypingCommand::setSmartDelete): Ditto.
2009         * kwq/WebCoreBridge.h:
2010         * kwq/WebCoreBridge.mm:
2011         (-[WebCoreBridge deleteKeyPressedWithSmartDelete:]): Add smart delete flag.
2012         (-[WebCoreBridge forwardDeleteKeyPressedWithSmartDelete:]): New method.
2013         * khtml/editing/jsediting.cpp: Add supprt for ForwardDelete command
2014         * layout-tests/editing/editing.js: Ditto.
2015         * layout-tests/editing/deleting/delete-and-undo-expected.txt: Added.
2016         * layout-tests/editing/deleting/delete-and-undo.html: Added.
2017         * layout-tests/editing/deleting/forward-delete-expected.txt: Added.
2018         * layout-tests/editing/deleting/forward-delete.html: Added.
2019
2020 === Safari-181 ===
2021
2022 2005-01-27  David Harrison  <harrison@apple.com>
2023
2024         Reviewed by Ken.
2025
2026         <rdar://problem/3962214> AX: AXLeftLineTextMarkerRangeForTextMarker returns incorrect range
2027
2028         * kwq/KWQAccObject.mm:
2029         (-[KWQAccObject doAXRightLineTextMarkerRangeForTextMarker:]):
2030         Allowed for selection to end of line including the linebreak.
2031
2032 2005-01-27  David Harrison  <harrison@apple.com>
2033
2034         Reviewed by Darin, Ken.
2035
2036         <rdar://problem/3964470> AX: Include attachments in AXAttributedStringForTextMarkerRange
2037
2038         * kwq/KWQAccObject.mm:
2039         (-[KWQAccObject isAttachment]):
2040         (-[KWQAccObject attachmentView]):
2041         (-[KWQAccObject role]):
2042         (-[KWQAccObject subrole]):
2043         (-[KWQAccObject roleDescription]):
2044         (-[KWQAccObject value]):
2045         (-[KWQAccObject title]):
2046         (-[KWQAccObject accessibilityDescription]):
2047         (-[KWQAccObject accessibilityIsIgnored]):
2048         (-[KWQAccObject accessibilityAttributeNames]):
2049         (-[KWQAccObject accessibilityAttributeValue:]):
2050         Make attachments accessible.
2051
2052         (CreateCGColorIfDifferent):
2053         (AXAttributeStringSetColor):
2054         (AXAttributeStringSetNumber):
2055         (AXAttributeStringSetFont):
2056         (AXAttributeStringSetStyle):
2057         (AXAttributeStringSetElement):
2058         (AXLinkElementForNode):
2059         (AXAttributedStringAppendText):
2060         (AXAttributedStringAppendReplaced):
2061         (-[KWQAccObject doAXAttributedStringForTextMarkerRange:]):
2062         Add attachments and more text attributes to AXAttributedStringForTextMarkerRange.
2063
2064 2005-01-27  Darin Adler  <darin@apple.com>
2065
2066         Reviewed by John.
2067
2068         - fixed <rdar://problem/3807935> DOM CSS computed style line-height is wrong in two ways
2069
2070         * khtml/css/css_computedstyle.cpp: (DOM::CSSComputedStyleDeclarationImpl::getPropertyCSSValue):
2071         Use specified size rather than computed size when dealing with a percentage for line height.
2072
2073 2005-01-27  Darin Adler  <darin@apple.com>
2074
2075         Reviewed by John.
2076
2077         - fixed <rdar://problem/3971372> SWB: template function DOM_cast() won't compile with gcc 4.0
2078         - fixed other gcc 4.0 compiling problems Patrick Beard pointed out on the phone
2079
2080         * kwq/DOMInternal.h: (DOM_cast): Change the "failToCompile()" trick to use a dependent name.
2081         Nowadays, if a name is not dependent, it is checked at template definition time. But we want
2082         a failure only at template instantiation time.
2083
2084         * ForwardingHeaders/editing/text_granularity.h: Added. Needed but a bug in the 3.3 compiler made
2085         it compile anyway without this.
2086
2087         - other changes
2088
2089         * khtml/html/html_miscimpl.h: For clarity, mark these functions virtual too. They are automatically
2090         virtual because the base class ones are virtual, but it's ugly to leave it this way.
2091
2092 2005-01-26  Ken Kocienda  <kocienda@apple.com>
2093
2094         Reviewed by Hyatt
2095
2096         Fix for this bug:
2097
2098         <rdar://problem/3971609> REGRESSION (Mail): up/down arrow navigation broken after rewrapping text by resizing window
2099         <rdar://problem/3975661> REGRESSION (Mail): left/right arrow navigation can place insertion point after last character on line
2100
2101         * khtml/editing/selection.cpp:
2102         (khtml::Selection::modifyAffinity): PARAGRAPH and LINE movements should not alter affinity. This was
2103         just a mistake before when it reset to upstream in some cases.
2104         * khtml/rendering/render_line.cpp:
2105         (khtml::RootInlineBox::closestLeafChildForXPos): Move this function down from InlineBox, and
2106         rework the implementation. This fixes 3971609.
2107         * khtml/rendering/render_line.h:
2108         * khtml/rendering/render_text.cpp:
2109         (RenderText::caretRect): Rework the algorithm to fix 3975661. This regressed when we began placing
2110         the spaces on the end of lines in text boxes when lines wrap.
2111
2112 2005-01-26  Richard Williamson   <rjw@apple.com>
2113
2114         Fixed <rdar://problem/3972522> (179-180) 40% slowdown on iBench JavaScript test
2115
2116         I added a member variable to ObjectImp.  This changed it's size and consequently
2117         hampered the optimizations built into the garbage collector.  Objects no longer
2118         fit within the allocators cell size, and thus allocation fell back to a slower
2119         allocator.
2120
2121         As a result of this fix I also dramatically cleaned up how runtime objects are
2122         accessed.  The path mostly *removes* code.
2123
2124         Reviewed by Chris.
2125
2126         * khtml/ecma/kjs_dom.cpp:
2127         (DOMDocumentProtoFunc::tryCall):
2128         (DOMElementProtoFunc::tryCall):
2129         (KJS::getRuntimeObject):
2130         * khtml/ecma/kjs_dom.h:
2131         * khtml/ecma/kjs_html.cpp:
2132         (KJS::HTMLDocument::tryGet):
2133         (KJS::HTMLElement::tryGet):
2134         (KJS::HTMLElement::implementsCall):
2135         (KJS::HTMLElement::call):
2136         (KJS::HTMLElement::tryPut):
2137         (KJS::HTMLCollection::tryGet):
2138         (KJS::HTMLCollection::getNamedItems):
2139         * khtml/ecma/kjs_html.h:
2140         * khtml/ecma/kjs_window.cpp:
2141         (Window::get):
2142
2143 2005-01-26  Richard Williamson   <rjw@apple.com>
2144
2145         Fixed <rdar://problem/3757712> REGRESSION (Mail): WebCore does not allow Devanagari ligature input
2146
2147         For now we are using ICU UBreakIterator to determine grapheme boundaries for
2148         cursor and deletion.  This does not match what Cocoa does exactly, but does match
2149         what Carbon does.  The areas are difference are obscure, but, according to
2150         Deborah Goldsmith, using the UBreakIterator is the reasonable approach.
2151
2152         Reviewed by Ken.
2153
2154         * khtml/rendering/render_text.cpp:
2155         (RenderText::previousOffset):
2156         (RenderText::nextOffset):
2157
2158 2005-01-25  David Harrison  <harrison@apple.com>
2159
2160         Reviewed by Maciej and Richard.
2161
2162         <rdar://problem/3963731> AX VO: Changing window sizes causes a crash using with Voice Over - KWQPtrDictImpl::clear
2163
2164         * khtml/html/html_miscimpl.cpp:
2165         (HTMLCollectionImpl::HTMLCollectionImpl):
2166         Ensure that isHTMLDocument before calling HTMLDocument-only method.
2167
2168 2005-01-25  Richard Williamson   <rjw@apple.com>
2169
2170         Part one of fix for <rdar://problem/3757712> REGRESSION (Mail): WebCore does not allow Devanagari ligature input
2171
2172         This patch changes the semantics of next() and previous() on
2173         VisiblePosition to move a grapheme (character cluster) at a
2174         time.  This means that cursor navigation with correctly move
2175         over an entire cluster.
2176
2177         However, the expected behavior for deleting a grapheme is to
2178         delete individual code points, thus decomposing the grapheme
2179         into it constituent parts.  That will be addressed in the next
2180         part of the fix.
2181
2182         Reviewed by Ken.
2183
2184         * khtml/editing/visible_position.cpp:
2185         (khtml::VisiblePosition::previousPosition):
2186         (khtml::VisiblePosition::nextPosition):
2187         * khtml/editing/visible_position.h:
2188         * khtml/rendering/render_object.cpp:
2189         (RenderObject::previousOffset):
2190         (RenderObject::nextOffset):
2191         * khtml/rendering/render_object.h:
2192         * khtml/rendering/render_text.cpp:
2193         (RenderText::previousOffset):
2194         (RenderText::nextOffset):
2195         (RenderText::findNextInlineTextBox):
2196         * khtml/rendering/render_text.h:
2197         * khtml/xml/dom_nodeimpl.cpp:
2198         (NodeImpl::previousOffset):
2199         (NodeImpl::nextOffset):
2200         * khtml/xml/dom_nodeimpl.h:
2201
2202 2005-01-25  David Harrison  <harrison@apple.com>
2203
2204         Reviewed by Maciej.
2205
2206         <rdar://problem/3973067> AXWebArea for http://apple.netscape.com/ has an empty AXLinkUIElements
2207
2208         * kwq/KWQAccObject.mm:
2209         (-[KWQAccObject accessibilityAttributeValue:]):
2210         Omit ignored AXLinks from AXLinkUIElements result
2211
2212 2005-01-25  Ken Kocienda  <kocienda@apple.com>
2213
2214         Reviewed by John
2215
2216         Fix for this bug:
2217
2218         <rdar://problem/3972851> REGRESSION (179-180+): Repro crash in ApplyStyleCommand::cleanUpEmptyStyleSpans
2219
2220         * khtml/editing/htmlediting.cpp:
2221         (khtml::ApplyStyleCommand::cleanUpEmptyStyleSpans): Added two null checks.
2222
2223 2005-01-25  Ken Kocienda  <kocienda@apple.com>
2224
2225         Reviewed by John
2226
2227         Fix for this bug:
2228
2229         <rdar://problem/3954710> Mail crashed while editing signatures - NodeImpl::isBlockFlow
2230
2231         The fix is more general than for this one bug, and may work to fix many crashers. The problem
2232         is that the ReplaceSelectionCommand never checked whether its starting selection is empty. If
2233         it is, then we need to bail before doing the work of the command, which we need to deref the
2234         start and end points of the selection in order to do its work. I think you can see the crash
2235         potential.
2236
2237         * khtml/editing/htmlediting.cpp:
2238         (khtml::ReplaceSelectionCommand::doApply): Assert selection is not empty.
2239         * kwq/WebCoreBridge.mm:
2240         (partHasSelection): New helper function to test that bridge has a part with a selection.
2241
2242         Use new helper function to test part and selection; return from these function if this test fails.
2243
2244         (-[WebCoreBridge rangeByExpandingSelectionWithGranularity:])
2245         (-[WebCoreBridge rangeByAlteringCurrentSelection:direction:granularity:])
2246         (-[WebCoreBridge alterCurrentSelection:direction:granularity:])
2247         (-[WebCoreBridge rangeByAlteringCurrentSelection:verticalDistance:])
2248         (-[WebCoreBridge alterCurrentSelection:verticalDistance:])
2249         (-[WebCoreBridge documentFragmentWithText:])
2250         (-[WebCoreBridge replaceSelectionWithFragment:selectReplacement:smartReplace:])
2251         (-[WebCoreBridge insertLineBreak])
2252         (-[WebCoreBridge insertParagraphSeparator])
2253         (-[WebCoreBridge insertParagraphSeparatorInQuotedContent])
2254         (-[WebCoreBridge insertText:selectInsertedText:])
2255         (-[WebCoreBridge deleteSelectionWithSmartDelete:])
2256         (-[WebCoreBridge ensureSelectionVisible])
2257
2258 2005-01-24  Kevin Decker  <kdecker@apple.com>
2259
2260         Reviewed by Darin.
2261
2262         Fixed <rdar://problem/3932374> REGRESSION: 'ReferenceError - Can't find variable' JavaScript error at webxpress.fidelity.com
2263
2264         * khtml/html/html_miscimpl.h: Missing virtual identifiers for namedItem() and nextNamedItem() exposed a flaw that prevented finding variables in form [HTMLFormCollectionImpl] scopes.
2265
2266
2267 2005-01-24  Vicki Murley <vicki@apple.com>
2268
2269         Reviewed by kocienda.
2270
2271         - fix <rdar://problem/3810661> computed style not handled for Apple extensions
2272
2273         * khtml/css/css_computedstyle.cpp: (DOM::):
2274         (DOM::CSSComputedStyleDeclarationImpl::getPropertyCSSValue): add -apple-line-clamp and -apple-text-size-adjust
2275
2276 2005-01-24  Darin Adler  <darin@apple.com>
2277
2278         Reviewed by Kevin.
2279
2280         - fixed <rdar://problem/3969884> REGRESSION (179-180): Typing password not echoed as "bullets" at paypal.com when field is first focused
2281
2282         * kwq/KWQTextField.mm: (-[KWQTextFieldController initWithTextField:QLineEdit:]):
2283         Sizing up the form editor, which fixes things for normal fields and search fields,
2284         screws things up for password fields. Eventually, I'll need to figure out why this is
2285         and file a bug so that the AppKit team fixes it. In the mean time, just check for the
2286         case of the secure text field, and don't size up in that case.
2287
2288 2005-01-24  Ken Kocienda  <kocienda@apple.com>
2289
2290         Reviewed by Hyatt
2291
2292         Fix for this bug:
2293
2294         <rdar://problem/3963560> 8A354: Inserting Japanese text to the line head causes line break
2295
2296         * khtml/editing/htmlediting.cpp:
2297         (khtml::ReplaceSelectionCommand::doApply): Add one more case where we merge content into the
2298         existing line. This covers the situation see using the steps to reproduce this bug.
2299
2300 2005-01-21  Maciej Stachowiak  <mjs@apple.com>
2301
2302         Reviewed by Darin.
2303
2304         <rdar://problem/3967572> Editing should split elements before removing style
2305
2306         * khtml/editing/htmlediting.cpp:
2307         (khtml::CompositeEditCommand::splitElement):
2308         (khtml::CompositeEditCommand::mergeIdenticalElements):
2309         (khtml::CompositeEditCommand::wrapContentsInDummySpan):
2310         (khtml::CompositeEditCommand::splitTextNodeContainingElement):
2311         (khtml::ApplyStyleCommand::applyInlineStyle):
2312         (khtml::ApplyStyleCommand::splitTextAtStartIfNeeded):
2313         (khtml::ApplyStyleCommand::splitTextAtEndIfNeeded):
2314         (khtml::ApplyStyleCommand::splitTextElementAtStartIfNeeded):
2315         (khtml::ApplyStyleCommand::splitTextElementAtEndIfNeeded):
2316         (khtml::areIdenticalElements):
2317         (khtml::ApplyStyleCommand::mergeStartWithPreviousIfIdentical):
2318         (khtml::ApplyStyleCommand::mergeEndWithNextIfIdentical):
2319         (khtml::ApplyStyleCommand::cleanUpEmptyStyleSpans):
2320         (khtml::ApplyStyleCommand::addInlineStyleIfNeeded):
2321         (khtml::InsertParagraphSeparatorCommand::doApply):
2322         (khtml::InsertParagraphSeparatorInQuotedContentCommand::doApply):
2323         (khtml::InsertTextCommand::prepareForTextInsertion):
2324         (khtml::SplitTextNodeCommand::doUnapply):
2325         (khtml::SplitElementCommand::SplitElementCommand):
2326         (khtml::SplitElementCommand::~SplitElementCommand):
2327         (khtml::SplitElementCommand::doApply):
2328         (khtml::SplitElementCommand::doUnapply):
2329         (khtml::MergeIdenticalElementsCommand::MergeIdenticalElementsCommand):
2330         (khtml::MergeIdenticalElementsCommand::~MergeIdenticalElementsCommand):
2331         (khtml::MergeIdenticalElementsCommand::doApply):
2332         (khtml::MergeIdenticalElementsCommand::doUnapply):
2333         (khtml::WrapContentsInDummySpanCommand::WrapContentsInDummySpanCommand):
2334         (khtml::WrapContentsInDummySpanCommand::~WrapContentsInDummySpanCommand):
2335         (khtml::WrapContentsInDummySpanCommand::doApply):
2336         (khtml::WrapContentsInDummySpanCommand::doUnapply):
2337         (khtml::SplitTextNodeContainingElementCommand::SplitTextNodeContainingElementCommand):
2338         (khtml::SplitTextNodeContainingElementCommand::~SplitTextNodeContainingElementCommand):
2339         (khtml::SplitTextNodeContainingElementCommand::doApply):
2340         * khtml/editing/htmlediting.h:
2341
2342         New layout test that shows the side benefit of this.
2343
2344         * layout-tests/editing/style/unbold-in-bold-expected.txt: Added.
2345         * layout-tests/editing/style/unbold-in-bold.html: Added.
2346
2347 2005-01-24  Darin Adler  <darin@apple.com>
2348
2349         Reviewed by John.
2350
2351         - fixed <rdar://problem/3933435> -[DOMRange cloneContents] sometimes alters the original DOMRange
2352
2353         * khtml/xml/dom2_rangeimpl.cpp: (DOM::RangeImpl::processContents): Fixed code that was collapsing
2354         the range after processing the contents. That's right for extract and delete, but not clone.
2355
2356         - fixed <rdar://problem/3714184> inline input works incorrectly when I type Korean slowly in textfield in form on www.google.co.kr
2357         - fixed <rdar://problem/3193848> Can't enter SSN to sign up for Chase visa online account (changing focus inside key press handler)
2358         - fixed <rdar://problem/3874683> REGRESSION (161-162): Crash after typing command-Z to undo after pasting text with newline in it into a text field <input type=text>
2359
2360         * kwq/KWQLineEdit.mm:
2361         (QLineEdit::text): Changed to call a new string method on the controller instead of calling
2362         stringValue, since stringValue has a bad side effect of ending inline input.
2363         (QLineEdit::selectAll): Changed to only call selectText: when the field is already selected.
2364         When you are giving the field focus, it automatically gets all selected, and this change
2365         is required in the case where we defer the focus change.
2366
2367         * kwq/KWQTextField.h: Added a new string method to the controller.
2368         * kwq/KWQTextField.mm:
2369         (-[KWQTextFieldController setMaximumLength:]): Call the new string method instead of using stringValue.
2370         (-[KWQTextFieldController controlTextDidChange:]): Removed the code to truncate at the first CR or LF.
2371         That's now handled in the shouldChangeTextInRange method instead.
2372         (-[KWQTextFieldController textView:shouldHandleEvent:]): Add new logic to defer responder changes until
2373         after the event is handled. This makes sure the key gets into the field that's already handling it at
2374         rather than ending up in the newly-focused field if the key press handler focuses a different element.
2375         (-[KWQTextFieldController textView:didHandleEvent:]): Turn off deferral here, now that the key press
2376         has been handled.
2377         (-[KWQTextFieldController string]): Added. Calls stringValue only if there's no editor. If there is
2378         an editor, calls string on that instead.
2379         (-[KWQTextFieldController textView:shouldChangeTextInRange:replacementString:]): Added. Called by
2380         the three field subclasses. Truncates incoming strings at the first CR or LF character. This works
2381         properly with Undo, and the old technique did not.
2382         (-[KWQTextFieldController preprocessString:]): Added. Helper used for all the setStringValue methods
2383         so they all truncate at the first CR or LF as well as truncating to the appropriate length.
2384         (-[KWQTextField setStringValue:]): Changed to call the controller.
2385         (-[KWQTextField textView:shouldChangeTextInRange:replacementString:]): Added. Calls the controller.
2386         (-[KWQSecureTextField setStringValue:]): More of the same.
2387         (-[KWQSecureTextField textView:shouldChangeTextInRange:replacementString:]): Ditto.
2388         (-[KWQSearchField setStringValue:]): Ditto.
2389         (-[KWQSearchField textView:shouldChangeTextInRange:replacementString:]): Ditto.
2390
2391         * kwq/KWQWidget.h: Added the new setDeferFirstResponderChanges function.
2392         * kwq/KWQWidget.mm:
2393         (QWidget::~QWidget): Clear out the deferredFirstResponder global just in case we are destroyed
2394         while we "have the ball".
2395         (QWidget::hasFocus): If we are in the deferred mode, use the global to decide who to say is focused.
2396         (QWidget::setFocus): If we are in the deferred mode, set the global instead of making something be
2397         the first responder.
2398         (QWidget::setDeferFirstResponderChanges): Set the boolean. If the boolean is being cleared, then
2399         get the deferred first responder and call setFocus on it again, which will do the work we didn't
2400         want to do earlier.
2401
2402 2005-01-24  Ken Kocienda  <kocienda@apple.com>
2403
2404         Reviewed by John
2405
2406         Fix for this bug:
2407
2408         <rdar://problem/3964350> Deleting all content and typing in a message gets one letter, then beeps
2409
2410         This is a problem with the way we turn key events, when a user starts typing, into editing commands.
2411         Right now, that initial delete key in an empty window is turned into an editing command, and much
2412         editing code runs that need not run. What's more, this is confusing the selection machinery. The
2413         solution is to detect the "delete in an empty window" case and don't turn this into an editing
2414         command at all.
2415
2416         * khtml/editing/htmlediting.cpp:
2417         (khtml::TypingCommand::deleteKeyPressed)
2418
2419 2005-01-23  Adele Amchan  <adele@apple.com>
2420
2421         Fixed <rdar://problem/3964286> REGRESSION (178-179): menus and "go to old site" link at Ameritrade's new site don't work
2422
2423         Removed check for document from checkCompleted.  In this case, WebCore never knew that a WebImageView was complete because
2424         it fell into this document check case in checkCompleted.
2425         Added check for document to tokenizerProcessedData (see <rdar://problem/3807144>)
2426
2427         Reviewed by Darin.
2428
2429         * khtml/khtml_part.cpp:
2430         (KHTMLPart::stop):
2431         (KHTMLPart::checkCompleted):
2432         * khtml/khtml_part.h:
2433         * kwq/KWQKHTMLPart.mm: (KWQKHTMLPart::tokenizerProcessedData):
2434
2435 2005-01-21  Richard Williamson   <rjw@apple.com>
2436
2437         Fixed <rdar://problem/3759399> Javascript / Liveconnect problems ((event handler):Undefined value)
2438
2439         Java applets specified with <object> or <embed> weren't scriptable.  Now they are.
2440
2441         Reviewed by Chris.
2442
2443         * khtml/html/html_objectimpl.cpp:
2444         (HTMLEmbedElementImpl::getEmbedInstance):
2445         (HTMLObjectElementImpl::getObjectInstance):
2446
2447 2005-01-21  Ken Kocienda  <kocienda@apple.com>
2448
2449         Reviewed by John
2450
2451         Fix for this bug:
2452
2453         <rdar://problem/3959464> REGRESSION (Mail): Insertion point goes back to beginning of document after deleting
2454
2455         * khtml/editing/htmlediting.cpp:
2456         (khtml::DeleteSelectionCommand::handleGeneralDelete): Add special case to handle retaining a fully-selected block.
2457         This fixes the bug.
2458         * layout-tests/editing/deleting/delete-3959464-fix-expected.txt: Added.
2459         * layout-tests/editing/deleting/delete-3959464-fix.html: Added.
2460
2461 2005-01-21  Richard Williamson   <rjw@apple.com>
2462
2463         Fixed <rdar://problem/3966998> REGRESSION(179-TOT) clicking on gmail message brings me to blank screen
2464
2465         getElementById() sometimes returns Undefined() instead of Null().
2466
2467         Reviewed by Hyatt.
2468
2469         * khtml/ecma/kjs_dom.cpp:
2470         (DOMDocumentProtoFunc::tryCall):
2471
2472 2005-01-21  David Hyatt  <hyatt@apple.com>
2473
2474         Fix for 3773809, make sure that overflow regions never end up at an invalid scroll offset because of a layout change.
2475
2476         Reviewed by kocienda
2477
2478         * khtml/rendering/render_layer.cpp:
2479         (RenderLayer::updateScrollInfoAfterLayout):
2480
2481 2005-01-21  David Hyatt  <hyatt@apple.com>
2482
2483         Fix for 3966349, hang loading page.  Make sure that loops using popOneBlock properly check for a null
2484         blockStack to avoid an infinite loop.
2485
2486         Reviewed by kocienda
2487
2488         * khtml/html/htmlparser.cpp:
2489         (KHTMLParser::parseToken):
2490         (KHTMLParser::insertNode):
2491         (KHTMLParser::popInlineBlocks):
2492
2493 2005-01-21  Ken Kocienda  <kocienda@apple.com>
2494
2495         Reviewed by John
2496
2497         Fix for this bug:
2498
2499         <rdar://problem/3957204> Mail crashed when sending a message in ~InsertParagraphSeparatorInQuotedContentCommand
2500
2501         * khtml/editing/htmlediting.cpp:
2502         (khtml::InsertParagraphSeparatorInQuotedContentCommand::InsertParagraphSeparatorInQuotedContentCommand):
2503         InsertParagraphSeparatorInQuotedContentCommand has this member variable: DOM::ElementImpl *m_breakNode;
2504         I failed to initialize this variable to null in the constructor, and there is an uncommonly-traveled
2505         code path which does not set this variable to something good. In the destructor, we check for null,
2506         and deref if non-null. Obvious "BOOM" potential. Fixed.
2507
2508 2005-01-21  Ken Kocienda  <kocienda@apple.com>
2509
2510         Reviewed by John
2511
2512         Fix for this bug:
2513
2514         <rdar://problem/3966311> REGRESSION (Mail): Hitting return makes space character disappear
2515
2516         * khtml/editing/htmlediting.cpp:
2517         (khtml::InsertParagraphSeparatorCommand::doApply): When writing the code to insert a block
2518         in response to the return key, I did not write code to cover the case described in the bug.
2519         Now I have.
2520         * layout-tests/editing/inserting/insert-div-025-expected.txt: Added.
2521         * layout-tests/editing/inserting/insert-div-025.html: Added.
2522
2523 2005-01-20  Maciej Stachowiak  <mjs@apple.com>
2524
2525         Reviewed by Ken.
2526
2527         <rdar://problem/3965196> security fix for javascript: exploit missed one case (already fixed in updates)
2528
2529         * khtml/ecma/kjs_window.cpp:
2530         (WindowFunc::tryCall): correct mistake in earlier fix for the following bug, caught by Adele:
2531
2532 2005-01-20  Maciej Stachowiak  <mjs@apple.com>
2533
2534         Reviewed by Ken.
2535
2536         <rdar://problem/3965466> editing needs to insert text before applying typing style
2537
2538         * khtml/editing/htmlediting.cpp:
2539         (khtml::ApplyStyleCommand::addInlineStyleIfNeeded): Remove a FIXME comment.
2540         (khtml::InsertLineBreakCommand::doApply): Apply style to the BR
2541         node after inserting it.
2542         (khtml::InsertTextCommand::prepareForTextInsertion): Don't try to
2543         apply style to any new nodes created before inserting.
2544         (khtml::InsertTextCommand::input): Apply style to the inserted range
2545         after doing the text insert.
2546         * khtml/editing/htmlediting.h:
2547
2548         This change leads to improved results on one of the layout tests:
2549
2550         * layout-tests/editing/inserting/insert-div-024-expected.txt:
2551
2552 === Safari-180 ===
2553
2554 2005-01-20  Ken Kocienda  <kocienda@apple.com>
2555
2556         Reviewed by John
2557
2558         Fix for this bug:
2559
2560         <rdar://problem/3964646> REGRESSION (179-180): Typing space at end of line makes following paragraph disappear
2561
2562         * khtml/editing/htmlediting.cpp:
2563         (khtml::CompositeEditCommand::removeBlockPlaceholderIfNeeded): Recent change to block placeholder removal code
2564         caused this regression. The code became too aggressive in removing block placeholders, and would remove them
2565         from blocks other than the block containing the selection.
2566         * layout-tests/editing/inserting/insert-div-023-expected.txt: This file had a spurious BR element in it that
2567         I did not notice earlier.
2568         * layout-tests/editing/inserting/insert-div-024-expected.txt: Ditto.
2569
2570 2005-01-20  David Hyatt  <hyatt@apple.com>
2571
2572         Fix for oddness on albertsons.com.  Make sure not to crash when setting/removing style properties on a node
2573         with no document.  The bug # is 3813900.
2574
2575         Reviewed by john
2576
2577 2005-01-20  David Harrison  <harrison@apple.com>
2578
2579         Reviewed by Darin.
2580
2581         PARTIAL fix for following bug.  Create attributed string with fonts and links.  Still need to add attachments.
2582         <rdar://problem/3942606> AX: Support kAXAttributedStringForTextMarkerRangeParameterizedAttribute
2583
2584         * kwq/KWQAccObject.mm:
2585         (-[KWQAccObject textUnderElement]):
2586         Touched up previous checkin to match code review comments.
2587
2588         (-[KWQAccObject value]):
2589         Use plainText for this one instead of obsolete attributedString.
2590
2591         (-[KWQAccObject accessibilityParameterizedAttributeNames]):
2592         Advertise AXAttributedStringForTextMarkerRange.
2593
2594         (AXAttributeStringAddFont):
2595         (AXAttributeStringAddElement):
2596         (-[KWQAccObject linkUIElementForNode:]):
2597         (-[KWQAccObject _addAttachmentForElement:URL:needsParagraph:usePlaceholder:]):
2598         (-[KWQAccObject accessibilityAttributedStringForRange:]):
2599         New routines to support AXAttributedStringForTextMarkerRange.
2600
2601         (-[KWQAccObject doAXAttributedStringForTextMarkerRange:]):
2602         Use new routines instead of obsolete attributedString.
2603
2604 2005-01-20  David Harrison  <harrison@apple.com>
2605
2606         Reviewed by Darin.
2607
2608         <rdar://problem/3960196> AX Crash in DOM::Range::setStartBefore
2609
2610         * khtml/xml/dom_docimpl.cpp:
2611         (DocumentImpl::getAccObjectCache):
2612         Adjust when detecting cache in non-top level WebArea.
2613         * kwq/KWQAccObject.mm:
2614         (-[KWQAccObject textUnderElement]):
2615         (-[KWQAccObject accessibilityAttributeValue:]):
2616         Make sure the elements document is the current one for the part.
2617
2618 2005-01-20  Darin Adler  <darin@apple.com>
2619
2620         Reviewed by John.
2621
2622         - fixed <rdar://problem/3922980> Mail not crashing, just quitting itself suddenly in -[WebHTMLView(MailExtras) findString:options:]
2623
2624         * khtml/dom/dom2_range.cpp: (DOM::operator==): Handle null and detached ranges without raising exceptions.
2625         The uncaught exception would make the entire program terminate.
2626
2627         - fixed <rdar://problem/3885729> attempting to modify a computed style does nothing, but should raise a DOM exception
2628
2629         * kwq/DOM-CSS.mm: (-[DOMCSSStyleDeclaration setProperty:::]): Uncommented this code which I had to disable
2630         back in November because Mail was stumbling over it.
2631
2632         - fixed <rdar://problem/3943049> focus() called during onload handler results in square text field on <input type=search>
2633
2634         * kwq/KWQLineEdit.mm: (QLineEdit::baselinePosition): Change computation so it will work even for fields
2635         that position their text in a way that depends on vertical size. In the case of a search field, the text
2636         is centered, so the old logic was broken.
2637         * kwq/KWQTextField.mm: (-[KWQTextFieldController initWithTextField:QLineEdit:]): Set the frame size to
2638         something large enough to accomodate the field editor. If we start the frame at size 0,0 we run into
2639         AppKit trouble when it insets the frame to figure out the frame for the field editor. In the case of this
2640         bug this happens because we become first responder before being sized and positioned by the HTML layout code.
2641
2642         - improved debugging output when using "po" from gdb with Objective-C DOM
2643
2644         * kwq/DOM.mm:
2645         (-[DOMNode description]): Added. Includes node name ("<tr>") and node value (e.g., string for text node).
2646         (-[DOMRange description]): Tweaked format.
2647
2648 2005-01-19  Richard Williamson   <rjw@apple.com>
2649
2650         After further discussion with Real we have decided to NOT include the
2651         additional CLSID for the real plugin.  See 3958601.
2652
2653         Reviewed by Darin.
2654
2655         * khtml/rendering/render_frames.cpp:
2656         (RenderPartObject::updateWidget):
2657
2658 2005-01-19  David Harrison  <harrison@apple.com>
2659
2660         Reviewed by Maciej.
2661
2662         <rdar://problem/3949429> AX: word marker routines returns incorrect data for empty line
2663
2664         Previous checkin for this bug was missing some of the patch.  Editing snafu with multiple changes in tree.
2665
2666         * kwq/KWQAccObject.mm:
2667         (-[KWQAccObject doAXRightLineTextMarkerRangeForTextMarker:]):
2668         (-[KWQAccObject doAXNextWordEndTextMarkerForTextMarker:]):
2669         (-[KWQAccObject doAXPreviousWordStartTextMarkerForTextMarker:]):
2670         (-[KWQAccObject doAXNextLineEndTextMarkerForTextMarker:]):
2671         (-[KWQAccObject doAXPreviousLineStartTextMarkerForTextMarker:]):
2672         (-[KWQAccObject doAXPreviousSentenceStartTextMarkerForTextMarker:]):
2673         (-[KWQAccObject doAXPreviousParagraphStartTextMarkerForTextMarker:]):
2674
2675 2005-01-19  Ken Kocienda  <kocienda@apple.com>
2676
2677         Reviewed by John
2678
2679         * khtml/editing/htmlediting.cpp:
2680         (khtml::InsertParagraphSeparatorCommand::applyStyleAfterInsertion): Small, cosmetic change John and I decided
2681         to do on my last checkin, but I forgot to do before landing.
2682
2683 2005-01-19  Ken Kocienda  <kocienda@apple.com>
2684
2685         Reviewed by John
2686
2687         Fix for this bug:
2688
2689         <rdar://problem/3959727> REGRESSION (Mail): Style not preserved on blank lines
2690
2691         * khtml/editing/htmlediting.cpp:
2692         (khtml::CompositeEditCommand::applyStyle):
2693         (khtml::CompositeEditCommand::insertBlockPlaceholder): New function that unconditionally adds a block placeholder.
2694         (khtml::CompositeEditCommand::insertBlockPlaceholderIfNeeded): Now returns bool based on whether
2695         placeholder was added or not.
2696         (khtml::CompositeEditCommand::removeBlockPlaceholderIfNeeded): Now searches all the descendents of a block
2697         looking for a placeholder. The old code, which just looked at the last child of a node, started missing
2698         once block placeholders became styled (which started happening with this patch).
2699         (khtml::DeleteSelectionCommand::calculateTypingStyleAfterDelete): Now handles applying typing style
2700         to a block placeholder at call time, rather than setting the typing style as a latent style that
2701         might be applied later. This is an important part of the bug fix.
2702         (khtml::DeleteSelectionCommand::doApply): Now uses bool return value from insertBlockPlaceholderIfNeeded()
2703         and passes it along to calculateStyleBeforeInsertion, so the case where a block placeholder needs to
2704         be styled can be detected.
2705         (khtml::InsertParagraphSeparatorCommand::InsertParagraphSeparatorCommand): Changed the way this class
2706         managed style. Before it would calculate and set typing style for the block added. This is not
2707         sufficient. Added blocks need to styled immediately. Some name changes to instance variables in
2708         this class due to the change to accommodate this change.
2709         (khtml::InsertParagraphSeparatorCommand::~InsertParagraphSeparatorCommand): Name changes, as above.
2710         (khtml::InsertParagraphSeparatorCommand::calculateStyleBeforeInsertion): Ditto.
2711         (khtml::InsertParagraphSeparatorCommand::applyStyleAfterInsertion): Ditto.
2712         (khtml::InsertParagraphSeparatorCommand::doApply): Ditto.
2713         (khtml::ReplaceSelectionCommand::doApply): Improve check for testing when a placeholder
2714         block can be removed in its entirety after the insertion.
2715         * khtml/editing/htmlediting.h: Update header accordingly.
2716         * khtml/khtml_part.cpp:
2717         (KHTMLPart::selectionComputedStyle): Move position for computed style check downstream before
2718         doing check when the position is in an empty block (this makes sure any style on any block
2719         placeholder is accounted for).
2720
2721         New layout tests to check bug fix.
2722
2723         * layout-tests/editing/style/block-style-004-expected.txt: Added.
2724         * layout-tests/editing/style/block-style-004.html: Added.
2725         * layout-tests/editing/style/block-style-005-expected.txt: Added.
2726         * layout-tests/editing/style/block-style-005.html: Added.
2727         * layout-tests/editing/style/block-style-006-expected.txt: Added.
2728         * layout-tests/editing/style/block-style-006.html: Added.
2729
2730         Results updated to reflect new block placeholder code.
2731
2732         * layout-tests/editing/inserting/insert-div-004-expected.txt
2733         * layout-tests/editing/inserting/insert-div-005-expected.txt
2734         * layout-tests/editing/inserting/insert-div-006-expected.txt
2735         * layout-tests/editing/inserting/insert-div-008-expected.txt
2736         * layout-tests/editing/inserting/insert-div-011-expected.txt
2737         * layout-tests/editing/inserting/insert-div-012-expected.txt
2738         * layout-tests/editing/inserting/insert-div-013-expected.txt
2739         * layout-tests/editing/inserting/insert-div-014-expected.txt
2740         * layout-tests/editing/inserting/insert-div-015-expected.txt
2741         * layout-tests/editing/inserting/insert-div-016-expected.txt
2742         * layout-tests/editing/inserting/insert-div-017-expected.txt
2743         * layout-tests/editing/inserting/insert-div-018-expected.txt
2744         * layout-tests/editing/inserting/insert-div-019-expected.txt
2745         * layout-tests/editing/inserting/insert-div-021-expected.txt
2746         * layout-tests/editing/inserting/insert-div-022-expected.txt
2747         * layout-tests/editing/inserting/insert-div-023-expected.txt
2748         * layout-tests/editing/inserting/insert-div-024-expected.txt
2749
2750 2005-01-19  David Hyatt  <hyatt@apple.com>
2751
2752         Dont null-check the renderer before submitting, since a script can set it to display:none and still expect the
2753         submission to occur.  Fixes bug #3477282.
2754
2755         Reviewed by kdecker
2756
2757         (DOM::HTMLInputElementImpl::defaultEventHandler):
2758
2759 2005-01-18  Richard Williamson   <rjw@apple.com>
2760
2761         Fixed several issues all arising from analysis of plugin detection code at ifilm.com:
2762
2763         Fixed <rdar://problem/3958592> can't script plug-ins if plug-in is invoked with <object> element instead of <embed>
2764         Fixed <rdar://problem/3958597> <object> elements with IDs do not show up as named properties of the document
2765         Fixed <rdar://problem/3960973> DOM objects for plugin elements are not accessible
2766         Fixed <rdar://problem/3958601> need an additional class ID in WebCore for the Real plug-in
2767
2768         We now support accessing scriptable plugin objects that are specified with <applet>, <embed>, or <object>
2769         tags.  Also, if any of these elements are named they can be accessed from the document or window objects.
2770         Finally, DOM methods are properties will be forwarded appropriately for the plugin's root scriptable object.
2771
2772         Reviewed by Chris.
2773
2774         * khtml/dom/html_document.cpp:
2775         (HTMLDocument::objects):
2776         * khtml/dom/html_document.h:
2777         * khtml/ecma/kjs_dom.cpp:
2778         (DOMDocumentProtoFunc::tryCall):
2779         (DOMElementProtoFunc::tryCall):
2780         (KJS::getRuntimeObject):
2781         * khtml/ecma/kjs_dom.h:
2782         * khtml/ecma/kjs_html.cpp:
2783         (KJS::HTMLDocument::tryGet):
2784         (KJS::HTMLElement::tryGet):
2785         (KJS::HTMLCollection::tryGet):
2786         (KJS::HTMLCollection::getNamedItems):
2787         * khtml/ecma/kjs_window.cpp:
2788         (Window::get):
2789         * khtml/html/html_miscimpl.cpp:
2790         (HTMLCollectionImpl::traverseNextItem):
2791         * khtml/html/html_miscimpl.h:
2792         (DOM::HTMLCollectionImpl::):
2793         * khtml/html/html_objectimpl.cpp:
2794         (HTMLAppletElementImpl::getAppletInstance):
2795         (HTMLObjectElementImpl::HTMLObjectElementImpl):
2796         (HTMLObjectElementImpl::getObjectInstance):
2797         * khtml/html/html_objectimpl.h:
2798         * khtml/rendering/render_frames.cpp:
2799         (RenderPartObject::updateWidget):
2800         * kwq/KWQKHTMLPart.h:
2801         * kwq/KWQKHTMLPart.mm:
2802         (KWQKHTMLPart::getObjectInstanceForView):
2803
2804 2005-01-18  David Hyatt  <hyatt@apple.com>
2805
2806         Fix for 3948123, rolling over link erases nearby text.  The repaint rect check for lines was wrong whenever
2807         two lines overlapped.
2808
2809         Reviewed by kocienda
2810
2811         * khtml/rendering/render_flow.cpp:
2812         (RenderFlow::paintLines):
2813
2814 2005-01-18  Ken Kocienda  <kocienda@apple.com>
2815
2816         Reviewed by Hyatt
2817
2818         Fix for this bug:
2819
2820         <rdar://problem/3960116> Focus rings paint incorrectly for contenteditable blocks in web pages
2821
2822         * khtml/rendering/render_flow.cpp:
2823         (RenderFlow::addFocusRingRects): Fix painting of focus rings so that ring only paints around
2824         outermost contenteditable elements.
2825
2826 2005-01-18  David Harrison  <harrison@apple.com>
2827
2828         Reviewed by Darin.
2829
2830         <rdar://problem/3959668> accessibilityFocusedUIElement sometimes returns an ignored element; it must not
2831
2832         * kwq/KWQAccObject.mm:
2833         (-[KWQAccObject accessibilityFocusedUIElement]):
2834         Return parentObjectUnignored if focused object is ignored.
2835
2836 2005-01-18  Ken Kocienda  <kocienda@apple.com>
2837
2838         Reviewed by John
2839
2840         Fix for this bug:
2841
2842         <rdar://problem/3952877> REGRESSION (Mail): Command-left/right-arrows don't work with file attachment
2843
2844         * khtml/editing/selection.cpp:
2845         (khtml::nodeForInlineBox): New helper function used in reimplementation of function below.
2846         (khtml::selectionForLine): Reimplemented using line box smarts. I originally wrote this code when
2847         I had a less than full understanding of line layout. I can do better now, and my new version no
2848         longer fails to notice attachments when doing the kind of navigation mentioned in the bug.
2849
2850 2005-01-17  David Harrison  <harrison@apple.com>
2851
2852         Reviewed by John Sullivan.
2853
2854         <rdar://problem/3949429> AX: word marker routines returns incorrect data for empty line
2855         <rdar://problem/3949848> AX: paragraph marker routines do not work when given a paragraph end marker
2856
2857         Also fixed sentence support in the same way.
2858
2859         * kwq/KWQAccObject.mm:
2860         (-[KWQAccObject doAXNextLineEndTextMarkerForTextMarker:]):
2861         (-[KWQAccObject doAXNextSentenceEndTextMarkerForTextMarker:]):
2862         (-[KWQAccObject doAXPreviousSentenceStartTextMarkerForTextMarker:]):
2863         (-[KWQAccObject doAXNextParagraphEndTextMarkerForTextMarker:]):
2864         (-[KWQAccObject doAXPreviousParagraphStartTextMarkerForTextMarker:]):
2865         Move one position in desired direction before calling visible_units.cpp code.
2866
2867 2005-01-17  David Harrison  <harrison@apple.com>
2868
2869         Updated expected test results.
2870
2871         <rdar://problem/3945880> line-ending space seems not to be present
2872
2873         * layout-tests/editing/deleting/delete-ws-fixup-001-expected.txt:
2874         * layout-tests/editing/deleting/delete-ws-fixup-002-expected.txt:
2875         * layout-tests/editing/inserting/insert-div-020-expected.txt:
2876         * layout-tests/editing/inserting/insert-div-021-expected.txt:
2877         * layout-tests/editing/inserting/insert-div-022-expected.txt:
2878         * layout-tests/editing/inserting/insert-div-023-expected.txt:
2879         * layout-tests/editing/inserting/insert-div-024-expected.txt:
2880
2881 2005-01-17  David Harrison  <harrison@apple.com>
2882
2883         Reviewed by Dave Hyatt (bidi.cpp) and Darin Adler (selection.cpp).
2884
2885         <rdar://problem/3945880> line-ending space seems not to be present
2886
2887         * khtml/editing/selection.cpp:
2888         (khtml::Selection::validate):
2889         Extend AFTER_WHITE_SPACE code to support white spac in the middle of paragraphs, not just the end.
2890         * khtml/rendering/bidi.cpp:
2891         (khtml::RenderBlock::findNextLineBreak):
2892         Pick left/rightness of word selection based on being at the end of paragraph (i.e. after a hard line break).
2893
2894 2005-01-17  Darin Adler  <darin@apple.com>
2895
2896         Reviewed by John Louch.
2897
2898         - fixed <rdar://problem/3958503> need screenX and screenY to use WebKit windowFrame delegate
2899
2900         * khtml/ecma/kjs_window.cpp: (Window::get): Change screenX and screenY to use frameGeometry instead
2901         of using mapToGlobal and screen in a complicated way.
2902         * kwq/KWQKHTMLView.mm: Removed unused mapToGlobal function.
2903         * kwq/KWQWindowWidget.h: Ditto.
2904         * kwq/KWQWindowWidget.mm: Ditto.
2905
2906 2005-01-17  David Hyatt  <hyatt@apple.com>
2907
2908         Fix a screwup in rightmost/lowets position computation.  3955207.  Make sure floats with layers are still checked.
2909
2910         Reviewed by kocienda
2911
2912         * khtml/rendering/render_block.cpp
2913
2914 2005-01-17  Ken Kocienda  <kocienda@apple.com>
2915
2916         Reviewed by John
2917
2918         <rdar://problem/3953366> Problems with typing attributes in HTML compose
2919
2920         * khtml/editing/htmlediting.cpp:
2921         (khtml::DeleteSelectionCommand::calculateTypingStyleAfterDelete): Remove misguided "optimization"
2922         that tried to sense when typing style could be cleared without actually doing a style diff between
2923         before-delete and after-delete positions. Removing this extra check and running the
2924         general-purpose code fixes the bug.
2925
2926 2005-01-17  Richard Williamson   <rjw@apple.com>
2927
2928         Fixed <rdar://problem/3753030> Need to ensure same origin for plugin binding invocations (origin security rules)
2929
2930         Keep track of originating execution context and target execution
2931         context for native JS object wrappers, and perform appropriate
2932         security checks.
2933
2934         Reviewed by David Harrison.
2935
2936         * khtml/ecma/kjs_binding.cpp:
2937         (ScriptInterpreter::isGlobalObject):
2938         (ScriptInterpreter::isSafeScript):
2939         (ScriptInterpreter::interpreterForGlobalObject):
2940         * khtml/ecma/kjs_binding.h:
2941         * khtml/ecma/kjs_window.cpp:
2942         (Window::interpreter):
2943         (Window::isSafeScript):
2944         * khtml/ecma/kjs_window.h:
2945         * kwq/DOMInternal.mm:
2946         (-[WebScriptObject _initializeScriptDOMNodeImp]):
2947         * kwq/KWQKHTMLPart.mm:
2948         (KWQKHTMLPart::windowScriptObject):
2949         (KWQKHTMLPart::windowScriptNPObject):
2950         * kwq/WebCoreBridge.mm:
2951         (rootForView):
2952         (-[WebCoreBridge init]):
2953
2954 2005-01-17  Ken Kocienda  <kocienda@apple.com>
2955
2956         Reviewed by John
2957
2958         Fix for this bug:
2959
2960         <rdar://problem/3786659> REGRESSION (Mail): editable WebViews don't work with
2961         "size up" and "size down" NSFontManager changes
2962
2963         * khtml/css/cssparser.cpp:
2964         (CSSParser::parseValue): Add support for parsing new font size delta property.
2965         * khtml/css/cssproperties.c: Generated file.
2966         * khtml/css/cssproperties.h: Ditto.
2967         * khtml/css/cssproperties.in: Add support for parsing new font size delta property.
2968         * khtml/editing/htmlediting.cpp:
2969         (khtml::isEmptyStyleSpan): New helper function. Code existed before, but now factored out for convenient use.
2970         (khtml::CompositeEditCommand::removeNodeAttribute): Does check on value to see it exists before creating
2971         and running command to do the removal.
2972         (khtml::ApplyStyleCommand::doApply): Now calls new applyRelativeFontStyleChange() function as part of its work.
2973         (khtml::ApplyStyleCommand::applyRelativeFontStyleChange): New function that does the "heavy lifting" to handle
2974         relative font size changes.
2975         (khtml::ApplyStyleCommand::applyInlineStyle): Range check the start and end positions to make sure the start is
2976         before or equal to the end. Swap them if not true. I ran into this problem in some similar code in
2977         applyRelativeFontStyleChange(). Moving that goodness here too.
2978         (khtml::ApplyStyleCommand::splitTextAtEndIfNeeded): Uses a local variable to save a value used often.
2979         Code is the same, but reads better now. Function now returns bool as well, just like splitTextAtStartIfNeeded.
2980         I use the bool return value now (I obviously did not need it before).
2981         (khtml::ApplyStyleCommand::computedFontSize): New helper function.
2982         (khtml::ApplyStyleCommand::joinChildTextNodes): Ditto.
2983         (khtml::createStyleSpanElement): Ditto.
2984         * khtml/editing/htmlediting.h: Update header accordingly.
2985         * khtml/editing/jsediting.cpp: Add new command string to enable relative font size changes.
2986         * kwq/DOM-CSS.mm:
2987         (-[DOMCSSStyleDeclaration _fontSizeDelta]): New convenience.
2988         (-[DOMCSSStyleDeclaration _setFontSizeDelta:]): Ditto.
2989         * kwq/DOMPrivate.h: Declare new conveniences.
2990         * layout-tests/editing/editing.js: Add new JS to enable relative font size changes, as well as explicit font size setting.
2991         * layout-tests/editing/style/relative-font-size-change-001-expected.txt: Added.
2992         * layout-tests/editing/style/relative-font-size-change-001.html: Added.
2993         * layout-tests/editing/style/relative-font-size-change-002-expected.txt: Added.
2994         * layout-tests/editing/style/relative-font-size-change-002.html: Added.
2995         * layout-tests/editing/style/relative-font-size-change-003-expected.txt: Added.
2996         * layout-tests/editing/style/relative-font-size-change-003.html: Added.
2997         * layout-tests/editing/style/relative-font-size-change-004-expected.txt: Added.
2998         * layout-tests/editing/style/relative-font-size-change-004.html: Added.
2999
3000 2005-01-14  Darin Adler  <darin@apple.com>
3001
3002         * khtml/css/cssproperties.c: Regenerated with gperf 3.0.1.
3003         * khtml/css/cssvalues.c: Regenerated with gperf 3.0.1.
3004         * khtml/misc/htmlattrs.c: Regenerated with gperf 3.0.1.
3005         * khtml/misc/htmltags.c: Regenerated with gperf 3.0.1.
3006
3007 2005-01-14  Chris Blumenberg  <cblu@apple.com>
3008
3009         Fixed: <rdar://problem/3886415> arrow keys don't work when the user hits Back to return to RSS page
3010
3011         Reviewed by hyatt.
3012
3013         (KWQKHTMLPart::openURLFromPageCache): restore mousePressNode
3014         (KWQKHTMLPart::mousePressNode): new
3015         * kwq/KWQPageState.h:
3016         * kwq/KWQPageState.mm:
3017         (-[KWQPageState initWithDocument:URL:windowProperties:locationProperties:interpreterBuiltins:]): save the mousePressNode
3018         (-[KWQPageState clear]): clear the mousePressNode
3019         (-[KWQPageState dealloc]): deref the mousePressNode
3020         (-[KWQPageState mousePressNode]): new
3021
3022 === Safari-179 ===
3023
3024 2005-01-13  Vicki Murley <vicki@apple.com>
3025
3026         Reviewed by Adele.
3027
3028         - fix <rdar://problem/3946836> Safari about box lists 2004 instead of 2005
3029
3030         * WebCore.pbproj/project.pbxproj: bump "2004" to "2005"
3031
3032 2005-01-13  David Harrison  <harrison@apple.com>
3033
3034         Reviewed by Ken Kocienda.
3035
3036         Better fix for 3905066.
3037
3038         * khtml/editing/htmlediting.cpp:
3039         (khtml::InsertParagraphSeparatorCommand::doApply):
3040
3041 2005-01-10  Maciej Stachowiak  <mjs@apple.com>
3042
3043         Reviewed by Darin.
3044
3045         <rdar://problem/3758033> REGRESSION (Mail): Support attributes in marked text (International input)
3046
3047         * khtml/rendering/render_text.cpp:
3048         (InlineTextBox::paint): Support painting custom underline markers for
3049         marked text in place of generic yellow.
3050         (InlineTextBox::paintMarkedTextUnderline): New method that handles this.
3051         * khtml/rendering/render_text.h:
3052         * kwq/KWQKHTMLPart.h: Declare new methods and structs.
3053         * kwq/KWQKHTMLPart.mm:
3054         (KWQKHTMLPart::clear): Clear marked test underlines.
3055         (KWQKHTMLPart::setMarkedTextRange): Takes attributes and ranges now.
3056         (convertAttributesToUnderlines): Converts NSAttributedString attributes
3057         to simplified and C++-friendly form.
3058         (KWQKHTMLPart::markedTextUsesUnderlines): New method.
3059         (KWQKHTMLPart::markedTextUnderlines): New method.
3060         * kwq/KWQPainter.mm:
3061         (QPainter::drawLineForText): Handle pen width.
3062         * kwq/WebCoreBridge.h:
3063         * kwq/WebCoreBridge.mm:
3064         (-[WebCoreBridge setMarkedTextDOMRange:customAttributes:ranges:]): Take attributes
3065         and ranges.
3066         * kwq/WebCoreTextRenderer.h:
3067
3068 2005-01-12  David Harrison  <harrison@apple.com>
3069
3070         Reviewed by Dave Hyatt.
3071
3072         <rdar://problem/3888973> AX: Parent AXWebArea of nested AXWebAreas is messed up
3073
3074         * kwq/KWQAccObject.mm:
3075         (-[KWQAccObject addChildrenToArray:]):
3076         Use the widget's outer view.
3077
3078 2005-01-12  David Harrison  <harrison@apple.com>
3079
3080         Reviewed by Darin Adler.
3081
3082         <rdar://problem/3949908> Crash when asking for the kAXLengthForTextMarkerRangeParameterizedAttribute
3083
3084         * khtml/xml/dom_nodeimpl.cpp:
3085         (NodeImpl::displayNode):
3086         Add quotes around text node content.
3087         (NodeBaseImpl::childNode):
3088         Add nil check to return nil rather than crash when child node not found.
3089         * kwq/KWQAccObject.mm:
3090         (-[KWQAccObject doAXStringForTextMarkerRange:]):
3091         Pass range compliant positions to TextIterator.
3092
3093 2005-01-12  David Hyatt  <hyatt@apple.com>
3094
3095         Fix for 3951203, CSS border style incorrectly clipped on inline elements.  Make sure the repaint rect set during
3096         line layout is smarter about including the overflow for both old states and new states.
3097
3098         Reviewed by john
3099
3100         * khtml/rendering/bidi.cpp:
3101         (khtml::RenderBlock::layoutInlineChildren):
3102
3103 2005-01-12  David Hyatt  <hyatt@apple.com>
3104
3105         Fix for bug 3937608, versiontracker.com flashes and displays the right column below content.  Make sure to move
3106         tables/overflows that dont fit within a block only in strict mode.
3107
3108         Fix for bug 3931049, characters dont show up when typing.  Make sure to dirty the right lines when this specific
3109         case in editing is hit.
3110
3111         Reviewed by darin (first one), kocienda (second one)
3112
3113         * khtml/rendering/render_block.cpp:
3114         (khtml::RenderBlock::getClearDelta):
3115         * khtml/rendering/render_flow.cpp:
3116         (RenderFlow::dirtyLinesFromChangedChild):
3117         * khtml/rendering/render_line.cpp:
3118         (khtml::InlineFlowBox::verticallyAlignBoxes):
3119
3120 2005-01-12  Ken Kocienda  <kocienda@apple.com>
3121
3122         Reviewed by John
3123
3124         Fix for this bug:
3125
3126         <rdar://problem/3926142> REGRESSION (Mail): Deleting text decreases quote level
3127
3128         * khtml/editing/htmlediting.cpp:
3129         (khtml::DeleteSelectionCommand::initializePositionData): Change test that will prevent block
3130         merging. End-of-line test was just wrong. Call new start-of and end-of paragraph functions
3131         added to visible_position files.
3132         (khtml::DeleteSelectionCommand::moveNodesAfterNode): Add tests for BR elements, which makes
3133         tests to determine when to stop moving nodes more complete and correct. Also improved comments.
3134         * khtml/editing/visible_position.cpp:
3135         (khtml::isFirstVisiblePositionInParagraph): New function.
3136         (khtml::isLastVisiblePositionInParagraph): New function.
3137         * khtml/editing/visible_position.h: Update header accordingly.
3138         * layout-tests/editing/deleting/delete-block-merge-contents-002-expected.txt: New results, slightly different from former
3139         results but still correct.
3140         * layout-tests/editing/deleting/delete-block-merge-contents-007-expected.txt: Ditto.
3141         * layout-tests/editing/deleting/delete-block-merge-contents-018-expected.txt: Added.
3142         * layout-tests/editing/deleting/delete-block-merge-contents-018.html: Added.
3143         * layout-tests/editing/deleting/delete-block-merge-contents-019-expected.txt: Added.
3144         * layout-tests/editing/deleting/delete-block-merge-contents-019.html: Added.
3145         * layout-tests/editing/deleting/delete-block-merge-contents-020-expected.txt: Added.
3146         * layout-tests/editing/deleting/delete-block-merge-contents-020.html: Added.
3147         * layout-tests/editing/deleting/delete-block-merge-contents-021-expected.txt: Added.
3148         * layout-tests/editing/deleting/delete-block-merge-contents-021.html: Added.
3149         * layout-tests/editing/deleting/delete-line-014-expected.txt: These new results are actually better, and fix a bug.
3150
3151 2005-01-11  Richard Williamson   <rjw@apple.com>
3152
3153         Fixed 3922875.  Fall back to DOM object is EMBED element
3154         has no associated runtime object.
3155
3156         Reviewed by Chris.
3157
3158         * khtml/ecma/kjs_dom.cpp:
3159         (KJS::getRuntimeObject):
3160         * khtml/ecma/kjs_html.cpp:
3161         (KJS::HTMLDocument::tryGet):
3162         (KJS::HTMLElement::tryGet):
3163         (KJS::HTMLCollection::tryGet):
3164         (KJS::HTMLCollection::getNamedItems):
3165
3166 2005-01-11  David Hyatt  <hyatt@apple.com>
3167
3168         Fix for 3882299, missing content on gibson.com.  Change our handling of " and ' in certain states of the parser to match
3169         other browsers.
3170
3171         Reviewed by Maciej
3172
3173         * khtml/html/htmltokenizer.cpp:
3174         (khtml::HTMLTokenizer::parseTag):
3175
3176 2005-01-11  Chris Blumenberg  <cblu@apple.com>
3177
3178         Fixed: <rdar://problem/3930733> Mail prints second page of email blank
3179
3180         Reviewed by dave.
3181
3182         * khtml/rendering/render_canvas.cpp:
3183         (RenderCanvas::paint): cache the print rect since the dirty rect can get changed during printing
3184         * khtml/rendering/render_flow.cpp:
3185         (RenderFlow::paintLines): removed null check since the print rect should never be null
3186         * khtml/rendering/render_list.cpp:
3187         (RenderListMarker::paint): ditto
3188         * kwq/KWQKHTMLPart.mm:
3189         (KWQKHTMLPart::adjustPageHeight): don't set the print rect here since this method is not called for the last page
3190
3191 2005-01-10  Ken Kocienda  <kocienda@apple.com>
3192
3193         Reviewed by Hyatt
3194
3195         Fix for this bug:
3196
3197         <rdar://problem/3943648> extra line is inserted after pressing return within quoted text of reply
3198
3199         * khtml/editing/htmlediting.cpp:
3200         (khtml::InsertParagraphSeparatorInQuotedContentCommand::doApply): This bug was the result of a
3201         simple coding mistake. A local variable was erroneously redefined in a deeper scope, and so the
3202         result calculated in that deeper scope was not available when tested.
3203
3204 2005-01-10  Ken Kocienda  <kocienda@apple.com>
3205
3206         Reviewed by John
3207
3208         Fix for this bug:
3209
3210         <rdar://problem/3946852> Option-e goes to next line
3211
3212         * khtml/editing/htmlediting.cpp:
3213         (khtml::ReplaceSelectionCommand::doApply): Tweak case used to determine when merging content
3214         into the start line is done. We plan to change pretty substantially soon to better handle
3215         the problem described in <rdar://problem/3937352> Quote level not maintained when copied
3216         and pasted within a Mail message. In the meantime, this change does no harm, and fixes the bug.
3217
3218 2005-01-10  Ken Kocienda  <kocienda@apple.com>
3219
3220         Reviewed by Darin
3221
3222         Fix for this bug:
3223
3224         <rdar://problem/3907005> Applying block styles to a line of text can unexpectedly affect other lines
3225
3226         * khtml/editing/htmlediting.cpp:
3227         (khtml::CompositeEditCommand::moveParagraphContentsToNewBlockIfNecessary): New function which moves "paragraphs"
3228         to their own blocks if needed so that a block style can be applied.
3229         (khtml::CompositeEditCommand::isMailBlockquote): Moved this function to base class so it can be used more generally.
3230         (khtml::ApplyStyleCommand::applyBlockStyle): Pass a node instead of a block to addBlockStyleIfNeeded().
3231         (khtml::ApplyStyleCommand::addBlockStyleIfNeeded): Now accepts a node instead of a block for styling.
3232         This function also now calls moveParagraphContentsToNewBlockIfNecessary() to make sure that only the current
3233         paragraph is styled.
3234         (khtml::SetNodeAttributeCommand::doUnapply): Should not assert on undo if old value of attribute was null.
3235         Should remove attributue instead.
3236         * khtml/editing/htmlediting.h: Touch function declarations accordingly.
3237         * layout-tests/editing/style/create-block-for-style-001-expected.txt: Added.
3238         * layout-tests/editing/style/create-block-for-style-001.html: Added.
3239         * layout-tests/editing/style/create-block-for-style-002-expected.txt: Added.
3240         * layout-tests/editing/style/create-block-for-style-002.html: Added.
3241         * layout-tests/editing/style/create-block-for-style-003-expected.txt: Added.
3242         * layout-tests/editing/style/create-block-for-style-003.html: Added.
3243         * layout-tests/editing/style/create-block-for-style-004-expected.txt: Added.
3244         * layout-tests/editing/style/create-block-for-style-004.html: Added.
3245         * layout-tests/editing/style/create-block-for-style-005-expected.txt: Added.
3246         * layout-tests/editing/style/create-block-for-style-005.html: Added.
3247         * layout-tests/editing/style/create-block-for-style-006-expected.txt: Added.
3248         * layout-tests/editing/style/create-block-for-style-006.html: Added.
3249         * layout-tests/editing/style/create-block-for-style-007-expected.txt: Added.
3250         * layout-tests/editing/style/create-block-for-style-007.html: Added.
3251         * layout-tests/editing/style/create-block-for-style-008-expected.txt: Added.
3252         * layout-tests/editing/style/create-block-for-style-008.html: Added.
3253         * layout-tests/editing/style/create-block-for-style-009-expected.txt: Added.
3254         * layout-tests/editing/style/create-block-for-style-009.html: Added.
3255         * layout-tests/editing/style/create-block-for-style-010-expected.txt: Added.
3256         * layout-tests/editing/style/create-block-for-style-010.html: Added.
3257         * layout-tests/editing/style/create-block-for-style-011-expected.txt: Added.
3258         * layout-tests/editing/style/create-block-for-style-011.html: Added.
3259         * layout-tests/editing/style/create-block-for-style-012-expected.txt: Added.
3260         * layout-tests/editing/style/create-block-for-style-012.html: Added.
3261         * layout-tests/editing/style/create-block-for-style-013-expected.txt: Added.
3262         * layout-tests/editing/style/create-block-for-style-013.html: Added.
3263
3264         Unrelated updates to these expected results.
3265         * layout-tests/editing/inserting/insert-div-007-expected.txt
3266         * layout-tests/editing/pasteboard/paste-text-013-expected.txt
3267
3268 2005-01-10  Chris Blumenberg  <cblu@apple.com>
3269
3270         Fixed: <rdar://problem/3936844> Mail: Messages with rich text do not print
3271
3272         Reviewed by john.
3273
3274         * khtml/rendering/render_flow.cpp:
3275         (RenderFlow::paintLines): don't do pagination work if printRect is not set
3276         * khtml/rendering/render_list.cpp:
3277         (RenderListMarker::paint): ditto
3278
3279 2005-01-10  David Harrison  <harrison@apple.com>
3280
3281         Reviewed by Darin.
3282
3283         * kwq/KWQTextUtilities.mm:
3284         (currentTextBreakLocaleID):
3285         Return empty string (AKA root locale) if locale pref can not be canonicalized.
3286
3287 2005-01-10  John Sullivan  <sullivan@apple.com>
3288
3289         Fixed broken Panther build.
3290
3291         * kwq/KWQTextUtilities.mm:
3292         (currentTextBreakLocaleID):
3293         This recently-added function was using code copy/pasted from CarbonCore UnicodeUtilities.
3294         That code used a Tiger-only function, CFLocaleCreateCanonicalLanguageIdentifierFromString.
3295         To fix the build, I added a BUILDING_ON_PANTHER #ifdef that avoids the call to the Tiger
3296         function. However, the Tiger-only code was wrong; the string generated using
3297         CFLocaleCreateCanonicalLanguageIdentifierFromString was not used at all, so I fixed that
3298         as well.
3299
3300 2005-01-09  David Harrison  <harrison@apple.com>
3301
3302         Reviewed by Ken Kocienda.
3303
3304         <rdar://problem/3786362> REGRESSION (Mail): pasted text loses one newline
3305
3306         * khtml/editing/htmlediting.cpp: