163911 2016-10-24 14:27:14 -0700 window.find does not scroll with overflow:auto content on mobile Safari 2025-07-07 03:19:42 -0700 1 1 1 Unclassified WebKit WebKit Misc. Safari 10 Unspecified Unspecified ASSIGNED https://bugs.webkit.org/show_bug.cgi?id=178789 InRadar P2 Normal --- 1 dvoytenko webkit-unassigned fred.wang malte rbuis simon.fraser thorton tonikitoo webkit-bug-importer oldest_to_newest 1243883 0 dvoytenko 2016-10-24 14:27:14 -0700 See http://output.jsbin.com/kufici. This page has main content wrapped into `overflow: auto; -webkit-overflow-scrolling: touch;` element. Executing `window.find('find')` on Desktop Safari finds the text and scrolls it into view as expected. Doing the same on mobile Safari (iPhone 6+) finds the text, but does NOT scroll it into view. Please advise. 1244001 1 webkit-bug-importer 2016-10-24 17:58:12 -0700 <rdar://problem/28927508> 1264127 2 malte 2017-01-05 08:25:31 -0800 Note that "Find in page" via UI in mobile Safari has the same problem. Here is a slightly simplified test case http://output.jsbin.com/piwuvu/quiet 1300972 3 308089 fred.wang 2017-04-25 04:02:48 -0700 Created attachment 308089 testcase 1302020 4 fred.wang 2017-04-27 03:36:19 -0700 So I debugged a bit what is happening here. Basically, in WebKit2/WebProcess/WebPage/FindController.cpp, FindController::findString sets a special DoNotRevealSelection flag under IOS: // iOS will reveal the selection through a different mechanism, and // we need to avoid sending the non-painted selection change to the UI process // so that it does not clear the selection out from under us. #if PLATFORM(IOS) coreOptions = static_cast<FindOptions>(coreOptions | DoNotRevealSelection); #endif As a consequence, Editor::findString will skip the call to FrameSelection::revealSelection where the actual scrolling request is performed: #if PLATFORM(IOS) if (RenderLayer* layer = start.deprecatedNode()->renderer()->enclosingLayer()) { if (!m_scrollingSuppressCount) { layer->setAdjustForIOSCaretWhenScrolling(true); layer->scrollRectToVisible(revealMode, rect, insideFixed, alignment, alignment); layer->setAdjustForIOSCaretWhenScrolling(false); updateAppearance(); if (m_frame->page()) m_frame->page()->chrome().client().notifyRevealedSelectionByScrollingFrame(*m_frame); } } #else If we force the call to FrameSelection::revealSelection then the scrolling happens as expected. Note that this behavior was added in bug 131510. 1302160 5 simon.fraser 2017-04-27 10:35:29 -0700 On the internal bug I said: The problem seems to be that FindController::updateFindIndicator() and UI-side code (oddly, SmartMagnificationController code) only know about scrolling the main page, and not about scrolling overflow. Why doesn't this use the same code path as scrollIntoView and friends? 1311436 6 simon.fraser 2017-05-22 11:07:10 -0700 <rdar://problem/26322565> 1312237 7 fred.wang 2017-05-24 02:52:11 -0700 (In reply to Simon Fraser (smfr) from comment #5) > On the internal bug I said: > > The problem seems to be that FindController::updateFindIndicator() and > UI-side code (oddly, SmartMagnificationController code) only know about > scrolling the main page, and not about scrolling overflow. > > Why doesn't this use the same code path as scrollIntoView and friends? Thank you for sharing this, Simon. I think it is useful piece of information. So my understanding is that: 1) iOS requires some specific code path to do the highlight of matched text and also this special "smart magnification" feature. 2) BUT, this currently only works for the main page and is completely broken for overflow:auto (or other situations e.g. iframes when bug 149264 is fixed) I am not exactly sure how important 1) is (maybe Timothy can provide more details here) but 2) definitely sounds serious to me. It is certainly worth discussing a bit more to see how we can make progress on it... Do you think we should fix the iOS code path to handle the non-main-frame-view case? How much effort would that require? What about the (maybe easier) alternative of falling back to the Mac code path in the non-main-frame-view case? I guess the "smart magnification" is not too important, but maybe that would cause issue for highlighting? (I also wonder if we should refactor a bit the code so that the two paths share as much code as possible) 1312285 8 simon.fraser 2017-05-24 09:54:52 -0700 (In reply to Frédéric Wang (:fredw) from comment #7) > (In reply to Simon Fraser (smfr) from comment #5) > > On the internal bug I said: > > > > The problem seems to be that FindController::updateFindIndicator() and > > UI-side code (oddly, SmartMagnificationController code) only know about > > scrolling the main page, and not about scrolling overflow. > > > > Why doesn't this use the same code path as scrollIntoView and friends? > > Thank you for sharing this, Simon. I think it is useful piece of > information. So my understanding is that: > > 1) iOS requires some specific code path to do the highlight of matched text > and also this special "smart magnification" feature. > 2) BUT, this currently only works for the main page and is completely broken > for overflow:auto (or other situations e.g. iframes when bug 149264 is fixed) > > I am not exactly sure how important 1) is (maybe Timothy can provide more > details here) but 2) definitely sounds serious to me. It is certainly worth > discussing a bit more to see how we can make progress on it... > > Do you think we should fix the iOS code path to handle the > non-main-frame-view case? How much effort would that require? We should, and we can re-use existing "scroll into view" code for this, although I think we do some zooming for Find results (but we also do this on focus, so should be similar to existing code). > What about the > (maybe easier) alternative of falling back to the Mac code path in the > non-main-frame-view case? I guess the "smart magnification" is not too > important, but maybe that would cause issue for highlighting? The "scroll into view" code is largely shared with Mac. > (I also wonder if we should refactor a bit the code so that the two paths > share as much code as possible) We should. Obviously Find needs to show the find indicator, but that should be done independently from scrolling into view. 1312296 9 thorton 2017-05-24 10:16:56 -0700 (In reply to Simon Fraser (smfr) from comment #8) > (In reply to Frédéric Wang (:fredw) from comment #7) > > (In reply to Simon Fraser (smfr) from comment #5) > > > > Do you think we should fix the iOS code path to handle the > > non-main-frame-view case? How much effort would that require? > > We should, and we can re-use existing "scroll into view" code for this, > although I think we do > some zooming for Find results (but we also do this on focus, so should be > similar to existing code). If the generic "good" scroll-into-view code supports zooming too, maybe we should use it for SmartMagnificationController too? (@frederic, "smart magnification" is "double tap to zoom") 1314184 10 fred.wang 2017-05-31 09:29:37 -0700 (In reply to Tim Horton from comment #9) > > We should, and we can re-use existing "scroll into view" code for this, > > although I think we do > > some zooming for Find results (but we also do this on focus, so should be > > similar to existing code). > > If the generic "good" scroll-into-view code supports zooming too, maybe we > should use it for SmartMagnificationController too? (@frederic, "smart > magnification" is "double tap to zoom") @Simon: So you are suggesting that zooming for focus is supported in the existing scroll-into-view code? Can you indicate when it is done? 1314566 11 fred.wang 2017-06-01 03:32:30 -0700 For the record, here are relevant backtraces for Focus (tab to an input field) and Find features. For the former feature, scrollRectToVisible is used for scrolling and _zoomToFocusRect is used for zooming. For the latter feature, scrollRectVisible is not called (as explained in comment 4). Instead, FindController::updateFindIndicator will rely on SmartMagnificationController to perform the scrolling and zooming... But this only works for the main frame :-( == Focus == WebProcess: WebCore::RenderLayer::scrollRectToVisible WebCore::FrameSelection::revealSelection WebCore::HTMLInputElement::updateFocusAppearance WebCore::Element::focus UIProcess: -[WKContentView _zoomToFocusRect] -[WKContentView _displayFormNodeInputView] == Find == WebProcess: WebCore::Editor::findString (FrameSelection::revealSelection and hence RenderLayer::scrollRectToVisible not called) WebCore::Page::findString WebKit::FindController::findString (DoNotRevealSelection is set) WebKit::WebPage::findString WebKit::FindController::updateFindIndicator (send magnify IPC message) WebKit::FindController::updateFindUIAfterPageScroll WebKit::DrawingArea::dispatchAfterEnsuringUpdatedScrollPosition WebKit::FindController::findString WebKit::WebPage::findString UIProcess: -[WKWebView _zoomToRect] (either _scrollToRect or _zoomToRect is called) -[WKContentView _zoomToRect] -[SmartMagnificationController::magnify] 1315676 12 312006 fred.wang 2017-06-05 05:02:26 -0700 Created attachment 312006 More complex testcase 1333388 13 fred.wang 2017-07-31 04:04:33 -0700 (In reply to Frédéric Wang (:fredw) from comment #11) > For the record, here are relevant backtraces for Focus (tab to an input > field) and Find features. Here are slightly more detailed traces: == Focus == WebProcess: WebKit::WebPage::elementDidFocus (sends StartAssistingNode IPC message) WebKit::WebChromeClient::elementDidFocus WebCore::Element::dispatchFocusEvent ... WebCore::Document::setFocusedElement WebCore::FocusControler::setFocusedElement WebCore::Element::focus followed by WebCore::RenderLayer::scrollRectToVisible WebCore::FrameSelection::revealSelection WebCore::HTMLInputElement::updateFocusAppearance WebCore::Element::focus UIProcess: -[WKContentView _zoomToFocusRect] -[WKContentView _displayFormNodeInputView] -[WKContentView _startAssistingNode] WebKit::PageClientImplIOS::startAssistingNode WebKit::WebPageProxy::startAssistingNode 1333826 14 316866 fred.wang 2017-08-01 09:52:06 -0700 Created attachment 316866 Patch So for the record, I'm uploading the experimental patch I had to try and do the same operations as "Focus" i.e. 1) IPC call to perform WKContentView _zoomTo*Rect (smart magnification + scrolling) 2) Followed by a call to WebCore::RenderLayer::scrollRectToVisible for the overflow nodes The following video demonstrates some bugs with attachment 312006: https://people.igalia.com/fwang/ios/2017-07-28/find.mov ; I had variants of the patch with the same issues... 1) Rendering may become wrong when the main frame is scrolled/zoomed. 2) Highlight is missing when the inner overflow node is scrolled. 3) When some text of the inner overflow node is highlighted, scrolling that node does not update the rendering of the highlighted text I am not sure how to handle all the cases but maybe we could agree on a first step to address the most important issue (no scrolling of inner frames) and use cases (only a single overflow:auto body)? For example, I'm moving (3) to bug 175032. 1363674 15 324671 fred.wang 2017-10-24 07:40:27 -0700 Created attachment 324671 Patch (WIP) New attempt. This seems to give better result. 1364180 16 324810 fred.wang 2017-10-25 05:28:20 -0700 Created attachment 324810 More complex testcase (with a button to call window.find) 1364181 17 324811 fred.wang 2017-10-25 05:32:16 -0700 Created attachment 324811 More complex testcase (with a button to call window.find) 1364183 18 324813 fred.wang 2017-10-25 06:05:35 -0700 Created attachment 324813 testcase without overflow nodes (with a button to call window.find) Here is a testcase that shows that window.find does not scroll for the main frame either. The scrolling works correctly for the main frame when the "Find UI" is used (Cmd+F on the simulator, see comment 2). Attachment 324671 fixes the cases of overflow nodes. Regarding window.find, the relevant stack trace is Editor::findString DOMWindow::find So indeed, the existing code in FindController::findString to call updateFindUIAfterPageScroll is not executed, nor is the changes made by attachment 324671. Actually, DoNotRevealSelection is not set and indeed the code in FrameSelection::revealSelection that I mentioned in comment 4 *is* executed. However, for some reason it does not seem to scroll the page (this would need to be debugged further). To summarize, window.find uses a different code path, won't use the FindController at all and does not set DoNotRevealSelection, so probably it should be handled separately. 1364767 19 fred.wang 2017-10-26 04:05:25 -0700 For the record, the trace for document.execCommand("FindString", ...) is WebCore::Editor::findString WebCore::executeFindString WebCore::Editor::Command::execute WebCore::Document::execCommand So we arrive at the same point as or window.find and we have the same bug. Passing showUI = true does not seem to have any effect. 1364850 20 simon.fraser 2017-10-26 08:48:53 -0700 The way to test this would be with new functions on UIScriptController. 1365350 21 fred.wang 2017-10-27 07:35:59 -0700 (In reply to Simon Fraser (smfr) from comment #20) > The way to test this would be with new functions on UIScriptController. @Simon: Thanks. I added a test to the patch for bug 178789 (iOS's Find UI). For the remaining cases of window.find and document.execCommand("FindString", ...), there is already a JS API to test them so UIScriptController should not be necessary. 2128092 22 fred.wang 2025-07-07 03:19:42 -0700 Unassigning myself since I'm not working on this. 308089 2017-04-25 04:02:48 -0700 2017-04-25 04:02:48 -0700 testcase 163911.html text/html 707 fred.wang PCFET0NUWVBFIGh0bWw+CjxoZWFkPgogIDxtZXRhIGNoYXJzZXQ9InV0Zi04Ij4KICA8dGl0bGU+ V2ViS2l0IEJ1ZyAxNjM5MTE8L3RpdGxlPgogIDxtZXRhIG5hbWU9InZpZXdwb3J0IiBjb250ZW50 PSJ3aWR0aD1kZXZpY2Utd2lkdGgiPgogIDxzdHlsZT4KICAgIC5zY3JvbGxhYmxlIHsKICAgICAg YmFja2dyb3VuZDogZ3JheTsKICAgICAgd2lkdGg6IDIwMHB4OwogICAgICBoZWlnaHQ6IDIwMHB4 OwogICAgICBvdmVyZmxvdy15OiBhdXRvOwogICAgICAtd2Via2l0LW92ZXJmbG93LXNjcm9sbGlu ZzogdG91Y2g7CiAgICB9CiAgICAuZ3JhZGllbnQgewogICAgICB3aWR0aDogMTAwcHg7CiAgICAg IGhlaWdodDogNTAwcHg7CiAgICAgIGJhY2tncm91bmQ6IGxpbmVhci1ncmFkaWVudCgxMzVkZWcs IHJlZCwgZ3JlZW4pOwogICAgfQogIDwvc3R5bGU+CjwvaGVhZD4KPGJvZHk+CiAgPHA+VXNlICJG aW5kIGluIHBhZ2UiIHRvIHNlYXJjaCAidGV4dCIuPC9wPgogIDxkaXYgY2xhc3M9InNjcm9sbGFi bGUiPgogICAgPGRpdiBjbGFzcz0iZ3JhZGllbnQiPjwvZGl2PgogICAgdGV4dDEKICAgIDxkaXYg Y2xhc3M9ImdyYWRpZW50Ij48L2Rpdj4KICAgIHRleHQyCiAgICA8ZGl2IGNsYXNzPSJncmFkaWVu dCI+PC9kaXY+CiAgICB0ZXh0MwogICAgPGRpdiBjbGFzcz0iZ3JhZGllbnQiPjwvZGl2PgogICAg dGV4dDQKICA8L2Rpdj4KPC9ib2R5Pgo= 312006 2017-06-05 05:02:26 -0700 2017-06-05 05:02:26 -0700 More complex testcase 163911.html text/html 1466 fred.wang PCFET0NUWVBFIGh0bWw+CjxoZWFkPgogIDxtZXRhIGNoYXJzZXQ9InV0Zi04Ij4KICA8dGl0bGU+ V2ViS2l0IEJ1ZyAxNjM5MTE8L3RpdGxlPgogIDxtZXRhIG5hbWU9InZpZXdwb3J0IiBjb250ZW50 PSJ3aWR0aD1kZXZpY2Utd2lkdGgiPgogIDxzdHlsZT4KICAgIC5zY3JvbGxhYmxlIHsKICAgICAg YmFja2dyb3VuZDogZ3JheTsKICAgICAgd2lkdGg6IDIwMHB4OwogICAgICBoZWlnaHQ6IDIwMHB4 OwogICAgICBvdmVyZmxvdy15OiBhdXRvOwogICAgICAtd2Via2l0LW92ZXJmbG93LXNjcm9sbGlu ZzogdG91Y2g7CiAgICB9CiAgICAucmVjdCB7CiAgICAgIHdpZHRoOiAxMDBweDsKICAgICAgaGVp Z2h0OiA1MDBweDsKICAgIH0KICA8L3N0eWxlPgo8L2hlYWQ+Cjxib2R5PgogIDxwPlVzZSAiRmlu ZCBpbiBwYWdlIiB0byBzZWFyY2ggInRleHQiLjwvcD4KICA8ZGl2IGNsYXNzPSJzY3JvbGxhYmxl Ij4KICAgIDxkaXYgY2xhc3M9InJlY3QiIHN0eWxlPSJiYWNrZ3JvdW5kOiBsaW5lYXItZ3JhZGll bnQoMTM1ZGVnLCByZWQsIGdyZWVuKSI+PC9kaXY+CiAgICB0ZXh0MWEgdGV4dDFiIHRleHQxYwog ICAgPGRpdiBjbGFzcz0icmVjdCIgc3R5bGU9ImJhY2tncm91bmQ6IGxpbmVhci1ncmFkaWVudCgx MzVkZWcsIGdyZWVuLCBibHVlKSI+PC9kaXY+CiAgICB0ZXh0MmEgdGV4dDJiIHRleHQyYwogICAg PGRpdiBjbGFzcz0icmVjdCIgc3R5bGU9ImJhY2tncm91bmQ6IGxpbmVhci1ncmFkaWVudCgxMzVk ZWcsIGJsdWUsIG1hZ2VudGEpIj48L2Rpdj4KICAgIHRleHQzYSB0ZXh0M2IgdGV4dDNjCiAgICA8 ZGl2IGNsYXNzPSJyZWN0IiBzdHlsZT0iYmFja2dyb3VuZDogbGluZWFyLWdyYWRpZW50KDEzNWRl ZywgbWFnZW50YSwgeWVsbG93KSI+PC9kaXY+CiAgICB0ZXh0NGEgdGV4dDRiIHRleHQ0YwogIDwv ZGl2PgogIDxkaXYgc3R5bGU9ImhlaWdodDogNTAwcHgiPnRleHQ1YSB0ZXh0NWIgdGV4dDVjPC9k aXY+CiAgPGRpdiBjbGFzcz0ic2Nyb2xsYWJsZSI+CiAgICA8ZGl2IGNsYXNzPSJyZWN0IiBzdHls ZT0iYmFja2dyb3VuZDogbGluZWFyLWdyYWRpZW50KDEzNWRlZywgcmVkLCBncmVlbikiPjwvZGl2 PgogICAgdGV4dDZhIHRleHQ2YiB0ZXh0NmMKICAgIDxkaXYgY2xhc3M9InJlY3QiIHN0eWxlPSJi YWNrZ3JvdW5kOiBsaW5lYXItZ3JhZGllbnQoMTM1ZGVnLCBncmVlbiwgYmx1ZSkiPjwvZGl2Pgog ICAgdGV4dDdhIHRleHQ3YiB0ZXh0N2MKICAgIDxkaXYgY2xhc3M9InJlY3QiIHN0eWxlPSJiYWNr Z3JvdW5kOiBsaW5lYXItZ3JhZGllbnQoMTM1ZGVnLCBibHVlLCBtYWdlbnRhKSI+PC9kaXY+CiAg ICB0ZXh0OGEgdGV4dDhiIHRleHQ4YwogICAgPGRpdiBjbGFzcz0icmVjdCIgc3R5bGU9ImJhY2tn cm91bmQ6IGxpbmVhci1ncmFkaWVudCgxMzVkZWcsIG1hZ2VudGEsIHllbGxvdykiPjwvZGl2Pgog ICAgdGV4dDlhIHRleHQ5YiB0ZXh0OWMKICA8L2Rpdj4KPC9ib2R5Pgo= 316866 2017-08-01 09:52:06 -0700 2017-08-01 09:52:06 -0700 Patch 163911.patch text/plain 894 fred.wang ZGlmZiAtLWdpdCBhL1NvdXJjZS9XZWJLaXQvV2ViUHJvY2Vzcy9XZWJQYWdlL2lvcy9GaW5kQ29u dHJvbGxlcklPUy5tbSBiL1NvdXJjZS9XZWJLaXQvV2ViUHJvY2Vzcy9XZWJQYWdlL2lvcy9GaW5k Q29udHJvbGxlcklPUy5tbQppbmRleCBkOTQ5YTFjY2QxZC4uMWQwNjIyZTYyZjQgMTAwNjQ0Ci0t LSBhL1NvdXJjZS9XZWJLaXQvV2ViUHJvY2Vzcy9XZWJQYWdlL2lvcy9GaW5kQ29udHJvbGxlcklP Uy5tbQorKysgYi9Tb3VyY2UvV2ViS2l0L1dlYlByb2Nlc3MvV2ViUGFnZS9pb3MvRmluZENvbnRy b2xsZXJJT1MubW0KQEAgLTExNiw2ICsxMTYsOCBAQCBib29sIEZpbmRDb250cm9sbGVyOjp1cGRh dGVGaW5kSW5kaWNhdG9yKEZyYW1lJiBzZWxlY3RlZEZyYW1lLCBib29sIGlzU2hvd2luZ092ZQog ICAgICAgICBJbnRSZWN0IHN0YXJ0UmVjdCA9IHZpc2libGVTZWxlY3Rpb24udmlzaWJsZVN0YXJ0 KCkuYWJzb2x1dGVDYXJldEJvdW5kcygpOwogCiAgICAgICAgIG1fd2ViUGFnZS0+c2VuZChNZXNz YWdlczo6U21hcnRNYWduaWZpY2F0aW9uQ29udHJvbGxlcjo6TWFnbmlmeShzdGFydFJlY3QuY2Vu dGVyKCksIHJlbmRlclJlY3QsIHZpc2libGVDb250ZW50UmVjdCwgbV93ZWJQYWdlLT5taW5pbXVt UGFnZVNjYWxlRmFjdG9yKCksIG1fd2ViUGFnZS0+bWF4aW11bVBhZ2VTY2FsZUZhY3RvcigpKSk7 CisKKyAgICAgICAgc2VsZWN0ZWRGcmFtZS5zZWxlY3Rpb24oKS5yZXZlYWxTZWxlY3Rpb24oU2Vs ZWN0aW9uUmV2ZWFsTW9kZTo6UmV2ZWFsVXBUb01haW5GcmFtZSwgU2Nyb2xsQWxpZ25tZW50Ojph bGlnblRvRWRnZUlmTmVlZGVkLCBXZWJDb3JlOjpEb05vdFJldmVhbEV4dGVudCk7CiAgICAgfQog CiAgICAgbV9pc1Nob3dpbmdGaW5kSW5kaWNhdG9yID0gdHJ1ZTsK 324671 2017-10-24 07:40:27 -0700 2017-10-24 07:40:27 -0700 Patch (WIP) 163911.patch text/plain 1106 fred.wang ZGlmZiAtLWdpdCBhL1NvdXJjZS9XZWJLaXQvV2ViUHJvY2Vzcy9XZWJQYWdlL0ZpbmRDb250cm9s bGVyLmNwcCBiL1NvdXJjZS9XZWJLaXQvV2ViUHJvY2Vzcy9XZWJQYWdlL0ZpbmRDb250cm9sbGVy LmNwcAppbmRleCBlNTY0MjNhZDYxLi5iMGViMjJjOWMwIDEwMDY0NAotLS0gYS9Tb3VyY2UvV2Vi S2l0L1dlYlByb2Nlc3MvV2ViUGFnZS9GaW5kQ29udHJvbGxlci5jcHAKKysrIGIvU291cmNlL1dl YktpdC9XZWJQcm9jZXNzL1dlYlBhZ2UvRmluZENvbnRyb2xsZXIuY3BwCkBAIC0yNTEsNiArMjUx LDExIEBAIHZvaWQgRmluZENvbnRyb2xsZXI6OmZpbmRTdHJpbmcoY29uc3QgU3RyaW5nJiBzdHJp bmcsIEZpbmRPcHRpb25zIG9wdGlvbnMsIHVuc2lnCiAgICAgfQogCiAgICAgUmVmUHRyPFdlYlBh Z2U+IHByb3RlY3RlZFdlYlBhZ2UgPSBtX3dlYlBhZ2U7CisjaWYgUExBVEZPUk0oSU9TKQorICAg IC8vIFNjcm9sbGluZyB0aGUgbWFpbiBmcmFtZSBpcyBoYW5kbGVkIGJ5IHRoZSBTbWFydE1hZ25p ZmljYXRpb25Db250cm9sbGVyIGNsYXNzIGJ1dCB3ZSBzdGlsbAorICAgIC8vIG5lZWQgdG8gY29u c2lkZXIgb3ZlcmZsb3cgbm9kZXMgYW5kIHN1YmZyYW1lcyBoZXJlLgorICAgIHByb3RlY3RlZFdl YlBhZ2UtPmNvcmVQYWdlKCktPmZvY3VzQ29udHJvbGxlcigpLmZvY3VzZWRPck1haW5GcmFtZSgp LnNlbGVjdGlvbigpLnJldmVhbFNlbGVjdGlvbihTZWxlY3Rpb25SZXZlYWxNb2RlOjpSZXZlYWxV cFRvTWFpbkZyYW1lLCBTY3JvbGxBbGlnbm1lbnQ6OmFsaWduVG9FZGdlSWZOZWVkZWQsIFdlYkNv cmU6OkRvTm90UmV2ZWFsRXh0ZW50KTsKKyNlbmRpZgogICAgIG1fd2ViUGFnZS0+ZHJhd2luZ0Fy ZWEoKS0+ZGlzcGF0Y2hBZnRlckVuc3VyaW5nVXBkYXRlZFNjcm9sbFBvc2l0aW9uKFtwcm90ZWN0 ZWRXZWJQYWdlLCBmb3VuZCwgc3RyaW5nLCBvcHRpb25zLCBtYXhNYXRjaENvdW50LCBkaWRXcmFw XSAoKSB7CiAgICAgICAgIHByb3RlY3RlZFdlYlBhZ2UtPmZpbmRDb250cm9sbGVyKCkudXBkYXRl RmluZFVJQWZ0ZXJQYWdlU2Nyb2xsKGZvdW5kLCBzdHJpbmcsIG9wdGlvbnMsIG1heE1hdGNoQ291 bnQsIGRpZFdyYXApOwogICAgIH0pOwo= 324810 2017-10-25 05:28:20 -0700 2017-10-25 05:32:16 -0700 More complex testcase (with a button to call window.find) window.find.html text/html 1532 fred.wang PCFET0NUWVBFIGh0bWw+CjxoZWFkPgogIDxtZXRhIGNoYXJzZXQ9InV0Zi04Ij4KICA8dGl0bGU+ V2ViS2l0IEJ1ZyAxNjM5MTE8L3RpdGxlPgogIDxtZXRhIG5hbWU9InZpZXdwb3J0IiBjb250ZW50 PSJ3aWR0aD1kZXZpY2Utd2lkdGgiPgogIDxzdHlsZT4KICAgIC5zY3JvbGxhYmxlIHsKICAgICAg YmFja2dyb3VuZDogZ3JheTsKICAgICAgd2lkdGg6IDIwMHB4OwogICAgICBoZWlnaHQ6IDIwMHB4 OwogICAgICBvdmVyZmxvdy15OiBhdXRvOwogICAgICAtd2Via2l0LW92ZXJmbG93LXNjcm9sbGlu ZzogdG91Y2g7CiAgICB9CiAgICAucmVjdCB7CiAgICAgIHdpZHRoOiAxMDBweDsKICAgICAgaGVp Z2h0OiA1MDBweDsKICAgIH0KICAgIGJ1dHRvbiB7CiAgICAgIHBvc2l0aW9uOiBmaXhlZDsKICAg IH0KICA8L3N0eWxlPgo8L2hlYWQ+Cjxib2R5PgogIDxwPjxidXR0b24gb25jbGljaz0id2luZG93 LmZpbmQoJ3RleHQnKSI+U2VhcmNoICJ0ZXh0IjwvYnV0dG9uPjwvcD4KICA8ZGl2IGNsYXNzPSJz Y3JvbGxhYmxlIj4KICAgIDxkaXYgY2xhc3M9InJlY3QiIHN0eWxlPSJiYWNrZ3JvdW5kOiBsaW5l YXItZ3JhZGllbnQoMTM1ZGVnLCByZWQsIGdyZWVuKSI+PC9kaXY+CiAgICB0ZXh0MWEgdGV4dDFi IHRleHQxYwogICAgPGRpdiBjbGFzcz0icmVjdCIgc3R5bGU9ImJhY2tncm91bmQ6IGxpbmVhci1n cmFkaWVudCgxMzVkZWcsIGdyZWVuLCBibHVlKSI+PC9kaXY+CiAgICB0ZXh0MmEgdGV4dDJiIHRl eHQyYwogICAgPGRpdiBjbGFzcz0icmVjdCIgc3R5bGU9ImJhY2tncm91bmQ6IGxpbmVhci1ncmFk aWVudCgxMzVkZWcsIGJsdWUsIG1hZ2VudGEpIj48L2Rpdj4KICAgIHRleHQzYSB0ZXh0M2IgdGV4 dDNjCiAgICA8ZGl2IGNsYXNzPSJyZWN0IiBzdHlsZT0iYmFja2dyb3VuZDogbGluZWFyLWdyYWRp ZW50KDEzNWRlZywgbWFnZW50YSwgeWVsbG93KSI+PC9kaXY+CiAgICB0ZXh0NGEgdGV4dDRiIHRl eHQ0YwogIDwvZGl2PgogIDxkaXYgc3R5bGU9ImhlaWdodDogNTAwcHgiPnRleHQ1YSB0ZXh0NWIg dGV4dDVjPC9kaXY+CiAgPGRpdiBjbGFzcz0ic2Nyb2xsYWJsZSI+CiAgICA8ZGl2IGNsYXNzPSJy ZWN0IiBzdHlsZT0iYmFja2dyb3VuZDogbGluZWFyLWdyYWRpZW50KDEzNWRlZywgcmVkLCBncmVl bikiPjwvZGl2PgogICAgdGV4dDZhIHRleHQ2YiB0ZXh0NmMKICAgIDxkaXYgY2xhc3M9InJlY3Qi IHN0eWxlPSJiYWNrZ3JvdW5kOiBsaW5lYXItZ3JhZGllbnQoMTM1ZGVnLCBncmVlbiwgYmx1ZSki PjwvZGl2PgogICAgdGV4dDdhIHRleHQ3YiB0ZXh0N2MKICAgIDxkaXYgY2xhc3M9InJlY3QiIHN0 eWxlPSJiYWNrZ3JvdW5kOiBsaW5lYXItZ3JhZGllbnQoMTM1ZGVnLCBibHVlLCBtYWdlbnRhKSI+ PC9kaXY+CiAgICB0ZXh0OGEgdGV4dDhiIHRleHQ4YwogICAgPGRpdiBjbGFzcz0icmVjdCIgc3R5 bGU9ImJhY2tncm91bmQ6IGxpbmVhci1ncmFkaWVudCgxMzVkZWcsIG1hZ2VudGEsIHllbGxvdyki PjwvZGl2PgogICAgdGV4dDlhIHRleHQ5YiB0ZXh0OWMKICA8L2Rpdj4KPC9ib2R5Pgo= 324811 2017-10-25 05:32:16 -0700 2017-10-25 05:32:16 -0700 More complex testcase (with a button to call window.find) window.find.html text/html 1550 fred.wang PCFET0NUWVBFIGh0bWw+CjxoZWFkPgogIDxtZXRhIGNoYXJzZXQ9InV0Zi04Ij4KICA8dGl0bGU+ V2ViS2l0IEJ1ZyAxNjM5MTE8L3RpdGxlPgogIDxtZXRhIG5hbWU9InZpZXdwb3J0IiBjb250ZW50 PSJ3aWR0aD1kZXZpY2Utd2lkdGgiPgogIDxzdHlsZT4KICAgIC5zY3JvbGxhYmxlIHsKICAgICAg YmFja2dyb3VuZDogZ3JheTsKICAgICAgd2lkdGg6IDIwMHB4OwogICAgICBoZWlnaHQ6IDIwMHB4 OwogICAgICBvdmVyZmxvdy15OiBhdXRvOwogICAgICAtd2Via2l0LW92ZXJmbG93LXNjcm9sbGlu ZzogdG91Y2g7CiAgICB9CiAgICAucmVjdCB7CiAgICAgIHdpZHRoOiAxMDBweDsKICAgICAgaGVp Z2h0OiA1MDBweDsKICAgIH0KICAgIGJ1dHRvbiB7CiAgICAgIHBvc2l0aW9uOiBmaXhlZDsKICAg ICAgei1pbmRleDogMTsKICAgIH0KICA8L3N0eWxlPgo8L2hlYWQ+Cjxib2R5PgogIDxwPjxidXR0 b24gb25jbGljaz0id2luZG93LmZpbmQoJ3RleHQnKSI+U2VhcmNoICJ0ZXh0IjwvYnV0dG9uPjwv cD4KICA8ZGl2IGNsYXNzPSJzY3JvbGxhYmxlIj4KICAgIDxkaXYgY2xhc3M9InJlY3QiIHN0eWxl PSJiYWNrZ3JvdW5kOiBsaW5lYXItZ3JhZGllbnQoMTM1ZGVnLCByZWQsIGdyZWVuKSI+PC9kaXY+ CiAgICB0ZXh0MWEgdGV4dDFiIHRleHQxYwogICAgPGRpdiBjbGFzcz0icmVjdCIgc3R5bGU9ImJh Y2tncm91bmQ6IGxpbmVhci1ncmFkaWVudCgxMzVkZWcsIGdyZWVuLCBibHVlKSI+PC9kaXY+CiAg ICB0ZXh0MmEgdGV4dDJiIHRleHQyYwogICAgPGRpdiBjbGFzcz0icmVjdCIgc3R5bGU9ImJhY2tn cm91bmQ6IGxpbmVhci1ncmFkaWVudCgxMzVkZWcsIGJsdWUsIG1hZ2VudGEpIj48L2Rpdj4KICAg IHRleHQzYSB0ZXh0M2IgdGV4dDNjCiAgICA8ZGl2IGNsYXNzPSJyZWN0IiBzdHlsZT0iYmFja2dy b3VuZDogbGluZWFyLWdyYWRpZW50KDEzNWRlZywgbWFnZW50YSwgeWVsbG93KSI+PC9kaXY+CiAg ICB0ZXh0NGEgdGV4dDRiIHRleHQ0YwogIDwvZGl2PgogIDxkaXYgc3R5bGU9ImhlaWdodDogNTAw cHgiPnRleHQ1YSB0ZXh0NWIgdGV4dDVjPC9kaXY+CiAgPGRpdiBjbGFzcz0ic2Nyb2xsYWJsZSI+ CiAgICA8ZGl2IGNsYXNzPSJyZWN0IiBzdHlsZT0iYmFja2dyb3VuZDogbGluZWFyLWdyYWRpZW50 KDEzNWRlZywgcmVkLCBncmVlbikiPjwvZGl2PgogICAgdGV4dDZhIHRleHQ2YiB0ZXh0NmMKICAg IDxkaXYgY2xhc3M9InJlY3QiIHN0eWxlPSJiYWNrZ3JvdW5kOiBsaW5lYXItZ3JhZGllbnQoMTM1 ZGVnLCBncmVlbiwgYmx1ZSkiPjwvZGl2PgogICAgdGV4dDdhIHRleHQ3YiB0ZXh0N2MKICAgIDxk aXYgY2xhc3M9InJlY3QiIHN0eWxlPSJiYWNrZ3JvdW5kOiBsaW5lYXItZ3JhZGllbnQoMTM1ZGVn LCBibHVlLCBtYWdlbnRhKSI+PC9kaXY+CiAgICB0ZXh0OGEgdGV4dDhiIHRleHQ4YwogICAgPGRp diBjbGFzcz0icmVjdCIgc3R5bGU9ImJhY2tncm91bmQ6IGxpbmVhci1ncmFkaWVudCgxMzVkZWcs IG1hZ2VudGEsIHllbGxvdykiPjwvZGl2PgogICAgdGV4dDlhIHRleHQ5YiB0ZXh0OWMKICA8L2Rp dj4KPC9ib2R5Pgo= 324813 2017-10-25 06:05:35 -0700 2017-10-25 06:05:35 -0700 testcase without overflow nodes (with a button to call window.find) window.find.html text/html 1315 fred.wang PCFET0NUWVBFIGh0bWw+CjxoZWFkPgogIDxtZXRhIGNoYXJzZXQ9InV0Zi04Ij4KICA8dGl0bGU+ V2ViS2l0IEJ1ZyAxNjM5MTE8L3RpdGxlPgogIDxtZXRhIG5hbWU9InZpZXdwb3J0IiBjb250ZW50 PSJ3aWR0aD1kZXZpY2Utd2lkdGgiPgogIDxzdHlsZT4KICAgIC5yZWN0IHsKICAgICAgd2lkdGg6 IDEwMHB4OwogICAgICBoZWlnaHQ6IDUwMHB4OwogICAgfQogICAgYnV0dG9uIHsKICAgICAgcG9z aXRpb246IGZpeGVkOwogICAgICB6LWluZGV4OiAxOwogICAgfQogIDwvc3R5bGU+CjwvaGVhZD4K PGJvZHk+CiAgPHA+PGJ1dHRvbiBvbmNsaWNrPSJ3aW5kb3cuZmluZCgndGV4dCcpIj5TZWFyY2gg InRleHQiPC9idXR0b24+PC9wPgogIDxkaXYgY2xhc3M9InJlY3QiIHN0eWxlPSJiYWNrZ3JvdW5k OiBsaW5lYXItZ3JhZGllbnQoMTM1ZGVnLCByZWQsIGdyZWVuKSI+PC9kaXY+CiAgdGV4dDFhIHRl eHQxYiB0ZXh0MWMKICA8ZGl2IGNsYXNzPSJyZWN0IiBzdHlsZT0iYmFja2dyb3VuZDogbGluZWFy LWdyYWRpZW50KDEzNWRlZywgZ3JlZW4sIGJsdWUpIj48L2Rpdj4KICB0ZXh0MmEgdGV4dDJiIHRl eHQyYwogIDxkaXYgY2xhc3M9InJlY3QiIHN0eWxlPSJiYWNrZ3JvdW5kOiBsaW5lYXItZ3JhZGll bnQoMTM1ZGVnLCBibHVlLCBtYWdlbnRhKSI+PC9kaXY+CiAgdGV4dDNhIHRleHQzYiB0ZXh0M2MK ICA8ZGl2IGNsYXNzPSJyZWN0IiBzdHlsZT0iYmFja2dyb3VuZDogbGluZWFyLWdyYWRpZW50KDEz NWRlZywgbWFnZW50YSwgeWVsbG93KSI+PC9kaXY+CiAgdGV4dDRhIHRleHQ0YiB0ZXh0NGMKICA8 ZGl2IHN0eWxlPSJoZWlnaHQ6IDUwMHB4Ij50ZXh0NWEgdGV4dDViIHRleHQ1YzwvZGl2PgogIDxk aXYgY2xhc3M9InJlY3QiIHN0eWxlPSJiYWNrZ3JvdW5kOiBsaW5lYXItZ3JhZGllbnQoMTM1ZGVn LCByZWQsIGdyZWVuKSI+PC9kaXY+CiAgICB0ZXh0NmEgdGV4dDZiIHRleHQ2YwogICAgPGRpdiBj bGFzcz0icmVjdCIgc3R5bGU9ImJhY2tncm91bmQ6IGxpbmVhci1ncmFkaWVudCgxMzVkZWcsIGdy ZWVuLCBibHVlKSI+PC9kaXY+CiAgICB0ZXh0N2EgdGV4dDdiIHRleHQ3YwogICAgPGRpdiBjbGFz cz0icmVjdCIgc3R5bGU9ImJhY2tncm91bmQ6IGxpbmVhci1ncmFkaWVudCgxMzVkZWcsIGJsdWUs IG1hZ2VudGEpIj48L2Rpdj4KICAgIHRleHQ4YSB0ZXh0OGIgdGV4dDhjCiAgICA8ZGl2IGNsYXNz PSJyZWN0IiBzdHlsZT0iYmFja2dyb3VuZDogbGluZWFyLWdyYWRpZW50KDEzNWRlZywgbWFnZW50 YSwgeWVsbG93KSI+PC9kaXY+CiAgICB0ZXh0OWEgdGV4dDliIHRleHQ5YwogIDwvZGl2Pgo8L2Jv ZHk+Cg==