7321 2006-02-16 21:48:05 -0800 REGRESSION: style tag in body causes two head elements to appear in the DOM 2006-03-07 20:29:59 -0800 1 1 1 Unclassified WebKit DOM 420+ Mac OS X 10.4 VERIFIED FIXED http://software.hixie.ch/utilities/js/live-dom-viewer/?%3Chtml%3E%3Chead%3E%3C/head%3E%3Cbody%3E%3Cstyle%3E%3C/style%3E%3C/body%3E%3C/html%3E%0A EasyFix, Regression P1 Normal --- 1 ddkilzer ddkilzer oldest_to_newest 33091 0 ddkilzer 2006-02-16 21:48:05 -0800 Summary: A <style> tag that appears in the <body> of an HTML document causes two <head> elements to appear in the DOM. Steps to reproduce: 1. Load an HTML document with a <style> tag in the <body> and a <head> element. 2. View the HTML document's DOM. Ian Hixie's Live DOM Inspector works well, too. http://software.hixie.ch/utilities/js/live-dom-viewer/ Expected results: The DOM should only have one HEAD element. Actual results: The DOM has two HEAD elements. Regression: This is a regression from Safari 2.0.3 (417.8) on Mac OS X 10.4.5, which only displays one HEAD element in the DOM. Tested on WebKit nightly build r12864. 33307 1 ddkilzer 2006-02-18 16:27:39 -0800 Checking the nightly builds, this bug is as old as the first nightly, WebKit-CVS-2005-10-01 03:27:01 GMT.dmg, so it's been around a while. 33314 2 darin 2006-02-18 17:40:02 -0800 The bug is in HTMLParser::headCreateErrorCheck. That function should always return false, and when it decides to create a head element it needs to store a pointer to it in "head". See formCreateErrorCheck and mapCreateErrorCheck for similar logic. The only part I'm not entirely sure about is the "current->localName() == htmlTag" clause. That means that you're allowed to create a second head if you're inside an <html> element; would be good to test that to see if it's the behavior we want. 33336 3 ddkilzer 2006-02-18 21:01:58 -0800 (In reply to comment #2) > The bug is in HTMLParser::headCreateErrorCheck. That function should always > return false, and when it decides to create a head element it needs to store a > pointer to it in "head". What about HTMLParser::createHead() and the duplicate head-creation code under "Step 2" in HTMLParser::handleError()? 33428 4 ddkilzer 2006-02-19 17:50:11 -0800 With Darin's proposed fix: bool HTMLParser::headCreateErrorCheck(Token* t, RefPtr<NodeImpl>& result) { if (!head || current->localName() == htmlTag) { head = new HTMLHeadElementImpl(document); result = head; } return false; } We must make sure that no more than one <title> and one <base> tag appear in the <head>. (This is not the case if only the above change is made.) Firefox 1.5 currently prevents multiple <title> tags, but it doesn't prevent multiple <base> tags. I haven't surveyed MSIE 6 yet. Regarding Comment #3, there seems like there might be a lot of unused code in the handleError() method, but I don't grok the HTML parser well enough (yet) to understand when all that code may be used. 33432 5 ddkilzer 2006-02-19 19:52:46 -0800 (In reply to comment #4) > bool HTMLParser::headCreateErrorCheck(Token* t, RefPtr<NodeImpl>& result) > { > if (!head || current->localName() == htmlTag) { > head = new HTMLHeadElementImpl(document); > result = head; > } > return false; > } This change also causes an HTML document like this to crash WebKit: <<head><style></style></head> I don't think this change will be quite as easy as Darin first expected. 33492 6 ddkilzer 2006-02-20 07:49:25 -0800 This fix won't crash with the test case from Comment #5, but it fails about 375 LayoutTests. More investigation will be needed. (I didn't try running LayoutTests after applying the change from Comment #5, yet.) bool HTMLParser::headCreateErrorCheck(Token* t, RefPtr<NodeImpl>& result) { createHead(); return false; } NOTE: The multiple <title> and <base> tags in <head> issue also occurs on Safari 2.0.3 on 10.4.5, so this isn't a regression. I will file a new bug (or bugs) for this issue. 33494 7 ddkilzer 2006-02-20 08:00:06 -0800 Darin's fix causes a crash (see Comment #5) but only fails one LayoutTest: editing/inserting/editing-empty-divs. 33754 8 ddkilzer 2006-02-22 22:25:25 -0800 I believe I have a fix for this. Also had to modify HTMLParser::parseToken() so that 'head' behaved the same way as 'map' and 'form'. Need a test case, though. 33764 9 6676 ddkilzer 2006-02-23 05:58:49 -0800 Created attachment 6676 Patch v1 Patch v1. Maciej commented on IRC that he wanted to see my next patch, so I'm giving him first dibs on the review. :) I assume I'll get a review- because there is no layout test, but I'm looking for feedback on the patch and on how to implement a layout test. Regarding a layout test, it seems to me that a layoutTestController.dumpDOMTree() method would be really handy for a test like this. Is this worth implementing, or is there a preferred method (such as JavaScript DOM inspection) to test that the DOM has but a single HEAD node? 33765 10 ddkilzer 2006-02-23 06:08:54 -0800 FWIW, the crashes I found from previous attempts at patches (Comment #5 and Comment #7) were because HTMLParser::parseToken() was not modified to set "head = 0". In those cases, the new head node was never inserted into the tree and subsequently caused a double-free error later in the parsing when an attempt was made to add a child node to it. 33767 11 ddkilzer 2006-02-23 07:19:01 -0800 (In reply to comment #9) > Created an attachment (id=6676) [edit] > Patch v1 I should also note that this patch passes all current layout tests as of r12944 (except for the ones that have known issues like some of the new http tests--failures are nearly identical to buildbot--and the timezone issue with fast/js/date-constructor). 33768 12 darin 2006-02-23 07:35:52 -0800 (In reply to comment #9) > Maciej commented on IRC that he wanted to see my next patch, so I'm > giving him first dibs on the review. :) I assume I'll get a review- because > there is no layout test, but I'm looking for feedback on the patch and on how > to implement a layout test. The patch looks perfect to me. And I also think it's straightforward enough that it doesn't need to be Maciej who reviews it. > Regarding a layout test, it seems to me that a > layoutTestController.dumpDOMTree() method would be really handy for a test like > this. Is this worth implementing, or is there a preferred method (such as > JavaScript DOM inspection) to test that the DOM has but a single HEAD node? It's very easy to find out how many <head> elements there are. document.getElementsByTagName("head").length() So I don't think you need a new layoutTestController feature. 33772 13 darin 2006-02-23 08:42:58 -0800 Given the other code to special-case <head>, I'd like to see layout tests that check all the different code paths. A great way to do that is to set a breakpoint on each one and make sure every one gets tickled. 33782 14 mjs 2006-02-23 10:49:34 -0800 I didn't mean I wanted to be the one to review necessarily, I just wanted to encourage you to send more patches. :-) The patch looks good to me too, marking r- for lack of test case. 33936 15 6721 ddkilzer 2006-02-24 22:42:48 -0800 Created attachment 6721 Patch v2 Changes since Patch v1: - Added a test that uses an iframe to load various HTML documents that test each code path in htmlparser.cpp. 34022 16 6721 darin 2006-02-25 20:44:17 -0800 Comment on attachment 6721 Patch v2 r=me 35479 17 ddkilzer 2006-03-07 20:29:59 -0800 Verified fixed as of r13202 (locally built). 6676 2006-02-23 05:58:49 -0800 2006-02-24 22:42:48 -0800 Patch v1 bug-7321-v1.diff text/plain 3002 ddkilzer SW5kZXg6IExheW91dFRlc3RzL2VkaXRpbmcvaW5zZXJ0aW5nL2VkaXRpbmctZW1wdHktZGl2cy1l eHBlY3RlZC50eHQKPT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PQotLS0gTGF5b3V0VGVzdHMvZWRpdGluZy9pbnNlcnRpbmcv ZWRpdGluZy1lbXB0eS1kaXZzLWV4cGVjdGVkLnR4dAkocmV2aXNpb24gMTI5NDQpCisrKyBMYXlv dXRUZXN0cy9lZGl0aW5nL2luc2VydGluZy9lZGl0aW5nLWVtcHR5LWRpdnMtZXhwZWN0ZWQudHh0 CSh3b3JraW5nIGNvcHkpCkBAIC04MCw0ICs4MCw0IEBAIGxheWVyIGF0ICgwLDApIHNpemUgODAw eDYwMAogICAgICAgICAgIFJlbmRlclRleHQge1RFWFR9IGF0ICgwLDApIHNpemUgN3gxOAogICAg ICAgICAgICAgdGV4dCBydW4gYXQgKDAsMCkgd2lkdGggNzogImMiCiAgICAgICAgIFJlbmRlckJs b2NrIHtQfSBhdCAoMSwzNSkgc2l6ZSA3MDV4MAotY2FyZXQ6IHBvc2l0aW9uIDEgb2YgY2hpbGQg MCB7VEVYVH0gb2YgY2hpbGQgMjEge0RJVn0gb2YgY2hpbGQgMyB7Qk9EWX0gb2YgY2hpbGQgMCB7 SFRNTH0gb2YgZG9jdW1lbnQKK2NhcmV0OiBwb3NpdGlvbiAxIG9mIGNoaWxkIDAge1RFWFR9IG9m IGNoaWxkIDIxIHtESVZ9IG9mIGNoaWxkIDIge0JPRFl9IG9mIGNoaWxkIDAge0hUTUx9IG9mIGRv Y3VtZW50CkluZGV4OiBMYXlvdXRUZXN0cy9DaGFuZ2VMb2cKPT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PQotLS0gTGF5b3V0 VGVzdHMvQ2hhbmdlTG9nCShyZXZpc2lvbiAxMjk0NCkKKysrIExheW91dFRlc3RzL0NoYW5nZUxv Zwkod29ya2luZyBjb3B5KQpAQCAtMSwzICsxLDE2IEBACisyMDA2LTAyLTIzICBEYXZpZCBLaWx6 ZXIgIDxkZGtpbHplckBraWx6ZXIubmV0PgorCisgICAgICAgIFJldmlld2VkIGJ5IE5PQk9EWSAo T09QUyEpLgorCisgICAgICAgIC0gVGVzdCBmb3IgaHR0cDovL2J1Z3ppbGxhLm9wZW5kYXJ3aW4u b3JnL3Nob3dfYnVnLmNnaT9pZD03MzIxCisgICAgICAgIFJFR1JFU1NJT046IHN0eWxlIHRhZyBp biBib2R5IGNhdXNlcyB0d28gaGVhZCBlbGVtZW50cyB0byBhcHBlYXIgaW4gdGhlIERPTQorCisg ICAgICAgIEZJWE1FIChPT1BTISk6IE5lZWQgYSBsYXlvdXQgdGVzdCBmb3IgdGhpcyBidWcuCisK KyAgICAgICAgKiBlZGl0aW5nL2luc2VydGluZy9lZGl0aW5nLWVtcHR5LWRpdnMtZXhwZWN0ZWQu dHh0OgorICAgICAgICBDb3JyZWN0ZWQgPGh0bWw+IGNoaWxkIGluZGV4IGZvciA8Ym9keT4gc2lu Y2UgdGhlIGRvY3VtZW50IHByZXZpb3VzbHkgaGFkCisgICAgICAgIHR3byA8aGVhZD4gbm9kZXMu CisKIDIwMDYtMDItMjIgIEFsZXhleSBQcm9za3VyeWFrb3YgIDxhcEBueXBvcC5jb20+CiAKICAg ICAgICAgUmV2aWV3ZWQgYnkgRGFyaW4uCkluZGV4OiBXZWJDb3JlL2todG1sL2h0bWwvaHRtbHBh cnNlci5jcHAKPT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PQotLS0gV2ViQ29yZS9raHRtbC9odG1sL2h0bWxwYXJzZXIuY3Bw CShyZXZpc2lvbiAxMjk0NCkKKysrIFdlYkNvcmUva2h0bWwvaHRtbC9odG1scGFyc2VyLmNwcAko d29ya2luZyBjb3B5KQpAQCAtMjMxLDcgKzIzMSwxMCBAQCBQYXNzUmVmUHRyPE5vZGVJbXBsPiBI VE1MUGFyc2VyOjpwYXJzZVRvCiAKICAgICAgICAgaWYgKGZvcm0gPT0gbikKICAgICAgICAgICAg IGZvcm0gPSAwOwotICAgICAgICAKKworICAgICAgICBpZiAoaGVhZCA9PSBuKQorICAgICAgICAg ICAgaGVhZCA9IDA7CisKICAgICAgICAgcmV0dXJuIDA7CiAgICAgfQogICAgIHJldHVybiBuOwpA QCAtNjA3LDcgKzYxMCwxMSBAQCBib29sIEhUTUxQYXJzZXI6OmNvbW1lbnRDcmVhdGVFcnJvckNo ZWNrCiAKIGJvb2wgSFRNTFBhcnNlcjo6aGVhZENyZWF0ZUVycm9yQ2hlY2soVG9rZW4qIHQsIFJl ZlB0cjxOb2RlSW1wbD4mIHJlc3VsdCkKIHsKLSAgICByZXR1cm4gKCFoZWFkIHx8IGN1cnJlbnQt PmxvY2FsTmFtZSgpID09IGh0bWxUYWcpOworICAgIGlmICghaGVhZCB8fCBjdXJyZW50LT5sb2Nh bE5hbWUoKSA9PSBodG1sVGFnKSB7CisgICAgICAgIGhlYWQgPSBuZXcgSFRNTEhlYWRFbGVtZW50 SW1wbChkb2N1bWVudCk7CisgICAgICAgIHJlc3VsdCA9IGhlYWQ7CisgICAgfQorICAgIHJldHVy biBmYWxzZTsKIH0KIAogYm9vbCBIVE1MUGFyc2VyOjpib2R5Q3JlYXRlRXJyb3JDaGVjayhUb2tl biogdCwgUmVmUHRyPE5vZGVJbXBsPiYgcmVzdWx0KQpJbmRleDogV2ViQ29yZS9DaGFuZ2VMb2cK PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PQotLS0gV2ViQ29yZS9DaGFuZ2VMb2cJKHJldmlzaW9uIDEyOTQ0KQorKysgV2Vi Q29yZS9DaGFuZ2VMb2cJKHdvcmtpbmcgY29weSkKQEAgLTEsMyArMSwxNCBAQAorMjAwNi0wMi0y MyAgRGF2aWQgS2lsemVyICA8ZGRraWx6ZXJAa2lsemVyLm5ldD4KKworICAgICAgICBSZXZpZXdl ZCBieSBOT0JPRFkgKE9PUFMhKS4KKworICAgICAgICAtIEZpeCBmb3IgaHR0cDovL2J1Z3ppbGxh Lm9wZW5kYXJ3aW4ub3JnL3Nob3dfYnVnLmNnaT9pZD03MzIxCisgICAgICAgIFJFR1JFU1NJT046 IHN0eWxlIHRhZyBpbiBib2R5IGNhdXNlcyB0d28gaGVhZCBlbGVtZW50cyB0byBhcHBlYXIgaW4g dGhlIERPTQorCisgICAgICAgICoga2h0bWwvaHRtbC9odG1scGFyc2VyLmNwcDogSGFuZGxlIGhl YWQgbm9kZSBsaWtlIGZvcm0gYW5kIG1hcCBub2Rlcy4KKyAgICAgICAgKFdlYkNvcmU6OkhUTUxQ YXJzZXI6OnBhcnNlVG9rZW4pOgorICAgICAgICAoV2ViQ29yZTo6SFRNTFBhcnNlcjo6aGVhZENy ZWF0ZUVycm9yQ2hlY2spOgorCiAyMDA2LTAyLTIyICBFcmljIFNlaWRlbCAgPGVzZWlkZWxAYXBw bGUuY29tPgogCiAgICAgICAgIFJldmlld2VkIGJ5IGRhcmluLgo= 6721 2006-02-24 22:42:48 -0800 2006-02-25 20:44:17 -0800 Patch v2 bug-7321-v2.diff text/plain 8919 ddkilzer SW5kZXg6IExheW91dFRlc3RzL2VkaXRpbmcvaW5zZXJ0aW5nL2VkaXRpbmctZW1wdHktZGl2cy1l eHBlY3RlZC50eHQKPT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PQotLS0gTGF5b3V0VGVzdHMvZWRpdGluZy9pbnNlcnRpbmcv ZWRpdGluZy1lbXB0eS1kaXZzLWV4cGVjdGVkLnR4dAkocmV2aXNpb24gMTI5ODgpCisrKyBMYXlv dXRUZXN0cy9lZGl0aW5nL2luc2VydGluZy9lZGl0aW5nLWVtcHR5LWRpdnMtZXhwZWN0ZWQudHh0 CSh3b3JraW5nIGNvcHkpCkBAIC04MCw0ICs4MCw0IEBAIGxheWVyIGF0ICgwLDApIHNpemUgODAw eDYwMAogICAgICAgICAgIFJlbmRlclRleHQge1RFWFR9IGF0ICgwLDApIHNpemUgN3gxOAogICAg ICAgICAgICAgdGV4dCBydW4gYXQgKDAsMCkgd2lkdGggNzogImMiCiAgICAgICAgIFJlbmRlckJs b2NrIHtQfSBhdCAoMSwzNSkgc2l6ZSA3MDV4MAotY2FyZXQ6IHBvc2l0aW9uIDEgb2YgY2hpbGQg MCB7VEVYVH0gb2YgY2hpbGQgMjEge0RJVn0gb2YgY2hpbGQgMyB7Qk9EWX0gb2YgY2hpbGQgMCB7 SFRNTH0gb2YgZG9jdW1lbnQKK2NhcmV0OiBwb3NpdGlvbiAxIG9mIGNoaWxkIDAge1RFWFR9IG9m IGNoaWxkIDIxIHtESVZ9IG9mIGNoaWxkIDIge0JPRFl9IG9mIGNoaWxkIDAge0hUTUx9IG9mIGRv Y3VtZW50CkluZGV4OiBMYXlvdXRUZXN0cy9DaGFuZ2VMb2cKPT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PQotLS0gTGF5b3V0 VGVzdHMvQ2hhbmdlTG9nCShyZXZpc2lvbiAxMjk4OCkKKysrIExheW91dFRlc3RzL0NoYW5nZUxv Zwkod29ya2luZyBjb3B5KQpAQCAtMSwzICsxLDI5IEBACisyMDA2LTAyLTI1ICBEYXZpZCBLaWx6 ZXIgIDxkZGtpbHplckBraWx6ZXIubmV0PgorCisgICAgICAgIFJldmlld2VkIGJ5IE5PQk9EWSAo T09QUyEpLgorCisgICAgICAgIC0gVGVzdHMgZm9yIGh0dHA6Ly9idWd6aWxsYS5vcGVuZGFyd2lu Lm9yZy9zaG93X2J1Zy5jZ2k/aWQ9NzMyMQorICAgICAgICBSRUdSRVNTSU9OOiBzdHlsZSB0YWcg aW4gYm9keSBjYXVzZXMgdHdvIGhlYWQgZWxlbWVudHMgdG8gYXBwZWFyIGluIHRoZSBET00KKwor ICAgICAgICBUaGUgaGVhZC1jaGVjay5odG1sIGRvY3VtZW50IHVzZXMgYW4gaWZyYW1lIGFzIGEg bGFib3JhdG9yeSBmb3IgbG9hZGluZyBlYWNoCisgICAgICAgIGhlYWQtY2hlY2stTi5odG1sIGZp bGUgYW5kIGNoZWNraW5nIGl0cyBET00sIHdpdGggaW5zcGlyYXRpb24gZnJvbSBJYW4gSGl4aWUn cworICAgICAgICBMaXZlIERPTSBWaWV3ZXI6ICBodHRwOi8vc29mdHdhcmUuaGl4aWUuY2gvdXRp bGl0aWVzL2pzL2xpdmUtZG9tLXZpZXdlci8KKworICAgICAgICBFYWNoIG9mIHRoZSBmb3VyIGhl YWQtY2hlY2stTi5odG1sIGZpbGVzIHRlc3RzIGEgZGlmZmVyZW50IGNvZGUgcGF0aCBmb3IKKyAg ICAgICAgY3JlYXRpbmcgYSBkb2N1bWVudCdzIGhlYWQgZWxlbWVudCBpbiBXZWJDb3JlL2todG1s L2h0bWwvaHRtbHBhcnNlci5jcHAuCisKKyAgICAgICAgKiBlZGl0aW5nL2luc2VydGluZy9lZGl0 aW5nLWVtcHR5LWRpdnMtZXhwZWN0ZWQudHh0OiBDb3JyZWN0ZWQgPGh0bWw+IGNoaWxkCisgICAg ICAgIGluZGV4IGZvciA8Ym9keT4gc2luY2UgdGhlIGRvY3VtZW50IHByZXZpb3VzbHkgaGFkIHR3 byA8aGVhZD4gbm9kZXMuCisgICAgICAgICogZmFzdC9kb20vSFRNTEhlYWRFbGVtZW50OiBBZGRl ZC4KKyAgICAgICAgKiBmYXN0L2RvbS9IVE1MSGVhZEVsZW1lbnQvaGVhZC1jaGVjay1leHBlY3Rl ZC50eHQ6IEFkZGVkLgorICAgICAgICAqIGZhc3QvZG9tL0hUTUxIZWFkRWxlbWVudC9oZWFkLWNo ZWNrLmh0bWw6IEFkZGVkLgorICAgICAgICAqIGZhc3QvZG9tL0hUTUxIZWFkRWxlbWVudC9yZXNv dXJjZXM6IEFkZGVkLgorICAgICAgICAqIGZhc3QvZG9tL0hUTUxIZWFkRWxlbWVudC9yZXNvdXJj ZXMvaGVhZC1jaGVjay0xLmh0bWw6IEFkZGVkLgorICAgICAgICAqIGZhc3QvZG9tL0hUTUxIZWFk RWxlbWVudC9yZXNvdXJjZXMvaGVhZC1jaGVjay0yLmh0bWw6IEFkZGVkLgorICAgICAgICAqIGZh c3QvZG9tL0hUTUxIZWFkRWxlbWVudC9yZXNvdXJjZXMvaGVhZC1jaGVjay0zLmh0bWw6IEFkZGVk LgorICAgICAgICAqIGZhc3QvZG9tL0hUTUxIZWFkRWxlbWVudC9yZXNvdXJjZXMvaGVhZC1jaGVj ay00Lmh0bWw6IEFkZGVkLgorICAgICAgICAqIGZhc3QvZG9tL0hUTUxIZWFkRWxlbWVudC9yZXNv dXJjZXMvaGVhZC1jaGVjay5qczogQWRkZWQuCisKIDIwMDYtMDItMjQgIEFudHRpIEtvaXZpc3Rv ICA8a29pdmlzdG9AaWtpLmZpPgogCiAgICAgICAgIFJldmlld2VkIGJ5IEh5YXR0LgpJbmRleDog TGF5b3V0VGVzdHMvZmFzdC9kb20vSFRNTEhlYWRFbGVtZW50L2hlYWQtY2hlY2suaHRtbAo9PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09Ci0tLSBMYXlvdXRUZXN0cy9mYXN0L2RvbS9IVE1MSGVhZEVsZW1lbnQvaGVhZC1jaGVj ay5odG1sCShyZXZpc2lvbiAwKQorKysgTGF5b3V0VGVzdHMvZmFzdC9kb20vSFRNTEhlYWRFbGVt ZW50L2hlYWQtY2hlY2suaHRtbAkocmV2aXNpb24gMCkKQEAgLTAsMCArMSwxNyBAQAorPCFET0NU WVBFIEhUTUwgUFVCTElDICItLy9JRVRGLy9EVEQgSFRNTC8vRU4iPgorPGh0bWw+Cis8aGVhZD4K KzxsaW5rIHJlbD0ic3R5bGVzaGVldCIgaHJlZj0iLi4vLi4vanMvcmVzb3VyY2VzL2pzLXRlc3Qt c3R5bGUuY3NzIj4KKzxzY3JpcHQgc3JjPSIuLi8uLi9qcy9yZXNvdXJjZXMvanMtdGVzdC1wcmUu anMiPjwvc2NyaXB0PgorPHNjcmlwdCBzcmM9InJlc291cmNlcy9oZWFkLWNoZWNrLmpzIj48L3Nj cmlwdD4KKzwvaGVhZD4KKzxib2R5PgorPHAgaWQ9ImRlc2NyaXB0aW9uIj4KK1RoaXMgdGVzdCBj aGVja3Mgd2hldGhlciBvbmUgKGFuZCBvbmx5IG9uZSkgSEVBRCBlbGVtZW50IGlzIGNyZWF0ZWQg Zm9yIHZhcmlvdXMgSFRNTCBkb2N1bWVudHMuCitTZWUgPGEgaHJlZj0iaHR0cDovL2J1Z3ppbGxh Lm9wZW5kYXJ3aW4ub3JnL3Nob3dfYnVnLmNnaT9pZD03MzIxIj5CdWcgNzMyMTwvYT4KK1JFR1JF U1NJT046IHN0eWxlIHRhZyBpbiBib2R5IGNhdXNlcyB0d28gaGVhZCBlbGVtZW50cyB0byBhcHBl YXIgaW4gdGhlIERPTS4KKzwvcD4KKzxkaXYgaWQ9ImNvbnNvbGUiPjwvZGl2PgorPGlmcmFtZSBp ZD0iaWZyYW1lIiBzcmM9ImRhdGE6dGV4dC9odG1sLCIgb25sb2FkPSJ0ZXN0QW5kTG9hZE5leHQo KSI+PC9pZnJhbWU+Cis8L2JvZHk+Cis8L2h0bWw+CkluZGV4OiBMYXlvdXRUZXN0cy9mYXN0L2Rv bS9IVE1MSGVhZEVsZW1lbnQvaGVhZC1jaGVjay1leHBlY3RlZC50eHQKPT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PQotLS0g TGF5b3V0VGVzdHMvZmFzdC9kb20vSFRNTEhlYWRFbGVtZW50L2hlYWQtY2hlY2stZXhwZWN0ZWQu dHh0CShyZXZpc2lvbiAwKQorKysgTGF5b3V0VGVzdHMvZmFzdC9kb20vSFRNTEhlYWRFbGVtZW50 L2hlYWQtY2hlY2stZXhwZWN0ZWQudHh0CShyZXZpc2lvbiAwKQpAQCAtMCwwICsxLDIwIEBACitU aGlzIHRlc3QgY2hlY2tzIHdoZXRoZXIgb25lIChhbmQgb25seSBvbmUpIEhFQUQgZWxlbWVudCBp cyBjcmVhdGVkIGZvciB2YXJpb3VzIEhUTUwgZG9jdW1lbnRzLiBTZWUgQnVnIDczMjEgUkVHUkVT U0lPTjogc3R5bGUgdGFnIGluIGJvZHkgY2F1c2VzIHR3byBoZWFkIGVsZW1lbnRzIHRvIGFwcGVh ciBpbiB0aGUgRE9NLgorCitUZXN0aW5nOiBoZWFkLWNoZWNrLTEuaHRtbAorUEFTUyBpZnJhbWUu Y29udGVudFdpbmRvdy5kb2N1bWVudC5nZXRFbGVtZW50c0J5VGFnTmFtZSgnaGVhZCcpLmxlbmd0 aCBpcyAxCitQQVNTIGlmcmFtZS5jb250ZW50V2luZG93LmRvY3VtZW50LmZpcnN0Q2hpbGQuZmly c3RDaGlsZC5ub2RlTmFtZS50b0xvd2VyQ2FzZSgpIGlzICdoZWFkJworCitUZXN0aW5nOiBoZWFk LWNoZWNrLTIuaHRtbAorUEFTUyBpZnJhbWUuY29udGVudFdpbmRvdy5kb2N1bWVudC5nZXRFbGVt ZW50c0J5VGFnTmFtZSgnaGVhZCcpLmxlbmd0aCBpcyAxCitQQVNTIGlmcmFtZS5jb250ZW50V2lu ZG93LmRvY3VtZW50LmZpcnN0Q2hpbGQuZmlyc3RDaGlsZC5ub2RlTmFtZS50b0xvd2VyQ2FzZSgp IGlzICdoZWFkJworCitUZXN0aW5nOiBoZWFkLWNoZWNrLTMuaHRtbAorUEFTUyBpZnJhbWUuY29u dGVudFdpbmRvdy5kb2N1bWVudC5nZXRFbGVtZW50c0J5VGFnTmFtZSgnaGVhZCcpLmxlbmd0aCBp cyAxCitQQVNTIGlmcmFtZS5jb250ZW50V2luZG93LmRvY3VtZW50LmZpcnN0Q2hpbGQuZmlyc3RD aGlsZC5ub2RlTmFtZS50b0xvd2VyQ2FzZSgpIGlzICdoZWFkJworCitUZXN0aW5nOiBoZWFkLWNo ZWNrLTQuaHRtbAorUEFTUyBpZnJhbWUuY29udGVudFdpbmRvdy5kb2N1bWVudC5nZXRFbGVtZW50 c0J5VGFnTmFtZSgnaGVhZCcpLmxlbmd0aCBpcyAxCitQQVNTIGlmcmFtZS5jb250ZW50V2luZG93 LmRvY3VtZW50LmZpcnN0Q2hpbGQuZmlyc3RDaGlsZC5ub2RlTmFtZS50b0xvd2VyQ2FzZSgpIGlz ICdoZWFkJworCitURVNUIENPTVBMRVRFCisKSW5kZXg6IExheW91dFRlc3RzL2Zhc3QvZG9tL0hU TUxIZWFkRWxlbWVudC9yZXNvdXJjZXMvaGVhZC1jaGVjay0zLmh0bWwKPT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PQotLS0g TGF5b3V0VGVzdHMvZmFzdC9kb20vSFRNTEhlYWRFbGVtZW50L3Jlc291cmNlcy9oZWFkLWNoZWNr LTMuaHRtbAkocmV2aXNpb24gMCkKKysrIExheW91dFRlc3RzL2Zhc3QvZG9tL0hUTUxIZWFkRWxl bWVudC9yZXNvdXJjZXMvaGVhZC1jaGVjay0zLmh0bWwJKHJldmlzaW9uIDApCkBAIC0wLDAgKzEg QEAKKzxodG1sPjxtZXRhPjwvaHRtbD4KSW5kZXg6IExheW91dFRlc3RzL2Zhc3QvZG9tL0hUTUxI ZWFkRWxlbWVudC9yZXNvdXJjZXMvaGVhZC1jaGVjay00Lmh0bWwKPT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PQotLS0gTGF5 b3V0VGVzdHMvZmFzdC9kb20vSFRNTEhlYWRFbGVtZW50L3Jlc291cmNlcy9oZWFkLWNoZWNrLTQu aHRtbAkocmV2aXNpb24gMCkKKysrIExheW91dFRlc3RzL2Zhc3QvZG9tL0hUTUxIZWFkRWxlbWVu dC9yZXNvdXJjZXMvaGVhZC1jaGVjay00Lmh0bWwJKHJldmlzaW9uIDApCkBAIC0wLDAgKzEgQEAK KzxodG1sPjxvYmplY3Q+PC9vYmplY3Q+PC9odG1sPgpJbmRleDogTGF5b3V0VGVzdHMvZmFzdC9k b20vSFRNTEhlYWRFbGVtZW50L3Jlc291cmNlcy9oZWFkLWNoZWNrLmpzCj09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT0KLS0t IExheW91dFRlc3RzL2Zhc3QvZG9tL0hUTUxIZWFkRWxlbWVudC9yZXNvdXJjZXMvaGVhZC1jaGVj ay5qcwkocmV2aXNpb24gMCkKKysrIExheW91dFRlc3RzL2Zhc3QvZG9tL0hUTUxIZWFkRWxlbWVu dC9yZXNvdXJjZXMvaGVhZC1jaGVjay5qcwkocmV2aXNpb24gMCkKQEAgLTAsMCArMSwzMCBAQAor dmFyIG1heFRlc3RzID0gNDsKK3ZhciBjdXJyZW50VGVzdCA9IDE7Cit2YXIgaWZyYW1lOworCitp ZiAod2luZG93LmxheW91dFRlc3RDb250cm9sbGVyKSB7CisgICAgd2luZG93LmxheW91dFRlc3RD b250cm9sbGVyLndhaXRVbnRpbERvbmUoKTsKK30KKworZnVuY3Rpb24gdGVzdEFuZExvYWROZXh0 KCkgeworICAgIGlmcmFtZSA9IGRvY3VtZW50LmdldEVsZW1lbnRCeUlkKCJpZnJhbWUiKTsKKyAg ICBpZiAoaWZyYW1lLnNyYy5zdWJzdHJpbmcoMCwgNSkgPT0gImRhdGE6IikgeworICAgICAgICBp ZnJhbWUuc3JjID0gInJlc291cmNlcy9oZWFkLWNoZWNrLSIgKyBjdXJyZW50VGVzdCArICIuaHRt bCI7CisgICAgICAgIHJldHVybjsKKyAgICB9CisgICAgZGVidWcoIlRlc3Rpbmc6ICIgKyBpZnJh bWUuc3JjLnN1YnN0cmluZyhpZnJhbWUuc3JjLmxhc3RJbmRleE9mKCIvIikgKyAxKSk7CisgICAg c2hvdWxkQmUoImlmcmFtZS5jb250ZW50V2luZG93LmRvY3VtZW50LmdldEVsZW1lbnRzQnlUYWdO YW1lKCdoZWFkJykubGVuZ3RoIiwgIjEiKTsKKyAgICBzaG91bGRCZSgiaWZyYW1lLmNvbnRlbnRX aW5kb3cuZG9jdW1lbnQuZmlyc3RDaGlsZC5maXJzdENoaWxkLm5vZGVOYW1lLnRvTG93ZXJDYXNl KCkiLCAiJ2hlYWQnIik7CisgICAgZGVidWcoIiIpOworICAgIGlmIChjdXJyZW50VGVzdCA8IG1h eFRlc3RzKSB7CisgICAgICAgIGN1cnJlbnRUZXN0Kys7CisgICAgICAgIGlmcmFtZS5zcmMgPSAi ZGF0YTp0ZXh0L2h0bWwsIjsKKyAgICB9CisgICAgZWxzZSB7CisgICAgICAgIGlmcmFtZS5zdHls ZS5kaXNwbGF5ID0gIm5vbmUiOworICAgICAgICBkZWJ1ZygnPHNwYW4gY2xhc3M9InBhc3MiPlRF U1QgQ09NUExFVEU8L3NwYW4+Jyk7CisgICAgICAgIGlmICh3aW5kb3cubGF5b3V0VGVzdENvbnRy b2xsZXIpIHsKKyAgICAgICAgICAgIHdpbmRvdy5sYXlvdXRUZXN0Q29udHJvbGxlci5ub3RpZnlE b25lKCk7CisgICAgICAgIH0KKyAgICB9Cit9CkluZGV4OiBMYXlvdXRUZXN0cy9mYXN0L2RvbS9I VE1MSGVhZEVsZW1lbnQvcmVzb3VyY2VzL2hlYWQtY2hlY2stMS5odG1sCj09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT0KLS0t IExheW91dFRlc3RzL2Zhc3QvZG9tL0hUTUxIZWFkRWxlbWVudC9yZXNvdXJjZXMvaGVhZC1jaGVj ay0xLmh0bWwJKHJldmlzaW9uIDApCisrKyBMYXlvdXRUZXN0cy9mYXN0L2RvbS9IVE1MSGVhZEVs ZW1lbnQvcmVzb3VyY2VzL2hlYWQtY2hlY2stMS5odG1sCShyZXZpc2lvbiAwKQpAQCAtMCwwICsx IEBACis8aHRtbD48aGVhZD48L2hlYWQ+PGJvZHk+PHN0eWxlPjwvc3R5bGU+PC9ib2R5PjwvaHRt bD4KSW5kZXg6IExheW91dFRlc3RzL2Zhc3QvZG9tL0hUTUxIZWFkRWxlbWVudC9yZXNvdXJjZXMv aGVhZC1jaGVjay0yLmh0bWwKPT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PQotLS0gTGF5b3V0VGVzdHMvZmFzdC9kb20vSFRN TEhlYWRFbGVtZW50L3Jlc291cmNlcy9oZWFkLWNoZWNrLTIuaHRtbAkocmV2aXNpb24gMCkKKysr IExheW91dFRlc3RzL2Zhc3QvZG9tL0hUTUxIZWFkRWxlbWVudC9yZXNvdXJjZXMvaGVhZC1jaGVj ay0yLmh0bWwJKHJldmlzaW9uIDApCkBAIC0wLDAgKzEgQEAKKzxodG1sPjxzdHlsZT48L3N0eWxl PjwvaHRtbD4KSW5kZXg6IFdlYkNvcmUva2h0bWwvaHRtbC9odG1scGFyc2VyLmNwcAo9PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09Ci0tLSBXZWJDb3JlL2todG1sL2h0bWwvaHRtbHBhcnNlci5jcHAJKHJldmlzaW9uIDEyOTg4 KQorKysgV2ViQ29yZS9raHRtbC9odG1sL2h0bWxwYXJzZXIuY3BwCSh3b3JraW5nIGNvcHkpCkBA IC0yMzEsNyArMjMxLDEwIEBAIFBhc3NSZWZQdHI8Tm9kZUltcGw+IEhUTUxQYXJzZXI6OnBhcnNl VG8KIAogICAgICAgICBpZiAoZm9ybSA9PSBuKQogICAgICAgICAgICAgZm9ybSA9IDA7Ci0gICAg ICAgIAorCisgICAgICAgIGlmIChoZWFkID09IG4pCisgICAgICAgICAgICBoZWFkID0gMDsKKwog ICAgICAgICByZXR1cm4gMDsKICAgICB9CiAgICAgcmV0dXJuIG47CkBAIC02MDcsNyArNjEwLDEx IEBAIGJvb2wgSFRNTFBhcnNlcjo6Y29tbWVudENyZWF0ZUVycm9yQ2hlY2sKIAogYm9vbCBIVE1M UGFyc2VyOjpoZWFkQ3JlYXRlRXJyb3JDaGVjayhUb2tlbiogdCwgUmVmUHRyPE5vZGVJbXBsPiYg cmVzdWx0KQogewotICAgIHJldHVybiAoIWhlYWQgfHwgY3VycmVudC0+bG9jYWxOYW1lKCkgPT0g aHRtbFRhZyk7CisgICAgaWYgKCFoZWFkIHx8IGN1cnJlbnQtPmxvY2FsTmFtZSgpID09IGh0bWxU YWcpIHsKKyAgICAgICAgaGVhZCA9IG5ldyBIVE1MSGVhZEVsZW1lbnRJbXBsKGRvY3VtZW50KTsK KyAgICAgICAgcmVzdWx0ID0gaGVhZDsKKyAgICB9CisgICAgcmV0dXJuIGZhbHNlOwogfQogCiBi b29sIEhUTUxQYXJzZXI6OmJvZHlDcmVhdGVFcnJvckNoZWNrKFRva2VuKiB0LCBSZWZQdHI8Tm9k ZUltcGw+JiByZXN1bHQpCkluZGV4OiBXZWJDb3JlL0NoYW5nZUxvZwo9PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09Ci0tLSBX ZWJDb3JlL0NoYW5nZUxvZwkocmV2aXNpb24gMTI5ODgpCisrKyBXZWJDb3JlL0NoYW5nZUxvZwko d29ya2luZyBjb3B5KQpAQCAtMSwzICsxLDE0IEBACisyMDA2LTAyLTI1ICBEYXZpZCBLaWx6ZXIg IDxkZGtpbHplckBraWx6ZXIubmV0PgorCisgICAgICAgIFJldmlld2VkIGJ5IE5PQk9EWSAoT09Q UyEpLgorCisgICAgICAgIC0gRml4IGZvciBodHRwOi8vYnVnemlsbGEub3BlbmRhcndpbi5vcmcv c2hvd19idWcuY2dpP2lkPTczMjEKKyAgICAgICAgUkVHUkVTU0lPTjogc3R5bGUgdGFnIGluIGJv ZHkgY2F1c2VzIHR3byBoZWFkIGVsZW1lbnRzIHRvIGFwcGVhciBpbiB0aGUgRE9NCisKKyAgICAg ICAgKiBraHRtbC9odG1sL2h0bWxwYXJzZXIuY3BwOiBIYW5kbGUgdGhlIGhlYWQgbm9kZSBsaWtl IGZvcm0gYW5kIG1hcCBub2Rlcy4KKyAgICAgICAgKFdlYkNvcmU6OkhUTUxQYXJzZXI6OnBhcnNl VG9rZW4pOgorICAgICAgICAoV2ViQ29yZTo6SFRNTFBhcnNlcjo6aGVhZENyZWF0ZUVycm9yQ2hl Y2spOgorCiAyMDA2LTAyLTI0ICBBbnR0aSBLb2l2aXN0byAgPGtvaXZpc3RvQGlraS5maT4KIAog ICAgICAgICBSZXZpZXdlZCBieSBIeWF0dC4K