116333 2013-05-17 10:59:21 -0700 [webkitpy] Explicitly specify the reference file extensions 2013-05-28 06:34:09 -0700 1 1 1 Unclassified WebKit New Bugs 528+ (Nightly build) Unspecified Unspecified RESOLVED FIXED P2 Normal --- 116312 1 zan zan benjamin commit-queue dpranke glenn rniwa santosh.mahto santosh.ma vivekg oldest_to_newest 890522 0 zan 2013-05-17 10:59:21 -0700 [webkitpy] Explicitly specify the reference file extensions 890524 1 202119 zan 2013-05-17 11:11:50 -0700 Created attachment 202119 Patch 890536 2 202125 zan 2013-05-17 11:24:49 -0700 Created attachment 202125 Patch 890634 3 202125 dpranke 2013-05-17 13:32:31 -0700 Comment on attachment 202125 Patch View in context: https://bugs.webkit.org/attachment.cgi?id=202125&action=review Why do we need or want to make this distinction? I'm not sure I understand ... > Tools/Scripts/webkitpy/port/base.py:617 > + filename_wihout_ext, ext = filesystem.splitext(filename) typo: "without". 890654 4 zan 2013-05-17 13:51:59 -0700 (In reply to comment #3) > (From update of attachment 202125 [details]) > View in context: https://bugs.webkit.org/attachment.cgi?id=202125&action=review > > Why do we need or want to make this distinction? I'm not sure I understand ... > There's undergoing work in bug #116312 that introduces running MHTML-dumping tests. The dumped output is then compared to the *-expected.mht MHTML archive baselines, much like the webarchive baselines under LayoutTests/webarchive. But .mht is also an extension used by various test files and is as such required to be listed in the _supported_file_extensions set (that I propose renaming). This same set is also used when searching for possible reference files in Port.reference_files, meaning that given a MHTML-dumping page.html layout test, its baseline, page-expected.mht, is actually recognized as a reference file, leading to considering the page.html test to be a reftest and comparing the pixel output etc. This is of course not welcomed as the .mht baseline should be treated as a text baseline, just like the .webarchive baselines. 890668 5 dpranke 2013-05-17 14:18:00 -0700 I see. Thanks for explaining. It seems like a mistake to me to have .mht be both a valid extension for a test file and a result that *isn't* a valid reference file. That requires the users to be aware that .html is fine for both tests and references, but .mht isn't, and that perhaps a better solution is to either dump the file as -expected.txt, or to pick a different suffix for the thing that is a baseline. I imagine this tradeoff was considered before; what do y'all think of this? 890926 6 zan 2013-05-18 12:25:57 -0700 CC-ing more people to hopefully get their opinion. I'll have to agree on the premise that a valid test file extension must also be a valid reference file extension, it makes total sense. Given that, I think it would make most sense to have the dumped MHTML archive compared against a *-expected.txt baseline. Another possibility would be to have the *.html test files and *.mht reference files. These would then have the pixel output compared, but I'm not sure how much coverage this would yield in regards to testing the proper format of the dumped MHTML archive. Another possibility that I would rather avoid is to have the test (and reference) files use .mht extension while the text baselines would use the .mhtml extension, but IMO that just brings additional complexity to the extensions support. FWIW, The current MHTML tests consist of .mht test files that are loaded and then dump text and pixel output. I think these can be easily converted into reference tests. Though, as mentioned before, I'm not sure of testing MHTML behavior the other way around (loading HTML test files and using MHTML files as the reference output). 892595 7 202125 benjamin 2013-05-22 17:03:59 -0700 Comment on attachment 202125 Patch Nothing against this, but why not differentiate based on the -expected? For MHTML, maybe we should have -expected.mht and -binary-expected.mht simultaneously for all tests don't you think? 892603 8 dpranke 2013-05-22 17:23:54 -0700 (In reply to comment #7) > (From update of attachment 202125 [details]) > Nothing against this, but why not differentiate based on the -expected? > > For MHTML, maybe we should have -expected.mht and -binary-expected.mht simultaneously for all tests don't you think? MHTML is a text-based format, so the word "binary" confuses me in this context. We need to distinguish between "here is the reference MHTML to compare against" and "here is the dumped MHTML archive from the test" (although I'm not actually sure we need reference MHTML files, we could probably just use regular html files). Which of those two types of files is which of yours? 892619 9 benjamin 2013-05-22 18:03:11 -0700 > MHTML is a text-based format, so the word "binary" confuses me in this context. We need to distinguish between "here is the reference MHTML to compare against" and "here is the dumped MHTML archive from the test" (although I'm not actually sure we need reference MHTML files, we could probably just use regular html files). I don't know much about MHTML but the API has a "useBinaryEncoding" boolean. Which is why I ask about that. > Which of those two types of files is which of yours? We have tests taking MHTML as the input test file (existing tests loading MHTML). What GTK/EFL/Qt need now is new tests that take HTML and generate MHTML. So we will have tests with ".mht" as input in some cases, and "-expected.mht" as output in some cases. 892626 10 dpranke 2013-05-22 18:22:47 -0700 (In reply to comment #9) > > MHTML is a text-based format, so the word "binary" confuses me in this context. We need to distinguish between "here is the reference MHTML to compare against" and "here is the dumped MHTML archive from the test" (although I'm not actually sure we need reference MHTML files, we could probably just use regular html files). > > I don't know much about MHTML but the API has a "useBinaryEncoding" boolean. Which is why I ask about that. > I see. I have no idea what that would do :) > > Which of those two types of files is which of yours? > > We have tests taking MHTML as the input test file (existing tests loading MHTML). > What GTK/EFL/Qt need now is new tests that take HTML and generate MHTML. > > So we will have tests with ".mht" as input in some cases, and "-expected.mht" as output in some cases. Right, so to recap my earlier objection, -expected.mht (a baseline) would conflict as a name with -expected.mht (the reference for a reftest). Given that the regular encoding for MHTML is as text (mime/multipart + base64-encoded subresources), I was suggesting that we just use -expected.txt for the baseline. If we were to use -expected.mht for the baseline we'd have to forbid it from also being used as a reftest (which this patch does), but that seems confusing to me, because then we'd have the same extension being used for two different things. I'm not sure what your opinion is of this dilemma yet. 892631 11 benjamin 2013-05-22 18:43:26 -0700 > Right, so to recap my earlier objection, -expected.mht (a baseline) would conflict as a name with -expected.mht (the reference for a reftest). Given that the regular encoding for MHTML is as text (mime/multipart + base64-encoded subresources), I was suggesting that we just use -expected.txt for the baseline. > > If we were to use -expected.mht for the baseline we'd have to forbid it from also being used as a reftest (which this patch does), but that seems confusing to me, because then we'd have the same extension being used for two different things. > > I'm not sure what your opinion is of this dilemma yet. What I don't like about "-expected.txt" for MHTML is it kills readability of the test. If we go down that path, people cannot just open the file in their browser to see the expectations live. Your objection about ref-tests is valid and I have no solution to propose for that issue. :( 892633 12 dpranke 2013-05-22 19:15:20 -0700 (In reply to comment #11) > > What I don't like about "-expected.txt" for MHTML is it kills readability of the test. If we go down that path, people cannot just open the file in their browser to see the expectations live. > That is a good point. Hrm. 892920 13 santosh.mahto 2013-05-23 09:27:09 -0700 It seems there is a confusion related to useBinaryEncoding First: MHTML is only text base format. and MHTML is independent of any type of encoding format. it just append all the resource in one file with header If the resource is not in Text based format( like images) then :--- By default resources data inside MHTML archive are encoded as base64 .setting useBinaryEncoding = true just override base64 encoding of resource data to binary encoding. But the format of MHTML archive is not changed. So useBinaryEncoding == true in only simple html file(no images) is useless as all content is in text format e.g if useBinaryEncoding == false then Image will be appened as base64 in MHTMl archive if useBinaryEncoding == true then Image will be appened as Binary in MHTML archive In webarchive format we append data in <data> </data> section with particular encoding but in general webarchive foramt is independent of encoding. So there is only one MHTML format(text, as webarchive is XML ) and encoding is just used to encode resource that are not in text format(pritnatble) so useBinaryEncoding(false) has not any dependancy on MHTML format. 892969 14 benjamin 2013-05-23 11:25:55 -0700 Thanks for clarifying the binary format. 892974 15 benjamin 2013-05-23 11:32:18 -0700 Back to the expectations: Unless I missed it, I did not find any ref-test using MHTML. I would go forward with Zan and Santosh work and use -expected.mht for a dump format for seralized MHTML using the public API. One can still write a ref-test by comparing a .mht with a -expected.html. <- Dirk, how much cannot be covered like this? Any opinion before I r-/r+ this? 892996 16 zan 2013-05-23 12:39:03 -0700 I'm OK with either way, really. Also, FWIW, bug #116443 has a patch that converts the current MHTML tests into reftests (meaning that the tested MHTML archive output is compared against the output of the originating on equivalent HTML). 893025 17 dpranke 2013-05-23 13:29:46 -0700 (In reply to comment #15) > Back to the expectations: > > Unless I missed it, I did not find any ref-test using MHTML. > > I would go forward with Zan and Santosh work and use -expected.mht for a dump format for seralized MHTML using the public API. > > One can still write a ref-test by comparing a .mht with a -expected.html. <- Dirk, how much cannot be covered like this? > I can't think of any reason you'd need a ref test to be mhtml, so in theory this patch's approach is fine, I just find it kind of confusing. > Any opinion before I r-/r+ this? another alternative we can consider is to not use -expected as the extension for reference tests. The W3C actually uses -ref instead. I know when we initially started supporting ref tests we wanted to use -expected for consistency with the text and pixel baselines, but that seems less compelling to me these days, and frankly following the W3C's conventions seems more valuable in the long run (I'm probably going to add support for it into Blink, for what that's worth). 893883 18 santosh.ma 2013-05-26 20:45:19 -0700 (In reply to comment #17) > (In reply to comment #15) > > Back to the expectations: > > > > Unless I missed it, I did not find any ref-test using MHTML. > > > > I would go forward with Zan and Santosh work and use -expected.mht for a dump format for seralized MHTML using the public API. > > > > One can still write a ref-test by comparing a .mht with a -expected.html. <- Dirk, how much cannot be covered like this? -expected.mht for a dump format for seralized MHTML using the public API : looks ok to me as same way we do for webarchive > I can't think of any reason you'd need a ref test to be mhtml, so in theory this patch's approach is fine, I just find it kind of confusing. > another alternative we can consider is to not use -expected as the extension for reference tests. The W3C actually uses -ref instead. I know when we initially started supporting ref tests we wanted to use -expected for consistency with the text and pixel baselines, but that seems less compelling to me these days, and frankly following the W3C's conventions seems more valuable in the long run (I'm probably going to add support for it into Blink, for what that's worth). This looks out of scope of this bug although seems good suggestion. Until anyone take this in his/her own hand in seperate bug i think we should follow the baseline comparsion with *.mht it makes sense to follow the WebArchive dump testing 894332 19 202125 zan 2013-05-28 06:34:02 -0700 Comment on attachment 202125 Patch Clearing flags on attachment: 202125 Committed r150800: <http://trac.webkit.org/changeset/150800> 894333 20 zan 2013-05-28 06:34:09 -0700 All reviewed patches have been landed. Closing bug. 202119 2013-05-17 11:11:50 -0700 2013-05-17 11:24:43 -0700 Patch bug-116333-20130517201043.patch text/plain 4008 zan U3VidmVyc2lvbiBSZXZpc2lvbjogMTUwMjYwCmRpZmYgLS1naXQgYS9Ub29scy9DaGFuZ2VMb2cg Yi9Ub29scy9DaGFuZ2VMb2cKaW5kZXggNWZjZTY4ZGU5NTkwY2U2NGI4ZDVjNTU2YjU2YWU0MGFm ZjNkZTAzYy4uMmQwNWUyY2MzMjdhNGM2YmJkYmY0MWJmZGU2MzhiZjA2NWFmNWFkMiAxMDA2NDQK LS0tIGEvVG9vbHMvQ2hhbmdlTG9nCisrKyBiL1Rvb2xzL0NoYW5nZUxvZwpAQCAtMSwzICsxLDIy IEBACisyMDEzLTA1LTE3ICBaYW4gRG9iZXJzZWsgIDx6ZG9iZXJzZWtAaWdhbGlhLmNvbT4KKwor ICAgICAgICBbd2Via2l0cHldIEV4cGxpY2l0bHkgc3BlY2lmeSB0aGUgcmVmZXJlbmNlIGZpbGUg ZXh0ZW5zaW9ucworICAgICAgICBodHRwczovL2J1Z3Mud2Via2l0Lm9yZy9zaG93X2J1Zy5jZ2k/ aWQ9MTE2MzMzCisKKyAgICAgICAgUmV2aWV3ZWQgYnkgTk9CT0RZIChPT1BTISkuCisKKyAgICAg ICAgRXhwbGljaXRseSBzcGVjaWZ5IHRoZSBleHRlbnNpb25zIHRoYXQgYXJlIHZhbGlkIGZvciB0 aGUgcmVmZXJlbmNlIGZpbGVzLiBUaGlzIGhlbHBzIHRvIGRpZmZlcmVudGlhdGUKKyAgICAgICAg YmV0d2VlbiB2YWxpZCByZWZlcmVuY2UgZmlsZXMgcmVxdWlyZWQgYnkgcmVmdGVzdHMgYW5kIHRl eHQgYmFzZWxpbmVzIHRoYXQgaGF2ZSB0aGUgc2FtZSBleHRlbnNpb24gYXMKKyAgICAgICAgcG90 ZW50aWFsIHRlc3QgZmlsZXMgKHNwZWNpZmljYWxseSB0aGUgY3VycmVudCBNSFRNTCB0ZXN0cyBp biB0aGUgLm1odCBmaWxlcyBhbmQgdGhlIGZ1dHVyZSAubWh0IGJhc2VsaW5lcworICAgICAgICB0 aGF0IHdlcmUgdW50aWwgdGhpcyBjaGFuZ2UgcmVjb2duaXplZCBhcyByZWZlcmVuY2UgZmlsZXMp LgorCisgICAgICAgICogU2NyaXB0cy93ZWJraXRweS9wb3J0L2Jhc2UucHk6CisgICAgICAgIChQ b3J0LnJlZmVyZW5jZV9maWxlcyk6IEl0ZXJhdGUgdGhyb3VnaCB0aGUgX3N1cHBvcnRlZF9yZWZl cmVuY2VfZXh0ZW5zaW9ucyBzZXQuCisgICAgICAgIChQb3J0KTogUmVuYW1lIHRoZSBfc3VwcG9y dGVkX2ZpbGVfZXh0ZW5zaW9ucyBzZXQgdG8gX3N1cHBvcnRlZF90ZXN0X2V4dGVuc2lvbnMuIEFk ZCB0aGUKKyAgICAgICAgX3N1cHBvcnRlZF9yZWZlcmVuY2VfZXh0ZW5zaW9ucyBzZXQsIGNvbnRh aW5pbmcgdGhlIGV4dGVuc2lvbnMgYWxsb3dlZCBmb3IgcmVmZXJlbmNlIGZpbGVzLgorICAgICAg ICAoUG9ydC5pc19yZWZlcmVuY2VfaHRtbF9maWxlKTogUmV0dXJuIEZhbHNlIGlmIHRoZSBmaWxl J3MgZXh0ZW5zaW9uIGlzIG5vdCBhIHN1cHBvcnRlZCByZWZlcmVuY2UgZmlsZSBleHRlbnNpb24u CisgICAgICAgIChQb3J0Ll9oYXNfc3VwcG9ydGVkX2V4dGVuc2lvbik6IFVzZSBfc3VwcG9ydGVk X3Rlc3RfZXh0ZW5zaW9ucy4KKwogMjAxMy0wNS0xNCAgU2FtIFdlaW5pZyAgPHNhbUB3ZWJraXQu b3JnPgogCiAgICAgICAgIEFkZCB2YXJpYW50cyBvZiB0aGUgbG9hZGluZyBBUElzIHRoYXQgdGFr ZSB1c2VyIGRhdGEgYW5kIGEgd2F5IGZvciB0aGUgaW5qZWN0ZWQgYnVuZGxlIHRvIGZpbmQgb3V0 IGFib3V0IHRoZW0KZGlmZiAtLWdpdCBhL1Rvb2xzL1NjcmlwdHMvd2Via2l0cHkvcG9ydC9iYXNl LnB5IGIvVG9vbHMvU2NyaXB0cy93ZWJraXRweS9wb3J0L2Jhc2UucHkKaW5kZXggYTkwYzQzNTg0 MmIwMjYwZDBjMDJkMGNmY2Q2OWQ3ZWFjZjgyYzVhMC4uNzhiOThhNGUxYjQ0NDhmZDdhNTQ1M2Y1 MjY2ZWRkMGEyNDVlNzhlOCAxMDA2NDQKLS0tIGEvVG9vbHMvU2NyaXB0cy93ZWJraXRweS9wb3J0 L2Jhc2UucHkKKysrIGIvVG9vbHMvU2NyaXB0cy93ZWJraXRweS9wb3J0L2Jhc2UucHkKQEAgLTU3 MSw3ICs1NzEsNyBAQCBjbGFzcyBQb3J0KG9iamVjdCk6CiAgICAgICAgIGlmIG5vdCByZWZ0ZXN0 X2xpc3Q6CiAgICAgICAgICAgICByZWZ0ZXN0X2xpc3QgPSBbXQogICAgICAgICAgICAgZm9yIGV4 cGVjdGF0aW9uLCBwcmVmaXggaW4gKCgnPT0nLCAnJyksICgnIT0nLCAnLW1pc21hdGNoJykpOgot ICAgICAgICAgICAgICAgIGZvciBleHRlbnRpb24gaW4gUG9ydC5fc3VwcG9ydGVkX2ZpbGVfZXh0 ZW5zaW9uczoKKyAgICAgICAgICAgICAgICBmb3IgZXh0ZW50aW9uIGluIFBvcnQuX3N1cHBvcnRl ZF9yZWZlcmVuY2VfZXh0ZW5zaW9uczoKICAgICAgICAgICAgICAgICAgICAgcGF0aCA9IHNlbGYu ZXhwZWN0ZWRfZmlsZW5hbWUodGVzdF9uYW1lLCBwcmVmaXggKyBleHRlbnRpb24pCiAgICAgICAg ICAgICAgICAgICAgIGlmIHNlbGYuX2ZpbGVzeXN0ZW0uZXhpc3RzKHBhdGgpOgogICAgICAgICAg ICAgICAgICAgICAgICAgcmVmdGVzdF9saXN0LmFwcGVuZCgoZXhwZWN0YXRpb24sIHBhdGgpKQpA QCAtNjA2LDE1ICs2MDYsMTcgQEAgY2xhc3MgUG9ydChvYmplY3QpOgogICAgICAgICByZXR1cm4g W3NlbGYucmVsYXRpdmVfdGVzdF9maWxlbmFtZShmKSBmb3IgZiBpbiBmaWxlc10KIAogICAgICMg V2hlbiBjb2xsZWN0aW5nIHRlc3QgY2FzZXMsIHdlIGluY2x1ZGUgYW55IGZpbGUgd2l0aCB0aGVz ZSBleHRlbnNpb25zLgotICAgIF9zdXBwb3J0ZWRfZmlsZV9leHRlbnNpb25zID0gc2V0KFsnLmh0 bWwnLCAnLnNodG1sJywgJy54bWwnLCAnLnhodG1sJywgJy5wbCcsCi0gICAgICAgICAgICAgICAg ICAgICAgICAgICAgICAgICAgICAgICcuaHRtJywgJy5waHAnLCAnLnN2ZycsICcubWh0J10pCisg ICAgX3N1cHBvcnRlZF90ZXN0X2V4dGVuc2lvbnMgPSBzZXQoWycuaHRtbCcsICcuc2h0bWwnLCAn LnhtbCcsICcueGh0bWwnLCAnLnBsJywgJy5odG0nLCAnLnBocCcsICcuc3ZnJywgJy5taHQnXSkK KyAgICBfc3VwcG9ydGVkX3JlZmVyZW5jZV9leHRlbnNpb25zID0gc2V0KFsnLmh0bWwnLCAnLnht bCcsICcueGh0bWwnLCAnLmh0bScsICcuc3ZnJ10pCiAKICAgICBAc3RhdGljbWV0aG9kCiAgICAg IyBJZiBhbnkgY2hhbmdlcyBhcmUgbWFkZSBoZXJlIGJlIHN1cmUgdG8gdXBkYXRlIHRoZSBpc1Vz ZWRJblJlZnRlc3QgbWV0aG9kIGluIG9sZC1ydW4td2Via2l0LXRlc3RzIGFzIHdlbGwuCiAgICAg ZGVmIGlzX3JlZmVyZW5jZV9odG1sX2ZpbGUoZmlsZXN5c3RlbSwgZGlybmFtZSwgZmlsZW5hbWUp OgogICAgICAgICBpZiBmaWxlbmFtZS5zdGFydHN3aXRoKCdyZWYtJykgb3IgZmlsZW5hbWUuc3Rh cnRzd2l0aCgnbm90cmVmLScpOgogICAgICAgICAgICAgcmV0dXJuIFRydWUKLSAgICAgICAgZmls ZW5hbWVfd2lob3V0X2V4dCwgdW51c2VkID0gZmlsZXN5c3RlbS5zcGxpdGV4dChmaWxlbmFtZSkK KyAgICAgICAgZmlsZW5hbWVfd2lob3V0X2V4dCwgZXh0ID0gZmlsZXN5c3RlbS5zcGxpdGV4dChm aWxlbmFtZSkKKyAgICAgICAgaWYgZXh0IG5vdCBpbiBQb3J0Ll9zdXBwb3J0ZWRfcmVmZXJlbmNl X2V4dGVuc2lvbnM6CisgICAgICAgICAgICByZXR1cm4gRmFsc2UKICAgICAgICAgZm9yIHN1ZmZp eCBpbiBbJy1leHBlY3RlZCcsICctZXhwZWN0ZWQtbWlzbWF0Y2gnLCAnLXJlZicsICctbm90cmVm J106CiAgICAgICAgICAgICBpZiBmaWxlbmFtZV93aWhvdXRfZXh0LmVuZHN3aXRoKHN1ZmZpeCk6 CiAgICAgICAgICAgICAgICAgcmV0dXJuIFRydWUKQEAgLTYyNCw3ICs2MjYsNyBAQCBjbGFzcyBQ b3J0KG9iamVjdCk6CiAgICAgZGVmIF9oYXNfc3VwcG9ydGVkX2V4dGVuc2lvbihmaWxlc3lzdGVt LCBmaWxlbmFtZSk6CiAgICAgICAgICIiIlJldHVybiB0cnVlIGlmIGZpbGVuYW1lIGlzIG9uZSBv ZiB0aGUgZmlsZSBleHRlbnNpb25zIHdlIHdhbnQgdG8gcnVuIGEgdGVzdCBvbi4iIiIKICAgICAg ICAgZXh0ZW5zaW9uID0gZmlsZXN5c3RlbS5zcGxpdGV4dChmaWxlbmFtZSlbMV0KLSAgICAgICAg cmV0dXJuIGV4dGVuc2lvbiBpbiBQb3J0Ll9zdXBwb3J0ZWRfZmlsZV9leHRlbnNpb25zCisgICAg ICAgIHJldHVybiBleHRlbnNpb24gaW4gUG9ydC5fc3VwcG9ydGVkX3Rlc3RfZXh0ZW5zaW9ucwog CiAgICAgQHN0YXRpY21ldGhvZAogICAgIGRlZiBfaXNfdGVzdF9maWxlKGZpbGVzeXN0ZW0sIGRp cm5hbWUsIGZpbGVuYW1lKToK 202125 2013-05-17 11:24:49 -0700 2013-05-28 06:34:02 -0700 Patch bug-116333-20130517202341.patch text/plain 5929 zan U3VidmVyc2lvbiBSZXZpc2lvbjogMTUwMjYwCmRpZmYgLS1naXQgYS9Ub29scy9DaGFuZ2VMb2cg Yi9Ub29scy9DaGFuZ2VMb2cKaW5kZXggNWZjZTY4ZGU5NTkwY2U2NGI4ZDVjNTU2YjU2YWU0MGFm ZjNkZTAzYy4uYzM1ZGUwMDAzYzY2Y2I3ZTU1YmY5YWNiN2U2NzYxZWEyMDE3ZmYwZCAxMDA2NDQK LS0tIGEvVG9vbHMvQ2hhbmdlTG9nCisrKyBiL1Rvb2xzL0NoYW5nZUxvZwpAQCAtMSwzICsxLDI3 IEBACisyMDEzLTA1LTE3ICBaYW4gRG9iZXJzZWsgIDx6ZG9iZXJzZWtAaWdhbGlhLmNvbT4KKwor ICAgICAgICBbd2Via2l0cHldIEV4cGxpY2l0bHkgc3BlY2lmeSB0aGUgcmVmZXJlbmNlIGZpbGUg ZXh0ZW5zaW9ucworICAgICAgICBodHRwczovL2J1Z3Mud2Via2l0Lm9yZy9zaG93X2J1Zy5jZ2k/ aWQ9MTE2MzMzCisKKyAgICAgICAgUmV2aWV3ZWQgYnkgTk9CT0RZIChPT1BTISkuCisKKyAgICAg ICAgRXhwbGljaXRseSBzcGVjaWZ5IHRoZSBleHRlbnNpb25zIHRoYXQgYXJlIHZhbGlkIGZvciB0 aGUgcmVmZXJlbmNlIGZpbGVzLiBUaGlzIGhlbHBzIHRvIGRpZmZlcmVudGlhdGUKKyAgICAgICAg YmV0d2VlbiB2YWxpZCByZWZlcmVuY2UgZmlsZXMgcmVxdWlyZWQgYnkgcmVmdGVzdHMgYW5kIHRl eHQgYmFzZWxpbmVzIHRoYXQgaGF2ZSB0aGUgc2FtZSBleHRlbnNpb24gYXMKKyAgICAgICAgcG90 ZW50aWFsIHRlc3QgZmlsZXMgKHNwZWNpZmljYWxseSB0aGUgY3VycmVudCBNSFRNTCB0ZXN0cyBp biB0aGUgLm1odCBmaWxlcyBhbmQgdGhlIGZ1dHVyZSAubWh0IGJhc2VsaW5lcworICAgICAgICB0 aGF0IHdlcmUgdW50aWwgdGhpcyBjaGFuZ2UgcmVjb2duaXplZCBhcyByZWZlcmVuY2UgZmlsZXMp LgorCisgICAgICAgIENvdmVyZWQgYnkgZXhpc3RpbmcgdGVzdHMgYW5kIHRoZSBuZXcgdGVzdF9p c19yZWZlcmVuY2VfaHRtbF9maWxlIHRlc3QgaW4gd2Via2l0cHkucG9ydC5iYXNlX3VuaXR0ZXN0 LgorCisgICAgICAgICogU2NyaXB0cy93ZWJraXRweS9wb3J0L2Jhc2UucHk6CisgICAgICAgIChQ b3J0LnJlZmVyZW5jZV9maWxlcyk6IEl0ZXJhdGUgdGhyb3VnaCB0aGUgX3N1cHBvcnRlZF9yZWZl cmVuY2VfZXh0ZW5zaW9ucyBzZXQuCisgICAgICAgIChQb3J0KTogUmVuYW1lIHRoZSBfc3VwcG9y dGVkX2ZpbGVfZXh0ZW5zaW9ucyBzZXQgdG8gX3N1cHBvcnRlZF90ZXN0X2V4dGVuc2lvbnMuIEFk ZCB0aGUKKyAgICAgICAgX3N1cHBvcnRlZF9yZWZlcmVuY2VfZXh0ZW5zaW9ucyBzZXQsIGNvbnRh aW5pbmcgdGhlIGV4dGVuc2lvbnMgYWxsb3dlZCBmb3IgcmVmZXJlbmNlIGZpbGVzLgorICAgICAg ICAoUG9ydC5pc19yZWZlcmVuY2VfaHRtbF9maWxlKTogUmV0dXJuIEZhbHNlIGlmIHRoZSBmaWxl J3MgZXh0ZW5zaW9uIGlzIG5vdCBhIHN1cHBvcnRlZCByZWZlcmVuY2UgZmlsZSBleHRlbnNpb24u CisgICAgICAgIChQb3J0Ll9oYXNfc3VwcG9ydGVkX2V4dGVuc2lvbik6IFVzZSBfc3VwcG9ydGVk X3Rlc3RfZXh0ZW5zaW9ucy4KKyAgICAgICAgKiBTY3JpcHRzL3dlYmtpdHB5L3BvcnQvYmFzZV91 bml0dGVzdC5weToKKyAgICAgICAgKFBvcnRUZXN0LnRlc3RfaXNfcmVmZXJlbmNlX2h0bWxfZmls ZSk6IEFkZCB0aGUgdW5pdCB0ZXN0LCB0ZXN0aW5nIHZhcmlvdXMgdGVzdCBjYXNlcyBhcm91bmQg dGhlCisgICAgICAgIFBvcnQuaXNfcmVmZXJlbmNlX2h0bWxfZmlsZSBtZXRob2QuCisKIDIwMTMt MDUtMTQgIFNhbSBXZWluaWcgIDxzYW1Ad2Via2l0Lm9yZz4KIAogICAgICAgICBBZGQgdmFyaWFu dHMgb2YgdGhlIGxvYWRpbmcgQVBJcyB0aGF0IHRha2UgdXNlciBkYXRhIGFuZCBhIHdheSBmb3Ig dGhlIGluamVjdGVkIGJ1bmRsZSB0byBmaW5kIG91dCBhYm91dCB0aGVtCmRpZmYgLS1naXQgYS9U b29scy9TY3JpcHRzL3dlYmtpdHB5L3BvcnQvYmFzZS5weSBiL1Rvb2xzL1NjcmlwdHMvd2Via2l0 cHkvcG9ydC9iYXNlLnB5CmluZGV4IGE5MGM0MzU4NDJiMDI2MGQwYzAyZDBjZmNkNjlkN2VhY2Y4 MmM1YTAuLjc4Yjk4YTRlMWI0NDQ4ZmQ3YTU0NTNmNTI2NmVkZDBhMjQ1ZTc4ZTggMTAwNjQ0Ci0t LSBhL1Rvb2xzL1NjcmlwdHMvd2Via2l0cHkvcG9ydC9iYXNlLnB5CisrKyBiL1Rvb2xzL1Njcmlw dHMvd2Via2l0cHkvcG9ydC9iYXNlLnB5CkBAIC01NzEsNyArNTcxLDcgQEAgY2xhc3MgUG9ydChv YmplY3QpOgogICAgICAgICBpZiBub3QgcmVmdGVzdF9saXN0OgogICAgICAgICAgICAgcmVmdGVz dF9saXN0ID0gW10KICAgICAgICAgICAgIGZvciBleHBlY3RhdGlvbiwgcHJlZml4IGluICgoJz09 JywgJycpLCAoJyE9JywgJy1taXNtYXRjaCcpKToKLSAgICAgICAgICAgICAgICBmb3IgZXh0ZW50 aW9uIGluIFBvcnQuX3N1cHBvcnRlZF9maWxlX2V4dGVuc2lvbnM6CisgICAgICAgICAgICAgICAg Zm9yIGV4dGVudGlvbiBpbiBQb3J0Ll9zdXBwb3J0ZWRfcmVmZXJlbmNlX2V4dGVuc2lvbnM6CiAg ICAgICAgICAgICAgICAgICAgIHBhdGggPSBzZWxmLmV4cGVjdGVkX2ZpbGVuYW1lKHRlc3RfbmFt ZSwgcHJlZml4ICsgZXh0ZW50aW9uKQogICAgICAgICAgICAgICAgICAgICBpZiBzZWxmLl9maWxl c3lzdGVtLmV4aXN0cyhwYXRoKToKICAgICAgICAgICAgICAgICAgICAgICAgIHJlZnRlc3RfbGlz dC5hcHBlbmQoKGV4cGVjdGF0aW9uLCBwYXRoKSkKQEAgLTYwNiwxNSArNjA2LDE3IEBAIGNsYXNz IFBvcnQob2JqZWN0KToKICAgICAgICAgcmV0dXJuIFtzZWxmLnJlbGF0aXZlX3Rlc3RfZmlsZW5h bWUoZikgZm9yIGYgaW4gZmlsZXNdCiAKICAgICAjIFdoZW4gY29sbGVjdGluZyB0ZXN0IGNhc2Vz LCB3ZSBpbmNsdWRlIGFueSBmaWxlIHdpdGggdGhlc2UgZXh0ZW5zaW9ucy4KLSAgICBfc3VwcG9y dGVkX2ZpbGVfZXh0ZW5zaW9ucyA9IHNldChbJy5odG1sJywgJy5zaHRtbCcsICcueG1sJywgJy54 aHRtbCcsICcucGwnLAotICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAnLmh0 bScsICcucGhwJywgJy5zdmcnLCAnLm1odCddKQorICAgIF9zdXBwb3J0ZWRfdGVzdF9leHRlbnNp b25zID0gc2V0KFsnLmh0bWwnLCAnLnNodG1sJywgJy54bWwnLCAnLnhodG1sJywgJy5wbCcsICcu aHRtJywgJy5waHAnLCAnLnN2ZycsICcubWh0J10pCisgICAgX3N1cHBvcnRlZF9yZWZlcmVuY2Vf ZXh0ZW5zaW9ucyA9IHNldChbJy5odG1sJywgJy54bWwnLCAnLnhodG1sJywgJy5odG0nLCAnLnN2 ZyddKQogCiAgICAgQHN0YXRpY21ldGhvZAogICAgICMgSWYgYW55IGNoYW5nZXMgYXJlIG1hZGUg aGVyZSBiZSBzdXJlIHRvIHVwZGF0ZSB0aGUgaXNVc2VkSW5SZWZ0ZXN0IG1ldGhvZCBpbiBvbGQt cnVuLXdlYmtpdC10ZXN0cyBhcyB3ZWxsLgogICAgIGRlZiBpc19yZWZlcmVuY2VfaHRtbF9maWxl KGZpbGVzeXN0ZW0sIGRpcm5hbWUsIGZpbGVuYW1lKToKICAgICAgICAgaWYgZmlsZW5hbWUuc3Rh cnRzd2l0aCgncmVmLScpIG9yIGZpbGVuYW1lLnN0YXJ0c3dpdGgoJ25vdHJlZi0nKToKICAgICAg ICAgICAgIHJldHVybiBUcnVlCi0gICAgICAgIGZpbGVuYW1lX3dpaG91dF9leHQsIHVudXNlZCA9 IGZpbGVzeXN0ZW0uc3BsaXRleHQoZmlsZW5hbWUpCisgICAgICAgIGZpbGVuYW1lX3dpaG91dF9l eHQsIGV4dCA9IGZpbGVzeXN0ZW0uc3BsaXRleHQoZmlsZW5hbWUpCisgICAgICAgIGlmIGV4dCBu b3QgaW4gUG9ydC5fc3VwcG9ydGVkX3JlZmVyZW5jZV9leHRlbnNpb25zOgorICAgICAgICAgICAg cmV0dXJuIEZhbHNlCiAgICAgICAgIGZvciBzdWZmaXggaW4gWyctZXhwZWN0ZWQnLCAnLWV4cGVj dGVkLW1pc21hdGNoJywgJy1yZWYnLCAnLW5vdHJlZiddOgogICAgICAgICAgICAgaWYgZmlsZW5h bWVfd2lob3V0X2V4dC5lbmRzd2l0aChzdWZmaXgpOgogICAgICAgICAgICAgICAgIHJldHVybiBU cnVlCkBAIC02MjQsNyArNjI2LDcgQEAgY2xhc3MgUG9ydChvYmplY3QpOgogICAgIGRlZiBfaGFz X3N1cHBvcnRlZF9leHRlbnNpb24oZmlsZXN5c3RlbSwgZmlsZW5hbWUpOgogICAgICAgICAiIiJS ZXR1cm4gdHJ1ZSBpZiBmaWxlbmFtZSBpcyBvbmUgb2YgdGhlIGZpbGUgZXh0ZW5zaW9ucyB3ZSB3 YW50IHRvIHJ1biBhIHRlc3Qgb24uIiIiCiAgICAgICAgIGV4dGVuc2lvbiA9IGZpbGVzeXN0ZW0u c3BsaXRleHQoZmlsZW5hbWUpWzFdCi0gICAgICAgIHJldHVybiBleHRlbnNpb24gaW4gUG9ydC5f c3VwcG9ydGVkX2ZpbGVfZXh0ZW5zaW9ucworICAgICAgICByZXR1cm4gZXh0ZW5zaW9uIGluIFBv cnQuX3N1cHBvcnRlZF90ZXN0X2V4dGVuc2lvbnMKIAogICAgIEBzdGF0aWNtZXRob2QKICAgICBk ZWYgX2lzX3Rlc3RfZmlsZShmaWxlc3lzdGVtLCBkaXJuYW1lLCBmaWxlbmFtZSk6CmRpZmYgLS1n aXQgYS9Ub29scy9TY3JpcHRzL3dlYmtpdHB5L3BvcnQvYmFzZV91bml0dGVzdC5weSBiL1Rvb2xz L1NjcmlwdHMvd2Via2l0cHkvcG9ydC9iYXNlX3VuaXR0ZXN0LnB5CmluZGV4IDUzYWE5MDI3MWNh MDQwN2E3NzhkYTgxN2FiZjliODVmY2UyMjM4NTkuLjFlODE4NTkxZGM3ZjgwZTUyMTQxMzM4ZWMx NjM5ZDUzZGVlY2JlMjQgMTAwNjQ0Ci0tLSBhL1Rvb2xzL1NjcmlwdHMvd2Via2l0cHkvcG9ydC9i YXNlX3VuaXR0ZXN0LnB5CisrKyBiL1Rvb2xzL1NjcmlwdHMvd2Via2l0cHkvcG9ydC9iYXNlX3Vu aXR0ZXN0LnB5CkBAIC0zNDcsNiArMzQ3LDE4IEBAIGNsYXNzIFBvcnRUZXN0KHVuaXR0ZXN0LlRl c3RDYXNlKToKICAgICAgICAgc2VsZi5hc3NlcnRGYWxzZShQb3J0Ll9pc190ZXN0X2ZpbGUoZmls ZXN5c3RlbSwgJycsICdyZWYtZm9vLmh0bWwnKSkKICAgICAgICAgc2VsZi5hc3NlcnRGYWxzZShQ b3J0Ll9pc190ZXN0X2ZpbGUoZmlsZXN5c3RlbSwgJycsICdub3RyZWYtZm9vLnhocicpKQogCisg ICAgZGVmIHRlc3RfaXNfcmVmZXJlbmNlX2h0bWxfZmlsZShzZWxmKToKKyAgICAgICAgZmlsZXN5 c3RlbSA9IE1vY2tGaWxlU3lzdGVtKCkKKyAgICAgICAgc2VsZi5hc3NlcnRUcnVlKFBvcnQuaXNf cmVmZXJlbmNlX2h0bWxfZmlsZShmaWxlc3lzdGVtLCAnJywgJ2Zvby1leHBlY3RlZC5odG1sJykp CisgICAgICAgIHNlbGYuYXNzZXJ0VHJ1ZShQb3J0LmlzX3JlZmVyZW5jZV9odG1sX2ZpbGUoZmls ZXN5c3RlbSwgJycsICdmb28tZXhwZWN0ZWQtbWlzbWF0Y2gueG1sJykpCisgICAgICAgIHNlbGYu YXNzZXJ0VHJ1ZShQb3J0LmlzX3JlZmVyZW5jZV9odG1sX2ZpbGUoZmlsZXN5c3RlbSwgJycsICdm b28tcmVmLnhodG1sJykpCisgICAgICAgIHNlbGYuYXNzZXJ0VHJ1ZShQb3J0LmlzX3JlZmVyZW5j ZV9odG1sX2ZpbGUoZmlsZXN5c3RlbSwgJycsICdmb28tbm90cmVmLnN2ZycpKQorICAgICAgICBz ZWxmLmFzc2VydEZhbHNlKFBvcnQuaXNfcmVmZXJlbmNlX2h0bWxfZmlsZShmaWxlc3lzdGVtLCAn JywgJ2Zvby5odG1sJykpCisgICAgICAgIHNlbGYuYXNzZXJ0RmFsc2UoUG9ydC5pc19yZWZlcmVu Y2VfaHRtbF9maWxlKGZpbGVzeXN0ZW0sICcnLCAnZm9vLWV4cGVjdGVkLnR4dCcpKQorICAgICAg ICBzZWxmLmFzc2VydEZhbHNlKFBvcnQuaXNfcmVmZXJlbmNlX2h0bWxfZmlsZShmaWxlc3lzdGVt LCAnJywgJ2Zvby1leHBlY3RlZC5zaHRtbCcpKQorICAgICAgICBzZWxmLmFzc2VydEZhbHNlKFBv cnQuaXNfcmVmZXJlbmNlX2h0bWxfZmlsZShmaWxlc3lzdGVtLCAnJywgJ2Zvby1leHBlY3RlZC5w aHAnKSkKKyAgICAgICAgc2VsZi5hc3NlcnRGYWxzZShQb3J0LmlzX3JlZmVyZW5jZV9odG1sX2Zp bGUoZmlsZXN5c3RlbSwgJycsICdmb28tZXhwZWN0ZWQubWh0JykpCisKICAgICBkZWYgdGVzdF9w YXJzZV9yZWZ0ZXN0X2xpc3Qoc2VsZik6CiAgICAgICAgIHBvcnQgPSBzZWxmLm1ha2VfcG9ydCh3 aXRoX3Rlc3RzPVRydWUpCiAgICAgICAgIHBvcnQuaG9zdC5maWxlc3lzdGVtLmZpbGVzWydiYXIv cmVmdGVzdC5saXN0J10gPSAiXG4iLmpvaW4oWyI9PSB0ZXN0Lmh0bWwgdGVzdC1yZWYuaHRtbCIs Cg==