84582 2012-04-23 03:11:49 -0700 webkitpy: changelog crashes when parsing authors containing + or , 2013-04-10 00:36:29 -0700 1 1 1 Unclassified WebKit Tools / Tests 528+ (Nightly build) Unspecified Unspecified RESOLVED DUPLICATE 110356 P2 Normal --- 1 cgarcia webkit-unassigned abarth dpranke eric mrobinson ojan rniwa webkit.review.bot oldest_to_newest 607578 0 cgarcia 2012-04-23 03:11:49 -0700 E...... ====================================================================== ERROR: test_parse_authors (webkitpy.common.checkout.changelog_unittest.ChangeLogTest) ---------------------------------------------------------------------- Traceback (most recent call last): File "webkitpy/common/checkout/changelog_unittest.py", line 433, in test_parse_authors [('James Simonsen', 'simonjam+webkit@google.com')]) File "webkitpy/common/checkout/changelog_unittest.py", line 417, in _assert_parse_authors parsed_authors = [(author['name'], author['email']) for author in self._entry_with_author(author_text).authors()] File "webkitpy/common/checkout/changelog_unittest.py", line 376, in _entry_with_author '''.replace("AUTHOR_TEXT", author_text)) File "webkitpy/common/checkout/changelog.py", line 104, in __init__ self._parse_entry() File "webkitpy/common/checkout/changelog.py", line 165, in _parse_entry self._authors = ChangeLogEntry._parse_author_text(self._author_text) File "webkitpy/common/checkout/changelog.py", line 156, in _parse_author_text return [ChangeLogEntry._parse_author_name_and_email(author) for author in authors] File "webkitpy/common/checkout/changelog.py", line 148, in _parse_author_name_and_email return {'name': match.group("name"), 'email': match.group("email")} AttributeError: 'NoneType' object has no attribute 'group' ---------------------------------------------------------------------- Ran 16 tests in 0.566s FAILED (errors=1) The problem is that _parse_author_name_and_email uses _split_contributor_names which splits the text using a set of separators when tey are found at any position of the text. Author names can contains ',' and email can contain +. 607581 1 138318 cgarcia 2012-04-23 03:16:08 -0700 Created attachment 138318 Patch 607685 2 138318 mrobinson 2012-04-23 07:57:22 -0700 Comment on attachment 138318 Patch I feel like the ChangeLog might be missing the "why." From your patch it looks like (unless I have misread) that you combined the regular expression to split things containing "and" or "&" into multiple parts with the regular expression that parses the name and email combination. I don't understand exactly why that fixes the problem. 607695 3 cgarcia 2012-04-23 08:10:38 -0700 (In reply to comment #2) > (From update of attachment 138318 [details]) > I feel like the ChangeLog might be missing the "why." From your patch it looks like (unless I have misread) that you combined the regular expression to split things containing "and" or "&" into multiple parts with the regular expression that parses the name and email combination. I don't understand exactly why that fixes the problem. As I said in the description _parse_author_name_and_email uses _split_contributor_names which splits the text using a set of separators when they are found *at any position of the text*. That means that a , or + in the inside an author name or email would split the name or email. Merging both, and using findall, it looks for name <email> patterns separated by the set of separators. 607696 4 mrobinson 2012-04-23 08:14:14 -0700 (In reply to comment #3) > As I said in the description _parse_author_name_and_email uses _split_contributor_names which splits the text using a set of separators when they are found *at any position of the text*. That means that a , or + in the inside an author name or email would split the name or email. Merging both, and using findall, it looks for name <email> patterns separated by the set of separators. Sorry, I missed that. Would it be possible to modify the regular expression to simply enforce spaces around '+'? That will allow us to keep the nice abstraction between splitting the name and email addresses and splitting things with "and", "&&", etc. 607700 5 cgarcia 2012-04-23 08:19:08 -0700 (In reply to comment #4) > (In reply to comment #3) > > > As I said in the description _parse_author_name_and_email uses _split_contributor_names which splits the text using a set of separators when they are found *at any position of the text*. That means that a , or + in the inside an author name or email would split the name or email. Merging both, and using findall, it looks for name <email> patterns separated by the set of separators. > > Sorry, I missed that. > > Would it be possible to modify the regular expression to simply enforce spaces around '+'? That will allow us to keep the nice abstraction between splitting the name and email addresses and splitting things with "and", "&&", etc. That would fix the case of +, but list of authors separated by , don't usually have spaces around. Foo Bar <foo@bar.com>, Baz <bazmail@webkit.org> 607708 6 mrobinson 2012-04-23 08:42:44 -0700 (In reply to comment #5) > That would fix the case of +, but list of authors separated by , don't usually have spaces around. > > Foo Bar <foo@bar.com>, Baz <bazmail@webkit.org> It would probably be enough only to make this a requirement for characters that can also exist in names and email addresses. Specifically, the words "and" and "+". 607719 7 cgarcia 2012-04-23 08:53:40 -0700 (In reply to comment #6) > (In reply to comment #5) > > > That would fix the case of +, but list of authors separated by , don't usually have spaces around. > > > > Foo Bar <foo@bar.com>, Baz <bazmail@webkit.org> > > It would probably be enough only to make this a requirement for characters that can also exist in names and email addresses. Specifically, the words "and" and "+". that wouldn't fix the real test case I added to the unit tests where there's a , in the author name. What's wrong with the proposed regexp? It fixes all test cases. 607869 8 mrobinson 2012-04-23 11:25:11 -0700 (In reply to comment #7) > that wouldn't fix the real test case I added to the unit tests where there's a , in the author name. What's wrong with the proposed regexp? It fixes all test cases. The new regular expressions look like it works great, but it has the downside of duplicating a fairly complicated regular expression in two places. Later if someone wants to add to the regex or fix a bug, they may miss one copy of it. Is there a way to reuse it in both places?Perhaps even just storing it as a string that you can use in both places. 608031 9 138318 dpranke 2012-04-23 14:45:16 -0700 Comment on attachment 138318 Patch View in context: https://bugs.webkit.org/attachment.cgi?id=138318&action=review > Tools/Scripts/webkitpy/common/checkout/changelog.py:147 > + return [{'name': name, 'email': email} for name, email in re.findall(r'(?P<name>.+?)\s+<(?P<email>[^>]+)>(?:\s*(?:,(?:\s+and\s+|&)?|(?:^|\s+)and\s+|&&|[/+&])\s*)?', text) if text] Is it possible to implement this correctly without merging the regexps? Frankly, both of these regexps were already verging on unmaintainable. I'd be surprised if this code is particularly performance critical, so if it isn't, I would suggest splitting this (or at least _split_contributor_names() ) up into multiple regexps bracketed by if/elses in order to make the code easier to comprehend. 608051 10 mrobinson 2012-04-23 15:01:05 -0700 (In reply to comment #9) > Is it possible to implement this correctly without merging the regexps? Frankly, both of these regexps were already verging on unmaintainable. This might be tricky. It's unclear what to do with a string like this "Reviewed by Pawel Hadjan, Jr., Martin Robinson, and Carlos Garcia Campos." You could make it work for the attribution line by only splitting names if the comma is prefixed by the end of an email address ala "Pawel Hadjan, Jr. <foo@foo.com>, Martin Robinson <mrobinson@webkit.org>, and Carlos Garcia Campos <cgarcia@igalia.com>." That means we'd need two copies of the regex though, I think. Perhaps it's easiest to just strip ", Jr." and the like from the string before processing. That would make things a lot simpler. 608062 11 dpranke 2012-04-23 15:15:26 -0700 (In reply to comment #10) > (In reply to comment #9) > > > Is it possible to implement this correctly without merging the regexps? Frankly, both of these regexps were already verging on unmaintainable. > > This might be tricky. It's unclear what to do with a string like this "Reviewed by Pawel Hadjan, Jr., Martin Robinson, and Carlos Garcia Campos." Isn't that case is hard to deal with regardless of whether the regexps are merged? 608119 12 mrobinson 2012-04-23 15:46:43 -0700 (In reply to comment #11) > > This might be tricky. It's unclear what to do with a string like this "Reviewed by Pawel Hadjan, Jr., Martin Robinson, and Carlos Garcia Campos." > > Isn't that case is hard to deal with regardless of whether the regexps are merged? It seems so. :/ 608385 13 cgarcia 2012-04-23 23:40:30 -0700 (In reply to comment #8) > (In reply to comment #7) > > > that wouldn't fix the real test case I added to the unit tests where there's a , in the author name. What's wrong with the proposed regexp? It fixes all test cases. > > The new regular expressions look like it works great, but it has the downside of duplicating a fairly complicated regular expression in two places. Later if someone wants to add to the regex or fix a bug, they may miss one copy of it. Is there a way to reuse it in both places?Perhaps even just storing it as a string that you can use in both places. Yes, that makes sense 608388 14 cgarcia 2012-04-23 23:45:27 -0700 (In reply to comment #11) > (In reply to comment #10) > > (In reply to comment #9) > > > > > Is it possible to implement this correctly without merging the regexps? Frankly, both of these regexps were already verging on unmaintainable. > > > > This might be tricky. It's unclear what to do with a string like this "Reviewed by Pawel Hadjan, Jr., Martin Robinson, and Carlos Garcia Campos." > > Isn't that case is hard to deal with regardless of whether the regexps are merged? Yes, this patch doesn't try to fix the reviewer text case, because it's impossible to know whether a , is part of a name or a separator. Fortunately multiple reviewers is not that common, so having a commit with multiple reviewers and one of them having a , in the name is very unlikely. Commits with multiple author is common though. I tried to fix the test cases I added to the unit tests using the existing functions, just playing with the regexp, but didn't manage to fix it until I merged both regexps, because we need the pattern name <email> in the regexp. 702420 15 eric 2012-08-22 15:26:45 -0700 Ryosuke was the last person in this code. :) 702449 16 138318 rniwa 2012-08-22 15:38:50 -0700 Comment on attachment 138318 Patch View in context: https://bugs.webkit.org/attachment.cgi?id=138318&action=review > Tools/Scripts/webkitpy/common/checkout/changelog.py:-154 > - authors = ChangeLogEntry._split_contributor_names(text) I don't think we want to duplicate reg exps here. 872632 17 cgarcia 2013-04-10 00:36:29 -0700 Tony fixed this already in r143504. *** This bug has been marked as a duplicate of bug 110356 *** 138318 2012-04-23 03:16:08 -0700 2012-08-22 15:38:50 -0700 Patch webkitpy-changelog-crash.diff text/plain 4745 cgarcia ZGlmZiAtLWdpdCBhL1Rvb2xzL0NoYW5nZUxvZyBiL1Rvb2xzL0NoYW5nZUxvZwppbmRleCAxNzEx Mzk1Li45MDNiNzcxIDEwMDY0NAotLS0gYS9Ub29scy9DaGFuZ2VMb2cKKysrIGIvVG9vbHMvQ2hh bmdlTG9nCkBAIC0xLDMgKzEsMjEgQEAKKzIwMTItMDQtMjMgIENhcmxvcyBHYXJjaWEgQ2FtcG9z ICA8Y2dhcmNpYUBpZ2FsaWEuY29tPgorCisgICAgICAgIHdlYmtpdHB5OiBjaGFuZ2Vsb2cgY3Jh c2hlcyB3aGVuIHBhcnNpbmcgYXV0aG9ycyBjb250YWluaW5nICsgb3IgLAorICAgICAgICBodHRw czovL2J1Z3Mud2Via2l0Lm9yZy9zaG93X2J1Zy5jZ2k/aWQ9ODQ1ODIKKworICAgICAgICBSZXZp ZXdlZCBieSBOT0JPRFkgKE9PUFMhKS4KKworICAgICAgICAqIFNjcmlwdHMvd2Via2l0cHkvY29t bW9uL2NoZWNrb3V0L2NoYW5nZWxvZy5weToKKyAgICAgICAgKENoYW5nZUxvZ0VudHJ5Ll9zcGxp dF9yZXZpZXdlcl9uYW1lcyk6IFRoaXMgaXMKKyAgICAgICAgX3NwbGl0X2NvbnRyaWJ1dG9yX25h bWVzIHJlbmFtZWQgYW5kIHVzZWQgb25seSB0byBzcGxpdCByZXZpZXdlcgorICAgICAgICBuYW1l cy4KKyAgICAgICAgKENoYW5nZUxvZ0VudHJ5Ll9wYXJzZV9yZXZpZXdlcl90ZXh0KTogVXNlIF9z cGxpdF9yZXZpZXdlcl9uYW1lcy4KKyAgICAgICAgKENoYW5nZUxvZ0VudHJ5Ll9wYXJzZV9hdXRo b3JfdGV4dCk6IE1lcmdlIHRoZSByZWd1bHRhciBleHByZXNzaW9uCisgICAgICAgIHRvIHNwbGl0 IHRoZSBjb250cmlidXRvcnMgbmFtZSB3aXRoIHRoZSBvbmUgdG8gZXh0cmF4Y3QgbmFtZSBhbmQK KyAgICAgICAgZW1haWwgdG8gYXZvaWQgc3BsaXR0aW5nIG5hbWVzIG9yIGVtYWlscy4KKyAgICAg ICAgKiBTY3JpcHRzL3dlYmtpdHB5L2NvbW1vbi9jaGVja291dC9jaGFuZ2Vsb2dfdW5pdHRlc3Qu cHk6CisgICAgICAgICh0ZXN0X3BhcnNlX2F1dGhvcnMpOiBBZGQgbmV3IHRlc3QgY2FzZXMuCisK IDIwMTItMDQtMjIgIFNyaXJhbSBOZWVsYWthbmRhbiAgPHNyaXJhbS5uZWVsYWthbmRhbkBnbWFp bC5jb20+CiAKICAgICAgICAgW0d0a10gQWRkZWQgTU9aX1gxMSBidWlsZCBmbGFnIGZvciBUQVJH RVRfWDExCmRpZmYgLS1naXQgYS9Ub29scy9TY3JpcHRzL3dlYmtpdHB5L2NvbW1vbi9jaGVja291 dC9jaGFuZ2Vsb2cucHkgYi9Ub29scy9TY3JpcHRzL3dlYmtpdHB5L2NvbW1vbi9jaGVja291dC9j aGFuZ2Vsb2cucHkKaW5kZXggM2NmZTI1Zi4uYjdmMDM5ZiAxMDA2NDQKLS0tIGEvVG9vbHMvU2Ny aXB0cy93ZWJraXRweS9jb21tb24vY2hlY2tvdXQvY2hhbmdlbG9nLnB5CisrKyBiL1Rvb2xzL1Nj cmlwdHMvd2Via2l0cHkvY29tbW9uL2NoZWNrb3V0L2NoYW5nZWxvZy5weQpAQCAtMTA0LDYgKzEw NCwxMCBAQCBjbGFzcyBDaGFuZ2VMb2dFbnRyeShvYmplY3QpOgogICAgICAgICBzZWxmLl9wYXJz ZV9lbnRyeSgpCiAKICAgICBAc3RhdGljbWV0aG9kCisgICAgZGVmIF9zcGxpdF9yZXZpZXdlcl9u YW1lcyh0ZXh0KToKKyAgICAgICAgcmV0dXJuIHJlLnNwbGl0KHInXHMqKD86LCg/OlxzK2FuZFxz K3wmKT98KD86XnxccyspYW5kXHMrfCYmfFsvKyZdKVxzKicsIHRleHQpCisKKyAgICBAc3RhdGlj bWV0aG9kCiAgICAgZGVmIF9wYXJzZV9yZXZpZXdlcl90ZXh0KHRleHQpOgogICAgICAgICBtYXRj aCA9IHJlLnNlYXJjaChDaGFuZ2VMb2dFbnRyeS5yZXZpZXdlZF9ieV9yZWdleHAsIHRleHQsIHJl Lk1VTFRJTElORSB8IHJlLklHTk9SRUNBU0UpCiAgICAgICAgIGlmIG5vdCBtYXRjaDoKQEAgLTEy Myw3ICsxMjcsNyBAQCBjbGFzcyBDaGFuZ2VMb2dFbnRyeShvYmplY3QpOgogICAgICAgICBpZiBu b3QgbGVuKHJldmlld2VyX3RleHQpOgogICAgICAgICAgICAgcmV0dXJuIE5vbmUsIE5vbmUKIAot ICAgICAgICByZXZpZXdlcl9saXN0ID0gQ2hhbmdlTG9nRW50cnkuX3NwbGl0X2NvbnRyaWJ1dG9y X25hbWVzKHJldmlld2VyX3RleHQpCisgICAgICAgIHJldmlld2VyX2xpc3QgPSBDaGFuZ2VMb2dF bnRyeS5fc3BsaXRfcmV2aWV3ZXJfbmFtZXMocmV2aWV3ZXJfdGV4dCkKIAogICAgICAgICAjIEdl dCByaWQgb2YgInJldmlld2VycyIgbGlrZSAiZXZlbiB0aG91Z2ggdGhpcyBpcyBqdXN0IGEuLi4i IGluICJSZXZpZXdlZCBieSBTYW0gV2VpbmlnLCBldmVuIHRob3VnaCB0aGlzIGlzIGp1c3QgYS4u LiIKICAgICAgICAgIyBhbmQgIndobyB3cm90ZSB0aGUgb3JpZ2luYWwgY29kZSIgaW4gIk5vYW0g Um9zZW50aGFsLCB3aG8gd3JvdGUgdGhlIG9yaWdpbmFsIGNvZGUiCkBAIC0xMzEsMTAgKzEzNSw2 IEBAIGNsYXNzIENoYW5nZUxvZ0VudHJ5KG9iamVjdCk6CiAKICAgICAgICAgcmV0dXJuIHJldmll d2VyX3RleHQsIHJldmlld2VyX2xpc3QKIAotICAgIEBzdGF0aWNtZXRob2QKLSAgICBkZWYgX3Nw bGl0X2NvbnRyaWJ1dG9yX25hbWVzKHRleHQpOgotICAgICAgICByZXR1cm4gcmUuc3BsaXQocidc cyooPzosKD86XHMrYW5kXHMrfCYpP3woPzpefFxzKylhbmRccyt8JiZ8Wy8rJl0pXHMqJywgdGV4 dCkKLQogICAgIGRlZiBfZnV6el9tYXRjaF9yZXZpZXdlcnMoc2VsZiwgcmV2aWV3ZXJzX3RleHRf bGlzdCk6CiAgICAgICAgIGlmIG5vdCByZXZpZXdlcnNfdGV4dF9saXN0OgogICAgICAgICAgICAg cmV0dXJuIFtdCkBAIC0xNDMsMTcgKzE0Myw4IEBAIGNsYXNzIENoYW5nZUxvZ0VudHJ5KG9iamVj dCk6CiAgICAgICAgIHJldHVybiBbcmV2aWV3ZXJzWzBdIGZvciByZXZpZXdlcnMgaW4gbGlzdF9v Zl9yZXZpZXdlcnMgaWYgbGVuKHJldmlld2VycykgPT0gMV0KIAogICAgIEBzdGF0aWNtZXRob2QK LSAgICBkZWYgX3BhcnNlX2F1dGhvcl9uYW1lX2FuZF9lbWFpbChhdXRob3JfbmFtZV9hbmRfZW1h aWwpOgotICAgICAgICBtYXRjaCA9IHJlLm1hdGNoKHInKD9QPG5hbWU+Lis/KVxzKzwoP1A8ZW1h aWw+W14+XSspPicsIGF1dGhvcl9uYW1lX2FuZF9lbWFpbCkKLSAgICAgICAgcmV0dXJuIHsnbmFt ZSc6IG1hdGNoLmdyb3VwKCJuYW1lIiksICdlbWFpbCc6IG1hdGNoLmdyb3VwKCJlbWFpbCIpfQot Ci0gICAgQHN0YXRpY21ldGhvZAogICAgIGRlZiBfcGFyc2VfYXV0aG9yX3RleHQodGV4dCk6Ci0g ICAgICAgIGlmIG5vdCB0ZXh0OgotICAgICAgICAgICAgcmV0dXJuIFtdCi0gICAgICAgIGF1dGhv cnMgPSBDaGFuZ2VMb2dFbnRyeS5fc3BsaXRfY29udHJpYnV0b3JfbmFtZXModGV4dCkKLSAgICAg ICAgYXNzZXJ0KGF1dGhvcnMgYW5kIGxlbihhdXRob3JzKSA+PSAxKQotICAgICAgICByZXR1cm4g W0NoYW5nZUxvZ0VudHJ5Ll9wYXJzZV9hdXRob3JfbmFtZV9hbmRfZW1haWwoYXV0aG9yKSBmb3Ig YXV0aG9yIGluIGF1dGhvcnNdCisgICAgICAgIHJldHVybiBbeyduYW1lJzogbmFtZSwgJ2VtYWls JzogZW1haWx9IGZvciBuYW1lLCBlbWFpbCBpbiByZS5maW5kYWxsKHInKD9QPG5hbWU+Lis/KVxz KzwoP1A8ZW1haWw+W14+XSspPig/OlxzKig/OiwoPzpccythbmRccyt8Jik/fCg/Ol58XHMrKWFu ZFxzK3wmJnxbLysmXSlccyopPycsIHRleHQpIGlmIHRleHRdCiAKICAgICBkZWYgX3BhcnNlX2Vu dHJ5KHNlbGYpOgogICAgICAgICBtYXRjaCA9IHJlLm1hdGNoKHNlbGYuZGF0ZV9saW5lX3JlZ2V4 cCwgc2VsZi5fY29udGVudHMsIHJlLk1VTFRJTElORSkKZGlmZiAtLWdpdCBhL1Rvb2xzL1Njcmlw dHMvd2Via2l0cHkvY29tbW9uL2NoZWNrb3V0L2NoYW5nZWxvZ191bml0dGVzdC5weSBiL1Rvb2xz L1NjcmlwdHMvd2Via2l0cHkvY29tbW9uL2NoZWNrb3V0L2NoYW5nZWxvZ191bml0dGVzdC5weQpp bmRleCA1MjI2ODQyLi4xMDIwYjc3IDEwMDY0NAotLS0gYS9Ub29scy9TY3JpcHRzL3dlYmtpdHB5 L2NvbW1vbi9jaGVja291dC9jaGFuZ2Vsb2dfdW5pdHRlc3QucHkKKysrIGIvVG9vbHMvU2NyaXB0 cy93ZWJraXRweS9jb21tb24vY2hlY2tvdXQvY2hhbmdlbG9nX3VuaXR0ZXN0LnB5CkBAIC00Mjcs NiArNDI3LDEwIEBAIGNsYXNzIENoYW5nZUxvZ1Rlc3QodW5pdHRlc3QuVGVzdENhc2UpOgogICAg ICAgICAgICAgWygnTmV3IENvbnRyaWJ1dG9yJywgJ25ld0B3ZWJraXQub3JnJyksICgnTm9vYics ICdub29iQHdlYmtpdC5vcmcnKV0pCiAgICAgICAgIHNlbGYuX2Fzc2VydF9wYXJzZV9hdXRob3Jz KCdBZGFtIEJhcnRoICA8YWJhcnRoQHdlYmtpdC5vcmc+ICYmIEJlbmphbWluIFBvdWxhaW4gIDxi cG91bGFpbkBhcHBsZS5jb20+JywKICAgICAgICAgICAgIFsoJ0FkYW0gQmFydGgnLCAnYWJhcnRo QHdlYmtpdC5vcmcnKSwgKCdCZW5qYW1pbiBQb3VsYWluJywgJ2Jwb3VsYWluQGFwcGxlLmNvbScp XSkKKyAgICAgICAgc2VsZi5fYXNzZXJ0X3BhcnNlX2F1dGhvcnMoJ0phbWVzIFNpbW9uc2VuICA8 c2ltb25qYW0rd2Via2l0QGdvb2dsZS5jb20+JywKKyAgICAgICAgICAgIFsoJ0phbWVzIFNpbW9u c2VuJywgJ3NpbW9uamFtK3dlYmtpdEBnb29nbGUuY29tJyldKQorICAgICAgICBzZWxmLl9hc3Nl cnRfcGFyc2VfYXV0aG9ycygnUGF3ZWwgSGFqZGFuLCBKci4gIDxwaGFqZGFuLmpyQGNocm9taXVt Lm9yZz4nLAorICAgICAgICAgICAgWygnUGF3ZWwgSGFqZGFuLCBKci4nLCAncGhhamRhbi5qckBj aHJvbWl1bS5vcmcnKV0pCiAKICAgICBkZWYgX2Fzc2VydF9oYXNfdmFsaWRfcmV2aWV3ZXIoc2Vs ZiwgcmV2aWV3ZXJfbGluZSwgZXhwZWN0ZWQpOgogICAgICAgICBzZWxmLmFzc2VydEVxdWFsKHNl bGYuX2VudHJ5X3dpdGhfcmV2aWV3ZXIocmV2aWV3ZXJfbGluZSkuaGFzX3ZhbGlkX3Jldmlld2Vy KCksIGV4cGVjdGVkKQo=