The issue here is that the two sides of the replacement are each passed in a variable.
$value =~ s/$regex->{search}/$regex->{replace}/
On the right side, we don't evaluate the variable, so anything there is treated as text. This was encountered during the callnumber split development and is why the code there stores the full regex in a single field.
https://blog.james.rcpt.to/2010/10/25/perl-search-and-replace-using-variables/
It seems maybe we can safely do this if we allow for /ee (double encoding) in the regex switch, then it will prevent malicious code, but allow desired code.
In the case of callnumber splitting I think we decided too that we should trust the end user. Someone who already has access to the staff client is hopefully someone trustworthy
I would also like to see substitution work in marc modification templates - or really anywhere we can use regex (such as batch item modification). This would really power-up the regex capabilities. Agree with Andrew that it would be hugely useful not just for date modification. Are you thinking of working on this Nick.. it sounds like you had a hold of the issues it may present? Created attachment 128430 [details] [review] Bug 23873: Enable capture groups in MarcModificationTemplates This patch adds support for regular expression capture groups in MarcModificationTemplates. We escape double quotes in the replacement string, then quote the whole string before applying the `ee` modifier to the final regex application. See https://blog.james.rcpt.to/2010/10/25/perl-search-and-replace-using-variables/ for further details. Test plan 1) Attempt to use capture groups in your regular expression based MarcModificationTemplate and note it does not work. 2) Apply the patch and try again, this time the capture group should yield the expected results. Created attachment 128431 [details] [review] Bug 23873: Enable capture groups in MarcModificationTemplates This patch adds support for regular expression capture groups in MarcModificationTemplates. We escape double quotes in the replacement string, then quote the whole string before applying the `ee` modifier to the final regex application. See https://blog.james.rcpt.to/2010/10/25/perl-search-and-replace-using-variables/ for further details. Test plan 1) Attempt to use capture groups in your regular expression based MarcModificationTemplate and note it does not work. 2) Apply the patch and try again, this time the capture group should yield the expected results. 3) Run t/SimpleMARC.t and confirm the tests still pass. Created attachment 128432 [details] [review] Bug 23873: Unit tests This patch adds a unit test for capture groups support in Koha::SimpleMARC. Test plan 1) Run the test prior to applying the capture groups fix, note it fails. 2) Apply teh capture groups fix patch 3) Run the test again, it should now pass Created attachment 128433 [details] [review] Bug 23873: Enable capture groups in MarcModificationTemplates This patch adds support for regular expression capture groups in MarcModificationTemplates. We escape double quotes in the replacement string, then quote the whole string before applying the `ee` modifier to the final regex application. See https://blog.james.rcpt.to/2010/10/25/perl-search-and-replace-using-variables/ for further details. Test plan 1) Attempt to use capture groups in your regular expression based MarcModificationTemplate and note it does not work. 2) Apply the patch and try again, this time the capture group should yield the expected results. 3) Run t/SimpleMARC.t and confirm the tests still pass. Adding some details to the test plan, to spare others having to think about it :) 1 - have a record with 245a "the best book ever" 2 - create a modification template to copy and replace the 245a to the 245a, using regex 3 - in your regex, match "(.+\b)best(.+)" and replace with "$1worst$2" 4 - perform a batch mod on your record, end up with "$1worst$2" as your literal title 5) Apply the patch and try again, this time the capture group should yield the expected results. 6) Run t/SimpleMARC.t and confirm the tests still pass. Created attachment 128980 [details] [review] Bug 23873: Unit tests This patch adds a unit test for capture groups support in Koha::SimpleMARC. Test plan 1) Run the test prior to applying the capture groups fix, note it fails. 2) Apply teh capture groups fix patch 3) Run the test again, it should now pass Signed-off-by: Andrew Fuerste-Henry <andrew@bywatersolutions.com> Created attachment 128981 [details] [review] Bug 23873: Enable capture groups in MarcModificationTemplates This patch adds support for regular expression capture groups in MarcModificationTemplates. We escape double quotes in the replacement string, then quote the whole string before applying the `ee` modifier to the final regex application. See https://blog.james.rcpt.to/2010/10/25/perl-search-and-replace-using-variables/ for further details. Test plan 1) Attempt to use capture groups in your regular expression based MarcModificationTemplate and note it does not work. 2) Apply the patch and try again, this time the capture group should yield the expected results. 3) Run t/SimpleMARC.t and confirm the tests still pass. Signed-off-by: Andrew Fuerste-Henry <andrew@bywatersolutions.com> Since this works - YAY, could this be applied to batch item modification (such as changing a captured group in the call number and changing it to upper case)? I tried in sandbox after patch applied and did not work, should it have? (In reply to AspenCat Team from comment #11) > Since this works - YAY, could this be applied to batch item modification > (such as changing a captured group in the call number and changing it to > upper case)? I tried in sandbox after patch applied and did not work, should > it have? Best to file a separate bug for this if there isn't one yet. They can be linked using 'see also'. (In reply to AspenCat Team from comment #11) > Since this works - YAY, could this be applied to batch item modification > (such as changing a captured group in the call number and changing it to > upper case)? I tried in sandbox after patch applied and did not work, should > it have? Good news! There'a already a separate bug for this. I just signed it off: https://bugs.koha-community.org/bugzilla3/show_bug.cgi?id=29781 Created attachment 130759 [details] [review] Bug 23873: Unit tests This patch adds a unit test for capture groups support in Koha::SimpleMARC. Test plan 1) Run the test prior to applying the capture groups fix, note it fails. 2) Apply teh capture groups fix patch 3) Run the test again, it should now pass Signed-off-by: Andrew Fuerste-Henry <andrew@bywatersolutions.com> Signed-off-by: Nick Clemens <nick@bywatersolutions.com> Created attachment 130760 [details] [review] Bug 23873: Enable capture groups in MarcModificationTemplates This patch adds support for regular expression capture groups in MarcModificationTemplates. We escape double quotes in the replacement string, then quote the whole string before applying the `ee` modifier to the final regex application. See https://blog.james.rcpt.to/2010/10/25/perl-search-and-replace-using-variables/ for further details. Test plan 1) Attempt to use capture groups in your regular expression based MarcModificationTemplate and note it does not work. 2) Apply the patch and try again, this time the capture group should yield the expected results. 3) Run t/SimpleMARC.t and confirm the tests still pass. Signed-off-by: Andrew Fuerste-Henry <andrew@bywatersolutions.com> Signed-off-by: Nick Clemens <nick@bywatersolutions.com> *** Bug 29014 has been marked as a duplicate of this bug. *** Pushed to master for 22.05, thanks to everybody involved [U+1F984] |
To test: - Find/create a bib record with a date formatted as YYYYMMDD - Create a marc modification template to Copy and Replace your field with this regex: s/(\d{4})(\d{2})(\d{2})/\1\-\2\-\3/ -- this should reformat your date as "YYYY-MM-DD" - Apply your template to your bib - Observe that the match worked, but the substitution failed -- the value "\1\-\2\-\3" is treated literally Apparently this fails due to some aspect of how regex works within perl? It'd be hugely useful if we could get it working.