Apache OpenOffice (AOO) Bugzilla – Issue 63849
Regex-support totally screwed up
Last modified: 2013-08-07 15:15:24 UTC
I compared the results of simple variations (leading digits) with advanced filter, Find/Replace, VLOOKUP and DCOUNT. No function gives the same result with the same set of regexes! I can not try each and every variation, but the attached spreadsheet with prepared scenarios may help finding more errors systematically.
Created attachment 35414 [details] regex analysis spreadsheet
Hi, a bit complicate the document. As we've found out, the option for 'Search criteria = and <> must a0pply to whole cells' is set. This option under tools-options-OpenOffice.org Calc-^calculate has also influences to regular expressions. If you untag this option, most of the errors you seem to see will be gone. Eike, this Issue is still open for the problem with the :digit: reg.exp. in the dcount function. Frank
Yes, complicated indeed. "Organically grown" from some numbers and a filter-range. Shame on me. I should have known about the setting you mentioned. I think this has to be mentioned in the online-help about filtering. Each cell-function with regex-support has a hint, leading to option "Enable Regexes in Formulae". This option should mentioned as well >> documentation-issue. With "Regex" and "Whole Cell" you can search for empty cells "<>.+", which should be documented as well. You are right. Now I get consistant results for filter, search and functions as well. Just let me mention another formula-error you might have overlooked: The variant VLOOKUP("[:digit:]";range;1;0) still gives #NA, unlike VLOOKUP("[0-9]";range;1;0) whereas "[:digit:].*" and other variants are consistant with their [0-9] counterparts.
<quote> VLOOKUP("[:digit:]";range;1;0) still gives #NA, [...] </quote> This single aspect relates to new issue 64368 (inconsistant named character-classes) There is also a new evaluation-spreadsheet (less complicated ;)
Wouldn't had been necessary to create a new issue for that, as I took this one for the [:digit:] class problem, but so I'm closing this one here and take the other one.
Closing.