Skip to main content
16 events
when toggle format what by license comment
Jun 12, 2021 at 22:53 comment added Daniel Kaplan @roaima true, admittedly, I didn't want to post every failed attempt I made 1) because I found something that kind of worked but felt clunky, 2) somehow i got it in my head that sed's s/... can't use */+ (I must have misunderstood a SO answer for one of the parts I was researching before I posted this). In my head that misinterpretation was an invariant. At that point, bringing that detail up seemed irrelevant, like bringing up which hand I used for the mouse as I tried to get this to work. TL;DR: When you assume, you make an...
Jun 12, 2021 at 11:52 history rollback cas
Rollback to Revision 3
Jun 12, 2021 at 9:17 history edited Stéphane Chazelas CC BY-SA 4.0
added 2 characters in body
Jun 12, 2021 at 8:34 history edited Stéphane Chazelas CC BY-SA 4.0
added 257 characters in body
Jun 12, 2021 at 7:45 comment added Kusalananda Ok, good, but I was mostly thinking about the PCRE-like expression available even if -P isn't used.
Jun 12, 2021 at 7:43 comment added cas @Kusalananda GNU grep -P uses libpcre and not some GNU implementation of it, so I expect it will behave identically to anything else using PCRE. Whether that's identical to perl's behaviour, I don't know. perl has good unicode support. man pcre says The current implementation of PCRE corresponds approximately with Perl 5.12, including support for UTF-8/16/32 encoded strings and Unicode general category properties. However, UTF-8/16/32 and Unicode support has to be explicitly enabled; it is not the default. The Unicode tables correspond to Unicode release 6.3.0.
Jun 12, 2021 at 7:38 comment added Kusalananda @cas Is \d and \s and the other PCRE-like shortcuts adopted by GNU identical to PCRE when used in GNU regular expressions? Do they behave the same in various locales as Perl REs would, for example?
Jun 12, 2021 at 7:08 history edited cas CC BY-SA 4.0
added 3 characters in body
Jun 12, 2021 at 7:03 comment added Chris Davies @DanielKaplan there was no mention of that anywhere in your question
Jun 12, 2021 at 7:01 comment added cas I'm kind of surprised that GNU sed doesn't do \d yet. It's not like they haven't copied other RE niceties from perl, like \s. and \d has advantages over [0-9] (valid digits aren't limited to just [0-9], depending on language). OTOH, [0-9.+-]+ is arguably easier than (?:\d|[.+-])+.
Jun 12, 2021 at 6:51 vote accept Daniel Kaplan
Jun 12, 2021 at 6:51 comment added Daniel Kaplan okay that explains the only problem I was having the whole time. Thanks!
Jun 12, 2021 at 6:50 comment added cas unfortunately, no. perl does, though :). perl -lne 'print if s/^\s*foo\s*=\s*(\d+).*/$1/' file
Jun 12, 2021 at 6:50 comment added Daniel Kaplan wait so does sed not have \d?
Jun 12, 2021 at 6:50 history edited cas CC BY-SA 4.0
dealing with whitespace
Jun 12, 2021 at 6:43 history answered cas CC BY-SA 4.0