REGEX - Match all the absolute path with more than one subfolder

Question

i have a series of string representing disks on a file system. I would like to match all the strings with more than one subfolder. Ex:

/dev/sda1 (no match)
/         (no match)
/dev      (no match)

/dev/mapper/usr (match)
/dev/test/local (match)

I've tried something like:

^\/[^\/]+\/[^\/]+\/[^\/]+$

But would like to have something more generic that matches all the strings that have this pattern /.* more than twice. Is there a more elegant way to approach this ?

Unfortunately the solution proposed by @Wiktor Stribiżew, while working on regex 101, does not seems to solve the problem.

To further expand on the topic i have this output from df:

...
cgroup                  0        0         0    - /sys/fs/cgroup/hugetlb
cgroup                  0        0         0    - /sys/fs/cgroup/devices
mqueue                  0        0         0    - /dev/mqueue
/dev/vda9        28056816 12475936  14381828  47% /hostroot
/dev/mapper/usr   1007760   880440     75304  93% /hostroot/usr
/dev/vda6          110576       96    101308   1% /hostroot/usr/share/oem
sysfs                   0        0         0    - /hostroot/sys
/dev/vda1          130798   106402     24396  82% /hostroot/boot
...

and this pattern for awk:

PATTERN='!/loop/ && /^\// && !/^(\/[^\/]+){3,}$/ {printf \
"\"'${NAME}'\":\"%s\","\
"\"mount\":\"%s\","\
"\"total\":%d,"\
"\"used\":%d,"\
"\"free\":%d,"\
"\"percentage\":%.2f\n", $1, $6, $2, $2-$4, $4, ($2-$4)/($2+1)*100}'

when launching the command: df -a | awk "$PATTERN"

i would expect to get this output:

"device":"/dev/vda6","mount":"/hostroot/usr/share/oem","total":110576,"used":9268,"free":101308,"percentage":8.38
"device":"/dev/vda1","mount":"/hostroot/boot","total":130798,"used":106402,"free":24396,"percentage":81.35
"device":"/dev/vda9","mount":"/docker-volumes.d/state","total":28056816,"used":13675036,"free":14381780,"percentage":48.74

But instead the /dev/mapper/usr does not get filtered out.

"device":"/dev/mapper/usr","mount":"/hostroot/usr","total":1007760,"used":932456,"free":75304,"percentage":92.53
"device":"/dev/vda6","mount":"/hostroot/usr/share/oem","total":110576,"used":9268,"free":101308,"percentage":8.38
"device":"/dev/vda1","mount":"/hostroot/boot","total":130798,"used":106402,"free":24396,"percentage":81.35
"device":"/dev/vda9","mount":"/docker-volumes.d/state","total":28056816,"used":13675036,"free":14381780,"percentage":48.74

any guess on why?

thanks, it worked like a charm. If you are willing to put it as an answer i can upvote and mark it as solved. — darkpirate, Sep 01 '22 at 15:26
sorry to bother you more, but i'm using the regex in awk like this: ```awk '!/loop/ && /^\// && !/^(?:\/[^\/]+){3,}$/ {printf...} and receiving: regular expression compile failed (missing operand), do you have any idea why? — darkpirate, Sep 01 '22 at 15:51
No, in POSIX ERE, you cannot use non-capturing groups, remove `?:` — Wiktor Stribiżew, Sep 01 '22 at 15:53
or not, but it is my bad, should've tested on more carefully. if you could look to the edited answer one last time, that would be great. — darkpirate, Sep 01 '22 at 16:32
But `/dev/mapper/usr` matches, why don't you want to extract it? — Wiktor Stribiżew, Sep 01 '22 at 16:39
it was a negation match, i want to exclude all the paths with more than two subpaths, a.k.a i'm only interested in /.../... — darkpirate, Sep 01 '22 at 16:40
Alright, but it is Field 1. Maybe you want `&& $1 !~ /^(\/[^\/]+){3,}$/`? See [this demo](https://ideone.com/lptq0j). — Wiktor Stribiżew, Sep 01 '22 at 16:42
You mean that otherwise i will loose data in field six as well? — darkpirate, Sep 01 '22 at 16:44
i got that and the demo seems to work fine, but when i run it inside my container (alpine, fluent-bit distro) does not work :(, meaning it does not filter out ```/dev/mapper/usr``` — darkpirate, Sep 01 '22 at 16:49
to anyone using alpine based distros: alpine doesn't "have awk installed," it's just that BusyBox that has an awk implementation. For real GNU awk you have to install the gawk package. The answer provided by Wiktor is correct. — darkpirate, Sep 02 '22 at 12:25

Wiktor Stribiżew · Accepted Answer · 2022-09-01T15:54:43.747

You can use

^(\/[^\/]+){3,}$

Details:

^ - start of string
(\/[^\/]+){3,} - three or more sequences of a / char followed with one or more (but as many as possible) chars other than /
$ - string end.

See the regex demo.

NOTE: The pattern above is POSIX ERE and PCRE compliant. When possible, in non-POSIX regex flavors, it is best to use non-capturing groups when you only use a grouping construct to quantify a pattern sequence. So, if you were to use it in JavaScript, I'd recommend using /^(?:\/[^\/]+){3,}$/. Also, if you even need to use it in POSIX BRE, you'd need ^$\/[^\/]\{1,\}$\{3,\}$.

REGEX - Match all the absolute path with more than one subfolder

1 Answers1