[SOLVED] Grep regex: bracket

fanoflq · 03-05-2017, 08:15 PM

Code:

#First grep result using regex '[c]ron', quotes not included.
#I got ONE result
 ~ $ ps -ef | grep '[c]ron'
root       877     1  0 Mar04 ?        00:00:00 /usr/sbin/cron -f

#Second grep result using regex 'cron', quotes not included.
#I got TWO results!
 ~ $ ps -ef | grep 'cron'                                                                                        
root       877     1  0 Mar04 ?        00:00:00 /usr/sbin/cron -f
user1    26692 11882  0 18:03 pts/2    00:00:00 grep --color=auto cron

Quote:

From man 7 regex:
A bracket expression is a list of characters enclosed in "[]".
It normally matches any single character from
the list (but see below). If the list begins with '^',
it matches any single character (but see below) not
from the rest of the list. If two characters in the list
are separated by '-', this is shorthand for the full
range of characters between those two (inclusive)
in the collating sequence, for example, "[0-9]" in ASCII
matches any decimal digit. It is illegal(!) for two ranges
to share an endpoint, for example, "a-c-e".
Ranges are very collating-sequence-dependent, and
portable programs should avoid relying on them.

Why do I get different results?

chrism01 · 03-05-2017, 11:57 PM

I admit I looked this one up https://unix.stackexchange.com/quest...rom-ps-results

fanoflq · 03-06-2017, 12:10 AM

Quote:

Originally Posted by chrism01

I admit I looked this one up https://unix.stackexchange.com/quest...rom-ps-results

Thanks.
I still do not get it.
Can you elaborate very sloooowly?

Turbocapitalist · 03-06-2017, 12:27 AM

You get different results because one is a pattern and the other is a string.
Both instances of grep are looking for a regex pattern among strings produces by ps.

Since that string is present in the lines for PID 877 and 26692, they will both be printed.

Code:

root       877     1  0 Mar04 ?        00:00:00 /usr/sbin/cron -f
user1    26692 11882  0 18:03 pts/2    00:00:00 grep --color=auto cron

Using the pattern [c]cron with grep, you'll find the following in your process list (if you could look at the right moment).

Code:

user1    26693 11882  0 18:03 pts/2    00:00:00 grep --color=auto [c]ron

That string doesn't match the pattern used in grep. The pattern only finds cron no square brackets. So 26693 is not shown. If you were to try [abcde]ron or [[:alpha:]]ron, you'd still find 26692 and not 26693.

However, if you were to try [\]]ron or [[:punct:]]ron, you'd find 26693. But note carefully what the colors show as the matched pattern. If you want the whole [c]ron thing you'd need a broader pattern like [c[:punct:]]ron to find it.

tldr; you've made a pattern that does not match itself by using a set

fanoflq · 03-06-2017, 01:20 AM

Quote:

Originally Posted by Turbocapitalist

You get different results because one is a pattern and the other is a string.
Both instances of grep are looking for a regex pattern among strings produces by ps.

Since that string is present in the lines for PID 877 and 26692, they will both be printed.

Code:

root       877     1  0 Mar04 ?        00:00:00 /usr/sbin/cron -f
user1    26692 11882  0 18:03 pts/2    00:00:00 grep --color=auto cron

Using the pattern [c]cron with grep, you'll find the following in your process list (if you could look at the right moment).

Code:

user1    26693 11882  0 18:03 pts/2    00:00:00 grep --color=auto [c]ron

That string doesn't match the pattern used in grep. The pattern only finds cron no square brackets. So 26693 is not shown. If you were to try [abcde]ron or [[:alpha:]]ron, you'd still find 26692 and not 26693.

tldr; you've made a pattern that does not match itself by using a set

OK I now get it.
Here is how I understood what you said.
When we do this:

Code:

~ $ ps -ef | grep "[[:punct:]]ron"

It prints out (as one of its many lines for all the processes) the grep process and its arguments.

Code:

user1     9197  8355  0 22:45 pts/1    00:00:00 grep --color=auto [[:punct:]]ron

In this case the grep command is actually an alias, grep --color=auto.

Likewise when we do this (as you have mentioned):

Code:

~ $ ps -ef | grep "[c]ron"

Two lines of interests (plus many lines for other processes)
were piped into grep, and they are:

Code:

... ... 
root       861     1  0 18:47 ?        00:00:00 /usr/sbin/cron -f
.... .... 
user1     9197  8355  0 22:45 pts/1    00:00:00 grep --color=auto [c]ron
... ...

But the match regex match pattern for grep is "[c]ron".
However regex "[c]ron", quotes excluded, only match string pattern "cron".
And there is no "cron" string in ps output string below:

Code:

user1     9197  8355  0 22:45 pts/1    00:00:00 grep --color=auto [c]ron

Which is why it did not show up as a grep output.

You wrote this:

Code:

However, if you were to try [\]]ron or [[: punct :]]ron, you'd find 26693. But note carefully what the colors show as the matched pattern. If you want the whole [c]ron thing you'd need a broader pattern like [c[: punct :]]ron to find it.

You made a mistake with [\]]ron, the "inside" closing bracket should be reposition like so:

Code:

~ $ ps -ef | grep "[]\]ron"
user1     9756  8355  0 23:12 pts/1    00:00:00 grep --color=auto []\]ron

How do you prevent conversion of ":<NO_SPACE_HERE>p" into smiley icon when we submit this post?
Thank you.

Turbocapitalist · 03-06-2017, 01:27 AM

Quote:

Originally Posted by fanoflq

You made a mistake with [\]]ron, the "inside" closing bracket should be reposition like so:
...

How do you prevent conversion of ":<NO_SPACE_HERE>p" into smiley icon when we submit this post?
Thank you.

Good catch. I should have tested more closely.

There are two ways to avoid the smilies. One is to put an emtpy tag pair inbetween them.

:[b][/b]P

But that's not efficient. Another way is to look in the section under the text box, below the Submit Reply button in the "Additional Options" menu. There is a heading "Miscellaneous Options" with a check box for "Disable smilies in text"

fanoflq · 03-06-2017, 02:03 AM

Thank you Turbocapitalist!