Prev: can a DHCP request recieved on eth0 also assign an IP to eth1?
Next: Simple line-printing program
From: Todd on 11 May 2010 13:13
On 05/10/2010 01:33 PM, A Watcher wrote:

> I hate awk. The commmands are so cryptic that whomever tries to read the
> code later has to work at it to understand it.

I hear you there. I keep a list of examples. That is
the only way I remember.
From: pk on 26 May 2010 04:43
Dave U. Random wrote:

> It can be done with 'sed' and a somewhat creative use of its hold space
> for matched patterns' score keeping. It does scale well. It will
> definitely be slower than the equivalent grep solution, but for small
> data sets it shouldn't matter much.
>
> The following example finds all positive integers up to 50 that
> contain the digits 2 and 4.
>
> seq 50 |
> sed -n '
> # find simultaneous occurrence of many search patterns on the same input =
> line
> #
> # store a N-character string in hold space;
> # the replacement string in the following s/// command must be N
> # chars long, where N is the number of patterns to match
> x;s/.*/12/;x
> ####################
> # add all your match patterns here
> # /pattern1/{x;s/.//;x;}
> # /pattern2/{x;s/.//;x;}
> # ...
> # /patternN/{x;s/.//;x;}
> /2/{x;s/.//;x;}
> /4/{x;s/.//;x;}
> ###################
> # ^------------- delete a char from hold space for each matched pattern
> {
> # make sure that any successful pattern matches above don`t
> # overshadow the results of the following s/// test; ie. clear test flag
> T
> # is the hold space empty, ie. did all patterns match?=20
> x;s/^$//;x
> T
> p # yes; print input line
> }
> '

That seems a bit overkill:

sed '/pattern1/{h;b end;}
sed '/pattern2/{h;b end;}
sed '/pattern3/{h;b end;}
....
:end
${x;s/\n//p;}
d'
From: pk on 26 May 2010 04:47
pk wrote:

>> seq 50 |
>> sed -n '
>> # find simultaneous occurrence of many search patterns on the same input
>> # =
>> line
>> #
>> # store a N-character string in hold space;
>> # the replacement string in the following s/// command must be N
>> # chars long, where N is the number of patterns to match
>> x;s/.*/12/;x
>> ####################
>> # add all your match patterns here
>> # /pattern1/{x;s/.//;x;}
>> # /pattern2/{x;s/.//;x;}
>> # ...
>> # /patternN/{x;s/.//;x;}
>> /2/{x;s/.//;x;}
>> /4/{x;s/.//;x;}
>> ###################
>> # ^------------- delete a char from hold space for each matched pattern
>> {
>> # make sure that any successful pattern matches above don`t
>> # overshadow the results of the following s/// test; ie. clear test flag
>> T
>> # is the hold space empty, ie. did all patterns match?=20
>> x;s/^$//;x
>> T
>> p # yes; print input line
>> }
>> '
>
> That seems a bit overkill:
>
> sed '/pattern1/{h;b end;}
> sed '/pattern2/{h;b end;}
> sed '/pattern3/{h;b end;}
> ...
> :end
> ${x;s/\n//p;}
> d'

Sorry, I misunderstood what you meant. Your code and mine do different
things.

 | 
Pages: 1
Prev: can a DHCP request recieved on eth0 also assign an IP to eth1?
Next: Simple line-printing program