Re: bash help please

From: Alan Connor (zzzzzz_at_xxx.yyy)
Date: 05/30/04 

Date: Sun, 30 May 2004 06:31:06 GMT

On Sun, 30 May 2004 06:01:53 GMT, Dave Brown <dhbrown@hobbes.dhbrown.net> wrote:
>
>
> In article <%wytc.14110$be.430@newsread2.news.pas.earthlink.net>,
> Alan Connor wrote:
>> On Fri, 28 May 2004 02:46:12 GMT, SuperDaemon <Super@DiskAndExecutionMonitor.biz> wrote:
>>>
>>> I know there are some good fellows around here who are well versed in using
>>> "grep", "sed" and "regexpr".
>>>
>>> I have large data file with entries like:
>>> ...
>>
>> ed -s file <<XXXX
>> g/^[0-1][0-9]\/[0-3][0-9]\/[0-9][0-9]/s/$/,/
>> g/.*([0-9][0-9]*%)/s/$/,/
>> g/^[0-1][0-9]\/[0-3][0-9]\/[0-9][0-9]/.,+2j
>> .
>> wq
>> XXXX
>
> And, of course, you can avoid the "leaning toothpicks", /\/\//,
> by using a different pattern delimiter, eg:
> g|^[0-1][0-9]/[0-3][0-9]/[0-9][0-9]|s|$|,|
>
>
> But the perl solution by John Krahn was the neatest!
>

No. It's the stupidest solution. Using perl for something
like this is like using a tactical nuke to kill houseflies.

It works no better than the ed solution and requires an
application and libraries ten thousand times the size
of ed. Can't imagine anything less efficient.

It's like a guy sitting around while hundreds of slaves
build his house and saying to his buddy: "Gee, building
a house is easier than I thought."

(and fails to notice that he has become fat and weak)

The best solution would be to just open the file in ed
and enter:

g/[%/]/s/$/,/
g/\//.,+2j