Very Computer

greetings,

strange behavior of egerp on rh 7.1 and 7.2  have been getting my
attention for some times but this is very strange.

Any ideas?

I have a file that at the end has some garbage meaning
strange non-printable characters. so I do this:

   egrep '^[a-z|A-Z|0-9]' d1  > junk

   junk now has the garbage also.
   egrep on rh 7.1 and 7.2 is at 2.4.2
   i.e. egrep --version

   I do not have access to rh 7.3

   I can NOT reproduce this on these following combinations:

   rh 6.2 whose  egrep is 2.4

   GNU egrep compiled on some AIX 4.3.3 system
   whose version is 2.1

thanks,
farid
unm

What it says to me (and egrep) is pass any string which begins with an
alpha or numeric character; (the "|" are superfluous, and simply say that
you can also have a string which starts with a "|"--you're not
really 'egrepping').  

grep processes strings separated by newline characters,
string by string.

So if the string being processed starts with an alphanumeric, then
regardless of the amount of 'garbage' contained in the string, the entire
string will be passed to the output.

If you're trying to get rid of "garbage characters", you might try

  tr -dc 'character_list_to_preserve' < in_file > out_file

(and don't forget to include punctuation, <space>, and "\n" in your list).

Eg:

  tr -dc 'A-Za-z0-9 ,.;:\t\n' <in_file.txt >output.txt

--
Dave Brown  Austin, TX

Quote:

>greetings,

>strange behavior of egerp on rh 7.1 and 7.2  have been getting my
>attention for some times but this is very strange.

>Any ideas?

>I have a file that at the end has some garbage meaning
>strange non-printable characters. so I do this:

>   egrep '^[a-z|A-Z|0-9]' d1  > junk

>   junk now has the garbage also.
>   egrep on rh 7.1 and 7.2 is at 2.4.2
>   i.e. egrep --version

I think the expression you want is
egrep -v '[^a-zA-Z0-9]' d1 >junk

This will filter out any line containing characters other than a-zA-Z0-9.
If you want to filter out garbage characters without losing entire lines,
use tr.