file maintenance

[anthony dillard_1]

i have a file that the contents looks like this

bl002amcvt187.federated.fds 0x00A0F8651EAF 2

i know how to delete (.federated.fds) using sed, but how do i delete the ending 2 when this value changes, any help would be nice

[Peter Nikitka]

Hi,

I assume, that by
>>
the ending 2 when this value changes
<<
you mean:
with a field delimiter 'space' the last field containing numbers only

Try
sed -e 's/[.]federated[.]fds / /' -e 's/ [0-9][0-9]*$//'

mfG Peter

[James R. Ferguson]

Hi Anthony:

# perl -anle 'print join " ", @F[0..(scalar @F-3)]' file

...will delete the last TWO fields (zero-relative) of any number of fields (as separated by whitespace) in each line of a file.

Regards!

...JRF...

[James R. Ferguson]

Hi (again) Anthony:

Well, a bit more *obvious* is this way:

perl -anle 'print join " ", @F[0..($#F-2)]' file

# ...will delete the last TWO fields (zero-relative) of any number of fields (as separated by whitespace) in each line of a file.

Regards!

...JRF...

[anthony dillard_1]

thanks folks but this is what i have

sed 's/.federated.fds//g' /home/b009021/host2mac1,
what ending do i need to used to delete the ending value in this case a number 0-9 for this entry in this file ie

before using sed
bl004amcvt031.federated.fds 0x00A0F83CD702 2

after sed
bl004amcvt031 0x00A0F83CD702 2
I still need to delete the 2

any ideas

[Ninad_1]

You can use Peter's suggestion with a bit of modification

sed -e 's/[.]federated[.]fds / /' -e 's/ [0-9]$//'

Regards,
Ninad

[anthony dillard_1]

tried, but didnt remove dittelly

[A. Clay Stephenson]

I think there must be some whitespace after the last digit. Try this sed:

sed -e 's/\.federated\.fds //' -e 's/ [0-9] *$//' < infile > outfile

This will match a single digit separated from the preceding data by a space and followed by zero or more spaces at the end of the line.

[Dennis Handly]

>Clay: sed -e 's/\.federated\.fds //' -e 's/ [0-9] *$//' < infile > outfile

This will delete the single digit on ALL lines.
Is this what Anthony wants?

[James R. Ferguson]

Hi:

Along the lines of Clay's thinking, perhaps *whitespace* (spaces and/or tabs) trails the last digit:

# sed -e 's/\.federated\.fds / /' -e 's/[0-9][ ]*$//' file > file.out

Note that the forum doesn't render spaces and/or tab characters nor the number thereof. Within the seemingly empty "[ ]" I typed a *space* and hit the *tab* key.

I'm sorry for not reading your original requirement better. I mistook "...delete the ending 2..." to mean "delete the ending 2 FIELDS...".

Regards!

...JRF...

[Peter Nikitka]

Hi,

additionally or instead of a space/TAB at the end of the line (after the ' 2') it may be, that there is a CR+LF at the end of the line (being a DOS borne file).
If you attach - perhaps only a part - of your input data we can verify this.

Extending my assumption about 'the ending' to:
"a field delimiter may be a TAB or space" I tend to some awk like this:

awk '/.federated.fds/ {sub("[.]federated[.]fds",""); printf $1;
for(i=2;i
mfG Peter

[A. Clay Stephenson]

Okay, to answer Dennis's concern (and frankly I don't know what the output should be), here's an all sed approach that should work (assuming there are zero or more trialing spaces).

sed -e 's/\(.*\)\(\.federated\.fds \)\(.*\)\( [0-9] *$\)/\1 \3/' < infile > outfile

It's using RE grouping \(...\) to grab everything before ".federated.fds " into group 1, and everything before the last space separated digit into group 3 and then substituting the saved group 1, a space, and group 3 for the line. Note that there are 4 RE Groups but we only want 2 of them when the entire RE is matched otherwise the line goes through unchanged.

If you want to filter out possible CR's in the input then the simplest way would probably be to use tr to feed this thing.


tr -d "\r" < infile | sed -e 's/\(.*\)\(\.federated\.fds \)\(.*\)\( [0-9] *$\)/\1 \3/' > outfile

Now tell me that ain't some ugly sed but it should work. I've carefully looked over the logic/syntax but not executed it.

[Dennis Handly]

>Clay: If you want to filter out possible CRs

do2ux(1) is made to do just that.

[Dennis Handly]

Oops:
>Clay: If you want to filter out possible CRs

dos2ux(1) is made to do just that.

[James R. Ferguson]

Hi (again):

At this point it would be most helpful if you posted representative lines from the file. In fact, posting :

# xd -tcx file

...output would settle the questions of whitespace and line endings (whitespace, LF, CRLF).

Are you committed to 'sed'? The syntax becomes clumsy and it's regular expressions are not as powerful nor easily deployed as Perl's as/if data becomes more variable and complex.

Regards!

...JRF...

[anthony dillard_1]

sorry about the non response to all of your help, I just had to take a weekend for myself, I will try all of your suggestions and get back to you, again thanks for all of your input

Anthony

[anthony dillard_1]

0007f70 8 \t 0 x 0 0 0 B 6 C 1 6 8 7 A 3
38093078 30303042 36433136 38374133
0007f80 \t 2 \n m c 2 1 3 a m c v t 1 0 8
9320a6d 63323133 616d6376 74313038
0007f90 \t 0 x 0 0 1 0 2 0 0 0 3 B 9 1 \t
9307830 30313032 30303033 42393109
0007fa0 2 \n m c 2 1 3 a m c v t 1 0 9 \t
320a6d63 32313361 6d637674 31303909
here you go, thanks

[Sandman! · Accepted solution]

Imho you could cat(1) the file to see the invisible characters embedded in the data before deciding on the command:

# cat -vet filename

[A. Clay Stephenson]

Ok, what you thought were spaces are actually tabs; You could use awk -F "\t" to separate the fields but since I went to all the trouble to write that incredibly ugly sed, I think I will stick to it after converting your to spaces via tr but Perl would probably be the best since it has a regular expression metacharacter that can represent whitespace.

tr "\t" " " < infile | sed -e 's/\(.*\)\(\.federated\.fds \)\(.*\)\( [0-9] *$\)/\1 \3/' > outfile

[anthony dillard_1]

found my soultion

sed 's/.federated.fds//g' /home/b009021/host2mac1 | awk '{print $1" "$2}' > /ho
me/b009021/host2mac2

[anthony dillard_1]

thanks for all your help

[anthony dillard_1]

sed 's/.federated.fds//g' /home/b009021/host2mac1 | awk '{print $1" "$2}' > /ho
me/b009021/host2mac2