Help with sed script

[petperson]

Hi,
I am learning UNIX and writing a sed script. I have a data file that has pipe symbol as the first character and the last character of each line. Every way I try to do this, I remove the entire line. Can someone help me please?

thanks
P.P.

 

[vgersh99]

what is it that you're trying to do?
explain and provide sample input and a desired output.

vlad
+----------------------------+
| #include<disclaimer.h> |
+----------------------------+

 

[petperson]

My data file looks like this:

|123-45-6789|Bob |Jones |10 |20 |30 |
|111-22-3333|Bill |Brown |15 |25 |35 |

I need to remove the first pipe and the last pipe on
each line.

I will then remove the rest of the pipes and replace
them with commas. I thought if I could first remove
the first and last pipe, I could then substitute the
comma's for the remaining pipes. But each time I try
the entire line gets deleted.

thanks

 

[vgersh99]

something to start with

sed -e 's/^|\(.*\)|$/\1/g;s/|/,/g' file.txt

vlad
+----------------------------+
| #include<disclaimer.h> |
+----------------------------+

 

[petperson]

thank you, that worked. One more question though...can you explain what the first part is doing when it deletes the pipe? It's the (.*\) that threw me off..and also the \1.

thanks for your help...
pp

 

[vgersh99]

s/^|\(.*\)|$/\1/g

^ - beginning of the line
| - followed by a pipe
\(.*\) - any number of characters captured - the \(...\) is called capture that can be back-referenced in the 'replacement' part

| - followed by a pipe
$ - end of line

\1 - back reference of the FIRST capture

vlad
+----------------------------+
| #include<disclaimer.h> |
+----------------------------+

 

[petperson]

thank you so much..

 

[PHV]

Another way, without back reference, so faster:
sed 's!^|!!;s!|!,!g;s!,$!!' /path/to/input

Hope This Helps, PH.
Want to get great answers to your Tek-Tips questions? Have a look at FAQ219-2884 or FAQ222-2244

 

[petperson]

thanks for all your help. I have another problem now. My file now looks like this:

id ,last ,first ,wk1,wk2,wk3
pts ,,,100,150,100
111-78-7878,Jones ,Ken ,95,143,79
123-45-6789,Rich ,Donald,99,123,89

I need to remove all the empty spaces. So I tried using this:

sed -e 's/\([,]\)[ ]*/\1/g' test

but it doesn't appear to do anything. What am I missing?

thanks

 

[vgersh99]

assuming there are no embedded spaces:

sed -e 's/[ ][ ]*//g' file

vlad
+----------------------------+
| #include<disclaimer.h> |
+----------------------------+

 

[petperson]

thanks..I was thinking I had to do something between the commas. Probably thinking too much!

 

[PHV]

A safer way:
sed 's! *,!,!g;s! *$!!' test

Hope This Helps, PH.
Want to get great answers to your Tek-Tips questions? Have a look at FAQ219-2884 or FAQ222-2244

 

[vgersh99]

and what about:
111-78-7878,Jones ,Ken ,95,143,79


vlad
+----------------------------+
| #include<disclaimer.h> |
+----------------------------+

 

[petperson]

Now I cannot seem to reverse the order of the names. My file now looks like this:

id,last,first,wk1,wk2,wk3
pts,,,100,150,100
111-78-7878,Jones,Ken,95,143,79
123-45-6789,Rich,Donald,99,123,89

I've tried several different things to see if I have it right but nothing works. I think the SSN in the front confuses me. Is that the first 'field'? What determines the breaks getween fields? I'm confused...

sed 's/\(..*\)\(..*\)/\3 \2/' test

sed 's/\(Jones\)\(Ken\)/\3 \2/' test

 

[PHV]

For field manipulations, you may consider awk:
awk -F',' 'BEGIN{OFS=","}{a=$3;$3=$2;$2=a;print}' test

Hope This Helps, PH.
Want to get great answers to your Tek-Tips questions? Have a look at FAQ219-2884 or FAQ222-2244

 

[vgersh99]

sed -e 's/^\([^,][^,]*\),\([^,][^,]*\),\([^,][^,]*\),\(.*\)/\1,\3,\2,\4/g' test

vlad
+----------------------------+
| #include<disclaimer.h> |
+----------------------------+

 

[PHV]

If awk is not an option:
sed 's!^\([^,]\{1,\}\),\([^,]\{1,\}\),\([^,]\{1,\}\),!\1,\3,\2,!' test

Hope This Helps, PH.
Want to get great answers to your Tek-Tips questions? Have a look at FAQ219-2884 or FAQ222-2244

 

[petperson]

your suggestions are working. I just need to keep practicing to understand all of this. thanks so much.

p.p.

 

[futurelet]

Let awk do it all.
awk -f petperson.awk file.txt

petperson.awk contains

BEGIN { FS="|"; OFS="," }
{ gsub( /^[|] *| *[|]$/, "" )
  gsub( / +[|]/, "|" )
  t=$2; $2=$3; $3=t
  print
}

Before reading file, set Field Separator and Output Field Separator:
[tt]BEGIN { FS="|"; OFS="," }
[/tt]
Remove | and spaces at beginning and end of line:
[tt]{ gsub( /^[|] *| *[|]$/, "" )
[/tt]
Remove trailing spaces in each field:
[tt] gsub( / +[|]/, "|" )
[/tt]
Swap field 2 and field 3; a side-effect is that [tt]$0[/tt] is rebuilt with [tt]OFS[/tt]
between the fields:
[tt] t=$2; $2=$3; $3=t
print
}
[/tt]

 

[futurelet]

There's a shorter way:

BEGIN { FS=" *[|] *"; OFS="," }
{ gsub( /^[|] *| *[|]$/, "" )
  t=$2; $2=$3; $3=t
  print
}

Before reading file, set Field Separator and Output Field Separator ([tt]FS[/tt] will
match [tt]|[/tt] and the spaces on either side of it);
[tt]BEGIN { FS=" *[|] *"; OFS="," }
[/tt]
Remove | and spaces at beginning and end of line:
[tt]{ gsub( /^[|] *| *[|]$/, "" )
[/tt]
Swap field 2 and field 3; a side-effect is that [tt]$0[/tt] is rebuilt with [tt]OFS[/tt]
replacing [tt]FS[/tt] (this eliminates trailing and leading spaces in each field):
[tt] t=$2; $2=$3; $3=t
[/tt]
Print [tt]$0[/tt] (the line):
[tt] print
}
[/tt]