Hi

I want to read a file, which has something like

5678 with error 4.5(ab)
1234 with error 6.7(fb)

and want to read the lines, and if (ab) is there replace it with something like

5678 10E-6 4.5 10E-6

and respectively , if fb is found

123410E-3 6.710E-3

might be trivial...not for me!

thanks for this!

You can use a construct like this:

sed '/keyword/s/^$/newstuff/' filename

(When keyword is found, replace the entire line with newstuff)

For a really good SED tutorial, go here: http://www.grymoire.com/Unix/Sed.html

the thing is that I do not want to replace the whole line, only the "with error" btwn the numbers, with
10E-6 if (ap) is found after the second number
10E=4 if (fb) is found aftet the second number as well

So, actually, that means two different cases - Is there any way to use if statement in sed?

Thanks

SED does not have an "if" statement---see the tutorial I linked.

To modify just part of line, you have to be able to uniquely describe it--eg "[0-9]*[^0-9]" means "any number of digits, followed by a non-digit". This still is not quite what you need--because of the decimal point in the 2nd number. For the second case, you can of course just match the (ab) and replace it with the new text.

example:
replaces the first space with a space + 10E-6, then replaces (ab) with a space + 10E-6

For multiple actions in SED, look at the -e flag.

Since your data is ordered in columns, you might be better off with AWK (not my strong suit). That same site has an AWK tutorial.

Hello Alkass,

The sed tool has a method where you can conditionally substitute text. Below is an example script that performs the substitutions. Two methods are shown, one method where the text that is to be processed is stored in variables, which are tested for a match and substituted, and the other method does not use variables. Note the slash at the end of the lines is used for a line continuation character to make this easier to read. Also make note that the characters \| within the sed statement are used to indicate an OR statement for sed. Hope this makes your day.

#!/bin/sh
#
# file : testsubs.sh
#
# Usage : `$ . testsubs.sh'
#


# the error messages set to variables
emsg1='5678 with error 4.5(ab)'
emsg2='1234 with error 6.7(fb)'

# write the error messages to a txt file named tx
echo -e "$emsg1\n$emsg2" > tx

# the args to be matched
argt=' with error'
arg1='(ab)'
arg2='(fb)'

# the substitution tokens
tk1=' 10E-6'
tk2=' 10E-3'

# # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # #
# display the before and after text
#
echo 'This is the error messages before substitution :'
cat tx

echo -e "\nThis is the error messages after substitution :"
cat tx \
| sed 's/\('"$argt"'\|'"$arg1"'\)/'"$tk1"'/g' \
| sed 's/\('"$argt"'\|'"$arg2"'\)/'"$tk2"'/g'

echo -e "\nThis is another method for substitution :"
cat tx \
| sed 's/\( with error\|(ab)\)/ 10E-6/g' \
| sed 's/\( with error\|(fb)\)/ 10E-3/g'

# EOF #

awk to the rescue

Hi grail,

Never used it before, but now you have me wanting to dive into awk.

I looked it up on : http://www.grymoire.com/Unix/Awk.html

I see how most of it works, except for the way gsub is performing the substituion.

From the example : {gsub(/\(.*/," "exponent)

From what I understand intuitively, the printed output is filtered through the gsub str subst function. The part I am trying to wrap my brain around is the brackets.

So far, since gsub is a function, I assume gsub() is the function text.

So the next part is to match the content within the slashes : /\(.*/
This grabs all text from the ( character through to the end of the line of text. The bracket is dereferenced as \( so it is interpeted correctly as text and not a function statement bracket.

The comma is the argument separator within the gsub function, then the following text that will be used for substitution is : " "exponent

Where the word exponent is a variable that has been set by the pattern matching in the first part of the awk statement. So the result is that a space character followed by the contents of variable named exponent - that has been set by the conditional substitution in the first part of the awk statement - is printed.

Do I have it right? Thanks, and nice solution.

@iPenguin - spot on

So with the field being omitted the gsub works on the entire line or $0.
So if we take the first line:

5678 with error 4.5(ab)

After the gsub it will become:

5678 with error 4.5 10E-6