[SOLVED] Replacing text with content from another file in an order

Sayan Acharjee · 06-14-2016, 02:40 AM

I am not that good and scripting and got stuck with a requirement while writing a script, I have been trying to replace contents of same text in a file that comes multiple times with the contents of another file which comes in each line. Below is an example
file1

Code:

REGEX Pattern1 FOLLOWS BaseEvent
^(?i)(.*(FAIL).*)$
ComponentID PRINTF("%s:Pattern1", log_name)
IdentifierKey $1
msg $1
ComponentArg $2
Severity 5
END

REGEX Pattern1 FOLLOWS BaseEvent
^(?i)(.*(FAIL).*)$
ComponentID PRINTF("%s:Pattern1", log_name)
IdentifierKey $1
msg $1
ComponentArg $2
Severity 5
END

REGEX Pattern1 FOLLOWS BaseEvent
^(?i)(.*(FAIL).*)$
ComponentID PRINTF("%s:Pattern1", log_name)
IdentifierKey $1
msg $1
ComponentArg $2
Severity 5
END

REGEX Pattern1 FOLLOWS BaseEvent
^(?i)(.*(FAIL).*)$
ComponentID PRINTF("%s:Pattern1", log_name)
IdentifierKey $1
msg $1
ComponentArg $2
Severity 5
END

file2

Code:

SUCCESS
DEFAULT
WRONG
CORRECT

I want to replace "FAIL" from the file1 with entries from file 2 so that first "FAIL" gets replaced with first entry from file 2 which is "SUCCESS" and this continues. The end result should be like following,

Code:

REGEX Pattern1 FOLLOWS BaseEvent
^(?i)(.*(SUCCESS).*)$
ComponentID PRINTF("%s:Pattern1", log_name)
IdentifierKey $1
msg $1
ComponentArg $2
Severity 5
END

REGEX Pattern1 FOLLOWS BaseEvent
^(?i)(.*(DEFAULT).*)$
ComponentID PRINTF("%s:Pattern1", log_name)
IdentifierKey $1
msg $1
ComponentArg $2
END

REGEX Pattern1 FOLLOWS BaseEvent
^(?i)(.*(WRONG).*)$
ComponentID PRINTF("%s:Pattern1", log_name)
IdentifierKey $1
msg $1
ComponentArg $2
END

REGEX Pattern1 FOLLOWS BaseEvent
^(?i)(.*(CORRECT).*)$
ComponentID PRINTF("%s:Pattern1", log_name)
IdentifierKey $1
msg $1
ComponentArg $2
END

I tried using a for loop in below format, but it doesn't work. How do increment to replace first occurrence with the first entry from file2 and second occurrence with second entry from file2, and so on?
Could you please help?

for a in `cat file2`; do sed 's/FAIL/$a/1' file1

HMW · 06-14-2016, 04:33 AM

Hi!

I used a tiny script for this, where 'fail.sh' is the name of the script, and source.txt is your infile:

Code:

$ ./fail.sh source.txt
REGEX Pattern1 FOLLOWS BaseEvent
^(?i)(.*(SUCCESS,).*)$
ComponentID PRINTF("%s:Pattern1", log_name)
IdentifierKey $1
msg $1
ComponentArg $2
Severity 5
END

REGEX Pattern1 FOLLOWS BaseEvent
^(?i)(.*(DEFAULT,).*)$
ComponentID PRINTF("%s:Pattern1", log_name)
IdentifierKey $1
msg $1
ComponentArg $2
Severity 5
END

REGEX Pattern1 FOLLOWS BaseEvent
^(?i)(.*(WRONG,).*)$
ComponentID PRINTF("%s:Pattern1", log_name)
IdentifierKey $1
msg $1
ComponentArg $2
Severity 5
END

REGEX Pattern1 FOLLOWS BaseEvent
^(?i)(.*(CORRECT).*)$
ComponentID PRINTF("%s:Pattern1", log_name)
IdentifierKey $1
msg $1
ComponentArg $2
Severity 5

What I used in the script:
- a while loop to read the file
- an array to hold the replacements (if you need to read these from a file, then do so)
- sed to do the actual replacing
- a counter
- echo to print the information to STDOUT

Being a Red Hat Certified Engineer, you will figure this out easily.

Best regards,
HMW

grail · 06-14-2016, 11:54 AM

I have 3 shell scripts, one is pure bash, second uses sed and third uses awk ... all seems fairly straight forward.

Your original solution is quite close ... may I suggest looking at the curly braces and /q options on the following page: http://www.grymoire.com/Unix/Sed.html

Sayan Acharjee · 06-17-2016, 12:09 AM

As grail pointed out, better to tell the solution even if it wasn't needed later as it was marked as solved.

Code:

declare -a arr;

#to store the texts to be replaced in the file in the array
arr=($(cat file1 |grep "(*)"|grep -v Comp|awk -F"(" {'print $4'}|awk -F")" {'print $1'}));



count=0;
while read -r i
do
sed -i "s/${arr[$count]}/$i/g" file1
(( count ++ ))
done < file2

grail · 06-17-2016, 05:23 AM

Interesting

Your initial example gave the impression that we knew the word FAIL and hence only had to replace it, but your solution points to not knowing what might be in the required spot ... which
is better as I initially was wondering how you would ever change FAIL for SUCCESS

Here is a bit of a tidy up for you:

Code:

arr=($(awk -F"[()]+" '/REGEX/{getline;print $4}' file1))

count=0

while read -r new
do
  sed -i "s/${arr[count++]}/$new/g" file1
done<file2

I would point out that I have tidied the code but it will not produce the results you were wanting .. this will update the original file (I presumed that "test" was a typo as there is no file with that name)
so that all items are changed to the first entry from second file and then no other changes get made as all the original names have now been changed.

Here is output after above has been run:

Code:

$ cat file1
REGEX Pattern1 FOLLOWS BaseEvent
^(?i)(.*(SUCCESS).*)$
ComponentID PRINTF("%s:Pattern1", log_name)
IdentifierKey $1
msg $1
ComponentArg $2
Severity 5
END

REGEX Pattern1 FOLLOWS BaseEvent
^(?i)(.*(SUCCESS).*)$
ComponentID PRINTF("%s:Pattern1", log_name)
IdentifierKey $1
msg $1
ComponentArg $2
Severity 5
END

REGEX Pattern1 FOLLOWS BaseEvent
^(?i)(.*(SUCCESS).*)$
ComponentID PRINTF("%s:Pattern1", log_name)
IdentifierKey $1
msg $1
ComponentArg $2
Severity 5
END

REGEX Pattern1 FOLLOWS BaseEvent
^(?i)(.*(SUCCESS).*)$
ComponentID PRINTF("%s:Pattern1", log_name)
IdentifierKey $1
msg $1
ComponentArg $2
Severity 5
END

So from the example, all FAIL words were changed to SUCCESS, but when loop performs further iterations the word FAIL is nowhere to be found so no future changes are made

Sayan Acharjee · 06-17-2016, 05:50 AM

There would not be multiple occurrence for the same string, so this works perfectly fine. Yes, that was a typo with the file name

grail · 06-17-2016, 02:26 PM

Maybe in the future you would like to present examples that actually reflect what you have and what you need then as it appears nothing about the original question was actually correct.