How to replace one specific line of many files from one instruction file in batches?-CodePudding

We hope to replace the Line 3 of file001 to file600 with corresponding line in the instruction.txt, and is there any possible solution in bash language? I thought maybe I could write a script with sed or awk command, but I couldn't come up with it. Any help would be appreciated. Thanks!

The file001 to file600 contain many cartesian coordinates and are of the same format. E.g. file001 is as follows:

3.02  5.46  8.94
4.55  3.22  4.35
0.00  0.00  0.00 # this is the wrong line and we want to get it replaced
2.34  3.32  5.47
...

The instruction.txt file is:

file001
3.25  1.13  6.10  #this means to replace the 3rd line of "file001" with "3.25  1.13  6.10"
file002
6.01  1.17  -0.32  #this means to replace the 3rd line of "file002" with "6.01  1.17  -0.32"
...

CodePudding user response：

My solution look pretty messed up, but it works :) you can try it, let me know if it works for you.

for i in `cat instruction.txt | awk 'NR % 2 == 1 { printf $0; printf "\n" }'`; do
        outp=`awk '/'"$i"'/{getline; print}' instruction.txt`;
        sed -i '3s/0.00  0.00  0.00/'"$outp"'/' $i;
done

First, I try to filter all the odd lines that contain filename, and assign it to the variable $i:

cat instruction.txt | awk 'NR % 2 == 1 { printf $0; printf "\n" }'

Then get the next line of every $i, which is fix value and assign it to the variable $outp:

outp=`awk '/'"$i"'/{getline; print}' instruction.txt`

And finally, replace the wrong 3rd line with 2 variables I just received.

sed -i '3s/0.00  0.00  0.00/'"$outp"'/' $i

CodePudding user response：

Here would be a solution with GNU awk >= 4.1.0 (I didn't test the inplace part because I don't have the correct version at hand):

awk -i inplace '
    BEGIN {
        while ( (getline filename < ARGV[1] > 0) && (getline data < ARGV[1] > 0) )
            instructions[filename] = data
        close(ARGV[1])
        delete ARGV[1]
    }
    FNR == 1 {
        filename = FILENAME
        sub(/.*\//,"",filename)
    }
    FNR == 3 && filename in instructions {
        print instructions[filename]
        next
    }
    1
' instructions.txt file[0-9][0-9][0-9]

The code first first loads instructions.txt then modifies the 3rd line of the other files if their "basename" is found in the instructions

CodePudding user response：

while read -r line; do 
  if [[ $(sed -E 's/^([^ ]*).*/\1/' <<< "$line") =~ "file" ]]; then 
    file=$line
  elif [[ $(sed -E 's/^([^ ]*).*/\1/' <<< "$line") =~ [0-9.] ]]; then 
    sed -i.bak "3c$line" "$file"
  fi
done < instructions.txt