How to find and extract all words appearing between brackets?

Question

How to put into array all words appearing between brackets in text file and replace it with random one from that array?

cat math.txt
First: {736|172|201|109} {+|-|*|%|/} {21|62|9|1|0}
Second: John had {22|12|15} apples and lost {2|4|3}

I need output like:

First: 172-9
Second: John had 15 apples and lost 4

Answer 1

This is trivial in awk:

$ cat tst.awk
BEGIN{ srand() }
{
    for (i=1; i<=NF; i++) {
        if ( match($i,/{[^}]+}/) ) {
            n = split(substr($i,RSTART+1,RLENGTH-2),arr,/\|/)
            idx = int(rand() * (n-1)) + 1
            $i  = arr[idx]
        }
        printf "%s%s", $i, (i<NF?OFS:ORS)
   }
}

$ awk -f tst.awk file
First: 172 - 9
Second: John had 22 apples and lost 2
$ awk -f tst.awk file
First: 201 - 9
Second: John had 12 apples and lost 2
$ awk -f tst.awk file
First: 201 + 62
Second: John had 12 apples and lost 2
$ awk -f tst.awk file
First: 201 + 1
Second: John had 12 apples and lost 4

Just check the math on the line where idx is set - I think it's right but didn't put much thought into it.

Answer 2

$ awk 'BEGIN{srand()} {for (i=1;i<=NF;i++) {if (substr($i,1,1)=="{") {split(substr($i,2,length($i)-2),a,"|"); j=1+int(rand()*length(a)); $i=a[j]}}; print}' math.txt
First: 172 + 1
Second: John had 12 apples and lost 3

How it works

• BEGIN{srand()}

  This initializes the random number generator.

• for (i=1;i<=NF;i++) {if (substr($i,1,1)=="{") {split(substr($i,2,length($i)-2),a,"|"); j=1+int(rand()*length(a)); $i=a[j]}

  This loops through each field. If any field starts with {, then substr is used to remove the first and last characters of the field and the remainder is split with | as the divider into array a. Then, a random index j into array a is chosen. Lastly, the field is replaced with a[j].

• print

  The line, as revised above, is printed.

The same code as above, but reformatted over multiple lines, is:

awk 'BEGIN{srand()}
{
    for (i=1;i<=NF;i++) {
        if (substr($i,1,1)=="{") {
            split(substr($i,2,length($i)-2),a,"|")
            j=1+int(rand()*length(a))
            $i=a[j]
        }
    }
    print
}' math.txt

Revised Problem with Spaces

Suppose that match.txt now looks like:

$ cat math.txt 
First: {736|172|201|109} {+|-|*|%|/} {21|62|9|1|0}
Second: John had {22|12|15} apples and lost {2|4|3}
Third: John had {22 22|12 12|15 15} apples and lost {2 2|4 4|3 3}

The last line has spaces inside the {...}. This changes how awk divides up the fields. For this situation, we can use:

$ awk -F'[{}]' 'BEGIN{srand()} {for (i=2;i<=NF;i+=2) {n=split($i,a,"|"); j=1+int(n*rand()); $i=a[j]}; print}' math.txt
First:  736   +   62 
Second: John had  12  apples and lost  3 
Third: John had  15 15  apples and lost  2 2

How it works:

• -F'[{}]'

  This tells awk to use either } or { as field separators.

• BEGIN{srand()}

  This initializes the random number generator

• {for (i=2;i<=NF;i+=2) {n=split($i,a,"|"); j=1+int(n*rand()); $i=a[j]}

  With our new definition for the field separator, the even numbered fields are the ones inside braces. Thus, we split these fields on | and randomly select one piece and assign the field to that piece: $i=a[j].

• print

  Having modified the line as above, we now print it.

Answer 3

You can try this awk:

awk -F'[{}]' 'function rand2(n) { 
   srand();
   return 1 + int(rand() * n);
}
{
   for (i=1; i<=NF; i++)
      if (i%2)
         printf $i OFS;
      else {
         split($i, arr, "|");
         printf arr[rand2(length(arr))]
      };
      printf ORS
}' math.txt

First:  736  -  62
Second: John had  22 apples and lost  2

---

More runs of above may produce:

First:  172  *  9
Second: John had  12 apples and lost  4

First:  109  /  0
Second: John had  15 apples and lost  3

First:  201  %  1
Second: John had  12 apples and lost  3

...
...