[SOLVED] Conversion from characters [A, H, B] to numeric [1, 0, -1]

Issue

My goal is to convert file with 2 1 0 to file with 1 0 -1 via A H B. For example,

Infile.txt        Temp.txt           Final.txt
2 2 2 1 1 1 0 0   A A A H H H B B    1 1 1 0 0 0 -1 -1

I was able to convert from numerics(2 1 0) to characters(A H B) using the code as follows:

cut -f2- Infile.txt | sed '1,1d' | sed 's/2/A/g' | sed 's/1/H/g' | sed 's/0/B/g' > Temp.txt

However, I could not convert from (A H B) to (1 0 -1). Hence, I got stuck with Temp.txt

So, I would appreciate any solution to figure it out. Thanks!

Solution

OP has mentioned in a comment the source file is a 20,000 x 500 (row x column) matrix of the digits 2, 1 and 0.

Create a 20,000 x 501 (row x column) matrix:

awk '
BEGIN { for (i=1;i<=20000;i++) {
            sep=""
            for (j=1;j<=167;j++) {
                printf "%s2 1 0", sep
                sep=" "
            }
            print ""
        }
      }
' > matrix.dat

$ head -5 matrix.dat | cut -c1-30
2 1 0 2 1 0 2 1 0 2 1 0 2 1 0
2 1 0 2 1 0 2 1 0 2 1 0 2 1 0
2 1 0 2 1 0 2 1 0 2 1 0 2 1 0
2 1 0 2 1 0 2 1 0 2 1 0 2 1 0
2 1 0 2 1 0 2 1 0 2 1 0 2 1 0

One awk/gsub() idea:

awk '{ gsub(/1/,9)
       gsub(/2/,1)
       gsub(/0/,-1);
       gsub(/9/,0)
     }
1
' matrix.dat > matrix.awk1.out

One awk/loop idea:

awk '{ for (i=1;i<=NF;i++) 
           $i=$i-1
     }
1
' matrix.dat > matrix.awk2.out

One sed idea:

sed 's/1/9/g;s/2/1/g;s/0/-1/g;s/9/0/g' matrix.dat > matrix.sed.out

These all generate the same result:

$ diff matrix.awk1.out matrix.awk2.out
$ diff matrix.awk2.out matrix.sed.out
$ head -5 matrix.awk1.out | cut -c1-35
1 0 -1 1 0 -1 1 0 -1 1 0 -1 1 0 -1
1 0 -1 1 0 -1 1 0 -1 1 0 -1 1 0 -1
1 0 -1 1 0 -1 1 0 -1 1 0 -1 1 0 -1
1 0 -1 1 0 -1 1 0 -1 1 0 -1 1 0 -1
1 0 -1 1 0 -1 1 0 -1 1 0 -1 1 0 -1

Run times:

• system: cygwin (in a VM), awk 5.1.1, sed 4.8
• 5.5 secs : awk/gsub()
• 3.9 secs : awk/loop
• 5.9 secs : sed

Answered By - markp-fuso

Answer Checked By - Clifford M. (WPSolving Volunteer)