I have the following file

A   B   C   D   E   F   G
-   -   -   -   -   -   -
1   2   3   4   5   6   7
1   3   3   4   8   6   8

I want to compare lines 2 and 3 column by column, and print the complete column when those two values are different.

In this case, the intended output would be

B   E   G
-   -   -
2   5   7
3   8   8

Is there a way to do that using awk or nawk?

The input file will always have 4 lines, but can have as many as 30 columns. I just want to be able to show the columns where the values changed.

Thanks

Jalr2003

You could adapt something like this:

{

    for (x=1 ; x <= NF ; x++) {
         if (NR > 2 && xold[x] && $x != xold[x]) {
            printf("At %d -- %d : %s ! %s\n",NR,x,xold[x], $x)
          }
         if ($x ~ /[0-9]+/) {xold[x] = $x}
    }
}



Test:
File = {
A   B   C   D   E   F   G
-   -   -   -   -   -   -
1   2   3   4   5   6   7
1   3   3   4   8   6   8
5   2   1   5   9   2   3
}

Output:
At 4 -- 2 : 2 ! 3
At 4 -- 5 : 5 ! 8
At 4 -- 7 : 7 ! 8
At 5 -- 1 : 1 ! 5
At 5 -- 2 : 3 ! 2
At 5 -- 3 : 3 ! 1
At 5 -- 4 : 4 ! 5
At 5 -- 5 : 8 ! 9
At 5 -- 6 : 6 ! 2
At 5 -- 7 : 8 ! 3

Or something like this?

BEGIN {
  getline
  n1 = split($0,a1)
  getline
  getline
  n2 = split($0,a2)
  getline
  n3 = split($0,a3)
  for(j=1;j<=n3;j++) if (a2[j]!=a3[j]) printf a1[j] "    "
  print ""
  for(j=1;j<=n3;j++) if (a2[j]!=a3[j]) printf "-    "
  print ""
  for(j=1;j<=n3;j++) if (a2[j]!=a3[j]) printf a2[j] "    "
  print ""
  for(j=1;j<=n3;j++) if (a2[j]!=a3[j]) printf a3[j] "    "
  print ""
  exit
}

You might need to check whether n1, n2 and n3 are equal and format the output better.

CaKiwi

"I love mankind, it's people I can't stand" - Linus Van Pelt

I tried caKiwi's suggestion, but I got blank lines in the output.

Any ideas why this happens?

Thanks for your answers,

Jalr2003

Put some debug print statements after the splits

printf n1
for (j=1;j<=n1;j++) printf " " a1[j]
print ""

CaKiwi

"I love mankind, it's people I can't stand" - Linus Van Pelt

Never mind,

I was missing the second line in my input file.

Thaks,

Jalr2003

One last question,

The lenght of the fields that I have in my input file is not the same.  What can I do to get a better organized output.

I have for input something like this:

SYS_ID    LIMIT      CODE    PART_NUM  
------       ----         ----       --------
16447     15          1NB        4A0001   
19631     14          1NB        4A0001

I changed one of your lines from:

for(j=1;j<=n3;j++) if (a2[j]!=a3[j]) printf "-    "

to:

for(j=1;j<=n3;j++) if (a2[j]!=a3[j]) printf a2[j] "    "

I get the disered output, but it is disorganized.

Thanks again,

Jalr2003

   

If the fields are all the same length, use something like

for(j=1;j<=n3;j++) if (a2[j]!=a3[j]) printf a2[j] substr("        ",1,8-length(a2[j]))

If they are not the same length, create an array, len say, containing the length of each and replace the above substr by

substr("          ",1,len[j]-length(a2[j]))

CaKiwi

"I love mankind, it's people I can't stand" - Linus Van Pelt

Thanks for your help,

I am starting learning about awk, and I am not sure how I can create the array in this case.

Could you help me with that?

Thanks,

Jalr2003

The simplest way would be

len[1]=6;len[2]=6;len[3]=6;....

or if you prefer

split("6,6,6,4,8,6,8",len,",")

CaKiwi

"I love mankind, it's people I can't stand" - Linus Van Pelt

Thanks CaKiwi

It certainly looks better!

However, I just noticed something.

If I have an imput like this

A  B  C  D  E  F  G  H
-  -  -  -  -  -  -  -
1     2  3  4     6  6
2     3  2  4     6  7

my output is

A  B  C  F
-  -  -  -
1  2  3  6
2  3  2  7

As you see, column B gets the values of C, because there are less columns in rows 2 and 3.

Is there a way to eliminate columns B and F, before doing what you suggested?

Thanks again,

Jalr2003

If your data can have blank fields, You will need to use substr instead of split to break them up.

BEGIN {
  split("4,4,4,4,4,4,4",len,",")
  getline
  n1 = split($0,a1)
  getline
  getline
  for (j=1;j<=n1;j++) a2[j] = substr($0,(j-1)*len[j]+1,len[j])
  getline
  for (j=1;j<=n1;j++) a3[j] = substr($0,(j-1)*len[j]+1,len[j])
  for(j=1;j<=n1;j++) if (a2[j]!=a3[j]) printf a1[j] "    "
  print ""
  for(j=1;j<=n1;j++) if (a2[j]!=a3[j]) printf "-    "
  print ""
  for(j=1;j<=n1;j++) if (a2[j]!=a3[j]) printf a2[j] " "
  print ""
  for(j=1;j<=n1;j++) if (a2[j]!=a3[j]) printf a3[j] " "
  print ""
  exit
}

CaKiwi

"I love mankind, it's people I can't stand" - Linus Van Pelt