2

I'm trying to match first letter, in this example "B", and second column "2". When match is found replace characters [38-41] with white spaces.

Here is the data I'm trying to modify:

A1234A123 1 2 12345.12345 1234.1234.112341234

B1234A123 2 2 12345.12345 1234.1234.112341234

A1234A123 2 2 12345.12345 1234.1234.112341234

I can match the conditions with awk using: 

awk '/^B/ && $2=="2" {print}'

and I can modify the lines with sed using: 

sed -r 's/^(.{37})(.{4})/\1    /'

I'm trying to find the lines in the file which contains the two conditions and then modify the characters, while still printing the entire line of lines that don't match. Can you combine the two commands in order to introduce some sort of if/then statement?

I've tried to combine the commands, but it edited all of the lines: 

awk '/^B/ && $2=="2" {print}' ¦ sed -r 's/^(.{37})(.{4})/\1    /' data

Resulting data should look like this: 

A1234A123 1 2 12345.12345 1234.1234.112341234

B1234A123 2 2 12345.12345 1234.1234.1    1234

A1234A123 2 2 12345.12345 1234.1234.112341234

Thanks in advance.

Tomas
  • 52,167
  • 46
  • 207
  • 345
fryman84
  • 37
  • 5
  • You never need to combine sed and awk (or grep and awk). sed is an excellent tool for simple substitutions on a single line, for any other text manipulation just use awk. – Ed Morton Dec 29 '13 at 12:37
  • OK @Ed, thanks for the advice and corrections to others posts. I was thinking the solution was more difficult than it ended up being. The more I read on AWK, the more I realize it's potential. I'll keep studying! Thanks again. – fryman84 Dec 29 '13 at 18:46
  • http://stackoverflow.com/questions/1632113/what-is-the-difference-between-sed-and-awk – fryman84 Dec 29 '13 at 19:09
  • The discussion on that link is all fine but THE important thing to know about the 2 tools is that sed was invented before awk. Once awk was invented in the mid-1970s most of seds language constructs became obsolete so today the only useful sed constructs are s, g, and p (with the -n option) and any time you're using hold space or pattern space or whatever other "space" sed supports, you are using the wrong tool. sed is an excellent tool for simple substitutions on a single line - that's it. – Ed Morton Dec 29 '13 at 19:17

4 Answers4

3

You can use single awk to combine both commands:

awk '/^B/ && $2=="2"{$0=substr($0, 1, 37) "    " substr($0, 38, 4)} 1' file
A1234A123 1 2 12345.12345 1234.1234.112341234
B1234A123 2 2 12345.12345 1234.1234.1    1234
A1234A123 2 2 12345.12345 1234.1234.112341234
anubhava
  • 664,788
  • 59
  • 469
  • 547
  • Got it! Thanks! I actually just changed the character location of the last substr so that it would retain the last four characters. `awk '/^B/ && $2=="2"{$0=substr($0, 1, 37) " " substr($0, 42, 4)} 1' file` – fryman84 Dec 29 '13 at 07:36
  • You're welcome, yes `substr($0, 42, 4)` will also return `1234` in output. – anubhava Dec 29 '13 at 07:48
  • You don't need the `, 4` arg in the last substr(). – Ed Morton Dec 29 '13 at 12:29
  • Yes if picking right most part of a string then 2nd argument isn't needed in substr – anubhava Dec 29 '13 at 12:34
2

You may instruct sed to replace only the matching line (/^B[^ ]* 2/) by prepending the regex:

sed -r '/^B[^\s]*\s2\s/s/^(.{37}).{4}/\1    /' data
Tomas
  • 52,167
  • 46
  • 207
  • 345
1

With GNU awk:

gawk '/^B/ && $2=="2" {print gensub(/(.{37}).{4}/,"\\1    ","")}' data
Ed Morton
  • 157,421
  • 15
  • 62
  • 152
0

In Gnu Awk version 4 you could try:

gawk 'BEGIN { FIELDWIDTHS = "1 9 1 26 4 20"; OFS="" }
$1=="B" && $3=="2" {
    $5="    "
} 1' file

with output:

A1234A123 1 2 12345.12345 1234.1234.112341234
B1234A123 2 2 12345.12345 1234.1234.1    1234
A1234A123 2 2 12345.12345 1234.1234.112341234
Håkon Hægland
  • 32,521
  • 18
  • 64
  • 139