replace command

Anuj12 · 01-21-2013, 09:48 AM

Hi

I have a file which looks like

I 23 value of I here=56
I 45 value of I here=36
I 45 value of I here=20
II 56 value of II here=22
III 56 value of III here=32
I 45 value of I here=11
II 11 value of II here=79
I 22 value of I here=74
II 3 value of II here=45
III 45 value of III here=43
BTR 12 value of BTR here=11
BT 11 value of BT here=23

I Want it to replace I by string1 and II by string2 and so on.... BT by string4 BTR by string5
I used

sed 's/I\t/string1\t/' infile >sortedfile
sed 's/II\t/string2\t/' infile >sortedfile
sed 's/III\t/string3\t/' infile >sortedfile
sed 's/BT\t/string4\t/' infile >sortedfile
sed 's/BTR\t/string5\t/' infile >sortedfile

but it replaced III as string1string1string1
whT AM i doing wrong.?

Also I want it to change only the first column/tab and not affect the other tabs.

I want it to look like this

string1 23 value of I here=56
string1 45 value of I here=36
string1 45 value of I here=20
string2 56 value of II here=22
string3 56 value of III here=32
.....

string5 12 value of BTR here=11
string4 11 value of BT here=23
PLease help.

divyashree · 01-21-2013, 10:35 AM

try this:

Code:

sed 's/^I /string1\t/' infile >sortedfile
sed 's/^II /string2\t/' infile >sortedfile
sed 's/^III /string3\t/' infile >sortedfile
sed 's/^BT /string4\t/' infile >sortedfile
sed 's/^BTR /string5\t/' infile >sortedfile

Anuj12 · 01-21-2013, 11:19 AM

Hi
I tried this mentioned command too.
sed 's/^I /string1\t/' infile >sortedfile
sed 's/^II /string2\t/' infile >sortedfile
sed 's/^III /string3\t/' infile >sortedfile
but it works on my III as

string1 II III 56 value of III here=32

instead of writing it as

string3 56 value of III here=32

shivaa · 01-21-2013, 11:24 AM

It's little tricky. but if your actual file is exactly like what you've mentioned, then following one-liner awk code could help:

Code:

~$ awk '{FS=" "; OFS=" "}; {gsub(/^[I]/,"string1",$1); gsub(/string1I/,"string2",$1);gsub(/string2I/,"string3",$1);gsub(/BTR/,"string4",$1);gsub(/BT/,"string5",$1); print $0}' infile.txt

But in order to solve the actual problem, specify that how large your file is, and how much strings or words you want to replace, so looping and regexp can be used. Anyway, provide little more clarification

David the H. · 01-22-2013, 11:29 AM

Please use ***[code][/code]*** tags around your code and data, to preserve the original formatting and to improve readability. Do not use quote tags, bolding, colors, "start/end" lines, or other creative techniques.

To limit your regexes, use "^" and "\b". "^" matches the start of the line, and \b is a zero-width anchor that matches the transition between a word and non-word character (word characters are [a-zA-Z0-9_]).

And there's no need to use separate sed commands. Just give one command multiple expressions.

Code:

sed 's/^I\b/1/ ; s/^II\b/2/ ; s/^III\b/3/ ; s/^BT\b/4/ ; s/^BTR\b/5/ ; s/^/string/' infile

And just for fun, here's another awk solution for you:

Code:

awk 'BEGIN{ a["I"]=1 ; a["II"]=2 ; a["III"]=3 ; a["BT"]=4 ; a["BTR"]=5 } { $1=a[$1]?"string"a[$1]:$1 }1' infile

@shivaa, there's no need to use sub/gsub if you want to replace the entire value of the field. Just set it as you would a variable (e.g. $1="newstring").

jpollard · 01-22-2013, 03:48 PM

The simple explanation is that you did it in the wrong order.

Because you tried the shortest replacement first, the s/I\t/string1\t/... matched EVERY I in the file.

Had you done the longest exchange first: sed 's/III /string3\t/' ..., then only sequences of III would be replaced - leaving the II and I sequences alone. Then you can do the s/II / sequence.

The same holds true for the 's/BT\ being done before the 's/BTR'. it replace the BT in all positions.

Do the s/BTR first, and it will only find those situations with BTR, and leave the BT entries alone.