Help with a "Expression" in Bash...

[iMisspell]

I have files with a bunch of lines...

Small example...


(ANYTHING IN PARENTHESES DO NOT TOUCH)
(UNTOUCHED TEST!@#TEST1234TEST)
N007T04
N008G54.1P01
N009S1000M03
N010G00X3.97Y7.3
N011G43H02Z0.75
N012Z0.45
N013G1Z0.01F75.M08
N014Y0.F20.
N015G2X3.97Y0.I-3.97J0.F42.5

And i need to seperate the line by adding a space between 'letters' and 'numbers' and possibly a 'period' ...


(ANYTHING IN PARENTHESES DO NOT TOUCH)
(UNTOUCHED TEST!@#TEST1234TEST)
N007 T04
N008 G54.1 P01
N009 S1000 M03
N010 G00 X3.97 Y7.3
N011 G43 H02 Z0.75
N012 Z0.45
N013 G1 Z0.01 F75. M08
N014 Y0. F20.
N015 G2 X3.97 Y0. I-3.97 J0. F42.5


As you will see, im horrible with Regular Expressions, ive tried the following, which is close, but still not working correctly...



sed 's/\([0-9]*\)*\([A-Z]\)/\1 \2/g' $file


Will produce the following.
The first 'space' before N### can not be there, and the 'spaces' added with-in the parentheses are not wanted.


( A N Y T H I N G I N P A R E N T H E S E S D O N O T T O U C H)
( U N T O U C H E D T E S T!@# T E S T1234 T E S T)
N007 T04
N008 G54.1 P01
N009 S1000 M03
N010 G00 X3.97 Y7.3
N011 G43 H02 Z0.75
N012 Z0.45
N013 G1 Z0.01 F75. M08
N014 Y0. F20.
N015 G2 X3.97 Y0. I-3.97 J0. F42.5



Any help would be great.


_

[Trumpen]

You just need to add a test for the presence of parentheses in the line:


sed '/^[^(]/ s/\([0-9]*\)*\([A-Z]\)/\1 \2/g' $file

/^[^(]/ is true if the line doesn't begin with '('

If you want to avoid adding a first space character you may
write:


sed '/^[^(]/ s/\([0-9.]\)\([A-Z]\)/\1 \2/g' $file

[DaithiF]

slight variation on how to avoid updating lines containing parentheses -- using the '!' negate operator.

sed '/(/!s/\([0-9.]\)\([A-Z]\)/\1 \2/g' testfile

[iMisspell]

Thanks to the both of you, its working correctly now.


Am i understanding this correctly...
The following will search the first character of the line, if that character is a 'open parentheses' ( then the expression will skip that line.


sed '/^[^(]/ s/\([0-9.]\)\([A-Z]\)/\1 \2/g' $file



And this will search the first character of the line and if that character is not a number or letter, it will skip ?


sed '/(/!s/\([0-9.]\)\([A-Z]\)/\1 \2/g' testfile



As for this ???
'\([0-9.]\)\([A-Z]\)'I see you's have removed the wild card form the middle and added a period to the end of the number range, why/how does that work ?



Again, thanks for the help.

_

[geirha]

Am i understanding this correctly...
The following will search the first character of the line, if that character is a 'open parentheses' ( then the expression will skip that line.


Yes, or you can say: For all lines that start with any character that is not «(», execute the following sed command.



And this will search the first character of the line and if that character is not a number or letter, it will skip ?


«sed '/(/COMMAND'» -> run COMMAND on all lines that match the regex «(».
«sed '/(/!COMMAND'» -> run COMMAND on all lines that does NOT match the regex «(».

And COMMAND in this case is a substitute command (s/regexp/replacement/)



As for this ???
'\([0-9.]\)\([A-Z]\)'I see you's have removed the wild card form the middle and added a period to the end of the number range, why/how does that work ?


It matches ONE digit or . (literal punctuation mark, not any char) directly followed by one uppercase letter.

E.g. in «N008G54.1P01» it'll match «8G» and «1P».

[iMisspell]

Thanks geirha
The first one with /^[^(]/ is alittle confusing for me (i see how its working, but still dont understand why), but the second with the "not" ! made more sence - being '!' is used in other languages.

Anyway... thanks everyone.

_