I would like to remove the "AS" part in the below string,
CONV.BUY AS BUY, CONV.CUSTOM1 AS CUSTOM1 , TP.BUY AS BUY, TP.CUSTOM1 AS CUSTOM1
should be changed to
CONV.BUY, CONV.CUSTOM1, TP.BUY, TP.CUSTOM1
What i did is
echo $string | sed -e 's/ AS.*\,/\,/g'
which gives
CONV.BUY, TP.CUSTOM1 AS CUSTOM1
Two fields are filtered off in that sed. Is there any better solution to achieve my objective without using a loop?
With sed:
sed 's/[[:blank:]]AS[^,]*//g'
[[:blank:]]AS matches AS preceded by a whitespace, then [^,]* matches upto the next ,. All such matched portions are removed by replacing with null.Example:
% sed 's/[[:blank:]]AS[^,]*//g' <<<'CONV.BUY AS BUY, CONV.CUSTOM1 AS CUSTOM1 , TP.BUY AS BUY, TP.CUSTOM1 AS CUSTOM1'
CONV.BUY, CONV.CUSTOM1, TP.BUY, TP.CUSTOM1
AS[^,]+ part in sed? and how it replaces even instances without a comma
AS[^,]* would do, no need for + (which is a Exteneded regex pattern, matches one or more of the preceding token). So in AS[^,]*, AS matches AS literally, [^,] matches any character except ,, the following * matches zero or more of such characters, so it eventually matches everything till next ,.AS case insensitive? i.e. match if it is AS or as?sed 's/[[:blank:]][AaSs][^,]*//g'You can use a negated character class as
sed 's/AS[^,]*//g'
[^,]* Matches zero or more non , characters. Which means that it matches till the immediate , and replace it with nullExample
>>> echo "EONV.BUY AS BUY, CONV.CUSTOM1 AS CUSTOM1 , TP.BUY AS BUY, TP.CUSTOM1 AS CUSTOM1" | gsed 's/AS[^,]*//g'
EONV.BUY , CONV.CUSTOM1 , TP.BUY , TP.CUSTOM1