remove a line from a comma delimited file that contains a single digit in position 2

j-me · 05-29-2013, 09:02 AM

I know this is very specific question. I have tried a few sed statements and grep with no luck. Here is my data:

Code:

Deploy_ACC_FBL.Legacy.ACV_Production,1,ACC_FBL.Legacy.ACV_2012.09.32,SUCCESS,
Deploy_ACC_FBL.Legacy.ACV_Production,2,ACC_FBL.Legacy.ACV_2012.09.44,FAILURE,
Deploy_ACC_FBL.Legacy.ACV_Production,2013-02-16_22-03-50,ACC_FBL.Legacy.ACV_2012.09.32,SUCCESS,

I need to remove the line that contains the single number in the second position while retaining the line that contains the valid date in second position.

Code:

sed '/[0-9]/d' filename.txt > fileout.txt

yielded nothing in the output file. I know why.
thank you

Ygrex · 05-29-2013, 09:24 AM

Code:

awk -F, '($2 ~ /^[0-9]$/) { next } { print $0 }'

Ygrex · 05-29-2013, 09:26 AM

Code:

sed -r '/^[^,]*,[0-9],/d'

Ygrex · 05-29-2013, 09:32 AM

with bash:

Code:

while IFS=, read a b rest ; do [[ "$b" =~ ^[0-9]$ ]] && continue ; printf "%s,%s,%s\n" "$a" "$b" "$rest" ; done

j-me · 05-29-2013, 09:54 AM

ok. that works but could you possibly take a few minutes and explain the command please? I would like to understand this better. specifically the ^,[^,]* . that is so confusing to me.

Also, I have some lines w/ 2 digit numeric in the field 2. I just modified and ran thru the output file to remove those with [0,9][0-9].

Very much appreciate your help ... and explanation.

Ygrex · 05-29-2013, 10:20 AM

seems like you'd better read something about regexps, but I'll try to explain this one: /^[^,]*,[0-9],/

values cannot include comma, then /[^,]*,/ is a general pattern for each column (excluding the last one, that does not have trailing comma: /[^,]*$/)

providing you need to match a single digit in the 13-th column it means that 12 columns go first followed by /[0-9],/

the line beginning matches with '^' at the 1st position in a whole regexp:
/^([^,]*,){12}[0-9],/

since you need to match the 2nd column, there is only one column goes before it, and no need to use repetition in this case:
/^([^,]*,)[0-9],/

and grouping as well:
/^[^,]*,[0-9],/

danielbmartin · 05-30-2013, 08:25 AM

With this InFile ...

Code:

Deploy_ACC_FBL.Legacy.ACV_Production,1,ACC_FBL.Legacy.ACV_2012.09.32,SUCCESS,
Deploy_ACC_FBL.Legacy.ACV_Production,2,ACC_FBL.Legacy.ACV_2012.09.44,FAILURE,
Deploy_ACC_FBL.Legacy.ACV_Production,2013-02-16_22-03-50,ACC_FBL.Legacy.ACV_2012.09.32,SUCCESS,

... this awk ...

Code:

awk -F, '($2!~"^[0-9]$")' $InFile >$OutFile

... produced this OutFile ...

Code:

Deploy_ACC_FBL.Legacy.ACV_Production,2013-02-16_22-03-50,ACC_FBL.Legacy.ACV_2012.09.32,SUCCESS,

Daniel B. Martin