Tricky grep command

Question

I have a text file containing lines of space separated data.

Example:

B  345678  2005-12-21  4174  1  62  11111 16543 1911  786543,45

However the spaces are kind of inconsistant and sometimes I suspect there are some added tabs too. I need to find a was to cut the line exactly before "62" which is a number that always occur in every line and it is somtimes followed by a 5-digit number and then always followed by another 5-digit number. I also add space before 62 to minimixe risk for mismatches.

So far I came up with this:

grep " 62 [1-9][0-9][0-9][0-9][0-9] " file

This will only give the lines where sometimes a 5-digit number shows up. I need a way to grep for 62 followed by an optional 5-digit numbers and then followed by a compulsary 5-digit number.

Can this be done?

/Paul

Answer 1

I need a way to grep for 62 followed by an optional 5-digit numbers and then followed by a compulsary 5-digit number.

That seems to be the same as saying that there's one or two 5-digit numbers after the 62, and then it's enough to just match on the first one. Handling a varying amount of spaces is easy, we can use ␣+, or [[:space:]]+ to include tabs too.

So,

grep -E ' 62 +[1-9][0-9]{4} '

or

grep -E '[[:space:]]62[[:space:]]+[1-9][0-9]{4}[[:space:]]'

This of course will print the entire line. If you want just the part starting from 62, add .* to match the end of the line and -o to print just the matching part:

grep -o -E '[[:space:]]62[[:space:]]+[1-9][0-9]{4}[[:space:]].*'

Answer 2

You could use awk for that purpose:

awk '{if (match($0, "(^[[:print:]]*[[:space:]]+)?(62([[:space:]]+[1-9][0-9]{4})?([[:space:]]+[[:print:]]*)?[[:space:]]+[1-9][0-9]{4}([[:space:]]*|[[:space:]]+[[:print:]]*)$)", a)!=0) print a[2];}' file.txt

This will match any line with an isolated 62, either at the beginning or preceded by any number of printable characters and then one or more space characters, then followed by

• optionally whitespace and a five-digit number
• optionally any number of printable characters
• compulsory whitespace and a five-digit number
• optionally at least one whitespace and any number of printable characters, or whitespace to the end of line

It will print the part starting with the 62 to the end of line if such a pattern was found.

Note that this will not squash the separating whitespaces, so the "inconsistent spaces" will be copied to the output as-is.

Answer 3

... remove everything before 62 (.....)

perl -pe 's/.*?(?= 62 +[1-9]\d{4} )//'

Where:

• s/.*?...// - means remove everything = substitute by nothing
• .*?(?= 62) - means something before 62...

Answer 4

I'd suggest

grep -E '\b62([[:blank:]]+[[:digit:]]{5}\b){1,2}'

where \b is a word boundary that will allow "62" at the start of the line or after a non-word character (like whitespace)