[SOLVED] grep for a string with both letters and digits...

hi,
is there a way to grep for a 'word' containing at least 1 letter and 1 digit (in any order)?
i mean something like this:

$ grep -iwo 'some_regex' <<< "this sample string n. 1: bu2bu"
bu2bu

i know this could be done with 3 separated grep commands:

$ grep -iwo '[^ ]\+' <<<"this sample string n. 1: bu2bu" \
| grep -i '[a-z]' \
| grep '[0-9]'
bu2bu

but i was wondering if it exists such a regex to perform this filtering with a single command...

Perhaps:

egrep '([0-9][a-z]|[a-z][0-9])'

Originally Posted by NevemTeve Perhaps: egrep '([0-9][a-z]|[a-z][0-9])'

much easier than i expected, thanks!

$ grep -Eio '[^ ]*([0-9][a-z]|[a-z][0-9])[^ ]*' <<< "this sample string n. 1: bu2bu"
bu2bu

^ Beware of what you call "word". Generally, in the regex world, a word is composed of characters [A-Za-z0-9_] but you seem to consider that a word is everything except space character.
So with your last regex, "5apples.3strawberries" and "1A|3B" and "1r,3d" are only one-word matches. Some people would have considered them being 2-word matches.
Moreover, as a side-effect, you include final punctuation signs inside your matches (for example: "t3st." matches inside "this is a t3st.", not just "t3st")

Depending on what you want, you may keep your regex or use one of the suggestions below:

grep -Eo '[[:alnum:]_]*(([[:digit:]][[:alpha:]])|([[:alpha:]][[:digit:]]))[[:alnum:]_]*'
grep -Eo '\w*(([[:digit:]][[:alpha:]])|([[:alpha:]][[:digit:]]))\w*'

Much simpler (and more readable) to use a more appropriate tool. awk for example is usually a better fit for field (i.e. "word") based processing.
And the regex handling is more flexible - use of logical AND for example.

^ Sir, at your command! Here is a suggestion with awk

... | awk 'BEGIN{FS="[^A-Za-z0-9_]"} {for(i=1;i<=NF;i++) {if($i ~ /([[:digit:]][[:alpha:]])|([[:alpha:]][[:digit:]])/) {print $i} } }'

Taking the lovely solution of NevemTeve in post #2 and recasting it in awk ...

With this InFile ...

aaabbb   toss this
cccggg   toss this one too
q3rrst   keep this
99wxyz   keep this
kkll88   keep this
122333   another toss

... this awk ...

awk '{if (match($0,/([0-9][a-z]|[a-z][0-9])/)) print}' $InFile >$OutFile

... produced this OutFile ...

q3rrst   keep this
99wxyz   keep this
kkll88   keep this

Daniel B. Martin

.

The OP only wants the matching "word", not the entire record. I was thinking more along the lines of ...

awk '{for (i=1; i<=NF; i++) {if ($i ~ /[0-9]/ && $i ~ /[[:alpha:]]/) {print $i}}}'

To me, Much more readable. Probably should use [:digit:] ... :shrug:

Originally Posted by syg00 The OP only wants the matching "word", not the entire record. ...

With this InFile ...

Four4 score and se7ven 777 years ago our fath22ers brought forth
on this continent a 88new nation, conceived in liberty,
and 99dedicated to the proposition99 that all men are cre5ated equal.

... this awk ...

awk 'BEGIN{RS=" |\n"}
  {if (match($0,/([0-9][a-z]|[a-z][0-9])/)) print}' $InFile >$OutFile

... produced this OutFile ...

Four4
se7ven
fath22ers
88new
99dedicated
proposition99
cre5ated

Daniel B. Martin

.