UNIX geek challenge

[chhotii]

UNIX Geek Of The Day bragging rights to whoever can propose the best solution to this problem: Identify all text files in the current directory which contain the string "jpg" on two consecutive lines. Using the typical command-line tools such as grep, not by writing a program. Extra points if the location (line number) of these consecutive lines of interest are output.

Posted via LiveJournal app for Android.

Comments

[ceo]

With GNU grep, the -P option enables Perl regular expressions, which handle newlines better than plain ol'regexes do.

grep -Pn '.*jpg.*\n.*jpg.*' * appears to work.

[chhotii.livejournal.com]

Whoa, clever! Thanks!

[achinhibitor.livejournal.com]

[edited for correctness]

Given that grep is defined to filter lines, and lines by definition do not contain regexps newlines, is there any guarantee that this works in general?

Also, what does it do if there are three jpg lines in sequence? There are *two* matches that should be reported.

[frobzwiththingz.livejournal.com]

Ok, Chip's got a good answer... Here's a truly horrible one:

for I in *; do test -f "$I" && awk ' BEGIN { last = 0 ; status = 1 } { if (/jpg/) { if (last == 1) { print $0, "line number ", NR ; status = 0 } else last = 1 } else last=0 } END { if (status == 1) exit 1 }' < "$I" && echo " file $I matches"; done

Don't use this. There's *got* to be a horrible bug in there somewhere. Unless you hate your workplace. In that case, make sure it gets into some important automated build process.

[whitebird.livejournal.com]

You are a nicely evil person. :)

[chhotii.livejournal.com]

So totally NOT what I was looking for.

[achinhibitor.livejournal.com]

I can do this without using anything that is programmable (like sed or awk):

for FILE in *
do
    paste <(head --lines=-1 $FILE) \
         <(while read X; do echo Throat-warbler-mangrove ; done <$FILE) \
         <(tail --lines=+2 $FILE) |
    grep -n $'jpg.*\tThroat-warbler-mangrove\t.*jpg' &&
    echo $FILE
done