I want to find all the files that has the words “Who”, “What”, “Why”, “How”, “When”. All of the words, in any order. Case Insensitive
I tried:
grep -rl --include='*.adoc' "Who" . | xargs grep -l "What" | xargs grep -l "Why" | xargs grep -l "How" | xargs grep -l "When" It is giving error like:
grep: Walkthrough/datatable/extras/Scroller/media/data/2500.adoc: No such file or directory The problem you are having is that some of your filenames contain spaces. xargs will split that into multiple "filenames".
Add the -0 option to the xargs to make them split on NULs instead of whitespace, and the --null or -Z option to the grep command line to make it use NULs instead of newlines. (but omit the --null on the final grep if you want to read the output...). So putting it all together:
grep -r -iwlZ --include='*.adoc' 'who' . | xargs -r0 grep -iwlZ 'what' | xargs -r0 grep -iwlZ 'why' | xargs -r0 grep -iwlZ 'how' | xargs -r0 grep -iwl 'when' Alternatively, eliminate the whitespace and other shell special characters from your filenames.
Otherwise, your solution is basically correct, though the answer by @James is correct that you need the -i option for case insensitive.
-w option to match whole words not compounds like whatever and Show grep -rl --null -w --include='*.adoc' "Who" . | xargs -0 grep -w --null -l "What" | xargs -0 grep -w --null -l "Why" | xargs -0 grep -w --null -l "How" | xargs -0 grep -w --null -l "When" | tr '\0' '\n'. Is this correct? --null on the final grep, then you won't need the tr. Based on your original post, all the greps would need -i as well to make the matches case insensitive. Using find and GNU awk in slurp mode to process whole files as single records and word boundaries \<, \> equivalent to grep's -w / --word-regexp option:
find . -name '*.adoc' -exec gawk -v RS='^$' -v IGNORECASE=1 ' /\<who\>/ && /\<what\>/ && /\<why\>/ && /\<how\>/ && /\<when\>/ {print FILENAME} ' {} + although this appears to be an order of magnitude slower than piping through multiple greps - I guess caching means there's very little overhead in grepping the same file multiple times.
-i is needed to do case-insensitive search. But this command will find any file that contains any of the words, not all of them (as the question asks). (2) For directory-based searches (e.g., recursive ones), it’s better to specify an argument of . rather than *. All of the words... in ONE file. Just realized this. Thanks -v to a BEGIN statement (without also moving the RS definition there).