Find command not working in for loop

Question

I have list of filename , I want to provide those files one by one to find command so that i can get the path. I have use the for loop and take all filename in variable using cat command, Then pass them to find command through variable one by one but its not working. can any one guide me?

for f in $(cat filelist.txt)
do 
find /home/abc/ -type f -name "$f"
done;

I am not getting any error but not getting output either I tried putting output in variable as well but no luck.

Answer 1

If you want to find the files whose name are any of the ones stored in filelist.txt one per line, one easy way is to use zsh's globbing:

names=( ${(f)"$(<filelist.txt)"} ) &&
  (( $#names )) &&
  print -rC1 -- /home/abc/**/${(~j[|])names}(D.)

As to why your approach doesn't work, it's not immediately obvious, though there are few problems with it:

• In bash, that unquoted $(cat filelist.txt) invokes split+glob (split only in zsh) on the contents of filelist.txt. With the default value of $IFS (which is used for the split part), that will split on newline characters but also on space and tab. So if you have a my file.txt line in there, it will look for files called my and files called file.txt instead of files called my file.txt
• If there is a file called * in filelist.txt, the glob part will expand it to the list of files in the directory the script is being called from.
• Also -name '*' will not look for a file called *, but will report every file as what's passed to -name is taken as a pattern. You'd need to escape the [, ?, * and \ with \ to disable the wildcard operators.

Now, your problem may also be that:

• filelist.txt is in MSDOS text format with CRLF delimiters instead of LF. You'd run dos2unix on it to fix it.
• that you have file paths in filelist.txt (such as dir/file.txt). -name matches on the file name only, so would never match on that.

Here, instead of a loop, you could use GNU xargs:

xargs -rd'\n' -a filelist.txt -I NAME find /home/abc/ -name NAME -type f

But that still means crawling the whole of /home/abc/ for each file name, which is inefficient.

Instead, you could do:

xargs -rd'\n' -n200 -a <(sed 's/[*?\\[]/\\&/g' filelist.txt) sh -c '
  for name do
    set -- -o -name "$name"
    shift
  done
  shift
  find /home/abc/ \( "$@" \) -type f' sh

Where we call find /home/abc/ \( -name name1 -o -name name2... \) -type f with up to 200 names in one find invocation, which means potentially 200 as few crawlings of /home/abc. Not as good as zsh which does only 1 in any case, but still a lot better.

We also use sed to escape the wildcard characters in the names. You could also through in a dos2unix in there.

Answer 2

Try also

find -iname $(sed -z 's/\n$//; s/\n/ -o -iname /g' list.txt)

The sed -z looks for lines terminated by a \000 character which should not exist in *nix text files, so it loads the entire file into memory. Then removes the trailing <NL> char, and substitutes the residual ones with -o -iname so find get served a long list of ored filenames.