if condition throws error

Question

In bash coding, line3 is a path taken from xyz/symlinks_paths.txt.

while read -r line3
do
    if [[ $(ls -lh $line3 | grep zzz.exe | grep '[8-9][0-9][0-9][MG]') -ne 0 ]] 
    then 
        echo $line3 >> xyz/size_list.txt
        exit 1
    fi
done < xyz/symlinks_paths.txt

The script throws me the following error. (h.sh is the script name.)

h.sh: line 20: [[: -r--r--r-- 1 syncmgr GX 838M Dec  1 21:55 zzz.txt: syntax error in expression (error token is "r--r-- 1 syncmgr GX 838M Dec  1 21:55 zzz.txt")

Answer 1

The problem here is that you're trying to parse the output of ls. This is always a bad idea. See Why *not* parse `ls`? for an explanation of why that is the case.

If you want the size of a file, then use stat. e.g.

minsize=$(( 800 * 1024 * 1024 ))

# alternatively, if you have `numfmt` from GNU coreutils, delete the line above
# and uncomment the following line:
#minsize=$(echo 800M | numfmt --from=iec)

while read -r line3 ; do
  if [ "$(stat -L --printf '%s' "$line3")" -gt "$minsize" ]; then
    echo "$line3" >>  xyz/size_list.txt
  fi
done < xyz/symlinks_paths.txt

Note: I've used stat's -L (aka --dereference) option above because the input filename implies that the filenames listed in it might be symbolic links. Without -L, stat won't follow a symlink, it would print the size of the symlink itself.

---

If you want the file size printed to the output file along with the filename, then the while loop would be more like the following:

while read -r line3 ; do
  fsize=$(stat -L --printf '%s' "$line3")

  if [ "$fsize" -gt "$minsize" ]; then
    fsize=$(echo "$fsize" | numfmt --to=iec)
    printf "%s\t%s\n" "$fsize" "$line3" >>  xyz/size_list.txt
  fi
done < xyz/symlinks_paths.txt

Answer 2

This can be done with find (and xargs), but it won’t win any beauty contests.

Write a script called check_files:

#!/bin/sh
find "$@" -size +800M –print

Then run

xargs -d '\n' < xyz/symlinks_paths.txt ./check_files

where

• You can move < xyz/symlinks_paths.txt redirection to the end of the command line, as in xargs -d '\n' ./check_files < xyz/symlinks_paths.txt, or to the beginning, or anywhere else.  Or you can replace it with -a xyz/symlinks_paths.txt.  Any of these mean that xargs will read from xyz/symlinks_paths.txt.
• You can replace ./check_files with an absolute pathname to your check_files script.

-d '\n' means use newline as the delimiter when reading xyz/symlinks_paths.txt.  You can probably leave this off if your filenames don’t contain whitespace (space(s) or tab(s)), quotes (remember that a single quote (') is the same character as an apostrophe) or backslashes, and you’re willing to wager a year’s salary that they never ever will.

This reads each line of the file and makes it an argument to the check_files script, which passes them to find as starting-point arguments.  Many people know that you can run find with multiple starting-point arguments; e.g.,

find dir1 dir2 dir3  search-expression

It’s not so well known that those arguments don’t have to be directories; they can be files; e.g.,

find file1 file2 file3  search-expression

(or a mixture of directories and files).  find will simply apply the expression to each file named as a starting-point.

So this checks each file whose name is listed in xyz/symlinks_paths.txt to see whether its size is 800M or more, and prints those that are.

If the filenames might refer to symbolic links (as the xyz/symlinks_paths.txt name suggests) and you want to look at the pointed-to files (which you surely do), change find to find -L.

You don’t need to have a separate check_files script; you can do

xargs -d '\n' < paths.txt sh -c 'find "$@" -size +800c -print' sh

Again, change find to find -L if desired.