Passing a variable from one function to another in bash script

Question

Iam new to linux and trying to pass a variable from one function to another in a same bash script.

Below is my code:

#!/bin/bash -x

FILES=$(mktemp)
FILESIZE=$(mktemp)

command_to_get_files(){
 aws s3 ls "s3://path1/path2/"| awk '{print $2}'  >>"$FILES"
}

command_to_get_filesizes(){

for file in `cat $FILES`
 do
 if [ -n "$file" ]
  then
  # echo $file
   s3cmd du -r s3://path1/path2/$file | awk '{print $1}'>>"$FILESIZE"

 fi
 done
}

files=( $(command_to_get_files) )

filesizes=( $(command_to_get_filesizes) )

So in the above code, in the first function $FILES variable is there with the output.

$FILES is passed as a input to the second function command_to_get_filesizes

But am getting error as a Broken Pipe.

Can anyone pls help me to pass a local varibale from one function to another.

Output of $FILES is

2016_01
2016_02
2016_03
2016_04

Output Kindly help!

Answer 1

It depends on your use-case on how to transfer data from one function into another one.

I could not reproduce your error - maybe it has something to do with aws or s3cmd. using backticks as subshell is deprecated - you should use $().

If you just want to pass data and you are not interested in storing them to your hard drive you could use global arrays (everything you don't declare otherwise is global):

#!/usr/bin/env bash

command_to_get_files() {
  local ifs
  # store the internal field separator in order to change it back once we used it in the for loop
  ifs=$IFS
    # change IFS in order to split only on newlines and not on spaces (this is to support filenames with spaces in them)
  IFS='
'
  # i dont know the output of this command but it should work with minor modifications
  # used for tests:
  # for i in *; do
  for file in $(aws s3 ls "s3://path1/path2/" | awk '{print $2}'); do
    # add $file as a new element to the end of the array
    files+=("${file}")
  done
  # restore IFS for the rest of the script to prevent possible issues at a later point in time
  IFS=${ifs}
}

# needs a non-empty files array
command_to_get_filesizes() {
  # check if the number of elements in the files-array is 0
  if (( 0 == ${#files[@]} )) then
    return 1
  fi
  local index
  # iterate over the indices of the files array
  for index in "${!files[@]}"; do
    # $(( )) converts the expression to an integer - so not found files are of size 0
    filesizes[${index}]=$(( $(s3cmd du -r "s3://path1/path2/${files[${index}]}" | awk '{print $1}') ))
    # used for testing:
    # filesizes[${index}]=$(( $(stat -c %s "${files[$i]}") ))
  done
}

command_to_get_files
command_to_get_filesizes

# loop over indices of array (in our case 0, 1, 2, ...)
for index in "${!files[@]}"; do
  echo "${files[${index}]}: ${filesizes[${index}]}"
done

notes about bash arrays:

• get the size of the array: ${#array[@]}
• get the size of the first element: ${#array[0]}
• get the indices of the array: ${!array[@]}
• get the first element of the array: ${array[0]}

for more information about arrays have a look here.

another method would be to just echo the names and provide them as parameters to the other function (this is difficult with multi-word filenames)

Using temporary files would result in something like this:

#!/usr/bin/env bash

readonly FILES=$(mktemp)
readonly FILESIZES=$(mktemp)

# at script exit remove temporary files
trap cleanup EXIT
cleanup() {
  rm -f "$FILES" "$FILESIZES"
}

command_to_get_files() {
  aws s3 ls "s3://path1/path2/" | awk '{print $2}' >> "$FILES"
}

command_to_get_filesizes() {
  while read -r file; do
    s3cmd du -r "s3://path1/path2/${file}" | awk '{print $1}' >> "$FILESIZES"
  done < "$FILES"
}

command_to_get_files
command_to_get_filesizes