Load environment variables from dotenv / .env file in Bash

load_dotenv.sh

# The initial version
if [ ! -f .env ]
then
  export $(cat .env | xargs)
fi

# My favorite from the comments. Thanks @richarddewit & others!
set -a && source .env && set +a

Posix compliant version built around set, [ ] and . Many thanks to the prior posters who brought up set -o a and set -a / set +a

This snippet will source a dotenv file, exporting the values into the environment. If allexport is already set, it leaves it set, otherwise it sets, reads, and unsets.

if [ -z "${-%%*a*}" ]; then
    set -a
    . ./.env
    set +a
else
    . ./.env
fi

double brackets [[, source, setopt are not available in posix. Nor is the test [[ -o a ]] to check for set options. And we need to quote our comparison strings to deal with empty vars.

The code to check if an option is set is a bit of a pain. It could be a case statement or a grep on set -o like set -o | grep allexport | grep -q yes, but blech. Instead I've used parameter expansion with pattern matching to remove a maximum match from the $- variable containing a single line of the set options.

${-%%*a*} uses %% parameter expansion to remove the longest suffix matching the pattern *a*. If $- contains a then this expansion produces and empty string which we can test with -z or -n.

subtle bug if no options are set, so the comparison "$-" = "${-%%a*}" will check that the expansion changed the string. allexport is set if the two strings differ. And even % will work as we don't need a maximal match and can remove the leading * from our pattern match.

if [ "$-" = "${-%a*}" ]; then
    # allexport is not set
    set -a
    . ./.env
    set +a
else
    . ./.env
fi

When the values have newline chars \n, spaces or quotes, it can get messy.

After a lot of trial and error, I ended up with a variation of what @bergkvist proposed in https://gist.github.com/mihow/9c7f559807069a03e302605691f85572?permalink_comment_id=4245050#gistcomment-4245050 (thank you very much!).

ENV_VARS="$(cat .env | awk '!/^\s*#/' | awk '!/^\s*$/')"

eval "$(
  printf '%s\n' "$ENV_VARS" | while IFS='' read -r line; do
    key=$(printf '%s\n' "$line"| sed 's/"/\\"/g' | cut -d '=' -f 1)
    value=$(printf '%s\n' "$line" | cut -d '=' -f 2- | sed 's/"/\\\"/g')
    printf '%s\n' "export $key=\"$value\""
  done
)"

env $(cat .env)
this does not work for me but this one works

env $(cat .env|xargs) CMD

my .env has some special value such as FOO='VPTO&wH7$^3ZHZX$o$udY4&i'
@NatoBoram

A simple solution that works for bash, zsh, and fish:

eval export $(cat .env)

Use this to create the file

export -p > .env

and just

. .env

to read it back in

From man export:

The shell shall format the output, including the proper use of quoting, so that it is suitable for reinput to the shell as commands that achieve the same exporting results

Although set -a; source .env; set +a is elegant and short, one feature which I missed is this overwrite existing exported variables.
I my use case I have a script, which connects to postgres with a predefined user. This user is stored in .env file as PG_USER=myuser. So the script does the magical set -a; source .env; set +a and everything works. But sometimes I need ad-hoc change the user. So what I'd do is PG_USER=postgres ./my_script.sh. In order not to over write the existing var I did this horrendous piece of code:

IFS=$'\n'
for l in $(cat /etc/my_service/.env); do
    IFS='=' read -ra VARVAL <<< "$l"
    # If variable with such name already exists, preserves it's value
    eval "export ${VARVAL[0]}=\${${VARVAL[0]}:-${VARVAL[1]}}"
done
unset IFS

The cleanest solution I found for this was using allexport and source like this

set -o allexport
source .env set
+o allexport

This was by far the best solution here for me, removed all the complexity around certain chars, spaces comments etc. Just needed a tweak on formatting to prevent others being tripped up, should be:

set -o allexport source .env set +o allexport

work like a charm. ty

this read line by line, allowing to use previous set variables

  while read -r LINE; do
    if [[ $LINE == *'='* ]] && [[ $LINE != '#'* ]]; then
      ENV_VAR="$(echo $LINE | envsubst)"
      eval "declare $ENV_VAR"
    fi
  done < .env

this was working the best for me but this still has 2 problems

1. code breaks if the value has () characters inside it
2. can not be used inside a function

here is my solution:

read_env() {
  local filename="${1:-.env}"

  if [ ! -f "$filename" ]; then
    echo "missing ${filename} file"
    exit 1
  fi

  echo "reading .env file..."
  while read -r LINE; do
    if [[ $LINE != '#'* ]] && [[ $LINE == *'='* ]]; then
      export "$LINE"
    fi
  done < "$filename"
}

UPDATED VERSION BELOW

Hi, here is mine solution to read vars from /etc/environemnt file which I used in /etc/profile or /etc/bash.bashrc

oneliner for easier validation if line exists in file
(I removed single quotes ' from conditions, so it could be easier parsed by grep)

while read -r LINE; do [[ ${LINE} =~ ^# || ${LINE} =~ ^PATH= || ! ${LINE} == *=* || ${LINE} =~ ^[0-9] || ${LINE} =~ ^[^a-zA-Z_] ]] || export "${LINE}"; done < "/etc/environment"

or formatad syntax:

while read -r LINE; do 
  if [[ ${LINE} =~ ^# || ${LINE} =~ ^PATH= || ! ${LINE} == *=* || ${LINE} =~ ^[0-9] || ${LINE} =~ ^[^a-zA-Z_] ]]; then
    continue
  else
    export "${LINE}"
  fi
done < "/etc/environment"

Hi,

Here is my solution to read vars from .env file and ignoring # and cleaning values from ' and ".

https://gist.github.com/shadiabuhilal/220aa09f9bb83caed93a1f87401fcc60

dot-env.sh File:

#!/bin/bash

# Specify the path to your .env file
ENV_FILE=".env"

# Check if the .env file exists
if [ -f "$ENV_FILE" ]; then

  echo "[INFO]: Reading $ENV_FILE file."

  # Read the .env file line by line
  while IFS= read -r line; do
    # Skip comments and empty lines
    if [[ "$line" =~ ^\s*#.*$ || -z "$line" ]]; then
      continue
    fi

    # Split the line into key and value
    key=$(echo "$line" | cut -d '=' -f 1)
    value=$(echo "$line" | cut -d '=' -f 2-)

    # Remove single quotes, double quotes, and leading/trailing spaces from the value
    value=$(echo "$value" | sed -e "s/^'//" -e "s/'$//" -e 's/^"//' -e 's/"$//' -e 's/^[ \t]*//;s/[ \t]*$//')

    # Export the key and value as environment variables
    export "$key=$value"
  done < "$ENV_FILE"
  echo "[DONE]: Reading $ENV_FILE file."
else
  echo "[ERROR]: $ENV_FILE not found."
fi

Enjoy :)

Thanks!
One-liner while running command a script (such as pnpm script)
I added parentheses not to pollute global environment vars. Not sure if it's needed though.

(export $(cat .env | xargs) && pnpm compile)

source .env

works for me

wont work if you have # in your .env

@abhidp it seems, that it works well - even with comments in .env! :D (cc: @muthugit)

this is the final version that im using, seems to work for all situations

read_env() {
  local filePath="${1:-.env}"

  if [ ! -f "$filePath" ]; then
    echo "missing ${filePath}"
    exit 1
  fi

  echo "Reading $filePath"
  while read -r LINE; do
    # Remove leading and trailing whitespaces, and carriage return
    CLEANED_LINE=$(echo "$LINE" | awk '{$1=$1};1' | tr -d '\r')

    if [[ $CLEANED_LINE != '#'* ]] && [[ $CLEANED_LINE == *'='* ]]; then
      export "$CLEANED_LINE"
    fi
  done < "$filePath"
}

Thanks @MansourM !

this is the final version that im using, seems to work for all situations

read_env() {
  local filePath="${1:-.env}"

  if [ ! -f "$filePath" ]; then
    echo "missing ${filePath}"
    exit 1
  fi

  log "Reading $filePath"
  while read -r LINE; do
    # Remove leading and trailing whitespaces, and carriage return
    CLEANED_LINE=$(echo "$LINE" | awk '{$1=$1};1' | tr -d '\r')

    if [[ $CLEANED_LINE != '#'* ]] && [[ $CLEANED_LINE == *'='* ]]; then
      export "$CLEANED_LINE"
    fi
  done < "$filePath"
}

Looks great, but it doesn't work if the .env file contains only 1 row (just one without br)

Looks great, but it doesn't work if the .env file contains only 1 row (just one without br)

how do you use it?
you get any errors?
btw u need to comment or remove this line as you don't have the log function
log "Reading $filePath"

I needed this and after reading the above realized none of the solutions quite worked for my .env file on a Mac... so I modified what @shadiabuhilal into this (which I put into a file called "readenv")

# quick bash function to read .env file
# use it via:
# source readenv
# readenv
#
# or
#
# readenv <filename>
#
# modified from https://gist.github.com/mihow/9c7f559807069a03e302605691f85572
# fixed for whitespace issues, posix compliance (e.g. \t on mac means t)
#
# NOT a standalone script as when used as a standalone script, it'll read in the ENV variables into a sub-process, not the
# calling process

readenv() {
  local filePath="${1:-.env}"

  if [ ! -f "$filePath" ]; then
    # silently be done
    # put some error / echo if you prefer non-silent errors
    return 0
  fi

#  echo "Reading $filePath"
  while read -r line; do
    if [[ "$line" =~ ^\s*#.*$ || -z "$line" ]]; then
      continue
    fi

     # Split the line into key and value. Trim whitespace on either side.
    key=$(echo "$line" | cut -d '=' -f 1 | sed -e 's/^[[:space:]]*//' -e 's/[[:space:]]*$//')
    value=$(echo "$line" | cut -d '=' -f 2- | sed -e 's/^[[:space:]]*//' -e 's/[[:space:]]*$//')

    # Leaving the below here... normally this works, but if you have something like
    # FOO="  string with leading and trailing  "
    # then the leading / trailing spaces are deleted. FOO="a word", FOO='a word', and FOO=a word all generally work
    # so leave the quotes
    # Remove single quotes, double quotes, and leading/trailing spaces from the value
    # value=$(echo "$value" | sed -e "s/^'//" -e "s/'$//" -e 's/^"//' -e 's/"$//' -e 's/^[[:space:]]*//;s/[[:space:]]*$//')

    # Export the key and value as environment variables
    # echo "$key=$value"
    export "$key=$value"

  done < "$filePath"
}

Also - recommend to use [[:space:]] rather than \s or [ \t] --- on Macs, \s isn't space, and \t isn't TAB but t. Yay standardization!

This is what I use:

# shellcheck disable=SC2046
[ -f .env ] && export $(grep -v '^#' .env | xargs)

No need to do a double negation with [ ! -f .env ] || when you can do [ -f .env ] &&

@bfontaine thanks, worked like a charm.

this is the final version that im using, seems to work for all situations

read_env() {
  local filePath="${1:-.env}"

  if [ ! -f "$filePath" ]; then
    echo "missing ${filePath}"
    exit 1
  fi

  echo "Reading $filePath"
  while read -r LINE; do
    # Remove leading and trailing whitespaces, and carriage return
    CLEANED_LINE=$(echo "$LINE" | awk '{$1=$1};1' | tr -d '\r')

    if [[ $CLEANED_LINE != '#'* ]] && [[ $CLEANED_LINE == *'='* ]]; then
      export "$CLEANED_LINE"
    fi
  done < "$filePath"
}

working for my case: Windows + WSL + bash + 7z password from env + export from .env

Thanks @MansourM !

this is the final version that im using, seems to work for all situations

read_env() {
  local filePath="${1:-.env}"

  if [ ! -f "$filePath" ]; then
    echo "missing ${filePath}"
    exit 1
  fi

  log "Reading $filePath"
  while read -r LINE; do
    # Remove leading and trailing whitespaces, and carriage return
    CLEANED_LINE=$(echo "$LINE" | awk '{$1=$1};1' | tr -d '\r')

    if [[ $CLEANED_LINE != '#'* ]] && [[ $CLEANED_LINE == *'='* ]]; then
      export "$CLEANED_LINE"
    fi
  done < "$filePath"
}

Looks great, but it doesn't work if the .env file contains only 1 row (just one without br)

You can adjust the while loop like this:

read_env() {
  local filePath="${1:-.env}"

  if [ ! -f "$filePath" ]; then
    echo "missing ${filePath}"
    exit 1
  fi

  echo "Reading $filePath"
-  while IFS= read -r LINE; do
+  while IFS= read -r LINE || [ -n "$LINE" ]; do
    # Remove leading and trailing whitespaces, and carriage return
    CLEANED_LINE=$(echo "$LINE" | awk '{$1=$1};1' | tr -d '\r')

    if [[ $CLEANED_LINE != '#'* ]] && [[ $CLEANED_LINE == *'='* ]]; then
      export "$CLEANED_LINE"
    fi
  done < "$filePath"
}