[Bug] Inline comment is included as value when key has empty value (`KEY= # comment`)

[Turtle-Hwan]

Summary

When a key has an empty value followed by an inline comment, the comment text becomes the value instead of being stripped.

Reproduction

from dotenv import dotenv_values

with open('.env.test', 'w') as f:
    f.write('''
WITH_VALUE=hello  # comment
EMPTY_VALUE=  # comment
JUST_EMPTY=
''')

values = dotenv_values('.env.test')

print("WITH_VALUE:", repr(values.get('WITH_VALUE')))
print("EMPTY_VALUE:", repr(values.get('EMPTY_VALUE')))
print("JUST_EMPTY:", repr(values.get('JUST_EMPTY')))

Output:

WITH_VALUE: 'hello'
EMPTY_VALUE: '# comment'
JUST_EMPTY: ''

Expected vs Actual

Input	Expected	Actual
KEY=value # comment	'value'	'value'
KEY= # comment	''	'# comment'
KEY=	''	''

Question

Is this the intended behavior? It seems inconsistent that inline comments are stripped when there's a value, but included when the value is empty.

Workaround

Use quoted empty values:

KEY=""  # comment

This correctly parses to ''.

Environment

• python-dotenv version: 1.2.1
• Python version: 3.11

---

Volunteering

If this is considered a bug rather than intended behavior, I'd be good to work on a fix and submit a PR.

[enzoxy]

Hi!

According to the documentation, a variable can have no associated value; in this case, the syntax is to write only the variable name (without the =).

https://saurabh-kumar.com/python-dotenv/#variable-without-a-value

However, a variable that does have an associated value, but is empty, is represented by writing VAR_NAME= or VAR_NAME=""; in both cases, the value is parsed as an empty string.

You could try something like this: VAR_NAME # comment; maybe it will work as you expect.

[Turtle-Hwan]

@enzonoba Thanks for comment with my issue!!

I tried myself is as follow code.

#!/usr/bin/env python3

import os
from dotenv import load_dotenv

load_dotenv()
print(os.getenv("VAR1"))
print(os.getenv("VAR2"))
print(os.getenv("VAR3"))
print(os.getenv("VAR4"))
print(os.getenv("VAR5"))

# .env file
VAR1
VAR2 # comment
VAR3=# test
VAR4="" # test
VAR5="value" # comment

# result of excute code
% python ./test.py
None
None
# test

value

You are right, but if there is no =, it is not recognized as a variable, but it is treated as a variable that does not exist at all, and None comes out.

I want to annotate it (as it is recognized as a variable) while giving an explicit empty string(""), but I think it's a different situation than None.

And it seems to be different from javascript's env behavior, so it left an issue.

1. VAR= # comment parsed well to "" in javascript & bash
2. VAR= # comment appears to be the color of annotation in syntax highlighters.

• VAR= # comment color is same of comments syntax color.

So, I think the annotation parsing code should be added in the same way as VAR= #comment.
What do you think about this?

[enzoxy]

Hi @Turtle-Hwan ! Sorry for the delay in responding.

Now I got it. It makes sense that VAR= # comment has to be parsed to value "".
And VAR=# comment to # comment.

I see your PR well👌🏼.
Another possibility would be to do this, in order to avoid modifying the regular expression of equal:

• Move line 157 to top of parse_value() function, and save its returned value (e.g. equal_match variable).
• Then, in the last elif statement, add an OR condition, so that "" is also returned if peek char is # and length of matched string is greater than 1 (e.g. "= ").

Maybe this could work too.
I haven't had time to test it, but if you want, try it and let me know how it goes!