Picky Parentheses

Picky Parentheses is a flake8 plugin that nitpicks all things parentheses, brackets and braces. The plugin has two components:

1. A checker that warns about redundant parentheses (with some exceptions).
2. A checker for parentheses, brackets, and braces alignment. This component is very opinionated but has its own error codes so you can easily disable it.

Table of Contents

• Installation and Usage
• Error Codes
• Details and Exceptions
• Additional Notes

Installation and Usage

This is a plugin for flake8. It supports Python 3.7 - 3.12.
Refer to the documentation of flake8 on how to run it on your code: https://flake8.pycqa.org/en/latest/

Two common options are to either install the plugin and then run flake8:

pip install flake8-picky-parentheses

flake8 '<path/to/your/code>'

Or to let flake8 fetch the plugin for you (requires flake8 >= 5.0.0):

flake8 --require-plugins flake8-picky-parentheses '<path/to/your/code>'

If you only want to run this plugin and bypass any other flake8 checks, you can use the --select option:

flake8 [other options] --select='PAR0,PAR1' '<path/to/your/code>'

Where PAR0 is the code for the redundant parentheses checker and PAR1 is the code for the parentheses alignment checker.

If you, in turn want to disable the opinionated parentheses alignment checker, you can use the --ignore or --extend-ignore option:

flake8 [other options] --extend-ignore='PAR1' '<path/to/your/code>'

Error Codes

These are the error codes which you can get using this plugin:

PAR0xx

These are the error codes for the redundant parentheses checker.

PAR001

It means that you use redundant parentheses, and they do not help readability. For example:

# BAD
a = (("a", "b"))

PAR002

It means that you use parentheses for an unpacking expression. For example:

# BAD
(a,) = "b"

PAR1xx

These are the error codes for the opinionated alignment checker.

PAR101

It means that the opening bracket is last in its line, but closing one is not on a new line. For example:

# BAD
if (
        a == b):
    c + d

# GOOD
if (
    a == b
):
    c + d

# BAD
a = [
    1, 2,
    3, 4]

# GOOD
a = [
    1, 2,
    3, 4
]

# GOOD
a = [1, 2,
     3, 4]

PAR102

It means that closing bracket is on new line, but there is a indentation mismatch. For example:

# BAD
if (
    a == b
        ):
    c + d

# GOOD
if (
    a == b
):
    c + d

# BAD
a = [
    1, 2,
    3, 4
    ]

# GOOD
a = [
    1, 2,
    3, 4
]

PAR103

It means that consecutive opening brackets at the end of a line must have consecutive closing brackets.

# BAD
answer = func((
    1, 2, 3, 4, 5,
    )
)

# GOOD
answer = func((
    1, 2, 3, 4, 5,
))

PAR104

Only operators and comments are allowed after a closing bracket on a new line.

# BAD
a = func(
    1, 2, 3, 4, 5
) + 6

# GOOD
a = (
    func(
        1, 2, 3, 4, 5
    )
    + 6
)

Details and Exceptions

The redundant parentheses checker uses Python's tokenize and ast module to try to remove each pair of parentheses and see if the code still compiles and yields the same AST (i.e., is semantically equivalent). If it does, a flake (lint error) is reported. However, there are two notable exceptions to this rule:

1. Parentheses for tuple literals.
2. A single pair or parentheses in expressions to highlight operator precedence. Even if these parentheses are redundant, they help to divide parts of expressions and show sequence of actions.
3. Parts of slices.
4. Multi-line¹⁾ expression, if and for parts in comprehensions.
5. Multi-line¹⁾ keyword arguments or argument defaults.
6. String concatenation over several lines in lists and tuples .

Exception type 1:

a = ("a",)     # GOOD
a = "a",       # GOOD
a = ("a")      # BAD
a = (("a"),)   # BAD
a = (("a",))   # BAD
foo(("a",))    # GOOD
foo("a",)      # BAD

Exception type 2:

a = (1 + 2) + 3            # GOOD
a = (1 + 2) % 3            # GOOD
a = 1 and (2 + 3)          # GOOD
a = (1 / 2) * 3            # GOOD
a = not (1 + 2)            # GOOD
a = (not 1) + 2            # GOOD
a = 1 + (2 if a else 3)    # GOOD
a = foo(*(a if b else c))  # GOOD
a = foo(*(a + b))          # GOOD
a = foo(**(a + b))         # GOOD
a = (1 + 2)                # BAD
a = 1 + (2)                # BAD
a = ((not 1)) + 2          # BAD
a = foo(*(a))              # BAD
a = foo(**(a))             # BAD

Exception type 3:

foo[(1 + 2):10]    # GOOD
foo[1:(1 + 2)]     # GOOD
foo[1:5:(1 + 1)]   # GOOD
foo[:(-bar)]       # GOOD
foo[(1):]          # BAD
foo[:(1)]          # BAD

Exception type 4:

# GOOD
a = (
    b for b in c
    if (
        some_thing == other_thing
        or whatever_but_long
    )
)

# GOOD
a = [
    b for b in c
    if (b
        in d)
]

# BAD
a = (
    b for b in c
    if (b in d)
)

# GOOD
a = (
    b for b in (c
                + d)
)

# BAD
a = (
    b for b in (c + d)
)

# GOOD
a = (
    (
        1
        + b
    )
    for b in c
)

# BAD
a = (
    (1 + b) for b in c
)

# GOOD
a = {
    (
        "foo%s"
        % b
    ): (
        b
        * 2
    )
    for b in c
}

Exception type 5:

# GOOD
foo(bar=(a
         in b))

# BAD
foo(bar=(a in b))

# GOOD
def foo(bar=(a
             is b)):
    ...

# BAD
def foo(bar=(a is b)):
    ...

Exception type 6:

# GOOD
[
    "a",
    (
        "b"
        "c"
    ),
    "d",
]

# This helps to avoid forgetting a comma at the end of a string spanning
# multiple lines. Compare with:
[
    "a",
    "b"
    "c"
    "d",
]
# Was the comma after "b" forgotten or was the string supposed to be "bc"?

# BAD
[
    (
        "a" "b"
    ),
]

Footnotes:

1. Multi-line means that either
   • the expression spans multiple lines, e.g.,

     (a
      + b)
     

   • or the first part of the expression is on a new line (e.g., if a name is very long), e.g.,

     (
         veeeeeeeeeeery_looooooong_name
     )
     

     but also

     (
         a
         + b
     )
     

   Multi-line expressions do not include

   (a + b
   )
   

Additional Notes

This plugin was developed to improve the code quality of Neo4j Python projects.