## Summary
fix#13602
Currently, `UP043` only applies to typing.Generator, but it should also
support collections.abc.Generator.
This update ensures `UP043` correctly handles both
`collections.abc.Generator` and `collections.abc.AsyncGenerator`
### UP043
> `UP043`
> Python 3.13 introduced the ability for type parameters to specify
default values. As such, the default type arguments for some types in
the standard library (e.g., Generator, AsyncGenerator) are now optional.
> Omitting type parameters that match the default values can make the
code more concise and easier to read.
```py
Generator[int, None, None] -> Generator[int]
```
Closes https://github.com/astral-sh/ruff/issues/13266
Avoids false negatives for shadowed bindings that aren't actually
references to the loop variable. There are some shadowed bindings we
need to support still, e.g., `del` requires the loop variable to exist.
## Summary
Add new rule and implement for `unnecessary default type arguments`
under the `UP` category (`UP043`).
```py
// < py313
Generator[int, None, None]
// >= py313
Generator[int]
```
I think that as Python 3.13 develops, there might be more default type
arguments added besides `Generator` and `AsyncGenerator`. So, I made
this more flexible to accommodate future changes.
related issue: #12286
## Test Plan
snapshot included..!
## Summary
This PR fixes a bug where the checker would require the tokens for an
invalid offset w.r.t. the source code.
Taking the source code from the linked issue as an example:
```py
relese_version :"0.0is 64"
```
Now, this isn't really a valid type annotation but that's what this PR
is fixing. Regardless of whether it's valid or not, Ruff shouldn't
panic.
The checker would visit the parsed type annotation (`0.0is 64`) and try
to detect any violations. Certain rule logic requests the tokens for the
same but it would fail because the lexer would only have the `String`
token considering original source code. This worked before because the
lexer was invoked again for each rule logic.
The solution is to store the parsed type annotation on the checker if
it's in a typing context and use the tokens from that instead if it's
available. This is enforced by creating a new API on the checker to get
the tokens.
But, this means that there are two ways to get the tokens via the
checker API. I want to restrict this in a follow-up PR (#11741) to only
expose `tokens` and `comment_ranges` as methods and restrict access to
the parsed source code.
fixes: #11736
## Test Plan
- [x] Add a test case for `F632` rule and update the snapshot
- [x] Check all affected rules
- [x] No ecosystem changes
## Summary
In an `__init__.py` file, it's not uncommon to lack a logical indent
(since it may just contain imports). In such cases, we were always
falling back to four-space indent. This PR adds detection for indents
within import groups.
Closes https://github.com/astral-sh/ruff/issues/11606.
## Summary
Similar to #11414, this PR extends `UP037` to flag quoted annotations
that are located in positions that won't be evaluated at runtime.
For example, the quotes on `Tuple` are unnecessary in:
```python
from typing import TYPE_CHECKING
if TYPE_CHECKING:
from typing import Tuple
def foo():
x: "Tuple[int, int]" = (0, 0)
foo()
```
Resolves#10187
<details>
<summary>Old PR description; accurate through commit e86dd7d; probably
best to leave this fold closed</summary>
## Description of change
In the case of a printf-style format string with only one %-placeholder
and a variable at right (e.g. `"%s" % var`):
* The new behavior attempts to dereference the variable and then match
on the bound expression to distinguish between a 1-tuple (fix), n-tuple
(bug 🐛), or a non-tuple (fix). Dereferencing is via
`analyze::typing::find_binding_value`.
* If the variable cannot be dereferenced, then the type-analysis routine
is called to distinguish only tuple (no-fix) or non-tuple (fix). Type
analysis is via `analyze::typing::is_tuple`.
* If any of the above fails, the rule still fires, but no fix is
offered.
## Alternatives
* If the reviewers think that singling out the 1-tuple case is too
complicated, I will remove that.
* The ecosystem results show that no new fixes are detected. So I could
probably delete all the variable dereferencing code and code that tries
to generate fixes, tbh.
## Changes to existing behavior
**All the previous rule-firings and fixes are unchanged except for** the
"false negatives" in
`crates/ruff_linter/resources/test/fixtures/pyupgrade/UP031_1.py`. Those
previous "false negatives" are now true positives and so I moved them to
`crates/ruff_linter/resources/test/fixtures/pyupgrade/UP031_0.py`.
<details>
<summary>Existing false negatives that are now true positives</summary>
```
crates/ruff_linter/resources/test/fixtures/pyupgrade/UP031_0.py:134:1: UP031 Use format specifiers instead of percent format
|
133 | # UP031 (no longer false negatives)
134 | 'Hello %s' % bar
| ^^^^^^^^^^^^^^^^ UP031
135 |
136 | 'Hello %s' % bar.baz
|
= help: Replace with format specifiers
crates/ruff_linter/resources/test/fixtures/pyupgrade/UP031_0.py:136:1: UP031 Use format specifiers instead of percent format
|
134 | 'Hello %s' % bar
135 |
136 | 'Hello %s' % bar.baz
| ^^^^^^^^^^^^^^^^^^^^ UP031
137 |
138 | 'Hello %s' % bar['bop']
|
= help: Replace with format specifiers
crates/ruff_linter/resources/test/fixtures/pyupgrade/UP031_0.py:138:1: UP031 Use format specifiers instead of percent format
|
136 | 'Hello %s' % bar.baz
137 |
138 | 'Hello %s' % bar['bop']
| ^^^^^^^^^^^^^^^^^^^^^^^ UP031
|
= help: Replace with format specifiers
```
One of them newly offers a fix.
```
# UP031 (no longer false negatives)
-'Hello %s' % bar
+'Hello {}'.format(bar)
```
This fix occurs because the new code dereferences `bar` to where it was
defined earlier in the file as a non-tuple:
```python
bar = {"bar": y}
```
---
</details>
## Behavior requiring new tests
Additionally, we now handle a few cases that we didn't previously test.
These cases are when a string has a single %-placeholder and the
righthand operand to the modulo operator is a variable **which can be
dereferenced.** One of those was shown in the previous section (the
"dereference non-tuple" case).
<details>
<summary>New cases handled</summary>
```
crates/ruff_linter/resources/test/fixtures/pyupgrade/UP031_0.py:126:1: UP031 [*] Use format specifiers instead of percent format
|
125 | t1 = (x,)
126 | "%s" % t1
| ^^^^^^^^^ UP031
127 | # UP031: deref t1 to 1-tuple, offer fix
|
= help: Replace with format specifiers
crates/ruff_linter/resources/test/fixtures/pyupgrade/UP031_0.py:130:1: UP031 Use format specifiers instead of percent format
|
129 | t2 = (x,y)
130 | "%s" % t2
| ^^^^^^^^^ UP031
131 | # UP031: deref t2 to n-tuple, this is a bug
|
= help: Replace with format specifiers
```
One of these offers a fix.
```
t1 = (x,)
-"%s" % t1
+"{}".format(t1[0])
# UP031: deref t1 to 1-tuple, offer fix
```
The other doesn't offer a fix because it's a bug.
---
</details>
---
</details>
## Changes to existing behavior
In the case of a string with a single %-placeholder and a single
ambiguous righthand argument to the modulo operator, (e.g. `"%s" % var`)
the rule now fires and offers a fix. We explain about this in the "fix
safety" section of the updated documentation.
## Documentation changes
I swapped the order of the "known problems" and the "examples" sections
so that the examples which describe the rule are first, before the
exceptions to the rule are described. I also tweaked the language to be
more explicit, as I had trouble understanding the documentation at
first. The "known problems" section is now "fix safety" but the content
is largely similar.
The diff of the documentation changes looks a little difficult unless
you look at the individual commits.
## Summary
Add new rule `pyupgrade - UP042` (I picked next available number).
Closes https://github.com/astral-sh/ruff/discussions/3867
Closes https://github.com/astral-sh/ruff/issues/9569
It should warn + provide a fix `class A(str, Enum)` -> `class
A(StrEnum)` for py311+.
## Test Plan
Added UP042.py test.
## Notes
I did not find a way to call `remove_argument` 2 times consecutively, so
the automatic fixing works only for classes that inherit exactly `str,
Enum` (regardless of the order).
I also plan to extend this rule to support IntEnum in next PR.
## Summary
We're seeing failures in https://github.com/astral-sh/ruff/issues/10470
because `resolve_qualified_import_name` isn't guaranteed to return a
specific import if a symbol is accessible in two ways (e.g., you have
both `import logging` and `from logging import error` in scope, and you
want `logging.error`). This PR breaks up the failing tests such that the
imports aren't in the same scope.
Closes https://github.com/astral-sh/ruff/issues/10470.
## Test Plan
I added a `bindings.reverse()` to `resolve_qualified_import_name` to
ensure that the tests pass regardless of the binding order.
## Summary
Given a format string like `"{x} {x}".format(x=foo())`, we should avoid
converting to an f-string, since doing so would require repeating the
function call (`f"{foo()} {foo()}"`), which could introduce side
effects.
Closes https://github.com/astral-sh/ruff/issues/10258.
## Summary
If a generic appears multiple times on the right-hand side, we should
only include it once on the left-hand side when rewriting.
Closes https://github.com/astral-sh/ruff/issues/9904.
## Summary
Often, when fixing, we need to dedent a block of code (e.g., if we
remove an `if` and dedent its body). Today, we use LibCST to parse and
adjust the indentation, which is really expensive -- but this is only
really necessary if the block contains a multiline string, since naively
adjusting the indentation for such a string can change the whitespace
_within_ the string.
This PR uses a simple dedent implementation for cases in which the block
doesn't intersect with a multi-line string (or an f-string, since we
don't support tracking multi-line strings for f-strings right now).
We could improve this even further by using the ranges to guide the
dedent function, such that we don't apply the dedent if the line starts
within a multiline string. But that would also need to take f-strings
into account, which is a little tricky.
## Test Plan
`cargo test`
## Summary
It seems like the range of an `ExprStringLiteral` can be somewhat
unreliable when the string is part of an implicit concatenation with an
f-string. Using the tokens themselves is more reliable.
Closes#8680.
Closes https://github.com/astral-sh/ruff/issues/7784.
## Summary
This PR resolves an issue raised in
https://github.com/astral-sh/ruff/discussions/7810, whereby we don't fix
an f-string that exceeds the line length _even if_ the resultant code is
_shorter_ than the current code.
As part of this change, I've also refactored and extracted some common
logic we use around "ensuring a fix isn't breaking the line length
rules".
## Test Plan
`cargo test`
**Summary** Check that `closefd` and `opener` aren't being used with
`builtin.open()` before suggesting `Path.open()` because pathlib doesn't
support these arguments.
Closes#7620
**Test Plan** New cases in the fixture.
## Summary
This is a follow-up to #7469 that attempts to achieve similar gains, but
without introducing malachite. Instead, this PR removes the `BigInt`
type altogether, instead opting for a simple enum that allows us to
store small integers directly and only allocate for values greater than
`i64`:
```rust
/// A Python integer literal. Represents both small (fits in an `i64`) and large integers.
#[derive(Clone, PartialEq, Eq, Hash)]
pub struct Int(Number);
#[derive(Debug, Clone, PartialEq, Eq, Hash)]
pub enum Number {
/// A "small" number that can be represented as an `i64`.
Small(i64),
/// A "large" number that cannot be represented as an `i64`.
Big(Box<str>),
}
impl std::fmt::Display for Number {
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
match self {
Number::Small(value) => write!(f, "{value}"),
Number::Big(value) => write!(f, "{value}"),
}
}
}
```
We typically don't care about numbers greater than `isize` -- our only
uses are comparisons against small constants (like `1`, `2`, `3`, etc.),
so there's no real loss of information, except in one or two rules where
we're now a little more conservative (with the worst-case being that we
don't flag, e.g., an `itertools.pairwise` that uses an extremely large
value for the slice start constant). For simplicity, a few diagnostics
now show a dedicated message when they see integers that are out of the
supported range (e.g., `outdated-version-block`).
An additional benefit here is that we get to remove a few dependencies,
especially `num-bigint`.
## Test Plan
`cargo test`
