Followup on #11168 and resolve#10391
# User facing changes
* F401 now recommends a fix to add unused import bindings to to
`__all__` if a single `__all__` list or tuple is found in `__init__.py`.
* If there are no `__all__` found in the file, fall back to recommending
redundant-aliases.
* If there are multiple `__all__` or only one but of the wrong type (non
list or tuple) then diagnostics are generated without fixes.
* `fix_title` is updated to reflect what the fix/recommendation is.
Subtlety: For a renamed import such as `import foo as bees`, we can
generate a fix to add `bees` to `__all__` but cannot generate a fix to
produce a redundant import (because that would break uses of the binding
`bees`).
# Implementation changes
* Add `name` field to `ImportBinding` to contain the name of the
_binding_ we want to add to `__all__` (important for the `import foo as
bees` case). It previously only contained the `AnyImport` which can give
us information about the import but not the binding.
* Add `binding` field to `UnusedImport` to contain the same. (Naming
note: the field `name` field already existed on `UnusedImport` and
contains the qualified name of the imported symbol/module)
* Change `fix_by_reexporting` to branch on the size of `dunder_all:
Vec<&Expr>`
* For length 0 call the edit-producing function `make_redundant_alias`.
* For length 1 call edit-producing function `add_to_dunder_all`.
* Otherwise, produce no fix.
* Implement the edit-producing function `add_to_dunder_all` and add unit
tests.
* Implement several fixture tests: empty `__all__ = []`, nonempty
`__all__ = ["foo"]`, mis-typed `__all__ = None`, plus-eq `__all__ +=
["foo"]`
* `UnusedImportContext::Init` variant now has two fields: whether the
fix is in `__init__.py` and how many `__all__` were found.
# Other changes
* Remove a spurious pattern match and instead use field lookups b/c the
addition of a field would have required changing the unrelated pattern.
* Tweak input type of `make_redundant_alias`
---------
Co-authored-by: Alex Waygood <Alex.Waygood@Gmail.com>
Resolves#10390 and starts to address #10391
# Changes to behavior
* In `__init__.py` we now offer some fixes for unused imports.
* If the import binding is first-party this PR suggests a fix to turn it
into a redundant alias.
* If the import binding is not first-party, this PR suggests a fix to
remove it from the `__init__.py`.
* The fix-titles are specific to these new suggested fixes.
* `checker.settings.ignore_init_module_imports` setting is
deprecated/ignored. There is probably a documentation change to make
that complete which I haven't done.
---
<details><summary>Old description of implementation changes</summary>
# Changes to the implementation
* In the body of the loop over import statements that contain unused
bindings, the bindings are partitioned into `to_reexport` and
`to_remove` (according to how we want to resolve the fact they're
unused) with the following predicate:
```rust
in_init && is_first_party(checker, &import.qualified_name().to_string())
// true means make it a reexport
```
* Instead of generating a single fix per import statement, we now
generate up to two fixes per import statement:
```rust
(fix_by_removing_imports(checker, node_id, &to_remove, in_init).ok(),
fix_by_reexporting(checker, node_id, &to_reexport, dunder_all).ok())
```
* The `to_remove` fixes are unsafe when `in_init`.
* The `to_explicit` fixes are safe. Currently, until a future PR, we
make them redundant aliases (e.g. `import a` would become `import a as
a`).
## Other changes
* `checker.settings.ignore_init_module_imports` is deprecated/ignored.
Instead, all fixes are gated on `checker.settings.preview.is_enabled()`.
* Got rid of the pattern match on the import-binding bound by the inner
loop because it seemed less readable than referencing fields on the
binding.
* [x] `// FIXME: rename "imports" to "bindings"` if reviewer agrees (see
code)
* [x] `// FIXME: rename "node_id" to "import_statement"` if reviewer
agrees (see code)
<details>
<summary><h2>Scope cut until a future PR</h2></summary>
* (Not implemented) The `to_explicit` fixes will be added to `__all__`
unless it doesn't exist. When `__all__` doesn't exist they're resolved
by converting to redundant aliases (e.g. `import a` would become `import
a as a`).
---
</details>
# Test plan
* [x] `crates/ruff_linter/resources/test/fixtures/pyflakes/F401_24`
contains an `__init__.py` with*out* `__all__` that exercises the
features in this PR, but it doesn't pass.
* [x]
`crates/ruff_linter/resources/test/fixtures/pyflakes/F401_25_dunder_all`
contains an `__init__.py` *with* `__all__` that exercises the features
in this PR, but it doesn't pass.
* [x] Write unit tests for the new edit functions in
`fix::edits::make_redundant_alias`.
</details>
---------
Co-authored-by: Micha Reiser <micha@reiser.io>
## Summary
Fixes#3011.
Type checkers currently allow forward references in all contexts in stub
files, and stubs frequently make use of this capability (although it
doesn't actually seem to be specc'd anywhere --neither in PEP 484, nor
https://typing.readthedocs.io/en/latest/source/stubs.html#id6, nor the
CPython typing docs). Implementing it so that Ruff allows forward
references in _all contexts_ in stub files seems non-trivial, however
(or at least, I couldn't figure out how to do it easily), so this PR
does not do that. Perhaps it _should_; if we think this apporach isn't
principled enough, I'm happy to close it and postpone changing anything
here.
However, this does reduce the number of F821 errors Ruff emits on
typeshed down from 76 to 2, which would mean that we could enable the
rule at typeshed. The remaining 2 F821 errors can be trivially fixed at
typeshed by moving definitions around; forward references in class bases
were really the only remaining places where there was a real _use case_
for forward references in stub files that Ruff wasn't yet allowing.
## Test plan
`cargo test`. I also ran this PR branch on typeshed to check to see if
there were any new false positives caused by the changes here; there
were none.
## Summary
Historically, given:
```python
__all__ = [ # noqa: F822
"Bernoulli",
"Beta",
"Binomial",
]
```
The F822 violations would be attached to the `__all__`, so this `# noqa`
would be enforced for _all_ definitions in the list. This changed in
https://github.com/astral-sh/ruff/pull/10525 for the better, in that we
now use the range of each string. But these `# noqa` directives stopped
working.
This PR sets the `__all__` as a parent range in the diagnostic, so that
these directives are respected once again.
Closes https://github.com/astral-sh/ruff/issues/10795.
## Test Plan
`cargo test`
## Summary
In https://github.com/astral-sh/ruff/pull/10341, we fixed some false
positives in `.pyi` files, but introduced others. This PR effectively
reverts the change in #10341 and fixes it in a slightly different way.
Instead of changing the _bindings_ we generate in the semantic model in
`.pyi` files, we instead change how we _resolve_ them.
Closes https://github.com/astral-sh/ruff/issues/10509.
This PR fixes the following false positive in a `.pyi` stub file:
```py
x: int
y = x # F821 currently emitted here, but shouldn't be in a stub file
```
In a `.py` file, this is invalid regardless of whether `from __future__ import annotations` is enabled or not. In a `.pyi` stub file, however, it's always valid, as an annotation counts as a binding in a stub file even if no value is assigned to the variable.
I also added more test coverage for `.pyi` stub files in various edge cases where ruff's behaviour is currently correct, but where `.pyi` stub files do slightly different things to `.py` files.
## Summary
This PR is a more holistic fix for
https://github.com/astral-sh/ruff/issues/9534 and
https://github.com/astral-sh/ruff/issues/9159.
When we visit the AST, we track nodes that we need to visit _later_
(deferred nodes). For example, when visiting a function, we defer the
function body, since we don't want to visit the body until we've visited
the rest of the statements in the containing scope.
However, deferred nodes can themselves contain deferred nodes... For
example, a function body can contain a lambda (which contains a deferred
body). And then there are rarer cases, like a lambda inside of a type
annotation.
The aforementioned issues were fixed by reordering the deferral visits
to catch common cases. But even with those fixes, we still fail on cases
like:
```python
from __future__ import annotations
import re
from typing import cast
cast(lambda: re.match, 1)
```
Since we don't expect lambdas to appear inside of type definitions.
This PR modifies the `Checker` to keep visiting until all the deferred
stacks are empty. We _already_ do this for any one kind of deferred
node; now, we do it for _all_ of them at a level above.
## Summary
This is effectively the same problem as
https://github.com/astral-sh/ruff/pull/9175. And this just papers over
it again, though I'm gonna try a more holistic fix in a follow-up PR.
The _real_ fix here is that we need to continue to visit deferred items
until they're exhausted since, e.g., we still get this case wrong
(flagging `re` as unused):
```python
import re
cast(lambda: re.match, 1)
```
Closes https://github.com/astral-sh/ruff/issues/9534.
## Summary
On `main`, we flag redefinitions in cases like:
```python
import os
x = 1
if x > 0:
import os
```
That is, we consider these to be in the "same branch", since they're not
in disjoint branches. This matches Flake8's behavior, but it seems to
lead to false positives.
## Summary
Ensures that any lint rules that include line locations render them as
relative to the cell (and include the cell number) when inside a Jupyter
notebook.
Closes https://github.com/astral-sh/ruff/issues/6672.
## Test Plan
`cargo test`
## Summary
`display` is a special-cased builtin in IPython. This PR adds it to the
builtin namespace when analyzing IPython notebooks.
Closes https://github.com/astral-sh/ruff/issues/8702.
<!--
Thank you for contributing to Ruff! To help us out with reviewing,
please consider the following:
- Does this pull request include a summary of the change? (See below.)
- Does this pull request include a descriptive title?
- Does this pull request include references to any relevant issues?
-->
## Summary
In https://github.com/astral-sh/ruff/pull/7968, I introduced a
regression whereby we started to treat imports used _only_ in type
annotation bounds (with `__future__` annotations) as unused.
The root of the issue is that I started using `visit_annotation` for
these bounds. So we'd queue up the bound in the list of deferred type
parameters, then when visiting, we'd further queue it up in the list of
deferred type annotations... Which we'd then never visit, since deferred
type annotations are visited _before_ deferred type parameters.
Anyway, the better solution here is to use a dedicated flag for these,
since they have slightly different behavior than type annotations.
I've also fixed what I _think_ is a bug whereby we previously failed to
resolve `Callable` in:
```python
type RecordCallback[R: Record] = Callable[[R], None]
from collections.abc import Callable
```
IIUC, the values in type aliases should be evaluated lazily, like type
parameters.
Closes https://github.com/astral-sh/ruff/issues/8017.
## Test Plan
`cargo test`
## Summary
Given:
```python
baz: Annotated[
str,
[qux for qux in foo],
]
```
We treat `baz` as `BindingKind::Annotation`, to ensure that references
to `baz` are marked as unbound. However, we were _also_ treating `qux`
as `BindingKind::Annotation`, which meant that the load in the
comprehension _also_ errored.
Closes https://github.com/astral-sh/ruff/issues/7879.
plredmond
Alex Waygood
Jelle Zijlstra
Charlie Marsh
Tom Kuson
Dhruv Manilawala
konsti