## Summary
Our current mechanism for handling deletions (e.g., `del x`) is to
remove the symbol from the scope's `bindings` table. This "does the
right thing", in that if we then reference a deleted symbol, we're able
to determine that it's unbound -- but it causes a variety of problems,
mostly in that it makes certain bindings and references unreachable
after-the-fact.
Consider:
```python
x = 1
print(x)
del x
```
If we analyze this code _after_ running the semantic model over the AST,
we'll have no way of knowing that `x` was ever introduced in the scope,
much less that it was bound to a value, read, and then deleted --
because we effectively erased `x` from the model entirely when we hit
the deletion.
In practice, this will make it impossible for us to support local symbol
renames. It also means that certain rules that we want to move out of
the model-building phase and into the "check dead scopes" phase wouldn't
work today, since we'll have lost important information about the source
code.
This PR introduces two new `BindingKind` variants to model deletions:
- `BindingKind::Deletion`, which represents `x = 1; del x`.
- `BindingKind::UnboundException`, which represents:
```python
try:
1 / 0
except Exception as e:
pass
```
In the latter case, `e` gets unbound after the exception handler
(assuming it's triggered), so we want to handle it similarly to a
deletion.
The main challenge here is auditing all of our existing `Binding` and
`Scope` usages to understand whether they need to accommodate deletions
or otherwise behave differently. If you look one commit back on this
branch, you'll see that the code is littered with `NOTE(charlie)`
comments that describe the reasoning behind changing (or not) each of
those call sites. I've also augmented our test suite in preparation for
this change over a few prior PRs.
### Alternatives
As an alternative, I considered introducing a flag to `BindingFlags`,
like `BindingFlags::UNBOUND`, and setting that at the appropriate time.
This turned out to be a much more difficult change, because we tend to
match on `BindingKind` all over the place (e.g., we have a bunch of code
blocks that only run when a `BindingKind` is
`BindingKind::Importation`). As a result, introducing these new
`BindingKind` variants requires only a few changes at the client sites.
Adding a flag would've required a much wider-reaching change.
## Summary
This PR (1) avoids flagging `TypedDict` and `NamedTuple` conversions
when attributes are dunder methods, like `__dict__`, and (2) avoids
flagging the `A003` shadowed-attribute rule for `TypedDict` classes at
all, where it doesn't really apply (since those attributes are only
accessed via subscripting anyway).
Closes#5027.
Improves the `ruff_parse_simple` fuzz harness by adding checks for
parsed locations to ensure they all lie on UTF-8 character boundaries.
This will allow for faster identification of issues like #5004.
This also adds additional details for Apple M1 users and clarifies the
importance of using `init-fuzzer.sh` (thanks for the feedback,
@jasikpark 🙂).
## Summary
The `RET504` rule, which looks for unnecessary assignments before return
statements, is a frequent source of issues (#4173, #4236, #4242, #1606,
#2950). Over time, we've tried to refine the logic to handle more cases.
For example, we now avoid analyzing any functions that contain any
function calls or attribute assignments, since those operations can
contain side effects (and so we mark them as a "read" on all variables
in the function -- we could do a better job with code graph analysis to
handle this limitation, but that'd be a more involved change.) We also
avoid flagging any variables that are the target of multiple
assignments. Ultimately, though, I'm not happy with the implementation
-- we just can't do sufficiently reliable analysis of arbitrary code
flow given the limited logic herein, and the existing logic is very hard
to reason about and maintain.
This PR refocuses the rule to only catch cases of the form:
```py
def f():
x = 1
return x
```
That is, we now only flag returns that are immediately preceded by an
assignment to the returned variable. While this is more limiting, in
some ways, it lets us flag more cases vis-a-vis the previous
implementation, since we no longer "fully eject" when functions contain
function calls and other effect-ful operations.
Closes#4173.
Closes#4236.
Closes#4242.
## Summary
We use `.trim()` and friends in a bunch of places, to strip whitespace
from source code. However, not all Unicode whitespace characters are
considered "whitespace" in Python, which only supports the standard
space, tab, and form-feed characters.
This PR audits our usages of `.trim()`, `.trim_start()`, `.trim_end()`,
and `char::is_whitespace`, and replaces them as appropriate with a new
`.trim_whitespace()` analogues, powered by a `PythonWhitespace` trait.
In general, the only place that should continue to use `.trim()` is
content within docstrings, which don't need to adhere to Python's
semantic definitions of whitespace.
Closes#4991.
## Summary
`ruff_newlines` becomes `ruff_python_whitespace`, and includes the
existing "universal newline" handlers alongside the Python
whitespace-specific utilities.
