## Summary
This was originally stacked on #19129, but some of the changes I made
for JSON also impacted the Azure format, so I went ahead and combined
them. The main changes here are:
- Implementing `FileResolver` for Ruff's `EmitterContext`
- Adding `FileResolver::notebook_index` and `FileResolver::is_notebook`
methods
- Adding a `DisplayDiagnostics` (with an "s") type for rendering a group
of diagnostics at once
- Adding `Azure`, `Json`, and `JsonLines` as new `DiagnosticFormat`s
I tried a couple of alternatives to the `FileResolver::notebook` methods
like passing down the `NotebookIndex` separately and trying to reparse a
`Notebook` from Ruff's `SourceFile`. The latter seemed promising, but
the `SourceFile` only stores the concatenated plain text of the
notebook, not the re-parsable JSON. I guess the current version is just
a variation on passing the `NotebookIndex`, but at least we can reuse
the existing `resolver` argument. I think a lot of this can be cleaned
up once Ruff has its own actual file resolver.
As suggested, I also tried deleting the corresponding `Emitter` files in
`ruff_linter`, but it doesn't look like git was able to follow this as a
rename. It did, however, track that the tests were moved, so the
snapshots should be easy to review.
## Test Plan
Existing Ruff tests ported to tests in `ruff_db`. I think some other
existing ruff tests also cover parts of this refactor.
---------
Co-authored-by: Micha Reiser <micha@reiser.io>
## Summary
Print the [new salsa memory usage
dumps](https://github.com/astral-sh/ruff/pull/18928) in mypy primer CI
runs to help us catch memory regressions. The numbers are rounded to the
nearest power of 1.1 (about a 5% threshold between buckets) to avoid overly sensitive diffs.
## Summary
Setting `TY_MEMORY_REPORT=full` will generate and print a memory usage
report to the CLI after a `ty check` run:
```
=======SALSA STRUCTS=======
`Definition` metadata=7.24MB fields=17.38MB count=181062
`Expression` metadata=4.45MB fields=5.94MB count=92804
`member_lookup_with_policy_::interned_arguments` metadata=1.97MB fields=2.25MB count=35176
...
=======SALSA QUERIES=======
`File -> ty_python_semantic::semantic_index::SemanticIndex`
metadata=11.46MB fields=88.86MB count=1638
`Definition -> ty_python_semantic::types::infer::TypeInference`
metadata=24.52MB fields=86.68MB count=146018
`File -> ruff_db::parsed::ParsedModule`
metadata=0.12MB fields=69.06MB count=1642
...
=======SALSA SUMMARY=======
TOTAL MEMORY USAGE: 577.61MB
struct metadata = 29.00MB
struct fields = 35.68MB
memo metadata = 103.87MB
memo fields = 409.06MB
```
Eventually, we should integrate these numbers into CI in some form. The
one limitation currently is that heap allocations in salsa structs (e.g.
interned values) are not tracked, but memoized values should have full
coverage. We may also want a peak memory usage counter (that accounts
for non-salsa memory), but that is relatively simple to profile manually
(e.g. `time -v ty check`) and would require a compile-time option to
avoid runtime overhead.
## Summary
Format conflicting declared types as
```
`str`, `int` and `bytes`
```
Thanks to @AlexWaygood for the initial draft.
@dcreager, looking forward to your one-character follow-up PR.
## Summary
Having a recursive type method to check whether a type is fully static
is inefficient, unnecessary, and makes us overly strict about subtyping
relations.
It's inefficient because we end up re-walking the same types many times
to check for fully-static-ness.
It's unnecessary because we can check relations involving the dynamic
type appropriately, depending whether the relation is subtyping or
assignability.
We use the subtyping relation to simplify unions and intersections. We
can usefully consider that `S <: T` for gradual types also, as long as
it remains true that `S | T` is equivalent to `T` and `S & T` is
equivalent to `S`.
One conservative definition (implemented here) that satisfies this
requirement is that we consider `S <: T` if, for every possible pair of
materializations `S'` and `T'`, `S' <: T'`. Or put differently the top
materialization of `S` (`S+` -- the union of all possible
materializations of `S`) is a subtype of the bottom materialization of
`T` (`T-` -- the intersection of all possible materializations of `T`).
In the most basic cases we can usefully say that `Any <: object` and
that `Never <: Any`, and we can handle more complex cases inductively
from there.
This definition of subtyping for gradual subtypes is not reflexive
(`Any` is not a subtype of `Any`).
As a corollary, we also remove `is_gradual_equivalent_to` --
`is_equivalent_to` now has the meaning that `is_gradual_equivalent_to`
used to have. If necessary, we could restore an
`is_fully_static_equivalent_to` or similar (which would not do an
`is_fully_static` pre-check of the types, but would instead pass a
relation-kind enum down through a recursive equivalence check, similar
to `has_relation_to`), but so far this doesn't appear to be necessary.
Credit to @JelleZijlstra for the observation that `is_fully_static` is
unnecessary and overly restrictive on subtyping.
There is another possible definition of gradual subtyping: instead of
requiring that `S+ <: T-`, we could instead require that `S+ <: T+` and
`S- <: T-`. In other words, instead of requiring all materializations of
`S` to be a subtype of every materialization of `T`, we just require
that every materialization of `S` be a subtype of _some_ materialization
of `T`, and that every materialization of `T` be a supertype of some
materialization of `S`. This definition also preserves the core
invariant that `S <: T` implies that `S | T = T` and `S & T = S`, and it
restores reflexivity: under this definition, `Any` is a subtype of
`Any`, and for any equivalent types `S` and `T`, `S <: T` and `T <: S`.
But unfortunately, this definition breaks transitivity of subtyping,
because nominal subclasses in Python use assignability ("consistent
subtyping") to define acceptable overrides. This means that we may have
a class `A` with `def method(self) -> Any` and a subtype `B(A)` with
`def method(self) -> int`, since `int` is assignable to `Any`. This
means that if we have a protocol `P` with `def method(self) -> Any`, we
would have `B <: A` (from nominal subtyping) and `A <: P` (`Any` is a
subtype of `Any`), but not `B <: P` (`int` is not a subtype of `Any`).
Breaking transitivity of subtyping is not tenable, so we don't use this
definition of subtyping.
## Test Plan
Existing tests (modified in some cases to account for updated
semantics.)
Stable property tests pass at a million iterations:
`QUICKCHECK_TESTS=1000000 cargo test -p ty_python_semantic -- --ignored
types::property_tests::stable`
### Changes to property test type generation
Since we no longer have a method of categorizing built types as
fully-static or not-fully-static, I had to add a previously-discussed
feature to the property tests so that some tests can build types that
are known by construction to be fully static, because there are still
properties that only apply to fully-static types (for example,
reflexiveness of subtyping.)
## Changes to handling of `*args, **kwargs` signatures
This PR "discovered" that, once we allow non-fully-static types to
participate in subtyping under the above definitions, `(*args: Any,
**kwargs: Any) -> Any` is now a subtype of `() -> object`. This is true,
if we take a literal interpretation of the former signature: all
materializations of the parameters `*args: Any, **kwargs: Any` can
accept zero arguments, making the former signature a subtype of the
latter. But the spec actually says that `*args: Any, **kwargs: Any`
should be interpreted as equivalent to `...`, and that makes a
difference here: `(...) -> Any` is not a subtype of `() -> object`,
because (unlike a literal reading of `(*args: Any, **kwargs: Any)`),
`...` can materialize to _any_ signature, including a signature with
required positional arguments.
This matters for this PR because it makes the "any two types are both
assignable to their union" property test fail if we don't implement the
equivalence to `...`. Because `FunctionType.__call__` has the signature
`(*args: Any, **kwargs: Any) -> Any`, and if we take that at face value
it's a subtype of `() -> object`, making `FunctionType` a subtype of `()
-> object)` -- but then a function with a required argument is also a
subtype of `FunctionType`, but not a subtype of `() -> object`. So I
went ahead and implemented the equivalence to `...` in this PR.
## Ecosystem analysis
* Most of the ecosystem report are cases of improved union/intersection
simplification. For example, we can now simplify a union like `bool |
(bool & Unknown) | Unknown` to simply `bool | Unknown`, because we can
now observe that every possible materialization of `bool & Unknown` is
still a subtype of `bool` (whereas before we would set aside `bool &
Unknown` as a not-fully-static type.) This is clearly an improvement.
* The `possibly-unresolved-reference` errors in sockeye, pymongo,
ignite, scrapy and others are true positives for conditional imports
that were formerly silenced by bogus conflicting-declarations (which we
currently don't issue a diagnostic for), because we considered two
different declarations of `Unknown` to be conflicting (we used
`is_equivalent_to` not `is_gradual_equivalent_to`). In this PR that
distinction disappears and all equivalence is gradual, so a declaration
of `Unknown` no longer conflicts with a declaration of `Unknown`, which
then results in us surfacing the possibly-unbound error.
* We will now issue "redundant cast" for casting from a typevar with a
gradual bound to the same typevar (the hydra-zen diagnostic). This seems
like an improvement.
* The new diagnostics in bandersnatch are interesting. For some reason
primer in CI seems to be checking bandersnatch on Python 3.10 (not yet
sure why; this doesn't happen when I run it locally). But bandersnatch
uses `enum.StrEnum`, which doesn't exist on 3.10. That makes the `class
SimpleDigest(StrEnum)` a class that inherits from `Unknown` (and
bypasses our current TODO handling for accessing attributes on enum
classes, since we don't recognize it as an enum class at all). This PR
improves our understanding of assignability to classes that inherit from
`Any` / `Unknown`, and we now recognize that a string literal is not
assignable to a class inheriting `Any` or `Unknown`.
## Summary
Fixes https://github.com/astral-sh/ty/issues/640. If a user passes
`--python=<some-virtual-environment>/bin/python`, we must avoid
canonicalizing the path until we've traversed upwards to find the
`sys.prefix` directory (`<some-virtual-environment>`). On Unix systems,
`<sys.prefix>/bin/python` is often a symlink to a system interpreter; if
we resolve the symlink too easily then we'll add the system
interpreter's `site-packages` directory as a search path rather than the
virtual environment's directory.
## Test Plan
I added an integration test to
`crates/ty/tests/cli/python_environment.rs` which fails on `main`. I
also manually tested locally that running `cargo run -p ty check foo.py
--python=.venv/bin/python -vv` now prints this log to the terminal
```
2025-06-20 18:35:24.57702 DEBUG Resolved site-packages directories for this virtual environment are: SitePackagesPaths({"/Users/alexw/dev/ruff/.venv/lib/python3.13/site-packages"})
```
Whereas it previously resolved `site-packages` to my system
intallation's `site-packages` directory
We already had support for homogeneous tuples (`tuple[int, ...]`). This
PR extends this to also support mixed tuples (`tuple[str, str,
*tuple[int, ...], str str]`).
A mixed tuple consists of a fixed-length (possibly empty) prefix and
suffix, and a variable-length portion in the middle. Every element of
the variable-length portion must be of the same type. A homogeneous
tuple is then just a mixed tuple with an empty prefix and suffix.
The new data representation uses different Rust types for a fixed-length
(aka heterogeneous) tuple. Another option would have been to use the
`VariableLengthTuple` representation for all tuples, and to wrap the
"variable + suffix" portion in an `Option`. I don't think that would
simplify the method implementations much, though, since we would still
have a 2×2 case analysis for most of them.
One wrinkle is that the definition of the `tuple` class in the typeshed
has a single typevar, and canonically represents a homogeneous tuple.
When getting the class of a tuple instance, that means that we have to
summarize our detailed mixed tuple type information into its
"homogeneous supertype". (We were already doing this for heterogeneous
types.)
A similar thing happens when concatenating two mixed tuples: the
variable-length portion and suffix of the LHS, and the prefix and
variable-length portion of the RHS, all get unioned into the
variable-length portion of the result. The LHS prefix and RHS suffix
carry through unchanged.
---------
Co-authored-by: Alex Waygood <Alex.Waygood@Gmail.com>
## Summary
Part of [#111](https://github.com/astral-sh/ty/issues/111).
After this change, dataclasses with two or more `KW_ONLY` field will be
reported as invalid. The duplicate fields will simply be ignored when
computing `__init__`'s signature.
## Test Plan
Markdown tests.
## Summary
Part of [#117](https://github.com/astral-sh/ty/issues/117).
`TypeIs[]` is a special form that allows users to define their own
narrowing functions. Despite the syntax, `TypeIs` is not a generic and,
on its own, it is meaningless as a type.
[Officially](https://typing.python.org/en/latest/spec/narrowing.html#typeis),
a function annotated as returning a `TypeIs[T]` is a <i>type narrowing
function</i>, where `T` is called the <i>`TypeIs` return type</i>.
A `TypeIs[T]` may or may not be bound to a symbol. Only bound types have
narrowing effect:
```python
def f(v: object = object()) -> TypeIs[int]: ...
a: str = returns_str()
if reveal_type(f()): # Unbound: TypeIs[int]
reveal_type(a) # str
if reveal_type(f(a)): # Bound: TypeIs[a, int]
reveal_type(a) # str & int
```
Delayed usages of a bound type has no effect, however:
```python
b = f(a)
if b:
reveal_type(a) # str
```
A `TypeIs[T]` type:
* Is fully static when `T` is fully static.
* Is a singleton/single-valued when it is bound.
* Has exactly two runtime inhabitants when it is unbound: `True` and
`False`.
In other words, an unbound type have ambiguous truthiness.
It is possible to infer more precise truthiness for bound types;
however, that is not part of this change.
`TypeIs[T]` is a subtype of or otherwise assignable to `bool`. `TypeIs`
is invariant with respect to the `TypeIs` return type: `TypeIs[int]` is
neither a subtype nor a supertype of `TypeIs[bool]`. When ty sees a
function marked as returning `TypeIs[T]`, its `return`s will be checked
against `bool` instead. ty will also report such functions if they don't
accept a positional argument. Addtionally, a type narrowing function
call with no positional arguments (e.g., `f()` in the example above)
will be considered invalid.
## Test Plan
Markdown tests.
---------
Co-authored-by: Carl Meyer <carl@astral.sh>
## Summary
Minor documentation update to make `mypy_primer` instructions a bit more
verbose/helpful for running against a local branch
## Test Plan
N/A
## Summary
https://github.com/astral-sh/ty/issues/214 will require a couple
invasive changes that I would like to get merged even before garbage
collection is fully implemented (to avoid rebasing):
- `ParsedModule` can no longer be dereferenced directly. Instead you
need to load a `ParsedModuleRef` to access the AST, which requires a
reference to the salsa database (as it may require re-parsing the AST if
it was collected).
- `AstNodeRef` can only be dereferenced with the `node` method, which
takes a reference to the `ParsedModuleRef`. This allows us to encode the
fact that ASTs do not live as long as the database and may be collected
as soon a given instance of a `ParsedModuleRef` is dropped. There are a
number of places where we currently merge the `'db` and `'ast`
lifetimes, so this requires giving some types/functions two separate
lifetime parameters.
## Summary
Fixes https://github.com/astral-sh/ty/issues/556.
On Windows, system installations have different layouts to virtual
environments. In Windows virtual environments, the Python executable is
found at `<sys.prefix>/Scripts/python.exe`. But in Windows system
installations, the Python executable is found at
`<sys.prefix>/python.exe`. That means that Windows users were able to
point to Python executables inside virtual environments with the
`--python` flag, but they weren't able to point to Python executables
inside system installations.
This PR fixes that issue. It also makes a couple of other changes:
- Nearly all `sys.prefix` resolution is moved inside `site_packages.rs`.
That was the original design of the `site-packages` resolution logic,
but features implemented since the initial implementation have added
some resolution and validation to `resolver.rs` inside the module
resolver. That means that we've ended up with a somewhat confusing code
structure and a situation where several checks are unnecessarily
duplicated between the two modules.
- I noticed that we had quite bad error messages if you e.g. pointed to
a path that didn't exist on disk with `--python` (we just gave a
somewhat impenetrable message saying that we "failed to canonicalize"
the path). I improved the error messages here and added CLI tests for
`--python` and the `environment.python` configuration setting.
## Test Plan
- Existing tests pass
- Added new CLI tests
- I manually checked that virtual-environment discovery still works if
no configuration is given
- Micha did some manual testing to check that pointing `--python` to a
system-installation executable now works on Windows
## Summary
This optimizes some of the logic added in
https://github.com/astral-sh/ruff/pull/18444. In general, we only
calculate information for subdiagnostics if we know we'll actually emit
the diagnostic. The check to see whether we'll emit the diagnostic is
work we'll definitely have to do whereas the the work to gather
information for a subdiagnostic isn't work we necessarily have to do if
the diagnostic isn't going to be emitted at all.
This PR makes us lazier about gathering the information we need for the
subdiagnostic, and moves all the subdiagnostic logic into one function
rather than having some `unresolved-reference` subdiagnostic logic in
`infer.rs` and some in `diagnostic.rs`.
## Test Plan
`cargo test -p ty_python_semantic`
This updates our representation of functions to more closely match our
representation of classes.
The new `OverloadLiteral` and `FunctionLiteral` classes represent a
function definition in the AST. If a function is generic, this is
unspecialized. `FunctionType` has been updated to represent a function
type, which is specialized if the function is generic. (These names are
chosen to match `ClassLiteral` and `ClassType` on the class side.)
This PR does not add a separate `Type` variant for `FunctionLiteral`.
Maybe we should? Possibly as a follow-on PR?
Part of https://github.com/astral-sh/ty/issues/462
---------
Co-authored-by: Micha Reiser <micha@reiser.io>
Jack O'Connor
Zanie Blue
Micha Reiser
renovate[bot]
Brent Westbrook
Charlie Marsh
Dhruv Manilawala
Ibraheem Ahmed
David Peter
Carl Meyer
Alex Waygood
Josiah Kane
Douglas Creager
InSync
justin
DetachHead