## Summary
This PR introduces a new mdtest option `system` that can either be
`in-memory` or `os`
where `in-memory` is the default.
The motivation for supporting `os` is so that we can write OS/system
specific tests
with mdtests. Specifically, I want to write mdtests for the module
resolver,
testing that module resolution is case sensitive.
## Test Plan
I tested that the case-sensitive module resolver test start failing when
setting `system = "os"`
## Summary
This PR updates the formatter and linter to use the `PythonVersion`
struct from the `ruff_python_ast` crate internally. While this doesn't
remove the need for the `linter::PythonVersion` enum, it does remove the
`formatter::PythonVersion` enum and limits the use in the linter to
deserializing from CLI arguments and config files and moves most of the
remaining methods to the `ast::PythonVersion` struct.
## Test Plan
Existing tests, with some inputs and outputs updated to reflect the new
(de)serialization format. I think these are test-specific and shouldn't
affect any external (de)serialization.
---------
Co-authored-by: Alex Waygood <Alex.Waygood@Gmail.com>
## Summary
This PR moves the `PythonVersion` struct from the
`red_knot_python_semantic` crate to the `ruff_python_ast` crate so that
it can be used more easily in the syntax error detection work. Compared
to that [prototype](https://github.com/astral-sh/ruff/pull/16090/) these
changes reduce us from 2 `PythonVersion` structs to 1.
This does not unify any of the `PythonVersion` *enums*, but I hope to
make some progress on that in a follow-up.
## Test Plan
Existing tests, this should not change any external behavior.
---------
Co-authored-by: Alex Waygood <Alex.Waygood@Gmail.com>
## Summary
This PR generalize the idea that we may want to emit diagnostics for
invalid or incompatible configuration values similar to how we already
do it for `rules`.
This PR introduces a new `Settings` struct that is similar to `Options`
but, unlike
`Options`, are fields have their default values filled in and they use a
representation optimized for reads.
The diagnostics created during loading the `Settings` are stored on the
`Project` so that we can emit them when calling `check`.
The motivation for this work is that it simplifies adding new settings.
That's also why I went ahead and added the `terminal.error-on-warning`
setting to demonstrate how new settings are added.
## Test Plan
Existing tests, new CLI test.
## Summary
- Add feature to specify a custom typeshed from within Markdown-based
tests
- Port "builtins" unit tests from `infer.rs` to Markdown tests, part of
#13696
## Test Plan
- Tests for the custom typeshed feature
- New Markdown tests for deleted Rust unit tests
## Summary
This PR adds support for configuring Red Knot in the `tool.knot` section
of the project's
`pyproject.toml` section. Options specified on the CLI precede the
options in the configuration file.
This PR only supports the `environment` and the `src.root` options for
now.
Other options will be added as separate PRs.
There are also a few concerns that I intentionally ignored as part of
this PR:
* Handling of relative paths: We need to anchor paths relative to the
current working directory (CLI), or the project (`pyproject.toml` or
`knot.toml`)
* Tracking the source of a value. Diagnostics would benefit from knowing
from which configuration a value comes so that we can point the user to
the right configuration file (or CLI) if the configuration is invalid.
* Schema generation and there's a lot more; see
https://github.com/astral-sh/ruff/issues/15491
This PR changes the default for first party codes: Our existing default
was to only add the project root. Now, Red Knot adds the project root
and `src` (if such a directory exists).
Theoretically, we'd have to add a file watcher event that changes the
first-party search paths if a user later creates a `src` directory. I
think this is pretty uncommon, which is why I ignored the complexity for
now but I can be persuaded to handle it if it's considered important.
Part of https://github.com/astral-sh/ruff/issues/15491
## Test Plan
Existing tests, new file watching test demonstrating that changing the
python version and platform is correctly reflected.
## Summary
This changeset adds support for precise type-inference and
boundness-handling of definitions inside control-flow branches with
statically-known conditions, i.e. test-expressions whose truthiness we
can unambiguously infer as *always false* or *always true*.
This branch also includes:
- `sys.platform` support
- statically-known branches handling for Boolean expressions and while
loops
- new `target-version` requirements in some Markdown tests which were
now required due to the understanding of `sys.version_info` branches.
closes#12700closes#15034
## Performance
### `tomllib`, -7%, needs to resolve one additional module (sys)
| Command | Mean [ms] | Min [ms] | Max [ms] | Relative |
|:---|---:|---:|---:|---:|
| `./red_knot_main --project /home/shark/tomllib` | 22.2 ± 1.3 | 19.1 |
25.6 | 1.00 |
| `./red_knot_feature --project /home/shark/tomllib` | 23.8 ± 1.6 | 20.8
| 28.6 | 1.07 ± 0.09 |
### `black`, -6%
| Command | Mean [ms] | Min [ms] | Max [ms] | Relative |
|:---|---:|---:|---:|---:|
| `./red_knot_main --project /home/shark/black` | 129.3 ± 5.1 | 119.0 |
137.8 | 1.00 |
| `./red_knot_feature --project /home/shark/black` | 136.5 ± 6.8 | 123.8
| 147.5 | 1.06 ± 0.07 |
## Test Plan
- New Markdown tests for the main feature in
`statically-known-branches.md`
- New Markdown tests for `sys.platform`
- Adapted tests for `EllipsisType`, `Never`, etc
## Summary
This PR renames the `--custom-typeshed-dir`, `target-version`, and
`--current-directory` cli options to `--typeshed`,
`--python-version`, and `--project` as discussed in the CLI proposal
document.
I added aliases for `--target-version` (for Ruff compat) and
`--custom-typeshed-dir` (for Alex)
## Test Plan
Long help
```
An extremely fast Python type checker.
Usage: red_knot [OPTIONS] [COMMAND]
Commands:
server Start the language server
help Print this message or the help of the given subcommand(s)
Options:
--project <PROJECT>
Run the command within the given project directory.
All `pyproject.toml` files will be discovered by walking up the directory tree from the project root, as will the project's virtual environment (`.venv`).
Other command-line arguments (such as relative paths) will be resolved relative to the current working directory."#,
--venv-path <PATH>
Path to the virtual environment the project uses.
If provided, red-knot will use the `site-packages` directory of this virtual environment to resolve type information for the project's third-party dependencies.
--typeshed-path <PATH>
Custom directory to use for stdlib typeshed stubs
--extra-search-path <PATH>
Additional path to use as a module-resolution source (can be passed multiple times)
--python-version <VERSION>
Python version to assume when resolving types
[possible values: 3.7, 3.8, 3.9, 3.10, 3.11, 3.12, 3.13]
-v, --verbose...
Use verbose output (or `-vv` and `-vvv` for more verbose output)
-W, --watch
Run in watch mode by re-running whenever files change
-h, --help
Print help (see a summary with '-h')
-V, --version
Print version
```
Short help
```
An extremely fast Python type checker.
Usage: red_knot [OPTIONS] [COMMAND]
Commands:
server Start the language server
help Print this message or the help of the given subcommand(s)
Options:
--project <PROJECT> Run the command within the given project directory
--venv-path <PATH> Path to the virtual environment the project uses
--typeshed-path <PATH> Custom directory to use for stdlib typeshed stubs
--extra-search-path <PATH> Additional path to use as a module-resolution source (can be passed multiple times)
--python-version <VERSION> Python version to assume when resolving types [possible values: 3.7, 3.8, 3.9, 3.10, 3.11, 3.12, 3.13]
-v, --verbose... Use verbose output (or `-vv` and `-vvv` for more verbose output)
-W, --watch Run in watch mode by re-running whenever files change
-h, --help Print help (see more with '--help')
-V, --version Print version
```
---------
Co-authored-by: Alex Waygood <Alex.Waygood@Gmail.com>
## Summary
This is the third and last PR in this stack that adds support for
toggling lints at a per-rule level.
This PR introduces a new `LintRegistry`, a central index of known lints.
The registry is required because we want to support lint rules from many
different crates but need a way to look them up by name, e.g., when
resolving a lint from a name in the configuration or analyzing a
suppression comment.
Adding a lint now requires two steps:
1. Declare the lint with `declare_lint`
2. Register the lint in the registry inside the `register_lints`
function.
I considered some more involved macros to avoid changes in two places.
Still, I ultimately decided against it because a) it's just two places
and b) I'd expect that registering a type checker lint will differ from
registering a lint that runs as a rule in the linter. I worry that any
more opinionated design could limit our options when working on the
linter, so I kept it simple.
The second part of this PR is the `RuleSelection`. It stores which lints
are enabled and what severity they should use for created diagnostics.
For now, the `RuleSelection` always gets initialized with all known
lints and it uses their default level.
## Linter crates
Each crate that defines lints should export a `register_lints` function
that accepts a `&mut LintRegistryBuilder` to register all its known
lints in the registry. This should make registering all known lints in a
top-level crate easy: Just call `register_lints` of every crate that
defines lint rules.
I considered defining a `LintCollection` trait and even some fancy
macros to accomplish the same but decided to go for this very simplistic
approach for now. We can add more abstraction once needed.
## Lint rules
This is a bit hand-wavy. I don't have a good sense for how our linter
infrastructure will look like, but I expect we'll need a way to register
the rules that should run as part of the red knot linter. One way is to
keep doing what Ruff does by having one massive `checker` and each lint
rule adds a call to itself in the relevant AST visitor methods. An
alternative is that we have a `LintRule` trait that provides common
hooks and implementations will be called at the "right time". Such a
design would need a way to register all known lint implementations,
possibly with the lint. This is where we'd probably want a dedicated
`register_rule` method. A third option is that lint rules are handled
separately from the `LintRegistry` and are specific to the linter crate.
The current design should be flexible enough to support the three
options.
## Documentation generation
The documentation for all known lints can be generated by creating a
factory, registering all lints by calling the `register_lints` methods,
and then querying the registry for the metadata.
## Deserialization and Schema generation
I haven't fully decided what the best approach is when it comes to
deserializing lint rule names:
* Reject invalid names in the deserializer. This gives us error messages
with line and column numbers (by serde)
* Don't validate lint rule names during deserialization; defer the
validation until the configuration is resolved. This gives us more
control over handling the error, e.g. emit a warning diagnostic instead
of aborting when a rule isn't known.
One technical challenge for both deserialization and schema generation
is that the `Deserialize` and `JSONSchema` traits do not allow passing
the `LintRegistry`, which is required to look up the lints by name. I
suggest that we either rely on the salsa db being set for the current
thread (`salsa::Attach`) or build our own thread-local storage for the
`LintRegistry`. It's the caller's responsibility to make the lint
registry available before calling `Deserialize` or `JSONSchema`.
## CLI support
I prefer deferring adding support for enabling and disabling lints from
the CLI for now because I think it will be easier
to add once I've figured out how to handle configurations.
## Bitset optimization
Ruff tracks the enabled rules using a cheap copyable `Bitset` instead of
a hash map. This helped improve performance by a few percent (see
https://github.com/astral-sh/ruff/pull/3606). However, this approach is
no longer possible because lints have no "cheap" way to compute their
index inside the registry (other than using a hash map).
We could consider doing something similar to Salsa where each
`LintMetadata` stores a `LazyLintIndex`.
```
pub struct LazyLintIndex {
cached: OnceLock<(Nonce, LintIndex)>
}
impl LazyLintIndex {
pub fn get(registry: &LintRegistry, lint: &'static LintMetadata) {
let (nonce, index) = self.cached.get_or_init(|| registry.lint_index(lint));
if registry.nonce() == nonce {
index
} else {
registry.lint_index(lint)
}
}
```
Each registry keeps a map from `LintId` to `LintIndex` where `LintIndex`
is in the range of `0...registry.len()`. The `LazyLintIndex` is based on
the assumption that every program has exactly **one** registry. This
assumption allows to cache the `LintIndex` directly on the
`LintMetadata`. The implementation falls back to the "slow" path if
there is more than one registry at runtime.
I was very close to implementing this optimization because it's kind of
fun to implement. I ultimately decided against it because it adds
complexity and I don't think it's worth doing in Red Knot today:
* Red Knot only queries the rule selection when deciding whether or not
to emit a diagnostic. It is rarely used to detect if a certain code
block should run. This is different from Ruff where the rule selection
is queried many times for every single AST node to determine which rules
*should* run.
* I'm not sure if a 2-3% performance improvement is worth the complexity
I suggest revisiting this decision when working on the linter where a
fast path for deciding if a rule is enabled might be more important (but
that depends on how lint rules are implemented)
## Test Plan
I removed a lint from the default rule registry, and the MD tests
started failing because the diagnostics were no longer emitted.
## Summary
- Instead of seven (more or less similar) `setup_db` functions, use just
one in a single central place.
- For every test that needs customization beyond that, offer a
`TestDbBuilder` that can control the Python target version, custom
typeshed, and pre-existing files.
The main motivation for this is that we're soon going to need
customization of the Python version, and I didn't feel like adding this
to each of the existing `setup_db` functions.
## Summary
This PR changes removes the typeshed stubs from the vendored file system
shipped with ruff
and instead ships an empty "typeshed".
Making the typeshed files optional required extracting the typshed files
into a new `ruff_vendored` crate. I do like this even if all our builds
always include typeshed because it means `red_knot_python_semantic`
contains less code that needs compiling.
This also allows us to use deflate because the compression algorithm
doesn't matter for an archive containing a single, empty file.
## Test Plan
`cargo test`
I verified with ` cargo tree -f "{p} {f}" -p <package> ` that:
* red_knot_wasm: enables `deflate` compression
* red_knot: enables `zstd` compression
* `ruff`: uses stored
I'm not quiet sure how to build the binary that maturin builds but
comparing the release artifact size with `strip = true` shows a `1.5MB`
size reduction
---------
Co-authored-by: Charlie Marsh <charlie.r.marsh@gmail.com>
Add a lint rule to detect if a name is definitely or possibly undefined
at a given usage.
If I create the file `undef/main.py` with contents:
```python
x = int
def foo():
z
return x
if flag:
y = x
y
```
And then run `cargo run --bin red_knot -- --current-directory
../ruff-examples/undef`, I get the output:
```
Name 'z' used when not defined.
Name 'flag' used when not defined.
Name 'y' used when possibly not defined.
```
If I modify the file to add `y = 0` at the top, red-knot re-checks it
and I get the new output:
```
Name 'z' used when not defined.
Name 'flag' used when not defined.
```
Note that `int` is not flagged, since it's a builtin, and `return x` in
the function scope is not flagged, since it refers to the global `x`.
Per comments in https://github.com/astral-sh/ruff/pull/12269, "module
global" is kind of long, and arguably redundant.
I tried just using "module" but there were too many cases where I felt
this was ambiguous. I like the way "global" works out better, though it
does require an understanding that in Python "global" generally means
"module global" not "globally global" (though in a sense module globals
are also globally global since modules are singletons).
Implements definition-level type inference, with basic control flow
(only if statements and if expressions so far) in Salsa.
There are a couple key ideas here:
1) We can do type inference queries at any of three region
granularities: an entire scope, a single definition, or a single
expression. These are represented by the `InferenceRegion` enum, and the
entry points are the salsa queries `infer_scope_types`,
`infer_definition_types`, and `infer_expression_types`. Generally
per-scope will be used for scopes that we are directly checking and
per-definition will be used anytime we are looking up symbol types from
another module/scope. Per-expression should be uncommon: used only for
the RHS of an unpacking or multi-target assignment (to avoid
re-inferring the RHS once per symbol defined in the assignment) and for
test nodes in type narrowing (e.g. the `test` of an `If` node). All
three queries return a `TypeInference` with a map of types for all
definitions and expressions within their region. If you do e.g.
scope-level inference, when it hits a definition, or an
independently-inferable expression, it should use the relevant query
(which may already be cached) to get all types within the smaller
region. This avoids double-inferring smaller regions, even though larger
regions encompass smaller ones.
2) Instead of building a control-flow graph and lazily traversing it to
find definitions which reach a use of a name (which is O(n^2) in the
worst case), instead semantic indexing builds a use-def map, where every
use of a name knows which definitions can reach that use. We also no
longer track all definitions of a symbol in the symbol itself; instead
the use-def map also records which defs remain visible at the end of the
scope, and considers these the publicly-visible definitions of the
symbol (see below).
Major items left as TODOs in this PR, to be done in follow-up PRs:
1) Free/global references aren't supported yet (only lookup based on
definitions in current scope), which means the override-check example
doesn't currently work. This is the first thing I'll fix as follow-up to
this PR.
2) Control flow outside of if statements and expressions.
3) Type narrowing.
There are also some smaller relevant changes here:
1) Eliminate `Option` in the return type of member lookups; instead
always return `Type::Unbound` for a name we can't find. Also use
`Type::Unbound` for modules we can't resolve (not 100% sure about this
one yet.)
2) Eliminate the use of the terms "public" and "root" to refer to
module-global scope or symbols. Instead consistently use the term
"module-global". It's longer, but it's the clearest, and the most
consistent with typical Python terminology. In particular I don't like
"public" for this use because it has other implications around author
intent (is an underscore-prefixed module-global symbol "public"?). And
"root" is just not commonly used for this in Python.
3) Eliminate the `PublicSymbol` Salsa ingredient. Many non-module-global
symbols can also be seen from other scopes (e.g. by a free var in a
nested scope, or by class attribute access), and thus need to have a
"public type" (that is, the type not as seen from a particular use in
the control flow of the same scope, but the type as seen from some other
scope.) So all symbols need to have a "public type" (here I want to keep
the use of the term "public", unless someone has a better term to
suggest -- since it's "public type of a symbol" and not "public symbol"
the confusion with e.g. initial underscores is less of an issue.) At
least initially, I would like to try not having special handling for
module-global symbols vs other symbols.
4) Switch to using "definitions that reach end of scope" rather than
"all definitions" in determining the public type of a symbol. I'm
convinced that in general this is the right way to go. We may want to
refine this further in future for some free-variable cases, but it can
be changed purely by making changes to the building of the use-def map
(the `public_definitions` index in it), without affecting any other
code. One consequence of combining this with no control-flow support
(just last-definition-wins) is that some inference tests now give more
wrong-looking results; I left TODO comments on these tests to fix them
when control flow is added.
And some potential areas for consideration in the future:
1) Should `symbol_ty` be a Salsa query? This would require making all
symbols a Salsa ingredient, and tracking even more dependencies. But it
would save some repeated reconstruction of unions, for symbols with
multiple public definitions. For now I'm not making it a query, but open
to changing this in future with actual perf evidence that it's better.
Intern types using Salsa interning instead of in the `TypeInference`
result.
This eliminates the need for `TypingContext`, and also paves the way for
finer-grained type inference queries.
