## Summary
This is another attempt at https://github.com/astral-sh/ruff/pull/21410
that fixes https://github.com/astral-sh/ruff/issues/19226.
@MichaReiser helped me get something working in a very helpful pairing
session. I pushed one additional commit moving the comments back from
leading comments to trailing comments, which I think retains more of the
input formatting.
I was inspired by Dylan's PR (#21185) to make one of these tables:
<table>
<thead>
<tr>
<th scope="col">Input</th>
<th scope="col">Main</th>
<th scope="col">PR</th>
</tr>
</thead>
<tbody>
<tr>
<td><pre lang="python">
if (
not
# comment
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa +
bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb
):
pass
</pre></td>
<td><pre lang="python">
if (
# comment
not aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
+ bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb
):
pass
</pre></td>
<td><pre lang="python">
if (
not
# comment
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
+ bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb
):
pass
</pre></td>
</tr>
<tr>
<td><pre lang="python">
if (
# unary comment
not
# operand comment
(
# comment
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
+ bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb
)
):
pass
</pre></td>
<td><pre lang="python">
if (
# unary comment
# operand comment
not (
# comment
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
+ bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb
)
):
pass
</pre></td>
<td><pre lang="python">
if (
# unary comment
not
# operand comment
(
# comment
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
+ bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb
)
):
pass
</pre></td>
</tr>
<tr>
<td><pre lang="python">
if (
not # comment
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
+ bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb
):
pass
</pre></td>
<td><pre lang="python">
if ( # comment
not aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
+ bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb
):
pass
</pre></td>
<td><pre lang="python">
if (
not aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa # comment
+ bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb
):
pass
</pre></td>
</tr>
</tbody>
</table>
hopefully it helps even though the snippets are much wider here.
The two main differences are (1) that we now retain own-line comments
between the unary operator and its operand instead of moving these to
leading comments on the operator itself, and (2) that we move
end-of-line comments between the operator and operand to dangling
end-of-line comments on the operand (the last example in the table).
## Test Plan
Existing tests, plus new ones based on the issue. As I noted below, I
also ran the output from main on the unary.py file back through this
branch to check that we don't reformat code from main. This made me feel
a bit better about not preview-gating the changes in this PR.
```shell
> git show main:crates/ruff_python_formatter/resources/test/fixtures/ruff/expression/unary.py | ruff format - | ./target/debug/ruff format --diff -
> echo $?
0
```
---------
Co-authored-by: Micha Reiser <micha@reiser.io>
Co-authored-by: Takayuki Maeda <takoyaki0316@gmail.com>
Closes#19350
This fixes a syntax error caused by formatting. However, the new tests reveal that there are some cases where formatting attributes with certain comments behaves strangely, both before and after this PR, so some more polish may be in order.
For example, without parentheses around the value, and both before and after this PR, we have:
```python
# unformatted
variable = (
something # a comment
.first_method("some string")
)
# formatted
variable = something.first_method("some string") # a comment
```
which is probably not where the comment ought to go.
Summary
--
This PR implements the black preview style from
https://github.com/psf/black/pull/4720. As of Python 3.14, you're
allowed to omit the parentheses around groups of exceptions, as long as
there's no `as` binding:
**3.13**
```pycon
Python 3.13.4 (main, Jun 4 2025, 17:37:06) [Clang 20.1.4 ] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> try: ...
... except (Exception, BaseException): ...
...
Ellipsis
>>> try: ...
... except Exception, BaseException: ...
...
File "<python-input-1>", line 2
except Exception, BaseException: ...
^^^^^^^^^^^^^^^^^^^^^^^^
SyntaxError: multiple exception types must be parenthesized
```
**3.14**
```pycon
Python 3.14.0rc2 (main, Sep 2 2025, 14:20:56) [Clang 20.1.4 ] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> try: ...
... except Exception, BaseException: ...
...
Ellipsis
>>> try: ...
... except (Exception, BaseException): ...
...
Ellipsis
>>> try: ...
... except Exception, BaseException as e: ...
...
File "<python-input-2>", line 2
except Exception, BaseException as e: ...
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
SyntaxError: multiple exception types must be parenthesized when using 'as'
```
I think this ended up being pretty straightforward, at least once Micha
showed me where to start :)
Test Plan
--
New tests
At first I thought we were deviating from black in how we handle
comments within the exception type tuple, but I think this applies to
how we format all tuples, not specifically with the new preview style.
## Summary
Garbage collect ASTs once we are done checking a given file. Queries
with a cross-file dependency on the AST will reparse the file on demand.
This reduces ty's peak memory usage by ~20-30%.
The primary change of this PR is adding a `node_index` field to every
AST node, that is assigned by the parser. `ParsedModule` can use this to
create a flat index of AST nodes any time the file is parsed (or
reparsed). This allows `AstNodeRef` to simply index into the current
instance of the `ParsedModule`, instead of storing a pointer directly.
The indices are somewhat hackily (using an atomic integer) assigned by
the `parsed_module` query instead of by the parser directly. Assigning
the indices in source-order in the (recursive) parser turns out to be
difficult, and collecting the nodes during semantic indexing is
impossible as `SemanticIndex` does not hold onto a specific
`ParsedModuleRef`, which the pointers in the flat AST are tied to. This
means that we have to do an extra AST traversal to assign and collect
the nodes into a flat index, but the small performance impact (~3% on
cold runs) seems worth it for the memory savings.
Part of https://github.com/astral-sh/ty/issues/214.
This PR implements template strings (t-strings) in the parser and
formatter for Ruff.
Minimal changes necessary to compile were made in other parts of the code (e.g. ty, the linter, etc.). These will be covered properly in follow-up PRs.
<!--
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
I decided to disable the new
[`needless_continue`](https://rust-lang.github.io/rust-clippy/master/index.html#needless_continue)
rule because I often found the explicit `continue` more readable over an
empty block or having to invert the condition of an other branch.
## Test Plan
`cargo test`
---------
Co-authored-by: Alex Waygood <Alex.Waygood@Gmail.com>
## Summary
This PR makes the following changes:
- It adjusts various callsites to use the new
`ast::StringLiteral::contents_range()` method that was introduced in
https://github.com/astral-sh/ruff/pull/16183. This is less verbose and
more type-safe than using the `ast::str::raw_contents()` helper
function.
- It adds a new `ast::ExprStringLiteral::as_unconcatenated_literal()`
helper method, and adjusts various callsites to use it. This addresses
@MichaReiser's review comment at
https://github.com/astral-sh/ruff/pull/16183#discussion_r1957334365.
There is no functional change here, but it helps readability to make it
clearer that we're differentiating between implicitly concatenated
strings and unconcatenated strings at various points.
- It renames the `StringLiteralValue::flags()` method to
`StringLiteralFlags::first_literal_flags()`. If you're dealing with an
implicitly concatenated string `string_node`,
`string_node.value.flags().closer_len()` could give an incorrect result;
this renaming makes it clearer that the `StringLiteralFlags` instance
returned by the method is only guaranteed to give accurate information
for the first `StringLiteral` contained in the `ExprStringLiteral` node.
- It deletes the unused `BytesLiteralValue::flags()` method. This seems
prone to misuse in the same way as `StringLiteralValue::flags()`: if
it's an implicitly concatenated bytestring, the `BytesLiteralFlags`
instance returned by the method would only give accurate information for
the first `BytesLiteral` in the bytestring.
## Test Plan
`cargo test`
The AST generator creates a reference enum for each syntax group — an
enum where each variant contains a reference to the relevant syntax
node. Previously you could customize the name of the reference enum for
a group — primarily because there was an existing `ExpressionRef` type
that wouldn't have lined up with the auto-derived name `ExprRef`. This
follow-up PR is a simple search/replace to switch over to the
auto-derived name, so that we can remove this customization point.
While looking into potential AST optimizations, I noticed the `AstNode`
trait and `AnyNode` type aren't used anywhere in Ruff or Red Knot. It
looks like they might be historical artifacts of previous ways of
consuming AST nodes?
- `AstNode::cast`, `AstNode::cast_ref`, and `AstNode::can_cast` are not
used anywhere.
- Since `cast_ref` isn't needed anymore, the `Ref` associated type isn't
either.
This is a pure refactoring, with no intended behavior changes.
## Summary
fixes: #13813
This PR fixes a bug in the formatting assignment statement when the
value is an f-string.
This is resolved by using custom best fit layouts if the f-string is (a)
not already a flat f-string (thus, cannot be multiline) and (b) is not a
multiline string (thus, cannot be flattened). So, it is used in cases
like the following:
```py
aaaaaaaaaaaaaaaaaa = f"testeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee{
expression}moreeeeeeeeeeeeeeeee"
```
Which is (a) `FStringLayout::Multiline` and (b) not a multiline.
There are various other examples in the PR diff along with additional
explanation and context as code comments.
## Test Plan
Add multiple test cases for various scenarios.
## Summary
This is just a small refactor to remove the `FormatFStringPart` as it's
only used in the case when the f-string is not implicitly concatenated
in which case the only part is going to be `FString`. In implicitly
concatenated f-strings, we use `StringLike` instead.
## Summary
This PR updates the parser to remove building the `CommentRanges` and
instead it'll be built by the linter and the formatter when it's
required.
For the linter, it'll be built and owned by the `Indexer` while for the
formatter it'll be built from the `Tokens` struct and passed as an
argument.
## Test Plan
`cargo insta test`
## Summary
This PR updates the entire parser stack in multiple ways:
### Make the lexer lazy
* https://github.com/astral-sh/ruff/pull/11244
* https://github.com/astral-sh/ruff/pull/11473
Previously, Ruff's lexer would act as an iterator. The parser would
collect all the tokens in a vector first and then process the tokens to
create the syntax tree.
The first task in this project is to update the entire parsing flow to
make the lexer lazy. This includes the `Lexer`, `TokenSource`, and
`Parser`. For context, the `TokenSource` is a wrapper around the `Lexer`
to filter out the trivia tokens[^1]. Now, the parser will ask the token
source to get the next token and only then the lexer will continue and
emit the token. This means that the lexer needs to be aware of the
"current" token. When the `next_token` is called, the current token will
be updated with the newly lexed token.
The main motivation to make the lexer lazy is to allow re-lexing a token
in a different context. This is going to be really useful to make the
parser error resilience. For example, currently the emitted tokens
remains the same even if the parser can recover from an unclosed
parenthesis. This is important because the lexer emits a
`NonLogicalNewline` in parenthesized context while a normal `Newline` in
non-parenthesized context. This different kinds of newline is also used
to emit the indentation tokens which is important for the parser as it's
used to determine the start and end of a block.
Additionally, this allows us to implement the following functionalities:
1. Checkpoint - rewind infrastructure: The idea here is to create a
checkpoint and continue lexing. At a later point, this checkpoint can be
used to rewind the lexer back to the provided checkpoint.
2. Remove the `SoftKeywordTransformer` and instead use lookahead or
speculative parsing to determine whether a soft keyword is a keyword or
an identifier
3. Remove the `Tok` enum. The `Tok` enum represents the tokens emitted
by the lexer but it contains owned data which makes it expensive to
clone. The new `TokenKind` enum just represents the type of token which
is very cheap.
This brings up a question as to how will the parser get the owned value
which was stored on `Tok`. This will be solved by introducing a new
`TokenValue` enum which only contains a subset of token kinds which has
the owned value. This is stored on the lexer and is requested by the
parser when it wants to process the data. For example:
8196720f80/crates/ruff_python_parser/src/parser/expression.rs (L1260-L1262)
[^1]: Trivia tokens are `NonLogicalNewline` and `Comment`
### Remove `SoftKeywordTransformer`
* https://github.com/astral-sh/ruff/pull/11441
* https://github.com/astral-sh/ruff/pull/11459
* https://github.com/astral-sh/ruff/pull/11442
* https://github.com/astral-sh/ruff/pull/11443
* https://github.com/astral-sh/ruff/pull/11474
For context,
https://github.com/RustPython/RustPython/pull/4519/files#diff-5de40045e78e794aa5ab0b8aacf531aa477daf826d31ca129467703855408220
added support for soft keywords in the parser which uses infinite
lookahead to classify a soft keyword as a keyword or an identifier. This
is a brilliant idea as it basically wraps the existing Lexer and works
on top of it which means that the logic for lexing and re-lexing a soft
keyword remains separate. The change here is to remove
`SoftKeywordTransformer` and let the parser determine this based on
context, lookahead and speculative parsing.
* **Context:** The transformer needs to know the position of the lexer
between it being at a statement position or a simple statement position.
This is because a `match` token starts a compound statement while a
`type` token starts a simple statement. **The parser already knows
this.**
* **Lookahead:** Now that the parser knows the context it can perform
lookahead of up to two tokens to classify the soft keyword. The logic
for this is mentioned in the PR implementing it for `type` and `match
soft keyword.
* **Speculative parsing:** This is where the checkpoint - rewind
infrastructure helps. For `match` soft keyword, there are certain cases
for which we can't classify based on lookahead. The idea here is to
create a checkpoint and keep parsing. Based on whether the parsing was
successful and what tokens are ahead we can classify the remaining
cases. Refer to #11443 for more details.
If the soft keyword is being parsed in an identifier context, it'll be
converted to an identifier and the emitted token will be updated as
well. Refer
8196720f80/crates/ruff_python_parser/src/parser/expression.rs (L487-L491).
The `case` soft keyword doesn't require any special handling because
it'll be a keyword only in the context of a match statement.
### Update the parser API
* https://github.com/astral-sh/ruff/pull/11494
* https://github.com/astral-sh/ruff/pull/11505
Now that the lexer is in sync with the parser, and the parser helps to
determine whether a soft keyword is a keyword or an identifier, the
lexer cannot be used on its own. The reason being that it's not
sensitive to the context (which is correct). This means that the parser
API needs to be updated to not allow any access to the lexer.
Previously, there were multiple ways to parse the source code:
1. Passing the source code itself
2. Or, passing the tokens
Now that the lexer and parser are working together, the API
corresponding to (2) cannot exists. The final API is mentioned in this
PR description: https://github.com/astral-sh/ruff/pull/11494.
### Refactor the downstream tools (linter and formatter)
* https://github.com/astral-sh/ruff/pull/11511
* https://github.com/astral-sh/ruff/pull/11515
* https://github.com/astral-sh/ruff/pull/11529
* https://github.com/astral-sh/ruff/pull/11562
* https://github.com/astral-sh/ruff/pull/11592
And, the final set of changes involves updating all references of the
lexer and `Tok` enum. This was done in two-parts:
1. Update all the references in a way that doesn't require any changes
from this PR i.e., it can be done independently
* https://github.com/astral-sh/ruff/pull/11402
* https://github.com/astral-sh/ruff/pull/11406
* https://github.com/astral-sh/ruff/pull/11418
* https://github.com/astral-sh/ruff/pull/11419
* https://github.com/astral-sh/ruff/pull/11420
* https://github.com/astral-sh/ruff/pull/11424
2. Update all the remaining references to use the changes made in this
PR
For (2), there were various strategies used:
1. Introduce a new `Tokens` struct which wraps the token vector and add
methods to query a certain subset of tokens. These includes:
1. `up_to_first_unknown` which replaces the `tokenize` function
2. `in_range` and `after` which replaces the `lex_starts_at` function
where the former returns the tokens within the given range while the
latter returns all the tokens after the given offset
2. Introduce a new `TokenFlags` which is a set of flags to query certain
information from a token. Currently, this information is only limited to
any string type token but can be expanded to include other information
in the future as needed. https://github.com/astral-sh/ruff/pull/11578
3. Move the `CommentRanges` to the parsed output because this
information is common to both the linter and the formatter. This removes
the need for `tokens_and_ranges` function.
## Test Plan
- [x] Update and verify the test snapshots
- [x] Make sure the entire test suite is passing
- [x] Make sure there are no changes in the ecosystem checks
- [x] Run the fuzzer on the parser
- [x] Run this change on dozens of open-source projects
### Running this change on dozens of open-source projects
Refer to the PR description to get the list of open source projects used
for testing.
Now, the following tests were done between `main` and this branch:
1. Compare the output of `--select=E999` (syntax errors)
2. Compare the output of default rule selection
3. Compare the output of `--select=ALL`
**Conclusion: all output were same**
## What's next?
The next step is to introduce re-lexing logic and update the parser to
feed the recovery information to the lexer so that it can emit the
correct token. This moves us one step closer to having error resilience
in the parser and provides Ruff the possibility to lint even if the
source code contains syntax errors.
(Supersedes #9152, authored by @LaBatata101)
## Summary
This PR replaces the current parser generated from LALRPOP to a
hand-written recursive descent parser.
It also updates the grammar for [PEP
646](https://peps.python.org/pep-0646/) so that the parser outputs the
correct AST. For example, in `data[*x]`, the index expression is now a
tuple with a single starred expression instead of just a starred
expression.
Beyond the performance improvements, the parser is also error resilient
and can provide better error messages. The behavior as seen by any
downstream tools isn't changed. That is, the linter and formatter can
still assume that the parser will _stop_ at the first syntax error. This
will be updated in the following months.
For more details about the change here, refer to the PR corresponding to
the individual commits and the release blog post.
## Test Plan
Write _lots_ and _lots_ of tests for both valid and invalid syntax and
verify the output.
## Acknowledgements
- @MichaReiser for reviewing 100+ parser PRs and continuously providing
guidance throughout the project
- @LaBatata101 for initiating the transition to a hand-written parser in
#9152
- @addisoncrump for implementing the fuzzer which helped
[catch](https://github.com/astral-sh/ruff/pull/10903)
[a](https://github.com/astral-sh/ruff/pull/10910)
[lot](https://github.com/astral-sh/ruff/pull/10966)
[of](https://github.com/astral-sh/ruff/pull/10896)
[bugs](https://github.com/astral-sh/ruff/pull/10877)
---------
Co-authored-by: Victor Hugo Gomes <labatata101@linuxmail.org>
Co-authored-by: Micha Reiser <micha@reiser.io>
The expression types in our AST are called `ExprYield`, `ExprAwait`,
`ExprStringLiteral` etc, except `ExprNamedExpr`, `ExprIfExpr` and
`ExprGenratorExpr`. This seems to align with [Python AST's
naming](https://docs.python.org/3/library/ast.html) but feels
inconsistent and excessive.
This PR removes the `Expr` postfix from `ExprNamedExpr`, `ExprIfExpr`,
and `ExprGeneratorExpr`.
## Summary
_This is preview only feature and is available using the `--preview`
command-line flag._
With the implementation of [PEP 701] in Python 3.12, f-strings can now
be broken into multiple lines, can contain comments, and can re-use the
same quote character. Currently, no other Python formatter formats the
f-strings so there's some discussion which needs to happen in defining
the style used for f-string formatting. Relevant discussion:
https://github.com/astral-sh/ruff/discussions/9785
The goal for this PR is to add minimal support for f-string formatting.
This would be to format expression within the replacement field without
introducing any major style changes.
### Newlines
The heuristics for adding newline is similar to that of
[Prettier](https://prettier.io/docs/en/next/rationale.html#template-literals)
where the formatter would only split an expression in the replacement
field across multiple lines if there was already a line break within the
replacement field.
In other words, the formatter would not add any newlines unless they
were already present i.e., they were added by the user. This makes
breaking any expression inside an f-string optional and in control of
the user. For example,
```python
# We wouldn't break this
aaaaaaaaaaa = f"asaaaaaaaaaaaaaaaa { aaaaaaaaaaaa + bbbbbbbbbbbb + ccccccccccccccc } cccccccccc"
# But, we would break the following as there's already a newline
aaaaaaaaaaa = f"asaaaaaaaaaaaaaaaa {
aaaaaaaaaaaa + bbbbbbbbbbbb + ccccccccccccccc } cccccccccc"
```
If there are comments in any of the replacement field of the f-string,
then it will always be a multi-line f-string in which case the formatter
would prefer to break expressions i.e., introduce newlines. For example,
```python
x = f"{ # comment
a }"
```
### Quotes
The logic for formatting quotes remains unchanged. The existing logic is
used to determine the necessary quote char and is used accordingly.
Now, if the expression inside an f-string is itself a string like, then
we need to make sure to preserve the existing quote and not change it to
the preferred quote unless it's 3.12. For example,
```python
f"outer {'inner'} outer"
# For pre 3.12, preserve the single quote
f"outer {'inner'} outer"
# While for 3.12 and later, the quotes can be changed
f"outer {"inner"} outer"
```
But, for triple-quoted strings, we can re-use the same quote char unless
the inner string is itself a triple-quoted string.
```python
f"""outer {"inner"} outer""" # valid
f"""outer {'''inner'''} outer""" # preserve the single quote char for the inner string
```
### Debug expressions
If debug expressions are present in the replacement field of a f-string,
then the whitespace needs to be preserved as they will be rendered as it
is (for example, `f"{ x = }"`. If there are any nested f-strings, then
the whitespace in them needs to be preserved as well which means that
we'll stop formatting the f-string as soon as we encounter a debug
expression.
```python
f"outer { x = !s :.3f}"
# ^^
# We can remove these whitespaces
```
Now, the whitespace doesn't need to be preserved around conversion spec
and format specifiers, so we'll format them as usual but we won't be
formatting any nested f-string within the format specifier.
### Miscellaneous
- The
[`hug_parens_with_braces_and_square_brackets`](https://github.com/astral-sh/ruff/issues/8279)
preview style isn't implemented w.r.t. the f-string curly braces.
- The
[indentation](https://github.com/astral-sh/ruff/discussions/9785#discussioncomment-8470590)
is always relative to the f-string containing statement
## Test Plan
* Add new test cases
* Review existing snapshot changes
* Review the ecosystem changes
[PEP 701]: https://peps.python.org/pep-0701/
## Summary
I always found it odd that we had to pass this in, since it's really
higher-level context for the error. The awkwardness is further evidenced
by the fact that we pass in fake values everywhere (even outside of
tests). The source path isn't actually used to display the error; it's
only accessed elsewhere to _re-display_ the error in certain cases. This
PR modifies to instead pass the path directly in those cases.
Brent Westbrook
Dylan
Micha Reiser
Ibraheem Ahmed
Carl Meyer
Alex Waygood
Douglas Creager
Dhruv Manilawala
Charlie Marsh
Dimitri Papadopoulos Orfanos