rust/compiler/rustc_expand/src/config.rs

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

451 lines
17 KiB
Rust
Raw Normal View History

2020-03-08 21:32:25 +00:00
//! Conditional compilation stripping.
use crate::errors::{
2023-12-10 17:34:13 +00:00
FeatureNotAllowed, FeatureRemoved, FeatureRemovedReason, InvalidCfg, MalformedFeatureAttribute,
MalformedFeatureAttributeHelp, RemoveExprNotSupported,
};
use rustc_ast::ptr::P;
use rustc_ast::token::{Delimiter, Token, TokenKind};
use rustc_ast::tokenstream::{AttrTokenStream, AttrTokenTree, DelimSpacing, DelimSpan, Spacing};
use rustc_ast::tokenstream::{LazyAttrTokenStream, TokenTree};
use rustc_ast::NodeId;
use rustc_ast::{self as ast, AttrStyle, Attribute, HasAttrs, HasTokens, MetaItem};
use rustc_attr as attr;
use rustc_data_structures::flat_map_in_place::FlatMapInPlace;
use rustc_feature::Features;
use rustc_feature::{ACCEPTED_FEATURES, REMOVED_FEATURES, UNSTABLE_FEATURES};
use rustc_parse::validate_attr;
use rustc_session::parse::feature_err;
use rustc_session::Session;
2020-01-02 11:33:56 +00:00
use rustc_span::symbol::{sym, Symbol};
use rustc_span::Span;
use thin_vec::ThinVec;
use tracing::instrument;
2019-02-06 17:33:01 +00:00
2016-06-01 02:13:45 +00:00
/// A folder that strips out items that do not belong in the current configuration.
pub struct StripUnconfigured<'a> {
pub sess: &'a Session,
2016-06-11 01:37:24 +00:00
pub features: Option<&'a Features>,
Implement token-based handling of attributes during expansion This PR modifies the macro expansion infrastructure to handle attributes in a fully token-based manner. As a result: * Derives macros no longer lose spans when their input is modified by eager cfg-expansion. This is accomplished by performing eager cfg-expansion on the token stream that we pass to the derive proc-macro * Inner attributes now preserve spans in all cases, including when we have multiple inner attributes in a row. This is accomplished through the following changes: * New structs `AttrAnnotatedTokenStream` and `AttrAnnotatedTokenTree` are introduced. These are very similar to a normal `TokenTree`, but they also track the position of attributes and attribute targets within the stream. They are built when we collect tokens during parsing. An `AttrAnnotatedTokenStream` is converted to a regular `TokenStream` when we invoke a macro. * Token capturing and `LazyTokenStream` are modified to work with `AttrAnnotatedTokenStream`. A new `ReplaceRange` type is introduced, which is created during the parsing of a nested AST node to make the 'outer' AST node aware of the attributes and attribute target stored deeper in the token stream. * When we need to perform eager cfg-expansion (either due to `#[derive]` or `#[cfg_eval]`), we tokenize and reparse our target, capturing additional information about the locations of `#[cfg]` and `#[cfg_attr]` attributes at any depth within the target. This is a performance optimization, allowing us to perform less work in the typical case where captured tokens never have eager cfg-expansion run.
2020-11-28 23:33:17 +00:00
/// If `true`, perform cfg-stripping on attached tokens.
/// This is only used for the input to derive macros,
/// which needs eager expansion of `cfg` and `cfg_attr`
pub config_tokens: bool,
pub lint_node_id: NodeId,
}
pub fn features(sess: &Session, krate_attrs: &[Attribute], crate_name: Symbol) -> Features {
fn feature_list(attr: &Attribute) -> ThinVec<ast::NestedMetaItem> {
if attr.has_name(sym::feature)
&& let Some(list) = attr.meta_item_list()
{
list
} else {
ThinVec::new()
}
}
2020-01-02 11:33:56 +00:00
let mut features = Features::default();
// Process all features declared in the code.
2020-01-02 11:33:56 +00:00
for attr in krate_attrs {
for mi in feature_list(attr) {
2020-01-02 11:33:56 +00:00
let name = match mi.ident() {
Some(ident) if mi.is_word() => ident.name,
Some(ident) => {
sess.dcx().emit_err(MalformedFeatureAttribute {
span: mi.span(),
help: MalformedFeatureAttributeHelp::Suggestion {
span: mi.span(),
suggestion: ident.name,
},
});
2020-01-02 11:33:56 +00:00
continue;
}
None => {
sess.dcx().emit_err(MalformedFeatureAttribute {
span: mi.span(),
help: MalformedFeatureAttributeHelp::Label { span: mi.span() },
});
2020-01-02 11:33:56 +00:00
continue;
}
};
// If the declared feature has been removed, issue an error.
if let Some(f) = REMOVED_FEATURES.iter().find(|f| name == f.feature.name) {
sess.dcx().emit_err(FeatureRemoved {
span: mi.span(),
reason: f.reason.map(|reason| FeatureRemovedReason { reason }),
});
continue;
2020-01-02 11:33:56 +00:00
}
// If the declared feature is stable, record it.
if let Some(f) = ACCEPTED_FEATURES.iter().find(|f| name == f.name) {
let since = Some(Symbol::intern(f.since));
features.set_declared_lang_feature(name, mi.span(), since);
2020-01-02 11:33:56 +00:00
continue;
}
// If `-Z allow-features` is used and the declared feature is
// unstable and not also listed as one of the allowed features,
// issue an error.
if let Some(allowed) = sess.opts.unstable_opts.allow_features.as_ref() {
if allowed.iter().all(|f| name.as_str() != f) {
sess.dcx().emit_err(FeatureNotAllowed { span: mi.span(), name });
2020-01-02 11:33:56 +00:00
continue;
}
}
// If the declared feature is unstable, record it.
if let Some(f) = UNSTABLE_FEATURES.iter().find(|f| name == f.feature.name) {
(f.set_enabled)(&mut features);
2024-04-30 23:27:20 +00:00
// When the ICE comes from core, alloc or std (approximation of the standard
// library), there's a chance that the person hitting the ICE may be using
// -Zbuild-std or similar with an untested target. The bug is probably in the
// standard library and not the compiler in that case, but that doesn't really
// matter - we want a bug report.
if features.internal(name)
&& ![sym::core, sym::alloc, sym::std].contains(&crate_name)
{
sess.using_internal_features.store(true, std::sync::atomic::Ordering::Relaxed);
}
features.set_declared_lang_feature(name, mi.span(), None);
2020-01-02 11:33:56 +00:00
continue;
}
// Otherwise, the feature is unknown. Record it as a lib feature.
// It will be checked later.
features.set_declared_lib_feature(name, mi.span());
2020-01-02 11:33:56 +00:00
}
}
features
}
pub fn pre_configure_attrs(sess: &Session, attrs: &[Attribute]) -> ast::AttrVec {
let strip_unconfigured = StripUnconfigured {
sess,
features: None,
config_tokens: false,
lint_node_id: ast::CRATE_NODE_ID,
};
attrs
.iter()
.flat_map(|attr| strip_unconfigured.process_cfg_attr(attr))
.take_while(|attr| !is_cfg(attr) || strip_unconfigured.cfg_true(attr).0)
.collect()
}
#[macro_export]
macro_rules! configure {
($this:ident, $node:ident) => {
match $this.configure($node) {
Some(node) => node,
None => return Default::default(),
}
};
}
impl<'a> StripUnconfigured<'a> {
pub fn configure<T: HasAttrs + HasTokens>(&self, mut node: T) -> Option<T> {
Overhaul `syntax::fold::Folder`. This commit changes `syntax::fold::Folder` from a functional style (where most methods take a `T` and produce a new `T`) to a more imperative style (where most methods take and modify a `&mut T`), and renames it `syntax::mut_visit::MutVisitor`. The first benefit is speed. The functional style does not require any reallocations, due to the use of `P::map` and `MoveMap::move_{,flat_}map`. However, every field in the AST must be overwritten; even those fields that are unchanged are overwritten with the same value. This causes a lot of unnecessary memory writes. The imperative style reduces instruction counts by 1--3% across a wide range of workloads, particularly incremental workloads. The second benefit is conciseness; the imperative style is usually more concise. E.g. compare the old functional style: ``` fn fold_abc(&mut self, abc: ABC) { ABC { a: fold_a(abc.a), b: fold_b(abc.b), c: abc.c, } } ``` with the imperative style: ``` fn visit_abc(&mut self, ABC { a, b, c: _ }: &mut ABC) { visit_a(a); visit_b(b); } ``` (The reductions get larger in more complex examples.) Overall, the patch removes over 200 lines of code -- even though the new code has more comments -- and a lot of the remaining lines have fewer characters. Some notes: - The old style used methods called `fold_*`. The new style mostly uses methods called `visit_*`, but there are a few methods that map a `T` to something other than a `T`, which are called `flat_map_*` (`T` maps to multiple `T`s) or `filter_map_*` (`T` maps to 0 or 1 `T`s). - `move_map.rs`/`MoveMap`/`move_map`/`move_flat_map` are renamed `map_in_place.rs`/`MapInPlace`/`map_in_place`/`flat_map_in_place` to reflect their slightly changed signatures. - Although this commit renames the `fold` module as `mut_visit`, it keeps it in the `fold.rs` file, so as not to confuse git. The next commit will rename the file.
2019-02-05 04:20:55 +00:00
self.process_cfg_attrs(&mut node);
self.in_cfg(node.attrs()).then(|| {
Implement token-based handling of attributes during expansion This PR modifies the macro expansion infrastructure to handle attributes in a fully token-based manner. As a result: * Derives macros no longer lose spans when their input is modified by eager cfg-expansion. This is accomplished by performing eager cfg-expansion on the token stream that we pass to the derive proc-macro * Inner attributes now preserve spans in all cases, including when we have multiple inner attributes in a row. This is accomplished through the following changes: * New structs `AttrAnnotatedTokenStream` and `AttrAnnotatedTokenTree` are introduced. These are very similar to a normal `TokenTree`, but they also track the position of attributes and attribute targets within the stream. They are built when we collect tokens during parsing. An `AttrAnnotatedTokenStream` is converted to a regular `TokenStream` when we invoke a macro. * Token capturing and `LazyTokenStream` are modified to work with `AttrAnnotatedTokenStream`. A new `ReplaceRange` type is introduced, which is created during the parsing of a nested AST node to make the 'outer' AST node aware of the attributes and attribute target stored deeper in the token stream. * When we need to perform eager cfg-expansion (either due to `#[derive]` or `#[cfg_eval]`), we tokenize and reparse our target, capturing additional information about the locations of `#[cfg]` and `#[cfg_attr]` attributes at any depth within the target. This is a performance optimization, allowing us to perform less work in the typical case where captured tokens never have eager cfg-expansion run.
2020-11-28 23:33:17 +00:00
self.try_configure_tokens(&mut node);
node
})
2016-06-01 02:13:45 +00:00
}
fn try_configure_tokens<T: HasTokens>(&self, node: &mut T) {
Implement token-based handling of attributes during expansion This PR modifies the macro expansion infrastructure to handle attributes in a fully token-based manner. As a result: * Derives macros no longer lose spans when their input is modified by eager cfg-expansion. This is accomplished by performing eager cfg-expansion on the token stream that we pass to the derive proc-macro * Inner attributes now preserve spans in all cases, including when we have multiple inner attributes in a row. This is accomplished through the following changes: * New structs `AttrAnnotatedTokenStream` and `AttrAnnotatedTokenTree` are introduced. These are very similar to a normal `TokenTree`, but they also track the position of attributes and attribute targets within the stream. They are built when we collect tokens during parsing. An `AttrAnnotatedTokenStream` is converted to a regular `TokenStream` when we invoke a macro. * Token capturing and `LazyTokenStream` are modified to work with `AttrAnnotatedTokenStream`. A new `ReplaceRange` type is introduced, which is created during the parsing of a nested AST node to make the 'outer' AST node aware of the attributes and attribute target stored deeper in the token stream. * When we need to perform eager cfg-expansion (either due to `#[derive]` or `#[cfg_eval]`), we tokenize and reparse our target, capturing additional information about the locations of `#[cfg]` and `#[cfg_attr]` attributes at any depth within the target. This is a performance optimization, allowing us to perform less work in the typical case where captured tokens never have eager cfg-expansion run.
2020-11-28 23:33:17 +00:00
if self.config_tokens {
if let Some(Some(tokens)) = node.tokens_mut() {
let attr_stream = tokens.to_attr_token_stream();
*tokens = LazyAttrTokenStream::new(self.configure_tokens(&attr_stream));
Implement token-based handling of attributes during expansion This PR modifies the macro expansion infrastructure to handle attributes in a fully token-based manner. As a result: * Derives macros no longer lose spans when their input is modified by eager cfg-expansion. This is accomplished by performing eager cfg-expansion on the token stream that we pass to the derive proc-macro * Inner attributes now preserve spans in all cases, including when we have multiple inner attributes in a row. This is accomplished through the following changes: * New structs `AttrAnnotatedTokenStream` and `AttrAnnotatedTokenTree` are introduced. These are very similar to a normal `TokenTree`, but they also track the position of attributes and attribute targets within the stream. They are built when we collect tokens during parsing. An `AttrAnnotatedTokenStream` is converted to a regular `TokenStream` when we invoke a macro. * Token capturing and `LazyTokenStream` are modified to work with `AttrAnnotatedTokenStream`. A new `ReplaceRange` type is introduced, which is created during the parsing of a nested AST node to make the 'outer' AST node aware of the attributes and attribute target stored deeper in the token stream. * When we need to perform eager cfg-expansion (either due to `#[derive]` or `#[cfg_eval]`), we tokenize and reparse our target, capturing additional information about the locations of `#[cfg]` and `#[cfg_attr]` attributes at any depth within the target. This is a performance optimization, allowing us to perform less work in the typical case where captured tokens never have eager cfg-expansion run.
2020-11-28 23:33:17 +00:00
}
}
}
/// Performs cfg-expansion on `stream`, producing a new `AttrTokenStream`.
Implement token-based handling of attributes during expansion This PR modifies the macro expansion infrastructure to handle attributes in a fully token-based manner. As a result: * Derives macros no longer lose spans when their input is modified by eager cfg-expansion. This is accomplished by performing eager cfg-expansion on the token stream that we pass to the derive proc-macro * Inner attributes now preserve spans in all cases, including when we have multiple inner attributes in a row. This is accomplished through the following changes: * New structs `AttrAnnotatedTokenStream` and `AttrAnnotatedTokenTree` are introduced. These are very similar to a normal `TokenTree`, but they also track the position of attributes and attribute targets within the stream. They are built when we collect tokens during parsing. An `AttrAnnotatedTokenStream` is converted to a regular `TokenStream` when we invoke a macro. * Token capturing and `LazyTokenStream` are modified to work with `AttrAnnotatedTokenStream`. A new `ReplaceRange` type is introduced, which is created during the parsing of a nested AST node to make the 'outer' AST node aware of the attributes and attribute target stored deeper in the token stream. * When we need to perform eager cfg-expansion (either due to `#[derive]` or `#[cfg_eval]`), we tokenize and reparse our target, capturing additional information about the locations of `#[cfg]` and `#[cfg_attr]` attributes at any depth within the target. This is a performance optimization, allowing us to perform less work in the typical case where captured tokens never have eager cfg-expansion run.
2020-11-28 23:33:17 +00:00
/// This is only used during the invocation of `derive` proc-macros,
/// which require that we cfg-expand their entire input.
/// Normal cfg-expansion operates on parsed AST nodes via the `configure` method
fn configure_tokens(&self, stream: &AttrTokenStream) -> AttrTokenStream {
fn can_skip(stream: &AttrTokenStream) -> bool {
stream.0.iter().all(|tree| match tree {
AttrTokenTree::Attributes(_) => false,
AttrTokenTree::Token(..) => true,
AttrTokenTree::Delimited(.., inner) => can_skip(inner),
Implement token-based handling of attributes during expansion This PR modifies the macro expansion infrastructure to handle attributes in a fully token-based manner. As a result: * Derives macros no longer lose spans when their input is modified by eager cfg-expansion. This is accomplished by performing eager cfg-expansion on the token stream that we pass to the derive proc-macro * Inner attributes now preserve spans in all cases, including when we have multiple inner attributes in a row. This is accomplished through the following changes: * New structs `AttrAnnotatedTokenStream` and `AttrAnnotatedTokenTree` are introduced. These are very similar to a normal `TokenTree`, but they also track the position of attributes and attribute targets within the stream. They are built when we collect tokens during parsing. An `AttrAnnotatedTokenStream` is converted to a regular `TokenStream` when we invoke a macro. * Token capturing and `LazyTokenStream` are modified to work with `AttrAnnotatedTokenStream`. A new `ReplaceRange` type is introduced, which is created during the parsing of a nested AST node to make the 'outer' AST node aware of the attributes and attribute target stored deeper in the token stream. * When we need to perform eager cfg-expansion (either due to `#[derive]` or `#[cfg_eval]`), we tokenize and reparse our target, capturing additional information about the locations of `#[cfg]` and `#[cfg_attr]` attributes at any depth within the target. This is a performance optimization, allowing us to perform less work in the typical case where captured tokens never have eager cfg-expansion run.
2020-11-28 23:33:17 +00:00
})
}
if can_skip(stream) {
return stream.clone();
}
Implement token-based handling of attributes during expansion This PR modifies the macro expansion infrastructure to handle attributes in a fully token-based manner. As a result: * Derives macros no longer lose spans when their input is modified by eager cfg-expansion. This is accomplished by performing eager cfg-expansion on the token stream that we pass to the derive proc-macro * Inner attributes now preserve spans in all cases, including when we have multiple inner attributes in a row. This is accomplished through the following changes: * New structs `AttrAnnotatedTokenStream` and `AttrAnnotatedTokenTree` are introduced. These are very similar to a normal `TokenTree`, but they also track the position of attributes and attribute targets within the stream. They are built when we collect tokens during parsing. An `AttrAnnotatedTokenStream` is converted to a regular `TokenStream` when we invoke a macro. * Token capturing and `LazyTokenStream` are modified to work with `AttrAnnotatedTokenStream`. A new `ReplaceRange` type is introduced, which is created during the parsing of a nested AST node to make the 'outer' AST node aware of the attributes and attribute target stored deeper in the token stream. * When we need to perform eager cfg-expansion (either due to `#[derive]` or `#[cfg_eval]`), we tokenize and reparse our target, capturing additional information about the locations of `#[cfg]` and `#[cfg_attr]` attributes at any depth within the target. This is a performance optimization, allowing us to perform less work in the typical case where captured tokens never have eager cfg-expansion run.
2020-11-28 23:33:17 +00:00
let trees: Vec<_> = stream
.0
.iter()
2023-11-13 13:24:55 +00:00
.flat_map(|tree| match tree.clone() {
AttrTokenTree::Attributes(mut data) => {
data.attrs.flat_map_in_place(|attr| self.process_cfg_attr(&attr));
Implement token-based handling of attributes during expansion This PR modifies the macro expansion infrastructure to handle attributes in a fully token-based manner. As a result: * Derives macros no longer lose spans when their input is modified by eager cfg-expansion. This is accomplished by performing eager cfg-expansion on the token stream that we pass to the derive proc-macro * Inner attributes now preserve spans in all cases, including when we have multiple inner attributes in a row. This is accomplished through the following changes: * New structs `AttrAnnotatedTokenStream` and `AttrAnnotatedTokenTree` are introduced. These are very similar to a normal `TokenTree`, but they also track the position of attributes and attribute targets within the stream. They are built when we collect tokens during parsing. An `AttrAnnotatedTokenStream` is converted to a regular `TokenStream` when we invoke a macro. * Token capturing and `LazyTokenStream` are modified to work with `AttrAnnotatedTokenStream`. A new `ReplaceRange` type is introduced, which is created during the parsing of a nested AST node to make the 'outer' AST node aware of the attributes and attribute target stored deeper in the token stream. * When we need to perform eager cfg-expansion (either due to `#[derive]` or `#[cfg_eval]`), we tokenize and reparse our target, capturing additional information about the locations of `#[cfg]` and `#[cfg_attr]` attributes at any depth within the target. This is a performance optimization, allowing us to perform less work in the typical case where captured tokens never have eager cfg-expansion run.
2020-11-28 23:33:17 +00:00
if self.in_cfg(&data.attrs) {
data.tokens = LazyAttrTokenStream::new(
self.configure_tokens(&data.tokens.to_attr_token_stream()),
Implement token-based handling of attributes during expansion This PR modifies the macro expansion infrastructure to handle attributes in a fully token-based manner. As a result: * Derives macros no longer lose spans when their input is modified by eager cfg-expansion. This is accomplished by performing eager cfg-expansion on the token stream that we pass to the derive proc-macro * Inner attributes now preserve spans in all cases, including when we have multiple inner attributes in a row. This is accomplished through the following changes: * New structs `AttrAnnotatedTokenStream` and `AttrAnnotatedTokenTree` are introduced. These are very similar to a normal `TokenTree`, but they also track the position of attributes and attribute targets within the stream. They are built when we collect tokens during parsing. An `AttrAnnotatedTokenStream` is converted to a regular `TokenStream` when we invoke a macro. * Token capturing and `LazyTokenStream` are modified to work with `AttrAnnotatedTokenStream`. A new `ReplaceRange` type is introduced, which is created during the parsing of a nested AST node to make the 'outer' AST node aware of the attributes and attribute target stored deeper in the token stream. * When we need to perform eager cfg-expansion (either due to `#[derive]` or `#[cfg_eval]`), we tokenize and reparse our target, capturing additional information about the locations of `#[cfg]` and `#[cfg_attr]` attributes at any depth within the target. This is a performance optimization, allowing us to perform less work in the typical case where captured tokens never have eager cfg-expansion run.
2020-11-28 23:33:17 +00:00
);
Some(AttrTokenTree::Attributes(data)).into_iter()
Implement token-based handling of attributes during expansion This PR modifies the macro expansion infrastructure to handle attributes in a fully token-based manner. As a result: * Derives macros no longer lose spans when their input is modified by eager cfg-expansion. This is accomplished by performing eager cfg-expansion on the token stream that we pass to the derive proc-macro * Inner attributes now preserve spans in all cases, including when we have multiple inner attributes in a row. This is accomplished through the following changes: * New structs `AttrAnnotatedTokenStream` and `AttrAnnotatedTokenTree` are introduced. These are very similar to a normal `TokenTree`, but they also track the position of attributes and attribute targets within the stream. They are built when we collect tokens during parsing. An `AttrAnnotatedTokenStream` is converted to a regular `TokenStream` when we invoke a macro. * Token capturing and `LazyTokenStream` are modified to work with `AttrAnnotatedTokenStream`. A new `ReplaceRange` type is introduced, which is created during the parsing of a nested AST node to make the 'outer' AST node aware of the attributes and attribute target stored deeper in the token stream. * When we need to perform eager cfg-expansion (either due to `#[derive]` or `#[cfg_eval]`), we tokenize and reparse our target, capturing additional information about the locations of `#[cfg]` and `#[cfg_attr]` attributes at any depth within the target. This is a performance optimization, allowing us to perform less work in the typical case where captured tokens never have eager cfg-expansion run.
2020-11-28 23:33:17 +00:00
} else {
None.into_iter()
}
}
AttrTokenTree::Delimited(sp, spacing, delim, mut inner) => {
Implement token-based handling of attributes during expansion This PR modifies the macro expansion infrastructure to handle attributes in a fully token-based manner. As a result: * Derives macros no longer lose spans when their input is modified by eager cfg-expansion. This is accomplished by performing eager cfg-expansion on the token stream that we pass to the derive proc-macro * Inner attributes now preserve spans in all cases, including when we have multiple inner attributes in a row. This is accomplished through the following changes: * New structs `AttrAnnotatedTokenStream` and `AttrAnnotatedTokenTree` are introduced. These are very similar to a normal `TokenTree`, but they also track the position of attributes and attribute targets within the stream. They are built when we collect tokens during parsing. An `AttrAnnotatedTokenStream` is converted to a regular `TokenStream` when we invoke a macro. * Token capturing and `LazyTokenStream` are modified to work with `AttrAnnotatedTokenStream`. A new `ReplaceRange` type is introduced, which is created during the parsing of a nested AST node to make the 'outer' AST node aware of the attributes and attribute target stored deeper in the token stream. * When we need to perform eager cfg-expansion (either due to `#[derive]` or `#[cfg_eval]`), we tokenize and reparse our target, capturing additional information about the locations of `#[cfg]` and `#[cfg_attr]` attributes at any depth within the target. This is a performance optimization, allowing us to perform less work in the typical case where captured tokens never have eager cfg-expansion run.
2020-11-28 23:33:17 +00:00
inner = self.configure_tokens(&inner);
Some(AttrTokenTree::Delimited(sp, spacing, delim, inner)).into_iter()
Implement token-based handling of attributes during expansion This PR modifies the macro expansion infrastructure to handle attributes in a fully token-based manner. As a result: * Derives macros no longer lose spans when their input is modified by eager cfg-expansion. This is accomplished by performing eager cfg-expansion on the token stream that we pass to the derive proc-macro * Inner attributes now preserve spans in all cases, including when we have multiple inner attributes in a row. This is accomplished through the following changes: * New structs `AttrAnnotatedTokenStream` and `AttrAnnotatedTokenTree` are introduced. These are very similar to a normal `TokenTree`, but they also track the position of attributes and attribute targets within the stream. They are built when we collect tokens during parsing. An `AttrAnnotatedTokenStream` is converted to a regular `TokenStream` when we invoke a macro. * Token capturing and `LazyTokenStream` are modified to work with `AttrAnnotatedTokenStream`. A new `ReplaceRange` type is introduced, which is created during the parsing of a nested AST node to make the 'outer' AST node aware of the attributes and attribute target stored deeper in the token stream. * When we need to perform eager cfg-expansion (either due to `#[derive]` or `#[cfg_eval]`), we tokenize and reparse our target, capturing additional information about the locations of `#[cfg]` and `#[cfg_attr]` attributes at any depth within the target. This is a performance optimization, allowing us to perform less work in the typical case where captured tokens never have eager cfg-expansion run.
2020-11-28 23:33:17 +00:00
}
2022-12-23 17:34:23 +00:00
AttrTokenTree::Token(ref token, _)
if let TokenKind::Interpolated(nt) = &token.kind =>
{
2021-08-16 15:29:49 +00:00
panic!("Nonterminal should have been flattened at {:?}: {:?}", token.span, nt);
}
AttrTokenTree::Token(token, spacing) => {
Some(AttrTokenTree::Token(token, spacing)).into_iter()
Implement token-based handling of attributes during expansion This PR modifies the macro expansion infrastructure to handle attributes in a fully token-based manner. As a result: * Derives macros no longer lose spans when their input is modified by eager cfg-expansion. This is accomplished by performing eager cfg-expansion on the token stream that we pass to the derive proc-macro * Inner attributes now preserve spans in all cases, including when we have multiple inner attributes in a row. This is accomplished through the following changes: * New structs `AttrAnnotatedTokenStream` and `AttrAnnotatedTokenTree` are introduced. These are very similar to a normal `TokenTree`, but they also track the position of attributes and attribute targets within the stream. They are built when we collect tokens during parsing. An `AttrAnnotatedTokenStream` is converted to a regular `TokenStream` when we invoke a macro. * Token capturing and `LazyTokenStream` are modified to work with `AttrAnnotatedTokenStream`. A new `ReplaceRange` type is introduced, which is created during the parsing of a nested AST node to make the 'outer' AST node aware of the attributes and attribute target stored deeper in the token stream. * When we need to perform eager cfg-expansion (either due to `#[derive]` or `#[cfg_eval]`), we tokenize and reparse our target, capturing additional information about the locations of `#[cfg]` and `#[cfg_attr]` attributes at any depth within the target. This is a performance optimization, allowing us to perform less work in the typical case where captured tokens never have eager cfg-expansion run.
2020-11-28 23:33:17 +00:00
}
})
.collect();
AttrTokenStream::new(trees)
}
/// Parse and expand all `cfg_attr` attributes into a list of attributes
/// that are within each `cfg_attr` that has a true configuration predicate.
///
2020-03-06 11:13:55 +00:00
/// Gives compiler warnings if any `cfg_attr` does not contain any
/// attributes and is in the original source code. Gives compiler errors if
/// the syntax of any `cfg_attr` is incorrect.
fn process_cfg_attrs<T: HasAttrs>(&self, node: &mut T) {
Overhaul `syntax::fold::Folder`. This commit changes `syntax::fold::Folder` from a functional style (where most methods take a `T` and produce a new `T`) to a more imperative style (where most methods take and modify a `&mut T`), and renames it `syntax::mut_visit::MutVisitor`. The first benefit is speed. The functional style does not require any reallocations, due to the use of `P::map` and `MoveMap::move_{,flat_}map`. However, every field in the AST must be overwritten; even those fields that are unchanged are overwritten with the same value. This causes a lot of unnecessary memory writes. The imperative style reduces instruction counts by 1--3% across a wide range of workloads, particularly incremental workloads. The second benefit is conciseness; the imperative style is usually more concise. E.g. compare the old functional style: ``` fn fold_abc(&mut self, abc: ABC) { ABC { a: fold_a(abc.a), b: fold_b(abc.b), c: abc.c, } } ``` with the imperative style: ``` fn visit_abc(&mut self, ABC { a, b, c: _ }: &mut ABC) { visit_a(a); visit_b(b); } ``` (The reductions get larger in more complex examples.) Overall, the patch removes over 200 lines of code -- even though the new code has more comments -- and a lot of the remaining lines have fewer characters. Some notes: - The old style used methods called `fold_*`. The new style mostly uses methods called `visit_*`, but there are a few methods that map a `T` to something other than a `T`, which are called `flat_map_*` (`T` maps to multiple `T`s) or `filter_map_*` (`T` maps to 0 or 1 `T`s). - `move_map.rs`/`MoveMap`/`move_map`/`move_flat_map` are renamed `map_in_place.rs`/`MapInPlace`/`map_in_place`/`flat_map_in_place` to reflect their slightly changed signatures. - Although this commit renames the `fold` module as `mut_visit`, it keeps it in the `fold.rs` file, so as not to confuse git. The next commit will rename the file.
2019-02-05 04:20:55 +00:00
node.visit_attrs(|attrs| {
attrs.flat_map_in_place(|attr| self.process_cfg_attr(&attr));
Overhaul `syntax::fold::Folder`. This commit changes `syntax::fold::Folder` from a functional style (where most methods take a `T` and produce a new `T`) to a more imperative style (where most methods take and modify a `&mut T`), and renames it `syntax::mut_visit::MutVisitor`. The first benefit is speed. The functional style does not require any reallocations, due to the use of `P::map` and `MoveMap::move_{,flat_}map`. However, every field in the AST must be overwritten; even those fields that are unchanged are overwritten with the same value. This causes a lot of unnecessary memory writes. The imperative style reduces instruction counts by 1--3% across a wide range of workloads, particularly incremental workloads. The second benefit is conciseness; the imperative style is usually more concise. E.g. compare the old functional style: ``` fn fold_abc(&mut self, abc: ABC) { ABC { a: fold_a(abc.a), b: fold_b(abc.b), c: abc.c, } } ``` with the imperative style: ``` fn visit_abc(&mut self, ABC { a, b, c: _ }: &mut ABC) { visit_a(a); visit_b(b); } ``` (The reductions get larger in more complex examples.) Overall, the patch removes over 200 lines of code -- even though the new code has more comments -- and a lot of the remaining lines have fewer characters. Some notes: - The old style used methods called `fold_*`. The new style mostly uses methods called `visit_*`, but there are a few methods that map a `T` to something other than a `T`, which are called `flat_map_*` (`T` maps to multiple `T`s) or `filter_map_*` (`T` maps to 0 or 1 `T`s). - `move_map.rs`/`MoveMap`/`move_map`/`move_flat_map` are renamed `map_in_place.rs`/`MapInPlace`/`map_in_place`/`flat_map_in_place` to reflect their slightly changed signatures. - Although this commit renames the `fold` module as `mut_visit`, it keeps it in the `fold.rs` file, so as not to confuse git. The next commit will rename the file.
2019-02-05 04:20:55 +00:00
});
2016-06-01 02:13:45 +00:00
}
fn process_cfg_attr(&self, attr: &Attribute) -> Vec<Attribute> {
if attr.has_name(sym::cfg_attr) {
self.expand_cfg_attr(attr, true)
} else {
vec![attr.clone()]
}
}
/// Parse and expand a single `cfg_attr` attribute into a list of attributes
/// when the configuration predicate is true, or otherwise expand into an
/// empty list of attributes.
///
2018-10-22 16:21:55 +00:00
/// Gives a compiler warning when the `cfg_attr` contains no attributes and
/// is in the original source file. Gives a compiler error if the syntax of
2019-02-08 13:53:55 +00:00
/// the attribute is incorrect.
#[allow(rustc::untranslatable_diagnostic)] // FIXME: make this translatable
pub(crate) fn expand_cfg_attr(&self, attr: &Attribute, recursive: bool) -> Vec<Attribute> {
2022-02-18 23:48:49 +00:00
let Some((cfg_predicate, expanded_attrs)) =
rustc_parse::parse_cfg_attr(attr, &self.sess.psess)
else {
2022-02-18 23:48:49 +00:00
return vec![];
};
2019-06-22 10:11:01 +00:00
// Lint on zero attributes in source.
if expanded_attrs.is_empty() {
self.sess.psess.buffer_lint(
rustc_lint_defs::builtin::UNUSED_ATTRIBUTES,
attr.span,
ast::CRATE_NODE_ID,
"`#[cfg_attr]` does not expand to any attributes",
);
2018-10-04 11:55:47 +00:00
}
if !attr::cfg_matches(&cfg_predicate, &self.sess, self.lint_node_id, self.features) {
2019-12-05 05:45:50 +00:00
return vec![];
}
if recursive {
// We call `process_cfg_attr` recursively in case there's a
// `cfg_attr` inside of another `cfg_attr`. E.g.
// `#[cfg_attr(false, cfg_attr(true, some_attr))]`.
expanded_attrs
.into_iter()
.flat_map(|item| self.process_cfg_attr(&self.expand_cfg_attr_item(attr, item)))
.collect()
} else {
expanded_attrs.into_iter().map(|item| self.expand_cfg_attr_item(attr, item)).collect()
}
}
#[allow(rustc::untranslatable_diagnostic)] // FIXME: make this translatable
fn expand_cfg_attr_item(
&self,
attr: &Attribute,
(item, item_span): (ast::AttrItem, Span),
) -> Attribute {
let orig_tokens = attr.tokens();
// We are taking an attribute of the form `#[cfg_attr(pred, attr)]`
// and producing an attribute of the form `#[attr]`. We
// have captured tokens for `attr` itself, but we need to
// synthesize tokens for the wrapper `#` and `[]`, which
// we do below.
// Use the `#` in `#[cfg_attr(pred, attr)]` as the `#` token
// for `attr` when we expand it to `#[attr]`
let mut orig_trees = orig_tokens.trees();
let TokenTree::Token(pound_token @ Token { kind: TokenKind::Pound, .. }, _) =
orig_trees.next().unwrap().clone()
else {
panic!("Bad tokens for attribute {attr:?}");
};
let pound_span = pound_token.span;
// We don't really have a good span to use for the synthesized `[]`
// in `#[attr]`, so just use the span of the `#` token.
let bracket_group = AttrTokenTree::Delimited(
DelimSpan::from_single(pound_span),
DelimSpacing::new(Spacing::JointHidden, Spacing::Alone),
Delimiter::Bracket,
item.tokens
.as_ref()
.unwrap_or_else(|| panic!("Missing tokens for {item:?}"))
.to_attr_token_stream(),
);
Improve `print_tts` by changing `tokenstream::Spacing`. `tokenstream::Spacing` appears on all `TokenTree::Token` instances, both punct and non-punct. Its current usage: - `Joint` means "can join with the next token *and* that token is a punct". - `Alone` means "cannot join with the next token *or* can join with the next token but that token is not a punct". The fact that `Alone` is used for two different cases is awkward. This commit augments `tokenstream::Spacing` with a new variant `JointHidden`, resulting in: - `Joint` means "can join with the next token *and* that token is a punct". - `JointHidden` means "can join with the next token *and* that token is a not a punct". - `Alone` means "cannot join with the next token". This *drastically* improves the output of `print_tts`. For example, this: ``` stringify!(let a: Vec<u32> = vec![];) ``` currently produces this string: ``` let a : Vec < u32 > = vec! [] ; ``` With this PR, it now produces this string: ``` let a: Vec<u32> = vec![] ; ``` (The space after the `]` is because `TokenTree::Delimited` currently doesn't have spacing information. The subsequent commit fixes this.) The new `print_tts` doesn't replicate original code perfectly. E.g. multiple space characters will be condensed into a single space character. But it's much improved. `print_tts` still produces the old, uglier output for code produced by proc macros. Because we have to translate the generated code from `proc_macro::Spacing` to the more expressive `token::Spacing`, which results in too much `proc_macro::Along` usage and no `proc_macro::JointHidden` usage. So `space_between` still exists and is used by `print_tts` in conjunction with the `Spacing` field. This change will also help with the removal of `Token::Interpolated`. Currently interpolated tokens are pretty-printed nicely via AST pretty printing. `Token::Interpolated` removal will mean they get printed with `print_tts`. Without this change, that would result in much uglier output for code produced by decl macro expansions. With this change, AST pretty printing and `print_tts` produce similar results. The commit also tweaks the comments on `proc_macro::Spacing`. In particular, it refers to "compound tokens" rather than "multi-char operators" because lifetimes aren't operators.
2023-08-08 01:43:44 +00:00
let trees = if attr.style == AttrStyle::Inner {
// For inner attributes, we do the same thing for the `!` in `#![some_attr]`
let TokenTree::Token(bang_token @ Token { kind: TokenKind::Not, .. }, _) =
orig_trees.next().unwrap().clone()
else {
panic!("Bad tokens for attribute {attr:?}");
};
vec![
AttrTokenTree::Token(pound_token, Spacing::Joint),
AttrTokenTree::Token(bang_token, Spacing::JointHidden),
bracket_group,
]
} else {
vec![AttrTokenTree::Token(pound_token, Spacing::JointHidden), bracket_group]
};
let tokens = Some(LazyAttrTokenStream::new(AttrTokenStream::new(trees)));
2022-09-02 08:29:40 +00:00
let attr = attr::mk_attr_from_item(
&self.sess.psess.attr_id_generator,
2022-09-02 08:29:40 +00:00
item,
tokens,
attr.style,
item_span,
);
if attr.has_name(sym::crate_type) {
self.sess.psess.buffer_lint(
rustc_lint_defs::builtin::DEPRECATED_CFG_ATTR_CRATE_TYPE_NAME,
attr.span,
ast::CRATE_NODE_ID,
"`crate_type` within an `#![cfg_attr] attribute is deprecated`",
);
}
if attr.has_name(sym::crate_name) {
self.sess.psess.buffer_lint(
rustc_lint_defs::builtin::DEPRECATED_CFG_ATTR_CRATE_TYPE_NAME,
attr.span,
ast::CRATE_NODE_ID,
"`crate_name` within an `#![cfg_attr] attribute is deprecated`",
);
}
attr
2019-12-05 05:45:50 +00:00
}
2019-02-08 13:53:55 +00:00
/// Determines if a node with the given attributes should be included in this configuration.
fn in_cfg(&self, attrs: &[Attribute]) -> bool {
attrs.iter().all(|attr| !is_cfg(attr) || self.cfg_true(attr).0)
}
pub(crate) fn cfg_true(&self, attr: &Attribute) -> (bool, Option<MetaItem>) {
let meta_item = match validate_attr::parse_meta(&self.sess.psess, attr) {
Ok(meta_item) => meta_item,
Make `DiagnosticBuilder::emit` consuming. This works for most of its call sites. This is nice, because `emit` very much makes sense as a consuming operation -- indeed, `DiagnosticBuilderState` exists to ensure no diagnostic is emitted twice, but it uses runtime checks. For the small number of call sites where a consuming emit doesn't work, the commit adds `DiagnosticBuilder::emit_without_consuming`. (This will be removed in subsequent commits.) Likewise, `emit_unless` becomes consuming. And `delay_as_bug` becomes consuming, while `delay_as_bug_without_consuming` is added (which will also be removed in subsequent commits.) All this requires significant changes to `DiagnosticBuilder`'s chaining methods. Currently `DiagnosticBuilder` method chaining uses a non-consuming `&mut self -> &mut Self` style, which allows chaining to be used when the chain ends in `emit()`, like so: ``` struct_err(msg).span(span).emit(); ``` But it doesn't work when producing a `DiagnosticBuilder` value, requiring this: ``` let mut err = self.struct_err(msg); err.span(span); err ``` This style of chaining won't work with consuming `emit` though. For that, we need to use to a `self -> Self` style. That also would allow `DiagnosticBuilder` production to be chained, e.g.: ``` self.struct_err(msg).span(span) ``` However, removing the `&mut self -> &mut Self` style would require that individual modifications of a `DiagnosticBuilder` go from this: ``` err.span(span); ``` to this: ``` err = err.span(span); ``` There are *many* such places. I have a high tolerance for tedious refactorings, but even I gave up after a long time trying to convert them all. Instead, this commit has it both ways: the existing `&mut self -> Self` chaining methods are kept, and new `self -> Self` chaining methods are added, all of which have a `_mv` suffix (short for "move"). Changes to the existing `forward!` macro lets this happen with very little additional boilerplate code. I chose to add the suffix to the new chaining methods rather than the existing ones, because the number of changes required is much smaller that way. This doubled chainging is a bit clumsy, but I think it is worthwhile because it allows a *lot* of good things to subsequently happen. In this commit, there are many `mut` qualifiers removed in places where diagnostics are emitted without being modified. In subsequent commits: - chaining can be used more, making the code more concise; - more use of chaining also permits the removal of redundant diagnostic APIs like `struct_err_with_code`, which can be replaced easily with `struct_err` + `code_mv`; - `emit_without_diagnostic` can be removed, which simplifies a lot of machinery, removing the need for `DiagnosticBuilderState`.
2024-01-03 01:17:35 +00:00
Err(err) => {
err.emit();
return (true, None);
2018-09-01 21:13:22 +00:00
}
};
(
parse_cfg(&meta_item, self.sess).map_or(true, |meta_item| {
attr::cfg_matches(meta_item, &self.sess, self.lint_node_id, self.features)
}),
Some(meta_item),
)
}
/// If attributes are not allowed on expressions, emit an error for `attr`
#[allow(rustc::untranslatable_diagnostic)] // FIXME: make this translatable
#[instrument(level = "trace", skip(self))]
pub(crate) fn maybe_emit_expr_attr_err(&self, attr: &Attribute) {
if self.features.is_some_and(|features| !features.stmt_expr_attributes)
&& !attr.span.allows_unstable(sym::stmt_expr_attributes)
{
let mut err = feature_err(
&self.sess,
sym::stmt_expr_attributes,
attr.span,
2019-11-30 01:40:28 +00:00
"attributes on expressions are experimental",
);
if attr.is_doc_comment() {
err.help("`///` is for documentation comments. For a plain comment, use `//`.");
2016-06-11 01:37:24 +00:00
}
err.emit();
2016-05-15 09:15:02 +00:00
}
}
#[instrument(level = "trace", skip(self))]
pub fn configure_expr(&self, expr: &mut P<ast::Expr>, method_receiver: bool) {
if !method_receiver {
for attr in expr.attrs.iter() {
self.maybe_emit_expr_attr_err(attr);
}
}
2016-06-11 01:37:24 +00:00
// If an expr is valid to cfg away it will have been removed by the
// outer stmt or expression folder before descending in here.
// Anything else is always required, and thus has to error out
// in case of a cfg attr.
//
Overhaul `syntax::fold::Folder`. This commit changes `syntax::fold::Folder` from a functional style (where most methods take a `T` and produce a new `T`) to a more imperative style (where most methods take and modify a `&mut T`), and renames it `syntax::mut_visit::MutVisitor`. The first benefit is speed. The functional style does not require any reallocations, due to the use of `P::map` and `MoveMap::move_{,flat_}map`. However, every field in the AST must be overwritten; even those fields that are unchanged are overwritten with the same value. This causes a lot of unnecessary memory writes. The imperative style reduces instruction counts by 1--3% across a wide range of workloads, particularly incremental workloads. The second benefit is conciseness; the imperative style is usually more concise. E.g. compare the old functional style: ``` fn fold_abc(&mut self, abc: ABC) { ABC { a: fold_a(abc.a), b: fold_b(abc.b), c: abc.c, } } ``` with the imperative style: ``` fn visit_abc(&mut self, ABC { a, b, c: _ }: &mut ABC) { visit_a(a); visit_b(b); } ``` (The reductions get larger in more complex examples.) Overall, the patch removes over 200 lines of code -- even though the new code has more comments -- and a lot of the remaining lines have fewer characters. Some notes: - The old style used methods called `fold_*`. The new style mostly uses methods called `visit_*`, but there are a few methods that map a `T` to something other than a `T`, which are called `flat_map_*` (`T` maps to multiple `T`s) or `filter_map_*` (`T` maps to 0 or 1 `T`s). - `move_map.rs`/`MoveMap`/`move_map`/`move_flat_map` are renamed `map_in_place.rs`/`MapInPlace`/`map_in_place`/`flat_map_in_place` to reflect their slightly changed signatures. - Although this commit renames the `fold` module as `mut_visit`, it keeps it in the `fold.rs` file, so as not to confuse git. The next commit will rename the file.
2019-02-05 04:20:55 +00:00
// N.B., this is intentionally not part of the visit_expr() function
// in order for filter_map_expr() to be able to avoid this check
2023-04-09 21:07:18 +00:00
if let Some(attr) = expr.attrs().iter().find(|a| is_cfg(a)) {
self.sess.dcx().emit_err(RemoveExprNotSupported { span: attr.span });
2016-06-11 01:37:24 +00:00
}
Implement token-based handling of attributes during expansion This PR modifies the macro expansion infrastructure to handle attributes in a fully token-based manner. As a result: * Derives macros no longer lose spans when their input is modified by eager cfg-expansion. This is accomplished by performing eager cfg-expansion on the token stream that we pass to the derive proc-macro * Inner attributes now preserve spans in all cases, including when we have multiple inner attributes in a row. This is accomplished through the following changes: * New structs `AttrAnnotatedTokenStream` and `AttrAnnotatedTokenTree` are introduced. These are very similar to a normal `TokenTree`, but they also track the position of attributes and attribute targets within the stream. They are built when we collect tokens during parsing. An `AttrAnnotatedTokenStream` is converted to a regular `TokenStream` when we invoke a macro. * Token capturing and `LazyTokenStream` are modified to work with `AttrAnnotatedTokenStream`. A new `ReplaceRange` type is introduced, which is created during the parsing of a nested AST node to make the 'outer' AST node aware of the attributes and attribute target stored deeper in the token stream. * When we need to perform eager cfg-expansion (either due to `#[derive]` or `#[cfg_eval]`), we tokenize and reparse our target, capturing additional information about the locations of `#[cfg]` and `#[cfg_attr]` attributes at any depth within the target. This is a performance optimization, allowing us to perform less work in the typical case where captured tokens never have eager cfg-expansion run.
2020-11-28 23:33:17 +00:00
self.process_cfg_attrs(expr);
self.try_configure_tokens(&mut *expr);
}
}
pub fn parse_cfg<'a>(meta_item: &'a MetaItem, sess: &Session) -> Option<&'a MetaItem> {
let span = meta_item.span;
match meta_item.meta_item_list() {
None => {
sess.dcx().emit_err(InvalidCfg::NotFollowedByParens { span });
None
}
Some([]) => {
sess.dcx().emit_err(InvalidCfg::NoPredicate { span });
None
}
Some([_, .., l]) => {
sess.dcx().emit_err(InvalidCfg::MultiplePredicates { span: l.span() });
None
}
Some([single]) => match single.meta_item() {
Some(meta_item) => Some(meta_item),
None => {
sess.dcx().emit_err(InvalidCfg::PredicateLiteral { span: single.span() });
None
}
},
}
}
fn is_cfg(attr: &Attribute) -> bool {
attr.has_name(sym::cfg)
}