#![feature(min_specialization)] #![deny(rustc::untranslatable_diagnostic)] #![deny(rustc::diagnostic_outside_of_impl)] #[macro_use] extern crate rustc_macros; pub use self::Level::*; use rustc_ast::node_id::{NodeId, NodeMap}; use rustc_ast::{AttrId, Attribute}; use rustc_data_structures::stable_hasher::{HashStable, StableHasher, ToStableHashKey}; use rustc_error_messages::{DiagnosticMessage, MultiSpan}; use rustc_hir::HashStableContext; use rustc_hir::HirId; use rustc_span::edition::Edition; use rustc_span::{sym, symbol::Ident, Span, Symbol}; use rustc_target::spec::abi::Abi; use serde::{Deserialize, Serialize}; pub mod builtin; #[macro_export] macro_rules! pluralize { ($x:expr) => { if $x != 1 { "s" } else { "" } }; ("is", $x:expr) => { if $x == 1 { "is" } else { "are" } }; ("was", $x:expr) => { if $x == 1 { "was" } else { "were" } }; ("this", $x:expr) => { if $x == 1 { "this" } else { "these" } }; } /// Indicates the confidence in the correctness of a suggestion. /// /// All suggestions are marked with an `Applicability`. Tools use the applicability of a suggestion /// to determine whether it should be automatically applied or if the user should be consulted /// before applying the suggestion. #[derive(Copy, Clone, Debug, Hash, Encodable, Decodable, Serialize, Deserialize)] #[derive(PartialEq, Eq, PartialOrd, Ord)] pub enum Applicability { /// The suggestion is definitely what the user intended, or maintains the exact meaning of the code. /// This suggestion should be automatically applied. /// /// In case of multiple `MachineApplicable` suggestions (whether as part of /// the same `multipart_suggestion` or not), all of them should be /// automatically applied. MachineApplicable, /// The suggestion may be what the user intended, but it is uncertain. The suggestion should /// result in valid Rust code if it is applied. MaybeIncorrect, /// The suggestion contains placeholders like `(...)` or `{ /* fields */ }`. The suggestion /// cannot be applied automatically because it will not result in valid Rust code. The user /// will need to fill in the placeholders. HasPlaceholders, /// The applicability of the suggestion is unknown. Unspecified, } /// Each lint expectation has a `LintExpectationId` assigned by the `LintLevelsBuilder`. /// Expected `Diagnostic`s get the lint level `Expect` which stores the `LintExpectationId` /// to match it with the actual expectation later on. /// /// The `LintExpectationId` has to be stable between compilations, as diagnostic /// instances might be loaded from cache. Lint messages can be emitted during an /// `EarlyLintPass` operating on the AST and during a `LateLintPass` traversing the /// HIR tree. The AST doesn't have enough information to create a stable id. The /// `LintExpectationId` will instead store the [`AttrId`] defining the expectation. /// These `LintExpectationId` will be updated to use the stable [`HirId`] once the /// AST has been lowered. The transformation is done by the `LintLevelsBuilder` /// /// Each lint inside the `expect` attribute is tracked individually, the `lint_index` /// identifies the lint inside the attribute and ensures that the IDs are unique. /// /// The index values have a type of `u16` to reduce the size of the `LintExpectationId`. /// It's reasonable to assume that no user will define 2^16 attributes on one node or /// have that amount of lints listed. `u16` values should therefore suffice. #[derive(Clone, Copy, PartialEq, PartialOrd, Eq, Ord, Debug, Hash, Encodable, Decodable)] pub enum LintExpectationId { /// Used for lints emitted during the `EarlyLintPass`. This id is not /// hash stable and should not be cached. Unstable { attr_id: AttrId, lint_index: Option }, /// The [`HirId`] that the lint expectation is attached to. This id is /// stable and can be cached. The additional index ensures that nodes with /// several expectations can correctly match diagnostics to the individual /// expectation. Stable { hir_id: HirId, attr_index: u16, lint_index: Option }, } impl LintExpectationId { pub fn is_stable(&self) -> bool { match self { LintExpectationId::Unstable { .. } => false, LintExpectationId::Stable { .. } => true, } } pub fn get_lint_index(&self) -> Option { let (LintExpectationId::Unstable { lint_index, .. } | LintExpectationId::Stable { lint_index, .. }) = self; *lint_index } pub fn set_lint_index(&mut self, new_lint_index: Option) { let (LintExpectationId::Unstable { ref mut lint_index, .. } | LintExpectationId::Stable { ref mut lint_index, .. }) = self; *lint_index = new_lint_index } } impl HashStable for LintExpectationId { #[inline] fn hash_stable(&self, hcx: &mut HCX, hasher: &mut StableHasher) { match self { LintExpectationId::Stable { hir_id, attr_index, lint_index: Some(lint_index) } => { hir_id.hash_stable(hcx, hasher); attr_index.hash_stable(hcx, hasher); lint_index.hash_stable(hcx, hasher); } _ => { unreachable!( "HashStable should only be called for filled and stable `LintExpectationId`" ) } } } } impl ToStableHashKey for LintExpectationId { type KeyType = (HirId, u16, u16); #[inline] fn to_stable_hash_key(&self, _: &HCX) -> Self::KeyType { match self { LintExpectationId::Stable { hir_id, attr_index, lint_index: Some(lint_index) } => { (*hir_id, *attr_index, *lint_index) } _ => { unreachable!("HashStable should only be called for a filled `LintExpectationId`") } } } } /// Setting for how to handle a lint. /// /// See: #[derive(Clone, Copy, PartialEq, PartialOrd, Eq, Ord, Debug, Hash, HashStable_Generic)] pub enum Level { /// The `allow` level will not issue any message. Allow, /// The `expect` level will suppress the lint message but in turn produce a message /// if the lint wasn't issued in the expected scope. `Expect` should not be used as /// an initial level for a lint. /// /// Note that this still means that the lint is enabled in this position and should /// be emitted, this will in turn fulfill the expectation and suppress the lint. /// /// See RFC 2383. /// /// The [`LintExpectationId`] is used to later link a lint emission to the actual /// expectation. It can be ignored in most cases. Expect(LintExpectationId), /// The `warn` level will produce a warning if the lint was violated, however the /// compiler will continue with its execution. Warn, /// This lint level is a special case of [`Warn`], that can't be overridden. This is used /// to ensure that a lint can't be suppressed. This lint level can currently only be set /// via the console and is therefore session specific. /// /// The [`LintExpectationId`] is intended to fulfill expectations marked via the /// `#[expect]` attribute, that will still be suppressed due to the level. ForceWarn(Option), /// The `deny` level will produce an error and stop further execution after the lint /// pass is complete. Deny, /// `Forbid` is equivalent to the `deny` level but can't be overwritten like the previous /// levels. Forbid, } impl Level { /// Converts a level to a lower-case string. pub fn as_str(self) -> &'static str { match self { Level::Allow => "allow", Level::Expect(_) => "expect", Level::Warn => "warn", Level::ForceWarn(_) => "force-warn", Level::Deny => "deny", Level::Forbid => "forbid", } } /// Converts a lower-case string to a level. This will never construct the expect /// level as that would require a [`LintExpectationId`] pub fn from_str(x: &str) -> Option { match x { "allow" => Some(Level::Allow), "warn" => Some(Level::Warn), "deny" => Some(Level::Deny), "forbid" => Some(Level::Forbid), "expect" | _ => None, } } /// Converts a symbol to a level. pub fn from_attr(attr: &Attribute) -> Option { match attr.name_or_empty() { sym::allow => Some(Level::Allow), sym::expect => Some(Level::Expect(LintExpectationId::Unstable { attr_id: attr.id, lint_index: None, })), sym::warn => Some(Level::Warn), sym::deny => Some(Level::Deny), sym::forbid => Some(Level::Forbid), _ => None, } } pub fn is_error(self) -> bool { match self { Level::Allow | Level::Expect(_) | Level::Warn | Level::ForceWarn(_) => false, Level::Deny | Level::Forbid => true, } } pub fn get_expectation_id(&self) -> Option { match self { Level::Expect(id) | Level::ForceWarn(Some(id)) => Some(*id), _ => None, } } } /// Specification of a single lint. #[derive(Copy, Clone, Debug)] pub struct Lint { /// A string identifier for the lint. /// /// This identifies the lint in attributes and in command-line arguments. /// In those contexts it is always lowercase, but this field is compared /// in a way which is case-insensitive for ASCII characters. This allows /// `declare_lint!()` invocations to follow the convention of upper-case /// statics without repeating the name. /// /// The name is written with underscores, e.g., "unused_imports". /// On the command line, underscores become dashes. /// /// See /// for naming guidelines. pub name: &'static str, /// Default level for the lint. /// /// See /// for guidelines on choosing a default level. pub default_level: Level, /// Description of the lint or the issue it detects. /// /// e.g., "imports that are never used" pub desc: &'static str, /// Starting at the given edition, default to the given lint level. If this is `None`, then use /// `default_level`. pub edition_lint_opts: Option<(Edition, Level)>, /// `true` if this lint is reported even inside expansions of external macros. pub report_in_external_macro: bool, pub future_incompatible: Option, pub is_plugin: bool, /// `Some` if this lint is feature gated, otherwise `None`. pub feature_gate: Option, pub crate_level_only: bool, } /// Extra information for a future incompatibility lint. #[derive(Copy, Clone, Debug)] pub struct FutureIncompatibleInfo { /// e.g., a URL for an issue/PR/RFC or error code pub reference: &'static str, /// The reason for the lint used by diagnostics to provide /// the right help message pub reason: FutureIncompatibilityReason, /// Whether to explain the reason to the user. /// /// Set to false for lints that already include a more detailed /// explanation. pub explain_reason: bool, } /// The reason for future incompatibility #[derive(Copy, Clone, Debug)] pub enum FutureIncompatibilityReason { /// This will be an error in a future release /// for all editions FutureReleaseError, /// This will be an error in a future release, and /// Cargo should create a report even for dependencies FutureReleaseErrorReportNow, /// Code that changes meaning in some way in a /// future release. FutureReleaseSemanticsChange, /// Previously accepted code that will become an /// error in the provided edition EditionError(Edition), /// Code that changes meaning in some way in /// the provided edition EditionSemanticsChange(Edition), /// A custom reason. Custom(&'static str), } impl FutureIncompatibilityReason { pub fn edition(self) -> Option { match self { Self::EditionError(e) => Some(e), Self::EditionSemanticsChange(e) => Some(e), _ => None, } } } impl FutureIncompatibleInfo { pub const fn default_fields_for_macro() -> Self { FutureIncompatibleInfo { reference: "", reason: FutureIncompatibilityReason::FutureReleaseError, explain_reason: true, } } } impl Lint { pub const fn default_fields_for_macro() -> Self { Lint { name: "", default_level: Level::Forbid, desc: "", edition_lint_opts: None, is_plugin: false, report_in_external_macro: false, future_incompatible: None, feature_gate: None, crate_level_only: false, } } /// Gets the lint's name, with ASCII letters converted to lowercase. pub fn name_lower(&self) -> String { self.name.to_ascii_lowercase() } pub fn default_level(&self, edition: Edition) -> Level { self.edition_lint_opts .filter(|(e, _)| *e <= edition) .map(|(_, l)| l) .unwrap_or(self.default_level) } } /// Identifies a lint known to the compiler. #[derive(Clone, Copy, Debug)] pub struct LintId { // Identity is based on pointer equality of this field. pub lint: &'static Lint, } impl PartialEq for LintId { fn eq(&self, other: &LintId) -> bool { std::ptr::eq(self.lint, other.lint) } } impl Eq for LintId {} impl std::hash::Hash for LintId { fn hash(&self, state: &mut H) { let ptr = self.lint as *const Lint; ptr.hash(state); } } impl LintId { /// Gets the `LintId` for a `Lint`. pub fn of(lint: &'static Lint) -> LintId { LintId { lint } } pub fn lint_name_raw(&self) -> &'static str { self.lint.name } /// Gets the name of the lint. pub fn to_string(&self) -> String { self.lint.name_lower() } } impl HashStable for LintId { #[inline] fn hash_stable(&self, hcx: &mut HCX, hasher: &mut StableHasher) { self.lint_name_raw().hash_stable(hcx, hasher); } } impl ToStableHashKey for LintId { type KeyType = &'static str; #[inline] fn to_stable_hash_key(&self, _: &HCX) -> &'static str { self.lint_name_raw() } } // This could be a closure, but then implementing derive trait // becomes hacky (and it gets allocated). #[derive(Debug)] pub enum BuiltinLintDiagnostics { Normal, AbsPathWithModule(Span), ProcMacroDeriveResolutionFallback(Span), MacroExpandedMacroExportsAccessedByAbsolutePaths(Span), ElidedLifetimesInPaths(usize, Span, bool, Span), UnknownCrateTypes(Span, String, String), UnusedImports(String, Vec<(Span, String)>, Option), RedundantImport(Vec<(Span, bool)>, Ident), DeprecatedMacro(Option, Span), MissingAbi(Span, Abi), UnusedDocComment(Span), UnusedBuiltinAttribute { attr_name: Symbol, macro_name: String, invoc_span: Span, }, PatternsInFnsWithoutBody(Span, Ident), LegacyDeriveHelpers(Span), ProcMacroBackCompat(String), OrPatternsBackCompat(Span, String), ReservedPrefix(Span), TrailingMacro(bool, Ident), BreakWithLabelAndLoop(Span), NamedAsmLabel(String), UnicodeTextFlow(Span, String), UnexpectedCfg((Symbol, Span), Option<(Symbol, Span)>), DeprecatedWhereclauseLocation(Span, String), SingleUseLifetime { /// Span of the parameter which declares this lifetime. param_span: Span, /// Span of the code that should be removed when eliding this lifetime. /// This span should include leading or trailing comma. deletion_span: Span, /// Span of the single use, or None if the lifetime is never used. /// If true, the lifetime will be fully elided. use_span: Option<(Span, bool)>, }, NamedArgumentUsedPositionally { /// Span where the named argument is used by position and will be replaced with the named /// argument name position_sp_to_replace: Option, /// Span where the named argument is used by position and is used for lint messages position_sp_for_msg: Option, /// Span where the named argument's name is (so we know where to put the warning message) named_arg_sp: Span, /// String containing the named arguments name named_arg_name: String, /// Indicates if the named argument is used as a width/precision for formatting is_formatting_arg: bool, }, } /// Lints that are buffered up early on in the `Session` before the /// `LintLevels` is calculated. pub struct BufferedEarlyLint { /// The span of code that we are linting on. pub span: MultiSpan, /// The lint message. pub msg: DiagnosticMessage, /// The `NodeId` of the AST node that generated the lint. pub node_id: NodeId, /// A lint Id that can be passed to /// `rustc_lint::early::EarlyContextAndPass::check_id`. pub lint_id: LintId, /// Customization of the `DiagnosticBuilder<'_>` for the lint. pub diagnostic: BuiltinLintDiagnostics, } #[derive(Default)] pub struct LintBuffer { pub map: NodeMap>, } impl LintBuffer { pub fn add_early_lint(&mut self, early_lint: BufferedEarlyLint) { let arr = self.map.entry(early_lint.node_id).or_default(); arr.push(early_lint); } pub fn add_lint( &mut self, lint: &'static Lint, node_id: NodeId, span: MultiSpan, msg: impl Into, diagnostic: BuiltinLintDiagnostics, ) { let lint_id = LintId::of(lint); let msg = msg.into(); self.add_early_lint(BufferedEarlyLint { lint_id, node_id, span, msg, diagnostic }); } pub fn take(&mut self, id: NodeId) -> Vec { self.map.remove(&id).unwrap_or_default() } pub fn buffer_lint( &mut self, lint: &'static Lint, id: NodeId, sp: impl Into, msg: impl Into, ) { self.add_lint(lint, id, sp.into(), msg, BuiltinLintDiagnostics::Normal) } pub fn buffer_lint_with_diagnostic( &mut self, lint: &'static Lint, id: NodeId, sp: impl Into, msg: impl Into, diagnostic: BuiltinLintDiagnostics, ) { self.add_lint(lint, id, sp.into(), msg, diagnostic) } } /// Declares a static item of type `&'static Lint`. /// /// See for /// documentation and guidelines on writing lints. /// /// The macro call should start with a doc comment explaining the lint /// which will be embedded in the rustc user documentation book. It should /// be written in markdown and have a format that looks like this: /// /// ```rust,ignore (doc-example) /// /// The `my_lint_name` lint detects [short explanation here]. /// /// /// /// ### Example /// /// /// /// ```rust /// /// [insert a concise example that triggers the lint] /// /// ``` /// /// /// /// {{produces}} /// /// /// /// ### Explanation /// /// /// /// This should be a detailed explanation of *why* the lint exists, /// /// and also include suggestions on how the user should fix the problem. /// /// Try to keep the text simple enough that a beginner can understand, /// /// and include links to other documentation for terminology that a /// /// beginner may not be familiar with. If this is "allow" by default, /// /// it should explain why (are there false positives or other issues?). If /// /// this is a future-incompatible lint, it should say so, with text that /// /// looks roughly like this: /// /// /// /// This is a [future-incompatible] lint to transition this to a hard /// /// error in the future. See [issue #xxxxx] for more details. /// /// /// /// [issue #xxxxx]: https://github.com/rust-lang/rust/issues/xxxxx /// ``` /// /// The `{{produces}}` tag will be automatically replaced with the output from /// the example by the build system. If the lint example is too complex to run /// as a simple example (for example, it needs an extern crate), mark the code /// block with `ignore` and manually replace the `{{produces}}` line with the /// expected output in a `text` code block. /// /// If this is a rustdoc-only lint, then only include a brief introduction /// with a link with the text `[rustdoc book]` so that the validator knows /// that this is for rustdoc only (see BROKEN_INTRA_DOC_LINKS as an example). /// /// Commands to view and test the documentation: /// /// * `./x.py doc --stage=1 src/doc/rustc --open`: Builds the rustc book and opens it. /// * `./x.py test src/tools/lint-docs`: Validates that the lint docs have the /// correct style, and that the code example actually emits the expected /// lint. /// /// If you have already built the compiler, and you want to make changes to /// just the doc comments, then use the `--keep-stage=0` flag with the above /// commands to avoid rebuilding the compiler. #[macro_export] macro_rules! declare_lint { ($(#[$attr:meta])* $vis: vis $NAME: ident, $Level: ident, $desc: expr) => ( $crate::declare_lint!( $(#[$attr])* $vis $NAME, $Level, $desc, ); ); ($(#[$attr:meta])* $vis: vis $NAME: ident, $Level: ident, $desc: expr, $(@feature_gate = $gate:expr;)? $(@future_incompatible = FutureIncompatibleInfo { $($field:ident : $val:expr),* $(,)* }; )? $($v:ident),*) => ( $(#[$attr])* $vis static $NAME: &$crate::Lint = &$crate::Lint { name: stringify!($NAME), default_level: $crate::$Level, desc: $desc, edition_lint_opts: None, is_plugin: false, $($v: true,)* $(feature_gate: Some($gate),)* $(future_incompatible: Some($crate::FutureIncompatibleInfo { $($field: $val,)* ..$crate::FutureIncompatibleInfo::default_fields_for_macro() }),)* ..$crate::Lint::default_fields_for_macro() }; ); ($(#[$attr:meta])* $vis: vis $NAME: ident, $Level: ident, $desc: expr, $lint_edition: expr => $edition_level: ident ) => ( $(#[$attr])* $vis static $NAME: &$crate::Lint = &$crate::Lint { name: stringify!($NAME), default_level: $crate::$Level, desc: $desc, edition_lint_opts: Some(($lint_edition, $crate::Level::$edition_level)), report_in_external_macro: false, is_plugin: false, }; ); } #[macro_export] macro_rules! declare_tool_lint { ( $(#[$attr:meta])* $vis:vis $tool:ident ::$NAME:ident, $Level: ident, $desc: expr $(, @feature_gate = $gate:expr;)? ) => ( $crate::declare_tool_lint!{$(#[$attr])* $vis $tool::$NAME, $Level, $desc, false $(, @feature_gate = $gate;)?} ); ( $(#[$attr:meta])* $vis:vis $tool:ident ::$NAME:ident, $Level:ident, $desc:expr, report_in_external_macro: $rep:expr $(, @feature_gate = $gate:expr;)? ) => ( $crate::declare_tool_lint!{$(#[$attr])* $vis $tool::$NAME, $Level, $desc, $rep $(, @feature_gate = $gate;)?} ); ( $(#[$attr:meta])* $vis:vis $tool:ident ::$NAME:ident, $Level:ident, $desc:expr, $external:expr $(, @feature_gate = $gate:expr;)? ) => ( $(#[$attr])* $vis static $NAME: &$crate::Lint = &$crate::Lint { name: &concat!(stringify!($tool), "::", stringify!($NAME)), default_level: $crate::$Level, desc: $desc, edition_lint_opts: None, report_in_external_macro: $external, future_incompatible: None, is_plugin: true, $(feature_gate: Some($gate),)? crate_level_only: false, ..$crate::Lint::default_fields_for_macro() }; ); } /// Declares a static `LintArray` and return it as an expression. #[macro_export] macro_rules! lint_array { ($( $lint:expr ),* ,) => { lint_array!( $($lint),* ) }; ($( $lint:expr ),*) => {{ vec![$($lint),*] }} } pub type LintArray = Vec<&'static Lint>; pub trait LintPass { fn name(&self) -> &'static str; } /// Implements `LintPass for $ty` with the given list of `Lint` statics. #[macro_export] macro_rules! impl_lint_pass { ($ty:ty => [$($lint:expr),* $(,)?]) => { impl $crate::LintPass for $ty { fn name(&self) -> &'static str { stringify!($ty) } } impl $ty { pub fn get_lints() -> $crate::LintArray { $crate::lint_array!($($lint),*) } } }; } /// Declares a type named `$name` which implements `LintPass`. /// To the right of `=>` a comma separated list of `Lint` statics is given. #[macro_export] macro_rules! declare_lint_pass { ($(#[$m:meta])* $name:ident => [$($lint:expr),* $(,)?]) => { $(#[$m])* #[derive(Copy, Clone)] pub struct $name; $crate::impl_lint_pass!($name => [$($lint),*]); }; }