compiler/rustc_parse/src/lib.rs

   1 //! The main parser interface.
   2
   3 #![feature(array_windows)]
   4 #![feature(box_patterns)]
   5 #![feature(crate_visibility_modifier)]
   6 #![feature(if_let_guard)]
   7 #![feature(let_chains)]
   8 #![feature(let_else)]
   9 #![feature(never_type)]
  10 #![recursion_limit = "256"]
  11
  12 #[macro_use]
  13 extern crate tracing;
  14
  15 use rustc_ast as ast;
  16 use rustc_ast::token::{self, Nonterminal, Token, TokenKind};
  17 use rustc_ast::tokenstream::{self, AttributesData, CanSynthesizeMissingTokens, LazyTokenStream};
  18 use rustc_ast::tokenstream::{AttrAnnotatedTokenStream, AttrAnnotatedTokenTree};
  19 use rustc_ast::tokenstream::{Spacing, TokenStream};
  20 use rustc_ast::AstLike;
  21 use rustc_ast::Attribute;
  22 use rustc_ast::{AttrItem, MetaItem};
  23 use rustc_ast_pretty::pprust;
  24 use rustc_data_structures::sync::Lrc;
  25 use rustc_errors::{Applicability, Diagnostic, FatalError, Level, PResult};
  26 use rustc_session::parse::ParseSess;
  27 use rustc_span::{FileName, SourceFile, Span};
  28
  29 use std::path::Path;
  30 use std::str;
  31
  32 pub const MACRO_ARGUMENTS: Option<&str> = Some("macro arguments");
  33
  34 #[macro_use]
  35 pub mod parser;
  36 use parser::{emit_unclosed_delims, make_unclosed_delims_error, Parser};
  37 pub mod lexer;
  38 pub mod validate_attr;
  39
  40 // A bunch of utility functions of the form `parse_<thing>_from_<source>`
  41 // where <thing> includes crate, expr, item, stmt, tts, and one that
  42 // uses a HOF to parse anything, and <source> includes file and
  43 // `source_str`.
  44
  45 /// A variant of 'panictry!' that works on a Vec<Diagnostic> instead of a single DiagnosticBuilder.
  46 macro_rules! panictry_buffer {
  47     ($handler:expr, $e:expr) => {{
  48         use rustc_errors::FatalError;
  49         use std::result::Result::{Err, Ok};
  50         match $e {
  51             Ok(e) => e,
  52             Err(errs) => {
  53                 for mut e in errs {
  54                     $handler.emit_diagnostic(&mut e);
  55                 }
  56                 FatalError.raise()
  57             }
  58         }
  59     }};
  60 }
  61
  62 pub fn parse_crate_from_file<'a>(input: &Path, sess: &'a ParseSess) -> PResult<'a, ast::Crate> {
  63     let mut parser = new_parser_from_file(sess, input, None);
  64     parser.parse_crate_mod()
  65 }
  66
  67 pub fn parse_crate_attrs_from_file<'a>(
  68     input: &Path,
  69     sess: &'a ParseSess,
  70 ) -> PResult<'a, Vec<ast::Attribute>> {
  71     let mut parser = new_parser_from_file(sess, input, None);
  72     parser.parse_inner_attributes()
  73 }
  74
  75 pub fn parse_crate_from_source_str(
  76     name: FileName,
  77     source: String,
  78     sess: &ParseSess,
  79 ) -> PResult<'_, ast::Crate> {
  80     new_parser_from_source_str(sess, name, source).parse_crate_mod()
  81 }
  82
  83 pub fn parse_crate_attrs_from_source_str(
  84     name: FileName,
  85     source: String,
  86     sess: &ParseSess,
  87 ) -> PResult<'_, Vec<ast::Attribute>> {
  88     new_parser_from_source_str(sess, name, source).parse_inner_attributes()
  89 }
  90
  91 pub fn parse_stream_from_source_str(
  92     name: FileName,
  93     source: String,
  94     sess: &ParseSess,
  95     override_span: Option<Span>,
  96 ) -> TokenStream {
  97     let (stream, mut errors) =
  98         source_file_to_stream(sess, sess.source_map().new_source_file(name, source), override_span);
  99     emit_unclosed_delims(&mut errors, &sess);
 100     stream
 101 }
 102
 103 /// Creates a new parser from a source string.
 104 pub fn new_parser_from_source_str(sess: &ParseSess, name: FileName, source: String) -> Parser<'_> {
 105     panictry_buffer!(&sess.span_diagnostic, maybe_new_parser_from_source_str(sess, name, source))
 106 }
 107
 108 /// Creates a new parser from a source string. Returns any buffered errors from lexing the initial
 109 /// token stream.
 110 pub fn maybe_new_parser_from_source_str(
 111     sess: &ParseSess,
 112     name: FileName,
 113     source: String,
 114 ) -> Result<Parser<'_>, Vec<Diagnostic>> {
 115     maybe_source_file_to_parser(sess, sess.source_map().new_source_file(name, source))
 116 }
 117
 118 /// Creates a new parser, handling errors as appropriate if the file doesn't exist.
 119 /// If a span is given, that is used on an error as the source of the problem.
 120 pub fn new_parser_from_file<'a>(sess: &'a ParseSess, path: &Path, sp: Option<Span>) -> Parser<'a> {
 121     source_file_to_parser(sess, file_to_source_file(sess, path, sp))
 122 }
 123
 124 /// Given a `source_file` and config, returns a parser.
 125 fn source_file_to_parser(sess: &ParseSess, source_file: Lrc<SourceFile>) -> Parser<'_> {
 126     panictry_buffer!(&sess.span_diagnostic, maybe_source_file_to_parser(sess, source_file))
 127 }
 128
 129 /// Given a `source_file` and config, return a parser. Returns any buffered errors from lexing the
 130 /// initial token stream.
 131 fn maybe_source_file_to_parser(
 132     sess: &ParseSess,
 133     source_file: Lrc<SourceFile>,
 134 ) -> Result<Parser<'_>, Vec<Diagnostic>> {
 135     let end_pos = source_file.end_pos;
 136     let (stream, unclosed_delims) = maybe_file_to_stream(sess, source_file, None)?;
 137     let mut parser = stream_to_parser(sess, stream, None);
 138     parser.unclosed_delims = unclosed_delims;
 139     if parser.token == token::Eof {
 140         parser.token.span = Span::new(end_pos, end_pos, parser.token.span.ctxt(), None);
 141     }
 142
 143     Ok(parser)
 144 }
 145
 146 // Base abstractions
 147
 148 /// Given a session and a path and an optional span (for error reporting),
 149 /// add the path to the session's source_map and return the new source_file or
 150 /// error when a file can't be read.
 151 fn try_file_to_source_file(
 152     sess: &ParseSess,
 153     path: &Path,
 154     spanopt: Option<Span>,
 155 ) -> Result<Lrc<SourceFile>, Diagnostic> {
 156     sess.source_map().load_file(path).map_err(|e| {
 157         let msg = format!("couldn't read {}: {}", path.display(), e);
 158         let mut diag = Diagnostic::new(Level::Fatal, &msg);
 159         if let Some(sp) = spanopt {
 160             diag.set_span(sp);
 161         }
 162         diag
 163     })
 164 }
 165
 166 /// Given a session and a path and an optional span (for error reporting),
 167 /// adds the path to the session's `source_map` and returns the new `source_file`.
 168 fn file_to_source_file(sess: &ParseSess, path: &Path, spanopt: Option<Span>) -> Lrc<SourceFile> {
 169     match try_file_to_source_file(sess, path, spanopt) {
 170         Ok(source_file) => source_file,
 171         Err(mut d) => {
 172             sess.span_diagnostic.emit_diagnostic(&mut d);
 173             FatalError.raise();
 174         }
 175     }
 176 }
 177
 178 /// Given a `source_file`, produces a sequence of token trees.
 179 pub fn source_file_to_stream(
 180     sess: &ParseSess,
 181     source_file: Lrc<SourceFile>,
 182     override_span: Option<Span>,
 183 ) -> (TokenStream, Vec<lexer::UnmatchedBrace>) {
 184     panictry_buffer!(&sess.span_diagnostic, maybe_file_to_stream(sess, source_file, override_span))
 185 }
 186
 187 /// Given a source file, produces a sequence of token trees. Returns any buffered errors from
 188 /// parsing the token stream.
 189 pub fn maybe_file_to_stream(
 190     sess: &ParseSess,
 191     source_file: Lrc<SourceFile>,
 192     override_span: Option<Span>,
 193 ) -> Result<(TokenStream, Vec<lexer::UnmatchedBrace>), Vec<Diagnostic>> {
 194     let src = source_file.src.as_ref().unwrap_or_else(|| {
 195         sess.span_diagnostic.bug(&format!(
 196             "cannot lex `source_file` without source: {}",
 197             sess.source_map().filename_for_diagnostics(&source_file.name)
 198         ));
 199     });
 200
 201     let (token_trees, unmatched_braces) =
 202         lexer::parse_token_trees(sess, src.as_str(), source_file.start_pos, override_span);
 203
 204     match token_trees {
 205         Ok(stream) => Ok((stream, unmatched_braces)),
 206         Err(err) => {
 207             let mut buffer = Vec::with_capacity(1);
 208             err.buffer(&mut buffer);
 209             // Not using `emit_unclosed_delims` to use `db.buffer`
 210             for unmatched in unmatched_braces {
 211                 if let Some(err) = make_unclosed_delims_error(unmatched, &sess) {
 212                     err.buffer(&mut buffer);
 213                 }
 214             }
 215             Err(buffer)
 216         }
 217     }
 218 }
 219
 220 /// Given a stream and the `ParseSess`, produces a parser.
 221 pub fn stream_to_parser<'a>(
 222     sess: &'a ParseSess,
 223     stream: TokenStream,
 224     subparser_name: Option<&'static str>,
 225 ) -> Parser<'a> {
 226     Parser::new(sess, stream, false, subparser_name)
 227 }
 228
 229 /// Runs the given subparser `f` on the tokens of the given `attr`'s item.
 230 pub fn parse_in<'a, T>(
 231     sess: &'a ParseSess,
 232     tts: TokenStream,
 233     name: &'static str,
 234     mut f: impl FnMut(&mut Parser<'a>) -> PResult<'a, T>,
 235 ) -> PResult<'a, T> {
 236     let mut parser = Parser::new(sess, tts, false, Some(name));
 237     let result = f(&mut parser)?;
 238     if parser.token != token::Eof {
 239         parser.unexpected()?;
 240     }
 241     Ok(result)
 242 }
 243
 244 // NOTE(Centril): The following probably shouldn't be here but it acknowledges the
 245 // fact that architecturally, we are using parsing (read on below to understand why).
 246
 247 pub fn nt_to_tokenstream(
 248     nt: &Nonterminal,
 249     sess: &ParseSess,
 250     synthesize_tokens: CanSynthesizeMissingTokens,
 251 ) -> TokenStream {
 252     // A `Nonterminal` is often a parsed AST item. At this point we now
 253     // need to convert the parsed AST to an actual token stream, e.g.
 254     // un-parse it basically.
 255     //
 256     // Unfortunately there's not really a great way to do that in a
 257     // guaranteed lossless fashion right now. The fallback here is to just
 258     // stringify the AST node and reparse it, but this loses all span
 259     // information.
 260     //
 261     // As a result, some AST nodes are annotated with the token stream they
 262     // came from. Here we attempt to extract these lossless token streams
 263     // before we fall back to the stringification.
 264
 265     let convert_tokens =
 266         |tokens: Option<&LazyTokenStream>| Some(tokens?.create_token_stream().to_tokenstream());
 267
 268     let tokens = match *nt {
 269         Nonterminal::NtItem(ref item) => prepend_attrs(&item.attrs, item.tokens.as_ref()),
 270         Nonterminal::NtBlock(ref block) => convert_tokens(block.tokens.as_ref()),
 271         Nonterminal::NtStmt(ref stmt) if let ast::StmtKind::Empty = stmt.kind => {
 272             let tokens = AttrAnnotatedTokenStream::new(vec![(
 273                 tokenstream::AttrAnnotatedTokenTree::Token(Token::new(
 274                     TokenKind::Semi,
 275                     stmt.span,
 276                 )),
 277                 Spacing::Alone,
 278             )]);
 279             prepend_attrs(&stmt.attrs(), Some(&LazyTokenStream::new(tokens)))
 280         }
 281         Nonterminal::NtStmt(ref stmt) => prepend_attrs(&stmt.attrs(), stmt.tokens()),
 282         Nonterminal::NtPat(ref pat) => convert_tokens(pat.tokens.as_ref()),
 283         Nonterminal::NtTy(ref ty) => convert_tokens(ty.tokens.as_ref()),
 284         Nonterminal::NtIdent(ident, is_raw) => {
 285             Some(tokenstream::TokenTree::token(token::Ident(ident.name, is_raw), ident.span).into())
 286         }
 287         Nonterminal::NtLifetime(ident) => {
 288             Some(tokenstream::TokenTree::token(token::Lifetime(ident.name), ident.span).into())
 289         }
 290         Nonterminal::NtMeta(ref attr) => convert_tokens(attr.tokens.as_ref()),
 291         Nonterminal::NtPath(ref path) => convert_tokens(path.tokens.as_ref()),
 292         Nonterminal::NtVis(ref vis) => convert_tokens(vis.tokens.as_ref()),
 293         Nonterminal::NtExpr(ref expr) | Nonterminal::NtLiteral(ref expr) => {
 294             prepend_attrs(&expr.attrs, expr.tokens.as_ref())
 295         }
 296     };
 297
 298     if let Some(tokens) = tokens {
 299         return tokens;
 300     } else if matches!(synthesize_tokens, CanSynthesizeMissingTokens::Yes) {
 301         return fake_token_stream(sess, nt);
 302     } else {
 303         panic!(
 304             "Missing tokens for nt {:?} at {:?}: {:?}",
 305             nt,
 306             nt.span(),
 307             pprust::nonterminal_to_string(nt)
 308         );
 309     }
 310 }
 311
 312 fn prepend_attrs(attrs: &[Attribute], tokens: Option<&LazyTokenStream>) -> Option<TokenStream> {
 313     let tokens = tokens?;
 314     if attrs.is_empty() {
 315         return Some(tokens.create_token_stream().to_tokenstream());
 316     }
 317     let attr_data = AttributesData { attrs: attrs.to_vec().into(), tokens: tokens.clone() };
 318     let wrapped = AttrAnnotatedTokenStream::new(vec![(
 319         AttrAnnotatedTokenTree::Attributes(attr_data),
 320         Spacing::Alone,
 321     )]);
 322     Some(wrapped.to_tokenstream())
 323 }
 324
 325 pub fn fake_token_stream(sess: &ParseSess, nt: &Nonterminal) -> TokenStream {
 326     let source = pprust::nonterminal_to_string(nt);
 327     let filename = FileName::macro_expansion_source_code(&source);
 328     parse_stream_from_source_str(filename, source, sess, Some(nt.span()))
 329 }
 330
 331 pub fn fake_token_stream_for_crate(sess: &ParseSess, krate: &ast::Crate) -> TokenStream {
 332     let source = pprust::crate_to_string_for_macros(krate);
 333     let filename = FileName::macro_expansion_source_code(&source);
 334     parse_stream_from_source_str(filename, source, sess, Some(krate.spans.inner_span))
 335 }
 336
 337 pub fn parse_cfg_attr(
 338     attr: &Attribute,
 339     parse_sess: &ParseSess,
 340 ) -> Option<(MetaItem, Vec<(AttrItem, Span)>)> {
 341     match attr.get_normal_item().args {
 342         ast::MacArgs::Delimited(dspan, delim, ref tts) if !tts.is_empty() => {
 343             let msg = "wrong `cfg_attr` delimiters";
 344             crate::validate_attr::check_meta_bad_delim(parse_sess, dspan, delim, msg);
 345             match parse_in(parse_sess, tts.clone(), "`cfg_attr` input", |p| p.parse_cfg_attr()) {
 346                 Ok(r) => return Some(r),
 347                 Err(mut e) => {
 348                     e.help(&format!("the valid syntax is `{}`", CFG_ATTR_GRAMMAR_HELP))
 349                         .note(CFG_ATTR_NOTE_REF)
 350                         .emit();
 351                 }
 352             }
 353         }
 354         _ => error_malformed_cfg_attr_missing(attr.span, parse_sess),
 355     }
 356     None
 357 }
 358
 359 const CFG_ATTR_GRAMMAR_HELP: &str = "#[cfg_attr(condition, attribute, other_attribute, ...)]";
 360 const CFG_ATTR_NOTE_REF: &str = "for more information, visit \
 361     <https://doc.rust-lang.org/reference/conditional-compilation.html\
 362     #the-cfg_attr-attribute>";
 363
 364 fn error_malformed_cfg_attr_missing(span: Span, parse_sess: &ParseSess) {
 365     parse_sess
 366         .span_diagnostic
 367         .struct_span_err(span, "malformed `cfg_attr` attribute input")
 368         .span_suggestion(
 369             span,
 370             "missing condition and attribute",
 371             CFG_ATTR_GRAMMAR_HELP.to_string(),
 372             Applicability::HasPlaceholders,
 373         )
 374         .note(CFG_ATTR_NOTE_REF)
 375         .emit();
 376 }