compiler/rustc_parse/src/lib.rs

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