src/librustc_parse/lib.rs

   1 //! The main parser interface.
   2
   3 #![feature(bool_to_option)]
   4 #![feature(crate_visibility_modifier)]
   5
   6 use syntax::ast;
   7 use syntax::print::pprust;
   8 use syntax::sess::ParseSess;
   9 use syntax::token::{self, Nonterminal};
  10 use syntax::tokenstream::{self, TokenStream, TokenTree};
  11
  12 use rustc_errors::{PResult, FatalError, Level, Diagnostic};
  13 use rustc_data_structures::sync::Lrc;
  14 use syntax_pos::{Span, SourceFile, FileName};
  15
  16 use std::borrow::Cow;
  17 use std::path::Path;
  18 use std::str;
  19
  20 use log::info;
  21
  22 pub const MACRO_ARGUMENTS: Option<&'static str> = Some("macro arguments");
  23
  24 #[macro_use]
  25 pub mod parser;
  26 use parser::{Parser, emit_unclosed_delims, make_unclosed_delims_error};
  27 pub mod lexer;
  28 pub mod validate_attr;
  29 #[macro_use]
  30 pub mod config;
  31
  32 #[derive(Clone)]
  33 pub struct Directory<'a> {
  34     pub path: Cow<'a, Path>,
  35     pub ownership: DirectoryOwnership,
  36 }
  37
  38 #[derive(Copy, Clone)]
  39 pub enum DirectoryOwnership {
  40     Owned {
  41         // None if `mod.rs`, `Some("foo")` if we're in `foo.rs`.
  42         relative: Option<ast::Ident>,
  43     },
  44     UnownedViaBlock,
  45     UnownedViaMod,
  46 }
  47
  48 // A bunch of utility functions of the form `parse_<thing>_from_<source>`
  49 // where <thing> includes crate, expr, item, stmt, tts, and one that
  50 // uses a HOF to parse anything, and <source> includes file and
  51 // `source_str`.
  52
  53 /// A variant of 'panictry!' that works on a Vec<Diagnostic> instead of a single DiagnosticBuilder.
  54 macro_rules! panictry_buffer {
  55     ($handler:expr, $e:expr) => ({
  56         use std::result::Result::{Ok, Err};
  57         use rustc_errors::FatalError;
  58         match $e {
  59             Ok(e) => e,
  60             Err(errs) => {
  61                 for e in errs {
  62                     $handler.emit_diagnostic(&e);
  63                 }
  64                 FatalError.raise()
  65             }
  66         }
  67     })
  68 }
  69
  70 pub fn parse_crate_from_file<'a>(input: &Path, sess: &'a ParseSess) -> PResult<'a, ast::Crate> {
  71     let mut parser = new_parser_from_file(sess, input);
  72     parser.parse_crate_mod()
  73 }
  74
  75 pub fn parse_crate_attrs_from_file<'a>(input: &Path, sess: &'a ParseSess)
  76                                        -> PResult<'a, Vec<ast::Attribute>> {
  77     let mut parser = new_parser_from_file(sess, input);
  78     parser.parse_inner_attributes()
  79 }
  80
  81 pub fn parse_crate_from_source_str(name: FileName, source: String, sess: &ParseSess)
  82                                        -> PResult<'_, ast::Crate> {
  83     new_parser_from_source_str(sess, name, source).parse_crate_mod()
  84 }
  85
  86 pub fn parse_crate_attrs_from_source_str(name: FileName, source: String, 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) = source_file_to_stream(
  98         sess,
  99         sess.source_map().new_source_file(name, source),
 100         override_span,
 101     );
 102     emit_unclosed_delims(&mut errors, &sess);
 103     stream
 104 }
 105
 106 /// Creates a new parser from a source string.
 107 pub fn new_parser_from_source_str(sess: &ParseSess, name: FileName, source: String) -> Parser<'_> {
 108     panictry_buffer!(&sess.span_diagnostic, maybe_new_parser_from_source_str(sess, name, source))
 109 }
 110
 111 /// Creates a new parser from a source string. Returns any buffered errors from lexing the initial
 112 /// token stream.
 113 pub fn maybe_new_parser_from_source_str(sess: &ParseSess, name: FileName, source: String)
 114     -> Result<Parser<'_>, Vec<Diagnostic>>
 115 {
 116     let mut parser = maybe_source_file_to_parser(sess,
 117                                                  sess.source_map().new_source_file(name, source))?;
 118     parser.recurse_into_file_modules = false;
 119     Ok(parser)
 120 }
 121
 122 /// Creates a new parser, handling errors as appropriate if the file doesn't exist.
 123 pub fn new_parser_from_file<'a>(sess: &'a ParseSess, path: &Path) -> Parser<'a> {
 124     source_file_to_parser(sess, file_to_source_file(sess, path, None))
 125 }
 126
 127 /// Creates a new parser, returning buffered diagnostics if the file doesn't exist,
 128 /// or from lexing the initial token stream.
 129 pub fn maybe_new_parser_from_file<'a>(sess: &'a ParseSess, path: &Path)
 130     -> Result<Parser<'a>, Vec<Diagnostic>> {
 131     let file = try_file_to_source_file(sess, path, None).map_err(|db| vec![db])?;
 132     maybe_source_file_to_parser(sess, file)
 133 }
 134
 135 /// Given a session, a crate config, a path, and a span, add
 136 /// the file at the given path to the `source_map`, and returns a parser.
 137 /// On an error, uses the given span as the source of the problem.
 138 pub fn new_sub_parser_from_file<'a>(sess: &'a ParseSess,
 139                                     path: &Path,
 140                                     directory_ownership: DirectoryOwnership,
 141                                     module_name: Option<String>,
 142                                     sp: Span) -> Parser<'a> {
 143     let mut p = source_file_to_parser(sess, file_to_source_file(sess, path, Some(sp)));
 144     p.directory.ownership = directory_ownership;
 145     p.root_module_name = module_name;
 146     p
 147 }
 148
 149 /// Given a `source_file` and config, returns a parser.
 150 fn source_file_to_parser(sess: &ParseSess, source_file: Lrc<SourceFile>) -> Parser<'_> {
 151     panictry_buffer!(&sess.span_diagnostic,
 152                      maybe_source_file_to_parser(sess, source_file))
 153 }
 154
 155 /// Given a `source_file` and config, return a parser. Returns any buffered errors from lexing the
 156 /// initial token stream.
 157 fn maybe_source_file_to_parser(
 158     sess: &ParseSess,
 159     source_file: Lrc<SourceFile>,
 160 ) -> Result<Parser<'_>, Vec<Diagnostic>> {
 161     let end_pos = source_file.end_pos;
 162     let (stream, unclosed_delims) = maybe_file_to_stream(sess, source_file, None)?;
 163     let mut parser = stream_to_parser(sess, stream, None);
 164     parser.unclosed_delims = unclosed_delims;
 165     if parser.token == token::Eof && parser.token.span.is_dummy() {
 166         parser.token.span = Span::new(end_pos, end_pos, parser.token.span.ctxt());
 167     }
 168
 169     Ok(parser)
 170 }
 171
 172 // Must preserve old name for now, because `quote!` from the *existing*
 173 // compiler expands into it.
 174 pub fn new_parser_from_tts(sess: &ParseSess, tts: Vec<TokenTree>) -> Parser<'_> {
 175     stream_to_parser(sess, tts.into_iter().collect(), crate::MACRO_ARGUMENTS)
 176 }
 177
 178
 179 // Base abstractions
 180
 181 /// Given a session and a path and an optional span (for error reporting),
 182 /// add the path to the session's source_map and return the new source_file or
 183 /// error when a file can't be read.
 184 fn try_file_to_source_file(sess: &ParseSess, path: &Path, spanopt: Option<Span>)
 185                    -> Result<Lrc<SourceFile>, Diagnostic> {
 186     sess.source_map().load_file(path)
 187     .map_err(|e| {
 188         let msg = format!("couldn't read {}: {}", path.display(), e);
 189         let mut diag = Diagnostic::new(Level::Fatal, &msg);
 190         if let Some(sp) = spanopt {
 191             diag.set_span(sp);
 192         }
 193         diag
 194     })
 195 }
 196
 197 /// Given a session and a path and an optional span (for error reporting),
 198 /// adds the path to the session's `source_map` and returns the new `source_file`.
 199 fn file_to_source_file(sess: &ParseSess, path: &Path, spanopt: Option<Span>)
 200                    -> Lrc<SourceFile> {
 201     match try_file_to_source_file(sess, path, spanopt) {
 202         Ok(source_file) => source_file,
 203         Err(d) => {
 204             sess.span_diagnostic.emit_diagnostic(&d);
 205             FatalError.raise();
 206         }
 207     }
 208 }
 209
 210 /// Given a `source_file`, produces a sequence of token trees.
 211 pub fn source_file_to_stream(
 212     sess: &ParseSess,
 213     source_file: Lrc<SourceFile>,
 214     override_span: Option<Span>,
 215 ) -> (TokenStream, Vec<lexer::UnmatchedBrace>) {
 216     panictry_buffer!(&sess.span_diagnostic, maybe_file_to_stream(sess, source_file, override_span))
 217 }
 218
 219 /// Given a source file, produces a sequence of token trees. Returns any buffered errors from
 220 /// parsing the token stream.
 221 pub fn maybe_file_to_stream(
 222     sess: &ParseSess,
 223     source_file: Lrc<SourceFile>,
 224     override_span: Option<Span>,
 225 ) -> Result<(TokenStream, Vec<lexer::UnmatchedBrace>), Vec<Diagnostic>> {
 226     let srdr = lexer::StringReader::new(sess, source_file, override_span);
 227     let (token_trees, unmatched_braces) = srdr.into_token_trees();
 228
 229     match token_trees {
 230         Ok(stream) => Ok((stream, unmatched_braces)),
 231         Err(err) => {
 232             let mut buffer = Vec::with_capacity(1);
 233             err.buffer(&mut buffer);
 234             // Not using `emit_unclosed_delims` to use `db.buffer`
 235             for unmatched in unmatched_braces {
 236                 if let Some(err) = make_unclosed_delims_error(unmatched, &sess) {
 237                     err.buffer(&mut buffer);
 238                 }
 239             }
 240             Err(buffer)
 241         }
 242     }
 243 }
 244
 245 /// Given a stream and the `ParseSess`, produces a parser.
 246 pub fn stream_to_parser<'a>(
 247     sess: &'a ParseSess,
 248     stream: TokenStream,
 249     subparser_name: Option<&'static str>,
 250 ) -> Parser<'a> {
 251     Parser::new(sess, stream, None, true, false, subparser_name)
 252 }
 253
 254 /// Given a stream, the `ParseSess` and the base directory, produces a parser.
 255 ///
 256 /// Use this function when you are creating a parser from the token stream
 257 /// and also care about the current working directory of the parser (e.g.,
 258 /// you are trying to resolve modules defined inside a macro invocation).
 259 ///
 260 /// # Note
 261 ///
 262 /// The main usage of this function is outside of rustc, for those who uses
 263 /// libsyntax as a library. Please do not remove this function while refactoring
 264 /// just because it is not used in rustc codebase!
 265 pub fn stream_to_parser_with_base_dir<'a>(
 266     sess: &'a ParseSess,
 267     stream: TokenStream,
 268     base_dir: Directory<'a>,
 269 ) -> Parser<'a> {
 270     Parser::new(sess, stream, Some(base_dir), true, false, None)
 271 }
 272
 273 /// Runs the given subparser `f` on the tokens of the given `attr`'s item.
 274 pub fn parse_in<'a, T>(
 275     sess: &'a ParseSess,
 276     tts: TokenStream,
 277     name: &'static str,
 278     mut f: impl FnMut(&mut Parser<'a>) -> PResult<'a, T>,
 279 ) -> PResult<'a, T> {
 280     let mut parser = Parser::new(sess, tts, None, false, false, Some(name));
 281     let result = f(&mut parser)?;
 282     if parser.token != token::Eof {
 283         parser.unexpected()?;
 284     }
 285     Ok(result)
 286 }
 287
 288 // NOTE(Centril): The following probably shouldn't be here but it acknowledges the
 289 // fact that architecturally, we are using parsing (read on below to understand why).
 290
 291 pub fn nt_to_tokenstream(nt: &Nonterminal, sess: &ParseSess, span: Span) -> TokenStream {
 292     // A `Nonterminal` is often a parsed AST item. At this point we now
 293     // need to convert the parsed AST to an actual token stream, e.g.
 294     // un-parse it basically.
 295     //
 296     // Unfortunately there's not really a great way to do that in a
 297     // guaranteed lossless fashion right now. The fallback here is to just
 298     // stringify the AST node and reparse it, but this loses all span
 299     // information.
 300     //
 301     // As a result, some AST nodes are annotated with the token stream they
 302     // came from. Here we attempt to extract these lossless token streams
 303     // before we fall back to the stringification.
 304     let tokens = match *nt {
 305         Nonterminal::NtItem(ref item) => {
 306             prepend_attrs(sess, &item.attrs, item.tokens.as_ref(), span)
 307         }
 308         Nonterminal::NtTraitItem(ref item) => {
 309             prepend_attrs(sess, &item.attrs, item.tokens.as_ref(), span)
 310         }
 311         Nonterminal::NtImplItem(ref item) => {
 312             prepend_attrs(sess, &item.attrs, item.tokens.as_ref(), span)
 313         }
 314         Nonterminal::NtIdent(ident, is_raw) => {
 315             Some(tokenstream::TokenTree::token(token::Ident(ident.name, is_raw), ident.span).into())
 316         }
 317         Nonterminal::NtLifetime(ident) => {
 318             Some(tokenstream::TokenTree::token(token::Lifetime(ident.name), ident.span).into())
 319         }
 320         Nonterminal::NtTT(ref tt) => {
 321             Some(tt.clone().into())
 322         }
 323         _ => None,
 324     };
 325
 326     // FIXME(#43081): Avoid this pretty-print + reparse hack
 327     let source = pprust::nonterminal_to_string(nt);
 328     let filename = FileName::macro_expansion_source_code(&source);
 329     let tokens_for_real = parse_stream_from_source_str(filename, source, sess, Some(span));
 330
 331     // During early phases of the compiler the AST could get modified
 332     // directly (e.g., attributes added or removed) and the internal cache
 333     // of tokens my not be invalidated or updated. Consequently if the
 334     // "lossless" token stream disagrees with our actual stringification
 335     // (which has historically been much more battle-tested) then we go
 336     // with the lossy stream anyway (losing span information).
 337     //
 338     // Note that the comparison isn't `==` here to avoid comparing spans,
 339     // but it *also* is a "probable" equality which is a pretty weird
 340     // definition. We mostly want to catch actual changes to the AST
 341     // like a `#[cfg]` being processed or some weird `macro_rules!`
 342     // expansion.
 343     //
 344     // What we *don't* want to catch is the fact that a user-defined
 345     // literal like `0xf` is stringified as `15`, causing the cached token
 346     // stream to not be literal `==` token-wise (ignoring spans) to the
 347     // token stream we got from stringification.
 348     //
 349     // Instead the "probably equal" check here is "does each token
 350     // recursively have the same discriminant?" We basically don't look at
 351     // the token values here and assume that such fine grained token stream
 352     // modifications, including adding/removing typically non-semantic
 353     // tokens such as extra braces and commas, don't happen.
 354     if let Some(tokens) = tokens {
 355         if tokens.probably_equal_for_proc_macro(&tokens_for_real) {
 356             return tokens
 357         }
 358         info!("cached tokens found, but they're not \"probably equal\", \
 359                 going with stringified version");
 360     }
 361     return tokens_for_real
 362 }
 363
 364 fn prepend_attrs(
 365     sess: &ParseSess,
 366     attrs: &[ast::Attribute],
 367     tokens: Option<&tokenstream::TokenStream>,
 368     span: syntax_pos::Span
 369 ) -> Option<tokenstream::TokenStream> {
 370     let tokens = tokens?;
 371     if attrs.len() == 0 {
 372         return Some(tokens.clone())
 373     }
 374     let mut builder = tokenstream::TokenStreamBuilder::new();
 375     for attr in attrs {
 376         assert_eq!(attr.style, ast::AttrStyle::Outer,
 377                    "inner attributes should prevent cached tokens from existing");
 378
 379         let source = pprust::attribute_to_string(attr);
 380         let macro_filename = FileName::macro_expansion_source_code(&source);
 381
 382         let item = match attr.kind {
 383             ast::AttrKind::Normal(ref item) => item,
 384             ast::AttrKind::DocComment(_) => {
 385                 let stream = parse_stream_from_source_str(macro_filename, source, sess, Some(span));
 386                 builder.push(stream);
 387                 continue
 388             }
 389         };
 390
 391         // synthesize # [ $path $tokens ] manually here
 392         let mut brackets = tokenstream::TokenStreamBuilder::new();
 393
 394         // For simple paths, push the identifier directly
 395         if item.path.segments.len() == 1 && item.path.segments[0].args.is_none() {
 396             let ident = item.path.segments[0].ident;
 397             let token = token::Ident(ident.name, ident.as_str().starts_with("r#"));
 398             brackets.push(tokenstream::TokenTree::token(token, ident.span));
 399
 400         // ... and for more complicated paths, fall back to a reparse hack that
 401         // should eventually be removed.
 402         } else {
 403             let stream = parse_stream_from_source_str(macro_filename, source, sess, Some(span));
 404             brackets.push(stream);
 405         }
 406
 407         brackets.push(item.args.outer_tokens());
 408
 409         // The span we list here for `#` and for `[ ... ]` are both wrong in
 410         // that it encompasses more than each token, but it hopefully is "good
 411         // enough" for now at least.
 412         builder.push(tokenstream::TokenTree::token(token::Pound, attr.span));
 413         let delim_span = tokenstream::DelimSpan::from_single(attr.span);
 414         builder.push(tokenstream::TokenTree::Delimited(
 415             delim_span, token::DelimToken::Bracket, brackets.build().into()));
 416     }
 417     builder.push(tokens.clone());
 418     Some(builder.build())
 419 }