rustc_parse/
lib.rs

1//! The main parser interface.
2
3// tidy-alphabetical-start
4#![allow(rustc::diagnostic_outside_of_impl)]
5#![allow(rustc::untranslatable_diagnostic)]
6#![feature(assert_matches)]
7#![feature(box_patterns)]
8#![feature(debug_closure_helpers)]
9#![feature(default_field_values)]
10#![feature(if_let_guard)]
11#![feature(iter_intersperse)]
12#![recursion_limit = "256"]
13// tidy-alphabetical-end
14
15use std::path::{Path, PathBuf};
16use std::str::Utf8Error;
17use std::sync::Arc;
18
19use rustc_ast as ast;
20use rustc_ast::tokenstream::{DelimSpan, TokenStream};
21use rustc_ast::{AttrItem, Attribute, MetaItemInner, token};
22use rustc_ast_pretty::pprust;
23use rustc_errors::{Diag, EmissionGuarantee, FatalError, PResult, pluralize};
24use rustc_session::parse::ParseSess;
25use rustc_span::source_map::SourceMap;
26use rustc_span::{FileName, SourceFile, Span};
27pub use unicode_normalization::UNICODE_VERSION as UNICODE_NORMALIZATION_VERSION;
28
29pub const MACRO_ARGUMENTS: Option<&str> = Some("macro arguments");
30
31#[macro_use]
32pub mod parser;
33use parser::Parser;
34use rustc_ast::token::Delimiter;
35
36use crate::lexer::StripTokens;
37
38pub mod lexer;
39
40mod errors;
41
42rustc_fluent_macro::fluent_messages! { "../messages.ftl" }
43
44// Unwrap the result if `Ok`, otherwise emit the diagnostics and abort.
45pub fn unwrap_or_emit_fatal<T>(expr: Result<T, Vec<Diag<'_>>>) -> T {
46    match expr {
47        Ok(expr) => expr,
48        Err(errs) => {
49            for err in errs {
50                err.emit();
51            }
52            FatalError.raise()
53        }
54    }
55}
56
57/// Creates a new parser from a source string.
58///
59/// On failure, the errors must be consumed via `unwrap_or_emit_fatal`, `emit`, `cancel`,
60/// etc., otherwise a panic will occur when they are dropped.
61pub fn new_parser_from_source_str(
62    psess: &ParseSess,
63    name: FileName,
64    source: String,
65    strip_tokens: StripTokens,
66) -> Result<Parser<'_>, Vec<Diag<'_>>> {
67    let source_file = psess.source_map().new_source_file(name, source);
68    new_parser_from_source_file(psess, source_file, strip_tokens)
69}
70
71/// Creates a new parser from a filename. On failure, the errors must be consumed via
72/// `unwrap_or_emit_fatal`, `emit`, `cancel`, etc., otherwise a panic will occur when they are
73/// dropped.
74///
75/// If a span is given, that is used on an error as the source of the problem.
76pub fn new_parser_from_file<'a>(
77    psess: &'a ParseSess,
78    path: &Path,
79    strip_tokens: StripTokens,
80    sp: Option<Span>,
81) -> Result<Parser<'a>, Vec<Diag<'a>>> {
82    let sm = psess.source_map();
83    let source_file = sm.load_file(path).unwrap_or_else(|e| {
84        let msg = format!("couldn't read `{}`: {}", path.display(), e);
85        let mut err = psess.dcx().struct_fatal(msg);
86        if let Ok(contents) = std::fs::read(path)
87            && let Err(utf8err) = String::from_utf8(contents.clone())
88        {
89            utf8_error(
90                sm,
91                &path.display().to_string(),
92                sp,
93                &mut err,
94                utf8err.utf8_error(),
95                &contents,
96            );
97        }
98        if let Some(sp) = sp {
99            err.span(sp);
100        }
101        err.emit();
102    });
103    new_parser_from_source_file(psess, source_file, strip_tokens)
104}
105
106pub fn utf8_error<E: EmissionGuarantee>(
107    sm: &SourceMap,
108    path: &str,
109    sp: Option<Span>,
110    err: &mut Diag<'_, E>,
111    utf8err: Utf8Error,
112    contents: &[u8],
113) {
114    // The file exists, but it wasn't valid UTF-8.
115    let start = utf8err.valid_up_to();
116    let note = format!("invalid utf-8 at byte `{start}`");
117    let msg = if let Some(len) = utf8err.error_len() {
118        format!(
119            "byte{s} `{bytes}` {are} not valid utf-8",
120            bytes = if len == 1 {
121                format!("{:?}", contents[start])
122            } else {
123                format!("{:?}", &contents[start..start + len])
124            },
125            s = pluralize!(len),
126            are = if len == 1 { "is" } else { "are" },
127        )
128    } else {
129        note.clone()
130    };
131    let contents = String::from_utf8_lossy(contents).to_string();
132    let source = sm.new_source_file(PathBuf::from(path).into(), contents);
133    let span = Span::with_root_ctxt(
134        source.normalized_byte_pos(start as u32),
135        source.normalized_byte_pos(start as u32),
136    );
137    if span.is_dummy() {
138        err.note(note);
139    } else {
140        if sp.is_some() {
141            err.span_note(span, msg);
142        } else {
143            err.span(span);
144            err.span_label(span, msg);
145        }
146    }
147}
148
149/// Given a session and a `source_file`, return a parser. Returns any buffered errors from lexing
150/// the initial token stream.
151fn new_parser_from_source_file(
152    psess: &ParseSess,
153    source_file: Arc<SourceFile>,
154    strip_tokens: StripTokens,
155) -> Result<Parser<'_>, Vec<Diag<'_>>> {
156    let end_pos = source_file.end_position();
157    let stream = source_file_to_stream(psess, source_file, None, strip_tokens)?;
158    let mut parser = Parser::new(psess, stream, None);
159    if parser.token == token::Eof {
160        parser.token.span = Span::new(end_pos, end_pos, parser.token.span.ctxt(), None);
161    }
162    Ok(parser)
163}
164
165/// Given a source string, produces a sequence of token trees.
166///
167/// NOTE: This only strips shebangs, not frontmatter!
168pub fn source_str_to_stream(
169    psess: &ParseSess,
170    name: FileName,
171    source: String,
172    override_span: Option<Span>,
173) -> Result<TokenStream, Vec<Diag<'_>>> {
174    let source_file = psess.source_map().new_source_file(name, source);
175    // FIXME(frontmatter): Consider stripping frontmatter in a future edition. We can't strip them
176    // in the current edition since that would be breaking.
177    // See also <https://github.com/rust-lang/rust/issues/145520>.
178    // Alternatively, stop stripping shebangs here, too, if T-lang and crater approve.
179    source_file_to_stream(psess, source_file, override_span, StripTokens::Shebang)
180}
181
182/// Given a source file, produces a sequence of token trees.
183///
184/// Returns any buffered errors from parsing the token stream.
185fn source_file_to_stream<'psess>(
186    psess: &'psess ParseSess,
187    source_file: Arc<SourceFile>,
188    override_span: Option<Span>,
189    strip_tokens: StripTokens,
190) -> Result<TokenStream, Vec<Diag<'psess>>> {
191    let src = source_file.src.as_ref().unwrap_or_else(|| {
192        psess.dcx().bug(format!(
193            "cannot lex `source_file` without source: {}",
194            psess.source_map().filename_for_diagnostics(&source_file.name)
195        ));
196    });
197
198    lexer::lex_token_trees(psess, src.as_str(), source_file.start_pos, override_span, strip_tokens)
199}
200
201/// Runs the given subparser `f` on the tokens of the given `attr`'s item.
202pub fn parse_in<'a, T>(
203    psess: &'a ParseSess,
204    tts: TokenStream,
205    name: &'static str,
206    mut f: impl FnMut(&mut Parser<'a>) -> PResult<'a, T>,
207) -> PResult<'a, T> {
208    let mut parser = Parser::new(psess, tts, Some(name));
209    let result = f(&mut parser)?;
210    if parser.token != token::Eof {
211        parser.unexpected()?;
212    }
213    Ok(result)
214}
215
216pub fn fake_token_stream_for_item(psess: &ParseSess, item: &ast::Item) -> TokenStream {
217    let source = pprust::item_to_string(item);
218    let filename = FileName::macro_expansion_source_code(&source);
219    unwrap_or_emit_fatal(source_str_to_stream(psess, filename, source, Some(item.span)))
220}
221
222pub fn fake_token_stream_for_crate(psess: &ParseSess, krate: &ast::Crate) -> TokenStream {
223    let source = pprust::crate_to_string_for_macros(krate);
224    let filename = FileName::macro_expansion_source_code(&source);
225    unwrap_or_emit_fatal(source_str_to_stream(
226        psess,
227        filename,
228        source,
229        Some(krate.spans.inner_span),
230    ))
231}
232
233pub fn parse_cfg_attr(
234    cfg_attr: &Attribute,
235    psess: &ParseSess,
236) -> Option<(MetaItemInner, Vec<(AttrItem, Span)>)> {
237    const CFG_ATTR_GRAMMAR_HELP: &str = "#[cfg_attr(condition, attribute, other_attribute, ...)]";
238    const CFG_ATTR_NOTE_REF: &str = "for more information, visit \
239        <https://doc.rust-lang.org/reference/conditional-compilation.html#the-cfg_attr-attribute>";
240
241    match cfg_attr.get_normal_item().args {
242        ast::AttrArgs::Delimited(ast::DelimArgs { dspan, delim, ref tokens })
243            if !tokens.is_empty() =>
244        {
245            check_cfg_attr_bad_delim(psess, dspan, delim);
246            match parse_in(psess, tokens.clone(), "`cfg_attr` input", |p| p.parse_cfg_attr()) {
247                Ok(r) => return Some(r),
248                Err(e) => {
249                    e.with_help(format!("the valid syntax is `{CFG_ATTR_GRAMMAR_HELP}`"))
250                        .with_note(CFG_ATTR_NOTE_REF)
251                        .emit();
252                }
253            }
254        }
255        _ => {
256            psess.dcx().emit_err(errors::MalformedCfgAttr {
257                span: cfg_attr.span,
258                sugg: CFG_ATTR_GRAMMAR_HELP,
259            });
260        }
261    }
262    None
263}
264
265fn check_cfg_attr_bad_delim(psess: &ParseSess, span: DelimSpan, delim: Delimiter) {
266    if let Delimiter::Parenthesis = delim {
267        return;
268    }
269    psess.dcx().emit_err(errors::CfgAttrBadDelim {
270        span: span.entire(),
271        sugg: errors::MetaBadDelimSugg { open: span.open, close: span.close },
272    });
273}
rustc_parse/lib.rs

rustc_parse/
lib.rs
