rustc_parse/
lib.rs

1//! The main parser interface.
2
3// tidy-alphabetical-start
4#![allow(internal_features)]
5#![allow(rustc::diagnostic_outside_of_impl)]
6#![allow(rustc::untranslatable_diagnostic)]
7#![cfg_attr(bootstrap, feature(let_chains))]
8#![feature(assert_matches)]
9#![feature(box_patterns)]
10#![feature(debug_closure_helpers)]
11#![feature(if_let_guard)]
12#![feature(iter_intersperse)]
13#![feature(string_from_utf8_lossy_owned)]
14#![recursion_limit = "256"]
15// tidy-alphabetical-end
16
17use std::path::{Path, PathBuf};
18use std::str::Utf8Error;
19use std::sync::Arc;
20
21use rustc_ast as ast;
22use rustc_ast::tokenstream::TokenStream;
23use rustc_ast::{AttrItem, Attribute, MetaItemInner, token};
24use rustc_ast_pretty::pprust;
25use rustc_errors::{Diag, EmissionGuarantee, FatalError, PResult, pluralize};
26use rustc_session::parse::ParseSess;
27use rustc_span::source_map::SourceMap;
28use rustc_span::{FileName, SourceFile, Span};
29pub use unicode_normalization::UNICODE_VERSION as UNICODE_NORMALIZATION_VERSION;
30
31pub const MACRO_ARGUMENTS: Option<&str> = Some("macro arguments");
32
33#[macro_use]
34pub mod parser;
35use parser::Parser;
36pub mod lexer;
37pub mod validate_attr;
38
39mod errors;
40
41rustc_fluent_macro::fluent_messages! { "../messages.ftl" }
42
43// Unwrap the result if `Ok`, otherwise emit the diagnostics and abort.
44pub fn unwrap_or_emit_fatal<T>(expr: Result<T, Vec<Diag<'_>>>) -> T {
45    match expr {
46        Ok(expr) => expr,
47        Err(errs) => {
48            for err in errs {
49                err.emit();
50            }
51            FatalError.raise()
52        }
53    }
54}
55
56/// Creates a new parser from a source string. On failure, the errors must be consumed via
57/// `unwrap_or_emit_fatal`, `emit`, `cancel`, etc., otherwise a panic will occur when they are
58/// dropped.
59pub fn new_parser_from_source_str(
60    psess: &ParseSess,
61    name: FileName,
62    source: String,
63) -> Result<Parser<'_>, Vec<Diag<'_>>> {
64    let source_file = psess.source_map().new_source_file(name, source);
65    new_parser_from_source_file(psess, source_file)
66}
67
68/// Creates a new parser from a filename. On failure, the errors must be consumed via
69/// `unwrap_or_emit_fatal`, `emit`, `cancel`, etc., otherwise a panic will occur when they are
70/// dropped.
71///
72/// If a span is given, that is used on an error as the source of the problem.
73pub fn new_parser_from_file<'a>(
74    psess: &'a ParseSess,
75    path: &Path,
76    sp: Option<Span>,
77) -> Result<Parser<'a>, Vec<Diag<'a>>> {
78    let sm = psess.source_map();
79    let source_file = sm.load_file(path).unwrap_or_else(|e| {
80        let msg = format!("couldn't read `{}`: {}", path.display(), e);
81        let mut err = psess.dcx().struct_fatal(msg);
82        if let Ok(contents) = std::fs::read(path)
83            && let Err(utf8err) = String::from_utf8(contents.clone())
84        {
85            utf8_error(
86                sm,
87                &path.display().to_string(),
88                sp,
89                &mut err,
90                utf8err.utf8_error(),
91                &contents,
92            );
93        }
94        if let Some(sp) = sp {
95            err.span(sp);
96        }
97        err.emit();
98    });
99    new_parser_from_source_file(psess, source_file)
100}
101
102pub fn utf8_error<E: EmissionGuarantee>(
103    sm: &SourceMap,
104    path: &str,
105    sp: Option<Span>,
106    err: &mut Diag<'_, E>,
107    utf8err: Utf8Error,
108    contents: &[u8],
109) {
110    // The file exists, but it wasn't valid UTF-8.
111    let start = utf8err.valid_up_to();
112    let note = format!("invalid utf-8 at byte `{start}`");
113    let msg = if let Some(len) = utf8err.error_len() {
114        format!(
115            "byte{s} `{bytes}` {are} not valid utf-8",
116            bytes = if len == 1 {
117                format!("{:?}", contents[start])
118            } else {
119                format!("{:?}", &contents[start..start + len])
120            },
121            s = pluralize!(len),
122            are = if len == 1 { "is" } else { "are" },
123        )
124    } else {
125        note.clone()
126    };
127    let contents = String::from_utf8_lossy(contents).to_string();
128    let source = sm.new_source_file(PathBuf::from(path).into(), contents);
129    let span = Span::with_root_ctxt(
130        source.normalized_byte_pos(start as u32),
131        source.normalized_byte_pos(start as u32),
132    );
133    if span.is_dummy() {
134        err.note(note);
135    } else {
136        if sp.is_some() {
137            err.span_note(span, msg);
138        } else {
139            err.span(span);
140            err.span_label(span, msg);
141        }
142    }
143}
144
145/// Given a session and a `source_file`, return a parser. Returns any buffered errors from lexing
146/// the initial token stream.
147fn new_parser_from_source_file(
148    psess: &ParseSess,
149    source_file: Arc<SourceFile>,
150) -> Result<Parser<'_>, Vec<Diag<'_>>> {
151    let end_pos = source_file.end_position();
152    let stream = source_file_to_stream(psess, source_file, None)?;
153    let mut parser = Parser::new(psess, stream, None);
154    if parser.token == token::Eof {
155        parser.token.span = Span::new(end_pos, end_pos, parser.token.span.ctxt(), None);
156    }
157    Ok(parser)
158}
159
160pub fn source_str_to_stream(
161    psess: &ParseSess,
162    name: FileName,
163    source: String,
164    override_span: Option<Span>,
165) -> Result<TokenStream, Vec<Diag<'_>>> {
166    let source_file = psess.source_map().new_source_file(name, source);
167    source_file_to_stream(psess, source_file, override_span)
168}
169
170/// Given a source file, produces a sequence of token trees. Returns any buffered errors from
171/// parsing the token stream.
172fn source_file_to_stream<'psess>(
173    psess: &'psess ParseSess,
174    source_file: Arc<SourceFile>,
175    override_span: Option<Span>,
176) -> Result<TokenStream, Vec<Diag<'psess>>> {
177    let src = source_file.src.as_ref().unwrap_or_else(|| {
178        psess.dcx().bug(format!(
179            "cannot lex `source_file` without source: {}",
180            psess.source_map().filename_for_diagnostics(&source_file.name)
181        ));
182    });
183
184    lexer::lex_token_trees(psess, src.as_str(), source_file.start_pos, override_span)
185}
186
187/// Runs the given subparser `f` on the tokens of the given `attr`'s item.
188pub fn parse_in<'a, T>(
189    psess: &'a ParseSess,
190    tts: TokenStream,
191    name: &'static str,
192    mut f: impl FnMut(&mut Parser<'a>) -> PResult<'a, T>,
193) -> PResult<'a, T> {
194    let mut parser = Parser::new(psess, tts, Some(name));
195    let result = f(&mut parser)?;
196    if parser.token != token::Eof {
197        parser.unexpected()?;
198    }
199    Ok(result)
200}
201
202pub fn fake_token_stream_for_item(psess: &ParseSess, item: &ast::Item) -> TokenStream {
203    let source = pprust::item_to_string(item);
204    let filename = FileName::macro_expansion_source_code(&source);
205    unwrap_or_emit_fatal(source_str_to_stream(psess, filename, source, Some(item.span)))
206}
207
208pub fn fake_token_stream_for_crate(psess: &ParseSess, krate: &ast::Crate) -> TokenStream {
209    let source = pprust::crate_to_string_for_macros(krate);
210    let filename = FileName::macro_expansion_source_code(&source);
211    unwrap_or_emit_fatal(source_str_to_stream(
212        psess,
213        filename,
214        source,
215        Some(krate.spans.inner_span),
216    ))
217}
218
219pub fn parse_cfg_attr(
220    cfg_attr: &Attribute,
221    psess: &ParseSess,
222) -> Option<(MetaItemInner, Vec<(AttrItem, Span)>)> {
223    const CFG_ATTR_GRAMMAR_HELP: &str = "#[cfg_attr(condition, attribute, other_attribute, ...)]";
224    const CFG_ATTR_NOTE_REF: &str = "for more information, visit \
225        <https://doc.rust-lang.org/reference/conditional-compilation.html#the-cfg_attr-attribute>";
226
227    match cfg_attr.get_normal_item().args {
228        ast::AttrArgs::Delimited(ast::DelimArgs { dspan, delim, ref tokens })
229            if !tokens.is_empty() =>
230        {
231            crate::validate_attr::check_cfg_attr_bad_delim(psess, dspan, delim);
232            match parse_in(psess, tokens.clone(), "`cfg_attr` input", |p| p.parse_cfg_attr()) {
233                Ok(r) => return Some(r),
234                Err(e) => {
235                    e.with_help(format!("the valid syntax is `{CFG_ATTR_GRAMMAR_HELP}`"))
236                        .with_note(CFG_ATTR_NOTE_REF)
237                        .emit();
238                }
239            }
240        }
241        _ => {
242            psess.dcx().emit_err(errors::MalformedCfgAttr {
243                span: cfg_attr.span,
244                sugg: CFG_ATTR_GRAMMAR_HELP,
245            });
246        }
247    }
248    None
249}
rustc_parse/lib.rs

rustc_parse/
lib.rs
