]> git.proxmox.com Git - rustc.git/blob - src/libsyntax/parse/mod.rs
New upstream version 1.20.0+dfsg1
[rustc.git] / src / libsyntax / parse / mod.rs
1 // Copyright 2012-2014 The Rust Project Developers. See the COPYRIGHT
2 // file at the top-level directory of this distribution and at
3 // http://rust-lang.org/COPYRIGHT.
4 //
5 // Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or
6 // http://www.apache.org/licenses/LICENSE-2.0> or the MIT license
7 // <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your
8 // option. This file may not be copied, modified, or distributed
9 // except according to those terms.
10
11 //! The main parser interface
12
13 use ast::{self, CrateConfig};
14 use codemap::{CodeMap, FilePathMapping};
15 use syntax_pos::{self, Span, FileMap, NO_EXPANSION};
16 use errors::{Handler, ColorConfig, DiagnosticBuilder};
17 use feature_gate::UnstableFeatures;
18 use parse::parser::Parser;
19 use ptr::P;
20 use str::char_at;
21 use symbol::Symbol;
22 use tokenstream::{TokenStream, TokenTree};
23
24 use std::cell::RefCell;
25 use std::collections::HashSet;
26 use std::iter;
27 use std::path::{Path, PathBuf};
28 use std::rc::Rc;
29 use std::str;
30
31 pub type PResult<'a, T> = Result<T, DiagnosticBuilder<'a>>;
32
33 #[macro_use]
34 pub mod parser;
35
36 pub mod lexer;
37 pub mod token;
38 pub mod attr;
39
40 pub mod common;
41 pub mod classify;
42 pub mod obsolete;
43
44 /// Info about a parsing session.
45 pub struct ParseSess {
46 pub span_diagnostic: Handler,
47 pub unstable_features: UnstableFeatures,
48 pub config: CrateConfig,
49 pub missing_fragment_specifiers: RefCell<HashSet<Span>>,
50 /// Used to determine and report recursive mod inclusions
51 included_mod_stack: RefCell<Vec<PathBuf>>,
52 code_map: Rc<CodeMap>,
53 }
54
55 impl ParseSess {
56 pub fn new(file_path_mapping: FilePathMapping) -> Self {
57 let cm = Rc::new(CodeMap::new(file_path_mapping));
58 let handler = Handler::with_tty_emitter(ColorConfig::Auto,
59 true,
60 false,
61 Some(cm.clone()));
62 ParseSess::with_span_handler(handler, cm)
63 }
64
65 pub fn with_span_handler(handler: Handler, code_map: Rc<CodeMap>) -> ParseSess {
66 ParseSess {
67 span_diagnostic: handler,
68 unstable_features: UnstableFeatures::from_environment(),
69 config: HashSet::new(),
70 missing_fragment_specifiers: RefCell::new(HashSet::new()),
71 included_mod_stack: RefCell::new(vec![]),
72 code_map: code_map
73 }
74 }
75
76 pub fn codemap(&self) -> &CodeMap {
77 &self.code_map
78 }
79 }
80
81 #[derive(Clone)]
82 pub struct Directory {
83 pub path: PathBuf,
84 pub ownership: DirectoryOwnership,
85 }
86
87 #[derive(Copy, Clone)]
88 pub enum DirectoryOwnership {
89 Owned,
90 UnownedViaBlock,
91 UnownedViaMod(bool /* legacy warnings? */),
92 }
93
94 // a bunch of utility functions of the form parse_<thing>_from_<source>
95 // where <thing> includes crate, expr, item, stmt, tts, and one that
96 // uses a HOF to parse anything, and <source> includes file and
97 // source_str.
98
99 pub fn parse_crate_from_file<'a>(input: &Path, sess: &'a ParseSess) -> PResult<'a, ast::Crate> {
100 let mut parser = new_parser_from_file(sess, input);
101 parser.parse_crate_mod()
102 }
103
104 pub fn parse_crate_attrs_from_file<'a>(input: &Path, sess: &'a ParseSess)
105 -> PResult<'a, Vec<ast::Attribute>> {
106 let mut parser = new_parser_from_file(sess, input);
107 parser.parse_inner_attributes()
108 }
109
110 pub fn parse_crate_from_source_str(name: String, source: String, sess: &ParseSess)
111 -> PResult<ast::Crate> {
112 new_parser_from_source_str(sess, name, source).parse_crate_mod()
113 }
114
115 pub fn parse_crate_attrs_from_source_str(name: String, source: String, sess: &ParseSess)
116 -> PResult<Vec<ast::Attribute>> {
117 new_parser_from_source_str(sess, name, source).parse_inner_attributes()
118 }
119
120 pub fn parse_expr_from_source_str(name: String, source: String, sess: &ParseSess)
121 -> PResult<P<ast::Expr>> {
122 new_parser_from_source_str(sess, name, source).parse_expr()
123 }
124
125 /// Parses an item.
126 ///
127 /// Returns `Ok(Some(item))` when successful, `Ok(None)` when no item was found, and`Err`
128 /// when a syntax error occurred.
129 pub fn parse_item_from_source_str(name: String, source: String, sess: &ParseSess)
130 -> PResult<Option<P<ast::Item>>> {
131 new_parser_from_source_str(sess, name, source).parse_item()
132 }
133
134 pub fn parse_meta_from_source_str(name: String, source: String, sess: &ParseSess)
135 -> PResult<ast::MetaItem> {
136 new_parser_from_source_str(sess, name, source).parse_meta_item()
137 }
138
139 pub fn parse_stmt_from_source_str(name: String, source: String, sess: &ParseSess)
140 -> PResult<Option<ast::Stmt>> {
141 new_parser_from_source_str(sess, name, source).parse_stmt()
142 }
143
144 pub fn parse_stream_from_source_str(name: String, source: String, sess: &ParseSess,
145 override_span: Option<Span>)
146 -> TokenStream {
147 filemap_to_stream(sess, sess.codemap().new_filemap(name, source), override_span)
148 }
149
150 // Create a new parser from a source string
151 pub fn new_parser_from_source_str(sess: &ParseSess, name: String, source: String)
152 -> Parser {
153 let mut parser = filemap_to_parser(sess, sess.codemap().new_filemap(name, source));
154 parser.recurse_into_file_modules = false;
155 parser
156 }
157
158 /// Create a new parser, handling errors as appropriate
159 /// if the file doesn't exist
160 pub fn new_parser_from_file<'a>(sess: &'a ParseSess, path: &Path) -> Parser<'a> {
161 filemap_to_parser(sess, file_to_filemap(sess, path, None))
162 }
163
164 /// Given a session, a crate config, a path, and a span, add
165 /// the file at the given path to the codemap, and return a parser.
166 /// On an error, use the given span as the source of the problem.
167 pub fn new_sub_parser_from_file<'a>(sess: &'a ParseSess,
168 path: &Path,
169 directory_ownership: DirectoryOwnership,
170 module_name: Option<String>,
171 sp: Span) -> Parser<'a> {
172 let mut p = filemap_to_parser(sess, file_to_filemap(sess, path, Some(sp)));
173 p.directory.ownership = directory_ownership;
174 p.root_module_name = module_name;
175 p
176 }
177
178 /// Given a filemap and config, return a parser
179 pub fn filemap_to_parser(sess: & ParseSess, filemap: Rc<FileMap>, ) -> Parser {
180 let end_pos = filemap.end_pos;
181 let mut parser = stream_to_parser(sess, filemap_to_stream(sess, filemap, None));
182
183 if parser.token == token::Eof && parser.span == syntax_pos::DUMMY_SP {
184 parser.span = Span { lo: end_pos, hi: end_pos, ctxt: NO_EXPANSION };
185 }
186
187 parser
188 }
189
190 // must preserve old name for now, because quote! from the *existing*
191 // compiler expands into it
192 pub fn new_parser_from_tts(sess: &ParseSess, tts: Vec<TokenTree>) -> Parser {
193 stream_to_parser(sess, tts.into_iter().collect())
194 }
195
196
197 // base abstractions
198
199 /// Given a session and a path and an optional span (for error reporting),
200 /// add the path to the session's codemap and return the new filemap.
201 fn file_to_filemap(sess: &ParseSess, path: &Path, spanopt: Option<Span>)
202 -> Rc<FileMap> {
203 match sess.codemap().load_file(path) {
204 Ok(filemap) => filemap,
205 Err(e) => {
206 let msg = format!("couldn't read {:?}: {}", path.display(), e);
207 match spanopt {
208 Some(sp) => panic!(sess.span_diagnostic.span_fatal(sp, &msg)),
209 None => panic!(sess.span_diagnostic.fatal(&msg))
210 }
211 }
212 }
213 }
214
215 /// Given a filemap, produce a sequence of token-trees
216 pub fn filemap_to_stream(sess: &ParseSess, filemap: Rc<FileMap>, override_span: Option<Span>)
217 -> TokenStream {
218 let mut srdr = lexer::StringReader::new(sess, filemap);
219 srdr.override_span = override_span;
220 srdr.real_token();
221 panictry!(srdr.parse_all_token_trees())
222 }
223
224 /// Given stream and the `ParseSess`, produce a parser
225 pub fn stream_to_parser(sess: &ParseSess, stream: TokenStream) -> Parser {
226 Parser::new(sess, stream, None, true, false)
227 }
228
229 /// Parse a string representing a character literal into its final form.
230 /// Rather than just accepting/rejecting a given literal, unescapes it as
231 /// well. Can take any slice prefixed by a character escape. Returns the
232 /// character and the number of characters consumed.
233 pub fn char_lit(lit: &str) -> (char, isize) {
234 use std::char;
235
236 // Handle non-escaped chars first.
237 if lit.as_bytes()[0] != b'\\' {
238 // If the first byte isn't '\\' it might part of a multi-byte char, so
239 // get the char with chars().
240 let c = lit.chars().next().unwrap();
241 return (c, 1);
242 }
243
244 // Handle escaped chars.
245 match lit.as_bytes()[1] as char {
246 '"' => ('"', 2),
247 'n' => ('\n', 2),
248 'r' => ('\r', 2),
249 't' => ('\t', 2),
250 '\\' => ('\\', 2),
251 '\'' => ('\'', 2),
252 '0' => ('\0', 2),
253 'x' => {
254 let v = u32::from_str_radix(&lit[2..4], 16).unwrap();
255 let c = char::from_u32(v).unwrap();
256 (c, 4)
257 }
258 'u' => {
259 assert_eq!(lit.as_bytes()[2], b'{');
260 let idx = lit.find('}').unwrap();
261 let v = u32::from_str_radix(&lit[3..idx], 16).unwrap();
262 let c = char::from_u32(v).unwrap();
263 (c, (idx + 1) as isize)
264 }
265 _ => panic!("lexer should have rejected a bad character escape {}", lit)
266 }
267 }
268
269 pub fn escape_default(s: &str) -> String {
270 s.chars().map(char::escape_default).flat_map(|x| x).collect()
271 }
272
273 /// Parse a string representing a string literal into its final form. Does
274 /// unescaping.
275 pub fn str_lit(lit: &str) -> String {
276 debug!("parse_str_lit: given {}", escape_default(lit));
277 let mut res = String::with_capacity(lit.len());
278
279 // FIXME #8372: This could be a for-loop if it didn't borrow the iterator
280 let error = |i| format!("lexer should have rejected {} at {}", lit, i);
281
282 /// Eat everything up to a non-whitespace
283 fn eat<'a>(it: &mut iter::Peekable<str::CharIndices<'a>>) {
284 loop {
285 match it.peek().map(|x| x.1) {
286 Some(' ') | Some('\n') | Some('\r') | Some('\t') => {
287 it.next();
288 },
289 _ => { break; }
290 }
291 }
292 }
293
294 let mut chars = lit.char_indices().peekable();
295 while let Some((i, c)) = chars.next() {
296 match c {
297 '\\' => {
298 let ch = chars.peek().unwrap_or_else(|| {
299 panic!("{}", error(i))
300 }).1;
301
302 if ch == '\n' {
303 eat(&mut chars);
304 } else if ch == '\r' {
305 chars.next();
306 let ch = chars.peek().unwrap_or_else(|| {
307 panic!("{}", error(i))
308 }).1;
309
310 if ch != '\n' {
311 panic!("lexer accepted bare CR");
312 }
313 eat(&mut chars);
314 } else {
315 // otherwise, a normal escape
316 let (c, n) = char_lit(&lit[i..]);
317 for _ in 0..n - 1 { // we don't need to move past the first \
318 chars.next();
319 }
320 res.push(c);
321 }
322 },
323 '\r' => {
324 let ch = chars.peek().unwrap_or_else(|| {
325 panic!("{}", error(i))
326 }).1;
327
328 if ch != '\n' {
329 panic!("lexer accepted bare CR");
330 }
331 chars.next();
332 res.push('\n');
333 }
334 c => res.push(c),
335 }
336 }
337
338 res.shrink_to_fit(); // probably not going to do anything, unless there was an escape.
339 debug!("parse_str_lit: returning {}", res);
340 res
341 }
342
343 /// Parse a string representing a raw string literal into its final form. The
344 /// only operation this does is convert embedded CRLF into a single LF.
345 pub fn raw_str_lit(lit: &str) -> String {
346 debug!("raw_str_lit: given {}", escape_default(lit));
347 let mut res = String::with_capacity(lit.len());
348
349 let mut chars = lit.chars().peekable();
350 while let Some(c) = chars.next() {
351 if c == '\r' {
352 if *chars.peek().unwrap() != '\n' {
353 panic!("lexer accepted bare CR");
354 }
355 chars.next();
356 res.push('\n');
357 } else {
358 res.push(c);
359 }
360 }
361
362 res.shrink_to_fit();
363 res
364 }
365
366 // check if `s` looks like i32 or u1234 etc.
367 fn looks_like_width_suffix(first_chars: &[char], s: &str) -> bool {
368 s.len() > 1 &&
369 first_chars.contains(&char_at(s, 0)) &&
370 s[1..].chars().all(|c| '0' <= c && c <= '9')
371 }
372
373 macro_rules! err {
374 ($opt_diag:expr, |$span:ident, $diag:ident| $($body:tt)*) => {
375 match $opt_diag {
376 Some(($span, $diag)) => { $($body)* }
377 None => return None,
378 }
379 }
380 }
381
382 pub fn lit_token(lit: token::Lit, suf: Option<Symbol>, diag: Option<(Span, &Handler)>)
383 -> (bool /* suffix illegal? */, Option<ast::LitKind>) {
384 use ast::LitKind;
385
386 match lit {
387 token::Byte(i) => (true, Some(LitKind::Byte(byte_lit(&i.as_str()).0))),
388 token::Char(i) => (true, Some(LitKind::Char(char_lit(&i.as_str()).0))),
389
390 // There are some valid suffixes for integer and float literals,
391 // so all the handling is done internally.
392 token::Integer(s) => (false, integer_lit(&s.as_str(), suf, diag)),
393 token::Float(s) => (false, float_lit(&s.as_str(), suf, diag)),
394
395 token::Str_(s) => {
396 let s = Symbol::intern(&str_lit(&s.as_str()));
397 (true, Some(LitKind::Str(s, ast::StrStyle::Cooked)))
398 }
399 token::StrRaw(s, n) => {
400 let s = Symbol::intern(&raw_str_lit(&s.as_str()));
401 (true, Some(LitKind::Str(s, ast::StrStyle::Raw(n))))
402 }
403 token::ByteStr(i) => {
404 (true, Some(LitKind::ByteStr(byte_str_lit(&i.as_str()))))
405 }
406 token::ByteStrRaw(i, _) => {
407 (true, Some(LitKind::ByteStr(Rc::new(i.to_string().into_bytes()))))
408 }
409 }
410 }
411
412 fn filtered_float_lit(data: Symbol, suffix: Option<Symbol>, diag: Option<(Span, &Handler)>)
413 -> Option<ast::LitKind> {
414 debug!("filtered_float_lit: {}, {:?}", data, suffix);
415 let suffix = match suffix {
416 Some(suffix) => suffix,
417 None => return Some(ast::LitKind::FloatUnsuffixed(data)),
418 };
419
420 Some(match &*suffix.as_str() {
421 "f32" => ast::LitKind::Float(data, ast::FloatTy::F32),
422 "f64" => ast::LitKind::Float(data, ast::FloatTy::F64),
423 suf => {
424 err!(diag, |span, diag| {
425 if suf.len() >= 2 && looks_like_width_suffix(&['f'], suf) {
426 // if it looks like a width, lets try to be helpful.
427 let msg = format!("invalid width `{}` for float literal", &suf[1..]);
428 diag.struct_span_err(span, &msg).help("valid widths are 32 and 64").emit()
429 } else {
430 let msg = format!("invalid suffix `{}` for float literal", suf);
431 diag.struct_span_err(span, &msg)
432 .help("valid suffixes are `f32` and `f64`")
433 .emit();
434 }
435 });
436
437 ast::LitKind::FloatUnsuffixed(data)
438 }
439 })
440 }
441 pub fn float_lit(s: &str, suffix: Option<Symbol>, diag: Option<(Span, &Handler)>)
442 -> Option<ast::LitKind> {
443 debug!("float_lit: {:?}, {:?}", s, suffix);
444 // FIXME #2252: bounds checking float literals is deferred until trans
445 let s = s.chars().filter(|&c| c != '_').collect::<String>();
446 filtered_float_lit(Symbol::intern(&s), suffix, diag)
447 }
448
449 /// Parse a string representing a byte literal into its final form. Similar to `char_lit`
450 pub fn byte_lit(lit: &str) -> (u8, usize) {
451 let err = |i| format!("lexer accepted invalid byte literal {} step {}", lit, i);
452
453 if lit.len() == 1 {
454 (lit.as_bytes()[0], 1)
455 } else {
456 assert_eq!(lit.as_bytes()[0], b'\\', "{}", err(0));
457 let b = match lit.as_bytes()[1] {
458 b'"' => b'"',
459 b'n' => b'\n',
460 b'r' => b'\r',
461 b't' => b'\t',
462 b'\\' => b'\\',
463 b'\'' => b'\'',
464 b'0' => b'\0',
465 _ => {
466 match u64::from_str_radix(&lit[2..4], 16).ok() {
467 Some(c) =>
468 if c > 0xFF {
469 panic!(err(2))
470 } else {
471 return (c as u8, 4)
472 },
473 None => panic!(err(3))
474 }
475 }
476 };
477 (b, 2)
478 }
479 }
480
481 pub fn byte_str_lit(lit: &str) -> Rc<Vec<u8>> {
482 let mut res = Vec::with_capacity(lit.len());
483
484 // FIXME #8372: This could be a for-loop if it didn't borrow the iterator
485 let error = |i| format!("lexer should have rejected {} at {}", lit, i);
486
487 /// Eat everything up to a non-whitespace
488 fn eat<I: Iterator<Item=(usize, u8)>>(it: &mut iter::Peekable<I>) {
489 loop {
490 match it.peek().map(|x| x.1) {
491 Some(b' ') | Some(b'\n') | Some(b'\r') | Some(b'\t') => {
492 it.next();
493 },
494 _ => { break; }
495 }
496 }
497 }
498
499 // byte string literals *must* be ASCII, but the escapes don't have to be
500 let mut chars = lit.bytes().enumerate().peekable();
501 loop {
502 match chars.next() {
503 Some((i, b'\\')) => {
504 let em = error(i);
505 match chars.peek().expect(&em).1 {
506 b'\n' => eat(&mut chars),
507 b'\r' => {
508 chars.next();
509 if chars.peek().expect(&em).1 != b'\n' {
510 panic!("lexer accepted bare CR");
511 }
512 eat(&mut chars);
513 }
514 _ => {
515 // otherwise, a normal escape
516 let (c, n) = byte_lit(&lit[i..]);
517 // we don't need to move past the first \
518 for _ in 0..n - 1 {
519 chars.next();
520 }
521 res.push(c);
522 }
523 }
524 },
525 Some((i, b'\r')) => {
526 let em = error(i);
527 if chars.peek().expect(&em).1 != b'\n' {
528 panic!("lexer accepted bare CR");
529 }
530 chars.next();
531 res.push(b'\n');
532 }
533 Some((_, c)) => res.push(c),
534 None => break,
535 }
536 }
537
538 Rc::new(res)
539 }
540
541 pub fn integer_lit(s: &str, suffix: Option<Symbol>, diag: Option<(Span, &Handler)>)
542 -> Option<ast::LitKind> {
543 // s can only be ascii, byte indexing is fine
544
545 let s2 = s.chars().filter(|&c| c != '_').collect::<String>();
546 let mut s = &s2[..];
547
548 debug!("integer_lit: {}, {:?}", s, suffix);
549
550 let mut base = 10;
551 let orig = s;
552 let mut ty = ast::LitIntType::Unsuffixed;
553
554 if char_at(s, 0) == '0' && s.len() > 1 {
555 match char_at(s, 1) {
556 'x' => base = 16,
557 'o' => base = 8,
558 'b' => base = 2,
559 _ => { }
560 }
561 }
562
563 // 1f64 and 2f32 etc. are valid float literals.
564 if let Some(suf) = suffix {
565 if looks_like_width_suffix(&['f'], &suf.as_str()) {
566 let err = match base {
567 16 => Some("hexadecimal float literal is not supported"),
568 8 => Some("octal float literal is not supported"),
569 2 => Some("binary float literal is not supported"),
570 _ => None,
571 };
572 if let Some(err) = err {
573 err!(diag, |span, diag| diag.span_err(span, err));
574 }
575 return filtered_float_lit(Symbol::intern(s), Some(suf), diag)
576 }
577 }
578
579 if base != 10 {
580 s = &s[2..];
581 }
582
583 if let Some(suf) = suffix {
584 if suf.as_str().is_empty() {
585 err!(diag, |span, diag| diag.span_bug(span, "found empty literal suffix in Some"));
586 }
587 ty = match &*suf.as_str() {
588 "isize" => ast::LitIntType::Signed(ast::IntTy::Is),
589 "i8" => ast::LitIntType::Signed(ast::IntTy::I8),
590 "i16" => ast::LitIntType::Signed(ast::IntTy::I16),
591 "i32" => ast::LitIntType::Signed(ast::IntTy::I32),
592 "i64" => ast::LitIntType::Signed(ast::IntTy::I64),
593 "i128" => ast::LitIntType::Signed(ast::IntTy::I128),
594 "usize" => ast::LitIntType::Unsigned(ast::UintTy::Us),
595 "u8" => ast::LitIntType::Unsigned(ast::UintTy::U8),
596 "u16" => ast::LitIntType::Unsigned(ast::UintTy::U16),
597 "u32" => ast::LitIntType::Unsigned(ast::UintTy::U32),
598 "u64" => ast::LitIntType::Unsigned(ast::UintTy::U64),
599 "u128" => ast::LitIntType::Unsigned(ast::UintTy::U128),
600 suf => {
601 // i<digits> and u<digits> look like widths, so lets
602 // give an error message along those lines
603 err!(diag, |span, diag| {
604 if looks_like_width_suffix(&['i', 'u'], suf) {
605 let msg = format!("invalid width `{}` for integer literal", &suf[1..]);
606 diag.struct_span_err(span, &msg)
607 .help("valid widths are 8, 16, 32, 64 and 128")
608 .emit();
609 } else {
610 let msg = format!("invalid suffix `{}` for numeric literal", suf);
611 diag.struct_span_err(span, &msg)
612 .help("the suffix must be one of the integral types \
613 (`u32`, `isize`, etc)")
614 .emit();
615 }
616 });
617
618 ty
619 }
620 }
621 }
622
623 debug!("integer_lit: the type is {:?}, base {:?}, the new string is {:?}, the original \
624 string was {:?}, the original suffix was {:?}", ty, base, s, orig, suffix);
625
626 Some(match u128::from_str_radix(s, base) {
627 Ok(r) => ast::LitKind::Int(r, ty),
628 Err(_) => {
629 // small bases are lexed as if they were base 10, e.g, the string
630 // might be `0b10201`. This will cause the conversion above to fail,
631 // but these cases have errors in the lexer: we don't want to emit
632 // two errors, and we especially don't want to emit this error since
633 // it isn't necessarily true.
634 let already_errored = base < 10 &&
635 s.chars().any(|c| c.to_digit(10).map_or(false, |d| d >= base));
636
637 if !already_errored {
638 err!(diag, |span, diag| diag.span_err(span, "int literal is too large"));
639 }
640 ast::LitKind::Int(0, ty)
641 }
642 })
643 }
644
645 #[cfg(test)]
646 mod tests {
647 use super::*;
648 use syntax_pos::{self, Span, BytePos, Pos, NO_EXPANSION};
649 use codemap::Spanned;
650 use ast::{self, Ident, PatKind};
651 use abi::Abi;
652 use attr::first_attr_value_str_by_name;
653 use parse;
654 use parse::parser::Parser;
655 use print::pprust::item_to_string;
656 use ptr::P;
657 use tokenstream::{self, TokenTree};
658 use util::parser_testing::{string_to_stream, string_to_parser};
659 use util::parser_testing::{string_to_expr, string_to_item, string_to_stmt};
660 use util::ThinVec;
661
662 // produce a syntax_pos::span
663 fn sp(a: u32, b: u32) -> Span {
664 Span {lo: BytePos(a), hi: BytePos(b), ctxt: NO_EXPANSION}
665 }
666
667 fn str2seg(s: &str, lo: u32, hi: u32) -> ast::PathSegment {
668 ast::PathSegment::from_ident(Ident::from_str(s), sp(lo, hi))
669 }
670
671 #[test] fn path_exprs_1() {
672 assert!(string_to_expr("a".to_string()) ==
673 P(ast::Expr{
674 id: ast::DUMMY_NODE_ID,
675 node: ast::ExprKind::Path(None, ast::Path {
676 span: sp(0, 1),
677 segments: vec![str2seg("a", 0, 1)],
678 }),
679 span: sp(0, 1),
680 attrs: ThinVec::new(),
681 }))
682 }
683
684 #[test] fn path_exprs_2 () {
685 assert!(string_to_expr("::a::b".to_string()) ==
686 P(ast::Expr {
687 id: ast::DUMMY_NODE_ID,
688 node: ast::ExprKind::Path(None, ast::Path {
689 span: sp(0, 6),
690 segments: vec![ast::PathSegment::crate_root(sp(0, 2)),
691 str2seg("a", 2, 3),
692 str2seg("b", 5, 6)]
693 }),
694 span: sp(0, 6),
695 attrs: ThinVec::new(),
696 }))
697 }
698
699 #[should_panic]
700 #[test] fn bad_path_expr_1() {
701 string_to_expr("::abc::def::return".to_string());
702 }
703
704 // check the token-tree-ization of macros
705 #[test]
706 fn string_to_tts_macro () {
707 let tts: Vec<_> =
708 string_to_stream("macro_rules! zip (($a)=>($a))".to_string()).trees().collect();
709 let tts: &[TokenTree] = &tts[..];
710
711 match (tts.len(), tts.get(0), tts.get(1), tts.get(2), tts.get(3)) {
712 (
713 4,
714 Some(&TokenTree::Token(_, token::Ident(name_macro_rules))),
715 Some(&TokenTree::Token(_, token::Not)),
716 Some(&TokenTree::Token(_, token::Ident(name_zip))),
717 Some(&TokenTree::Delimited(_, ref macro_delimed)),
718 )
719 if name_macro_rules.name == "macro_rules"
720 && name_zip.name == "zip" => {
721 let tts = &macro_delimed.stream().trees().collect::<Vec<_>>();
722 match (tts.len(), tts.get(0), tts.get(1), tts.get(2)) {
723 (
724 3,
725 Some(&TokenTree::Delimited(_, ref first_delimed)),
726 Some(&TokenTree::Token(_, token::FatArrow)),
727 Some(&TokenTree::Delimited(_, ref second_delimed)),
728 )
729 if macro_delimed.delim == token::Paren => {
730 let tts = &first_delimed.stream().trees().collect::<Vec<_>>();
731 match (tts.len(), tts.get(0), tts.get(1)) {
732 (
733 2,
734 Some(&TokenTree::Token(_, token::Dollar)),
735 Some(&TokenTree::Token(_, token::Ident(ident))),
736 )
737 if first_delimed.delim == token::Paren && ident.name == "a" => {},
738 _ => panic!("value 3: {:?}", *first_delimed),
739 }
740 let tts = &second_delimed.stream().trees().collect::<Vec<_>>();
741 match (tts.len(), tts.get(0), tts.get(1)) {
742 (
743 2,
744 Some(&TokenTree::Token(_, token::Dollar)),
745 Some(&TokenTree::Token(_, token::Ident(ident))),
746 )
747 if second_delimed.delim == token::Paren
748 && ident.name == "a" => {},
749 _ => panic!("value 4: {:?}", *second_delimed),
750 }
751 },
752 _ => panic!("value 2: {:?}", *macro_delimed),
753 }
754 },
755 _ => panic!("value: {:?}",tts),
756 }
757 }
758
759 #[test]
760 fn string_to_tts_1() {
761 let tts = string_to_stream("fn a (b : i32) { b; }".to_string());
762
763 let expected = TokenStream::concat(vec![
764 TokenTree::Token(sp(0, 2), token::Ident(Ident::from_str("fn"))).into(),
765 TokenTree::Token(sp(3, 4), token::Ident(Ident::from_str("a"))).into(),
766 TokenTree::Delimited(
767 sp(5, 14),
768 tokenstream::Delimited {
769 delim: token::DelimToken::Paren,
770 tts: TokenStream::concat(vec![
771 TokenTree::Token(sp(6, 7), token::Ident(Ident::from_str("b"))).into(),
772 TokenTree::Token(sp(8, 9), token::Colon).into(),
773 TokenTree::Token(sp(10, 13), token::Ident(Ident::from_str("i32"))).into(),
774 ]).into(),
775 }).into(),
776 TokenTree::Delimited(
777 sp(15, 21),
778 tokenstream::Delimited {
779 delim: token::DelimToken::Brace,
780 tts: TokenStream::concat(vec![
781 TokenTree::Token(sp(17, 18), token::Ident(Ident::from_str("b"))).into(),
782 TokenTree::Token(sp(18, 19), token::Semi).into(),
783 ]).into(),
784 }).into()
785 ]);
786
787 assert_eq!(tts, expected);
788 }
789
790 #[test] fn ret_expr() {
791 assert!(string_to_expr("return d".to_string()) ==
792 P(ast::Expr{
793 id: ast::DUMMY_NODE_ID,
794 node:ast::ExprKind::Ret(Some(P(ast::Expr{
795 id: ast::DUMMY_NODE_ID,
796 node:ast::ExprKind::Path(None, ast::Path{
797 span: sp(7, 8),
798 segments: vec![str2seg("d", 7, 8)],
799 }),
800 span:sp(7,8),
801 attrs: ThinVec::new(),
802 }))),
803 span:sp(0,8),
804 attrs: ThinVec::new(),
805 }))
806 }
807
808 #[test] fn parse_stmt_1 () {
809 assert!(string_to_stmt("b;".to_string()) ==
810 Some(ast::Stmt {
811 node: ast::StmtKind::Expr(P(ast::Expr {
812 id: ast::DUMMY_NODE_ID,
813 node: ast::ExprKind::Path(None, ast::Path {
814 span:sp(0,1),
815 segments: vec![str2seg("b", 0, 1)],
816 }),
817 span: sp(0,1),
818 attrs: ThinVec::new()})),
819 id: ast::DUMMY_NODE_ID,
820 span: sp(0,1)}))
821
822 }
823
824 fn parser_done(p: Parser){
825 assert_eq!(p.token.clone(), token::Eof);
826 }
827
828 #[test] fn parse_ident_pat () {
829 let sess = ParseSess::new(FilePathMapping::empty());
830 let mut parser = string_to_parser(&sess, "b".to_string());
831 assert!(panictry!(parser.parse_pat())
832 == P(ast::Pat{
833 id: ast::DUMMY_NODE_ID,
834 node: PatKind::Ident(ast::BindingMode::ByValue(ast::Mutability::Immutable),
835 Spanned{ span:sp(0, 1),
836 node: Ident::from_str("b")
837 },
838 None),
839 span: sp(0,1)}));
840 parser_done(parser);
841 }
842
843 // check the contents of the tt manually:
844 #[test] fn parse_fundecl () {
845 // this test depends on the intern order of "fn" and "i32"
846 assert_eq!(string_to_item("fn a (b : i32) { b; }".to_string()),
847 Some(
848 P(ast::Item{ident:Ident::from_str("a"),
849 attrs:Vec::new(),
850 id: ast::DUMMY_NODE_ID,
851 node: ast::ItemKind::Fn(P(ast::FnDecl {
852 inputs: vec![ast::Arg{
853 ty: P(ast::Ty{id: ast::DUMMY_NODE_ID,
854 node: ast::TyKind::Path(None, ast::Path{
855 span:sp(10,13),
856 segments: vec![str2seg("i32", 10, 13)],
857 }),
858 span:sp(10,13)
859 }),
860 pat: P(ast::Pat {
861 id: ast::DUMMY_NODE_ID,
862 node: PatKind::Ident(
863 ast::BindingMode::ByValue(ast::Mutability::Immutable),
864 Spanned{
865 span: sp(6,7),
866 node: Ident::from_str("b")},
867 None
868 ),
869 span: sp(6,7)
870 }),
871 id: ast::DUMMY_NODE_ID
872 }],
873 output: ast::FunctionRetTy::Default(sp(15, 15)),
874 variadic: false
875 }),
876 ast::Unsafety::Normal,
877 Spanned {
878 span: sp(0,2),
879 node: ast::Constness::NotConst,
880 },
881 Abi::Rust,
882 ast::Generics{ // no idea on either of these:
883 lifetimes: Vec::new(),
884 ty_params: Vec::new(),
885 where_clause: ast::WhereClause {
886 id: ast::DUMMY_NODE_ID,
887 predicates: Vec::new(),
888 },
889 span: syntax_pos::DUMMY_SP,
890 },
891 P(ast::Block {
892 stmts: vec![ast::Stmt {
893 node: ast::StmtKind::Semi(P(ast::Expr{
894 id: ast::DUMMY_NODE_ID,
895 node: ast::ExprKind::Path(None,
896 ast::Path{
897 span:sp(17,18),
898 segments: vec![str2seg("b", 17, 18)],
899 }),
900 span: sp(17,18),
901 attrs: ThinVec::new()})),
902 id: ast::DUMMY_NODE_ID,
903 span: sp(17,19)}],
904 id: ast::DUMMY_NODE_ID,
905 rules: ast::BlockCheckMode::Default, // no idea
906 span: sp(15,21),
907 })),
908 vis: ast::Visibility::Inherited,
909 span: sp(0,21)})));
910 }
911
912 #[test] fn parse_use() {
913 let use_s = "use foo::bar::baz;";
914 let vitem = string_to_item(use_s.to_string()).unwrap();
915 let vitem_s = item_to_string(&vitem);
916 assert_eq!(&vitem_s[..], use_s);
917
918 let use_s = "use foo::bar as baz;";
919 let vitem = string_to_item(use_s.to_string()).unwrap();
920 let vitem_s = item_to_string(&vitem);
921 assert_eq!(&vitem_s[..], use_s);
922 }
923
924 #[test] fn parse_extern_crate() {
925 let ex_s = "extern crate foo;";
926 let vitem = string_to_item(ex_s.to_string()).unwrap();
927 let vitem_s = item_to_string(&vitem);
928 assert_eq!(&vitem_s[..], ex_s);
929
930 let ex_s = "extern crate foo as bar;";
931 let vitem = string_to_item(ex_s.to_string()).unwrap();
932 let vitem_s = item_to_string(&vitem);
933 assert_eq!(&vitem_s[..], ex_s);
934 }
935
936 fn get_spans_of_pat_idents(src: &str) -> Vec<Span> {
937 let item = string_to_item(src.to_string()).unwrap();
938
939 struct PatIdentVisitor {
940 spans: Vec<Span>
941 }
942 impl<'a> ::visit::Visitor<'a> for PatIdentVisitor {
943 fn visit_pat(&mut self, p: &'a ast::Pat) {
944 match p.node {
945 PatKind::Ident(_ , ref spannedident, _) => {
946 self.spans.push(spannedident.span.clone());
947 }
948 _ => {
949 ::visit::walk_pat(self, p);
950 }
951 }
952 }
953 }
954 let mut v = PatIdentVisitor { spans: Vec::new() };
955 ::visit::walk_item(&mut v, &item);
956 return v.spans;
957 }
958
959 #[test] fn span_of_self_arg_pat_idents_are_correct() {
960
961 let srcs = ["impl z { fn a (&self, &myarg: i32) {} }",
962 "impl z { fn a (&mut self, &myarg: i32) {} }",
963 "impl z { fn a (&'a self, &myarg: i32) {} }",
964 "impl z { fn a (self, &myarg: i32) {} }",
965 "impl z { fn a (self: Foo, &myarg: i32) {} }",
966 ];
967
968 for &src in &srcs {
969 let spans = get_spans_of_pat_idents(src);
970 let Span{ lo, hi, .. } = spans[0];
971 assert!("self" == &src[lo.to_usize()..hi.to_usize()],
972 "\"{}\" != \"self\". src=\"{}\"",
973 &src[lo.to_usize()..hi.to_usize()], src)
974 }
975 }
976
977 #[test] fn parse_exprs () {
978 // just make sure that they parse....
979 string_to_expr("3 + 4".to_string());
980 string_to_expr("a::z.froob(b,&(987+3))".to_string());
981 }
982
983 #[test] fn attrs_fix_bug () {
984 string_to_item("pub fn mk_file_writer(path: &Path, flags: &[FileFlag])
985 -> Result<Box<Writer>, String> {
986 #[cfg(windows)]
987 fn wb() -> c_int {
988 (O_WRONLY | libc::consts::os::extra::O_BINARY) as c_int
989 }
990
991 #[cfg(unix)]
992 fn wb() -> c_int { O_WRONLY as c_int }
993
994 let mut fflags: c_int = wb();
995 }".to_string());
996 }
997
998 #[test] fn crlf_doc_comments() {
999 let sess = ParseSess::new(FilePathMapping::empty());
1000
1001 let name = "<source>".to_string();
1002 let source = "/// doc comment\r\nfn foo() {}".to_string();
1003 let item = parse_item_from_source_str(name.clone(), source, &sess)
1004 .unwrap().unwrap();
1005 let doc = first_attr_value_str_by_name(&item.attrs, "doc").unwrap();
1006 assert_eq!(doc, "/// doc comment");
1007
1008 let source = "/// doc comment\r\n/// line 2\r\nfn foo() {}".to_string();
1009 let item = parse_item_from_source_str(name.clone(), source, &sess)
1010 .unwrap().unwrap();
1011 let docs = item.attrs.iter().filter(|a| a.path == "doc")
1012 .map(|a| a.value_str().unwrap().to_string()).collect::<Vec<_>>();
1013 let b: &[_] = &["/// doc comment".to_string(), "/// line 2".to_string()];
1014 assert_eq!(&docs[..], b);
1015
1016 let source = "/** doc comment\r\n * with CRLF */\r\nfn foo() {}".to_string();
1017 let item = parse_item_from_source_str(name, source, &sess).unwrap().unwrap();
1018 let doc = first_attr_value_str_by_name(&item.attrs, "doc").unwrap();
1019 assert_eq!(doc, "/** doc comment\n * with CRLF */");
1020 }
1021
1022 #[test]
1023 fn ttdelim_span() {
1024 let sess = ParseSess::new(FilePathMapping::empty());
1025 let expr = parse::parse_expr_from_source_str("foo".to_string(),
1026 "foo!( fn main() { body } )".to_string(), &sess).unwrap();
1027
1028 let tts: Vec<_> = match expr.node {
1029 ast::ExprKind::Mac(ref mac) => mac.node.stream().trees().collect(),
1030 _ => panic!("not a macro"),
1031 };
1032
1033 let span = tts.iter().rev().next().unwrap().span();
1034
1035 match sess.codemap().span_to_snippet(span) {
1036 Ok(s) => assert_eq!(&s[..], "{ body }"),
1037 Err(_) => panic!("could not get snippet"),
1038 }
1039 }
1040
1041 // This tests that when parsing a string (rather than a file) we don't try
1042 // and read in a file for a module declaration and just parse a stub.
1043 // See `recurse_into_file_modules` in the parser.
1044 #[test]
1045 fn out_of_line_mod() {
1046 let sess = ParseSess::new(FilePathMapping::empty());
1047 let item = parse_item_from_source_str(
1048 "foo".to_owned(),
1049 "mod foo { struct S; mod this_does_not_exist; }".to_owned(),
1050 &sess,
1051 ).unwrap().unwrap();
1052
1053 if let ast::ItemKind::Mod(ref m) = item.node {
1054 assert!(m.items.len() == 2);
1055 } else {
1056 panic!();
1057 }
1058 }
1059 }