]> git.proxmox.com Git - rustc.git/blob - src/libsyntax/parse/mod.rs
Merge tag 'upstream/1.5.0+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;
14 use codemap::{self, Span, CodeMap, FileMap};
15 use diagnostic::{SpanHandler, Handler, Auto, FatalError};
16 use parse::attr::ParserAttr;
17 use parse::parser::Parser;
18 use parse::token::InternedString;
19 use ptr::P;
20 use str::char_at;
21
22 use std::cell::RefCell;
23 use std::io::Read;
24 use std::iter;
25 use std::path::{Path, PathBuf};
26 use std::rc::Rc;
27 use std::str;
28
29 pub type PResult<T> = Result<T, FatalError>;
30
31 #[macro_use]
32 pub mod parser;
33
34 pub mod lexer;
35 pub mod token;
36 pub mod attr;
37
38 pub mod common;
39 pub mod classify;
40 pub mod obsolete;
41
42 /// Info about a parsing session.
43 pub struct ParseSess {
44 pub span_diagnostic: SpanHandler, // better be the same as the one in the reader!
45 /// Used to determine and report recursive mod inclusions
46 included_mod_stack: RefCell<Vec<PathBuf>>,
47 }
48
49 impl ParseSess {
50 pub fn new() -> ParseSess {
51 let handler = SpanHandler::new(Handler::new(Auto, None, true), CodeMap::new());
52 ParseSess::with_span_handler(handler)
53 }
54
55 pub fn with_span_handler(sh: SpanHandler) -> ParseSess {
56 ParseSess {
57 span_diagnostic: sh,
58 included_mod_stack: RefCell::new(vec![])
59 }
60 }
61
62 pub fn codemap(&self) -> &CodeMap {
63 &self.span_diagnostic.cm
64 }
65 }
66
67 // a bunch of utility functions of the form parse_<thing>_from_<source>
68 // where <thing> includes crate, expr, item, stmt, tts, and one that
69 // uses a HOF to parse anything, and <source> includes file and
70 // source_str.
71
72 pub fn parse_crate_from_file(
73 input: &Path,
74 cfg: ast::CrateConfig,
75 sess: &ParseSess
76 ) -> ast::Crate {
77 panictry!(new_parser_from_file(sess, cfg, input).parse_crate_mod())
78 // why is there no p.abort_if_errors here?
79 }
80
81 pub fn parse_crate_attrs_from_file(
82 input: &Path,
83 cfg: ast::CrateConfig,
84 sess: &ParseSess
85 ) -> Vec<ast::Attribute> {
86 new_parser_from_file(sess, cfg, input).parse_inner_attributes()
87 }
88
89 pub fn parse_crate_from_source_str(name: String,
90 source: String,
91 cfg: ast::CrateConfig,
92 sess: &ParseSess)
93 -> ast::Crate {
94 let mut p = new_parser_from_source_str(sess,
95 cfg,
96 name,
97 source);
98 maybe_aborted(panictry!(p.parse_crate_mod()),p)
99 }
100
101 pub fn parse_crate_attrs_from_source_str(name: String,
102 source: String,
103 cfg: ast::CrateConfig,
104 sess: &ParseSess)
105 -> Vec<ast::Attribute> {
106 let mut p = new_parser_from_source_str(sess,
107 cfg,
108 name,
109 source);
110 maybe_aborted(p.parse_inner_attributes(), p)
111 }
112
113 pub fn parse_expr_from_source_str(name: String,
114 source: String,
115 cfg: ast::CrateConfig,
116 sess: &ParseSess)
117 -> P<ast::Expr> {
118 let mut p = new_parser_from_source_str(sess, cfg, name, source);
119 maybe_aborted(p.parse_expr(), p)
120 }
121
122 pub fn parse_item_from_source_str(name: String,
123 source: String,
124 cfg: ast::CrateConfig,
125 sess: &ParseSess)
126 -> Option<P<ast::Item>> {
127 let mut p = new_parser_from_source_str(sess, cfg, name, source);
128 maybe_aborted(p.parse_item(),p)
129 }
130
131 pub fn parse_meta_from_source_str(name: String,
132 source: String,
133 cfg: ast::CrateConfig,
134 sess: &ParseSess)
135 -> P<ast::MetaItem> {
136 let mut p = new_parser_from_source_str(sess, cfg, name, source);
137 maybe_aborted(p.parse_meta_item(),p)
138 }
139
140 pub fn parse_stmt_from_source_str(name: String,
141 source: String,
142 cfg: ast::CrateConfig,
143 sess: &ParseSess)
144 -> Option<P<ast::Stmt>> {
145 let mut p = new_parser_from_source_str(
146 sess,
147 cfg,
148 name,
149 source
150 );
151 maybe_aborted(p.parse_stmt(), p)
152 }
153
154 // Warning: This parses with quote_depth > 0, which is not the default.
155 pub fn parse_tts_from_source_str(name: String,
156 source: String,
157 cfg: ast::CrateConfig,
158 sess: &ParseSess)
159 -> Vec<ast::TokenTree> {
160 let mut p = new_parser_from_source_str(
161 sess,
162 cfg,
163 name,
164 source
165 );
166 p.quote_depth += 1;
167 // right now this is re-creating the token trees from ... token trees.
168 maybe_aborted(panictry!(p.parse_all_token_trees()),p)
169 }
170
171 // Create a new parser from a source string
172 pub fn new_parser_from_source_str<'a>(sess: &'a ParseSess,
173 cfg: ast::CrateConfig,
174 name: String,
175 source: String)
176 -> Parser<'a> {
177 filemap_to_parser(sess, sess.codemap().new_filemap(name, source), cfg)
178 }
179
180 /// Create a new parser, handling errors as appropriate
181 /// if the file doesn't exist
182 pub fn new_parser_from_file<'a>(sess: &'a ParseSess,
183 cfg: ast::CrateConfig,
184 path: &Path) -> Parser<'a> {
185 filemap_to_parser(sess, file_to_filemap(sess, path, None), cfg)
186 }
187
188 /// Given a session, a crate config, a path, and a span, add
189 /// the file at the given path to the codemap, and return a parser.
190 /// On an error, use the given span as the source of the problem.
191 pub fn new_sub_parser_from_file<'a>(sess: &'a ParseSess,
192 cfg: ast::CrateConfig,
193 path: &Path,
194 owns_directory: bool,
195 module_name: Option<String>,
196 sp: Span) -> Parser<'a> {
197 let mut p = filemap_to_parser(sess, file_to_filemap(sess, path, Some(sp)), cfg);
198 p.owns_directory = owns_directory;
199 p.root_module_name = module_name;
200 p
201 }
202
203 /// Given a filemap and config, return a parser
204 pub fn filemap_to_parser<'a>(sess: &'a ParseSess,
205 filemap: Rc<FileMap>,
206 cfg: ast::CrateConfig) -> Parser<'a> {
207 let end_pos = filemap.end_pos;
208 let mut parser = tts_to_parser(sess, filemap_to_tts(sess, filemap), cfg);
209
210 if parser.token == token::Eof && parser.span == codemap::DUMMY_SP {
211 parser.span = codemap::mk_sp(end_pos, end_pos);
212 }
213
214 parser
215 }
216
217 // must preserve old name for now, because quote! from the *existing*
218 // compiler expands into it
219 pub fn new_parser_from_tts<'a>(sess: &'a ParseSess,
220 cfg: ast::CrateConfig,
221 tts: Vec<ast::TokenTree>) -> Parser<'a> {
222 tts_to_parser(sess, tts, cfg)
223 }
224
225
226 // base abstractions
227
228 /// Given a session and a path and an optional span (for error reporting),
229 /// add the path to the session's codemap and return the new filemap.
230 fn file_to_filemap(sess: &ParseSess, path: &Path, spanopt: Option<Span>)
231 -> Rc<FileMap> {
232 match sess.codemap().load_file(path) {
233 Ok(filemap) => filemap,
234 Err(e) => {
235 let msg = format!("couldn't read {:?}: {}", path.display(), e);
236 match spanopt {
237 Some(sp) => panic!(sess.span_diagnostic.span_fatal(sp, &msg)),
238 None => sess.span_diagnostic.handler().fatal(&msg)
239 }
240 }
241 }
242 }
243
244 /// Given a filemap, produce a sequence of token-trees
245 pub fn filemap_to_tts(sess: &ParseSess, filemap: Rc<FileMap>)
246 -> Vec<ast::TokenTree> {
247 // it appears to me that the cfg doesn't matter here... indeed,
248 // parsing tt's probably shouldn't require a parser at all.
249 let cfg = Vec::new();
250 let srdr = lexer::StringReader::new(&sess.span_diagnostic, filemap);
251 let mut p1 = Parser::new(sess, cfg, Box::new(srdr));
252 panictry!(p1.parse_all_token_trees())
253 }
254
255 /// Given tts and cfg, produce a parser
256 pub fn tts_to_parser<'a>(sess: &'a ParseSess,
257 tts: Vec<ast::TokenTree>,
258 cfg: ast::CrateConfig) -> Parser<'a> {
259 let trdr = lexer::new_tt_reader(&sess.span_diagnostic, None, None, tts);
260 let mut p = Parser::new(sess, cfg, Box::new(trdr));
261 panictry!(p.check_unknown_macro_variable());
262 p
263 }
264
265 /// Abort if necessary
266 pub fn maybe_aborted<T>(result: T, p: Parser) -> T {
267 p.abort_if_errors();
268 result
269 }
270
271 /// Parse a string representing a character literal into its final form.
272 /// Rather than just accepting/rejecting a given literal, unescapes it as
273 /// well. Can take any slice prefixed by a character escape. Returns the
274 /// character and the number of characters consumed.
275 pub fn char_lit(lit: &str) -> (char, isize) {
276 use std::char;
277
278 let mut chars = lit.chars();
279 let c = match (chars.next(), chars.next()) {
280 (Some(c), None) if c != '\\' => return (c, 1),
281 (Some('\\'), Some(c)) => match c {
282 '"' => Some('"'),
283 'n' => Some('\n'),
284 'r' => Some('\r'),
285 't' => Some('\t'),
286 '\\' => Some('\\'),
287 '\'' => Some('\''),
288 '0' => Some('\0'),
289 _ => { None }
290 },
291 _ => panic!("lexer accepted invalid char escape `{}`", lit)
292 };
293
294 match c {
295 Some(x) => return (x, 2),
296 None => { }
297 }
298
299 let msg = format!("lexer should have rejected a bad character escape {}", lit);
300 let msg2 = &msg[..];
301
302 fn esc(len: usize, lit: &str) -> Option<(char, isize)> {
303 u32::from_str_radix(&lit[2..len], 16).ok()
304 .and_then(char::from_u32)
305 .map(|x| (x, len as isize))
306 }
307
308 let unicode_escape = || -> Option<(char, isize)> {
309 if lit.as_bytes()[2] == b'{' {
310 let idx = lit.find('}').expect(msg2);
311 let subslice = &lit[3..idx];
312 u32::from_str_radix(subslice, 16).ok()
313 .and_then(char::from_u32)
314 .map(|x| (x, subslice.chars().count() as isize + 4))
315 } else {
316 esc(6, lit)
317 }
318 };
319
320 // Unicode escapes
321 return match lit.as_bytes()[1] as char {
322 'x' | 'X' => esc(4, lit),
323 'u' => unicode_escape(),
324 'U' => esc(10, lit),
325 _ => None,
326 }.expect(msg2);
327 }
328
329 /// Parse a string representing a string literal into its final form. Does
330 /// unescaping.
331 pub fn str_lit(lit: &str) -> String {
332 debug!("parse_str_lit: given {}", lit.escape_default());
333 let mut res = String::with_capacity(lit.len());
334
335 // FIXME #8372: This could be a for-loop if it didn't borrow the iterator
336 let error = |i| format!("lexer should have rejected {} at {}", lit, i);
337
338 /// Eat everything up to a non-whitespace
339 fn eat<'a>(it: &mut iter::Peekable<str::CharIndices<'a>>) {
340 loop {
341 match it.peek().map(|x| x.1) {
342 Some(' ') | Some('\n') | Some('\r') | Some('\t') => {
343 it.next();
344 },
345 _ => { break; }
346 }
347 }
348 }
349
350 let mut chars = lit.char_indices().peekable();
351 loop {
352 match chars.next() {
353 Some((i, c)) => {
354 match c {
355 '\\' => {
356 let ch = chars.peek().unwrap_or_else(|| {
357 panic!("{}", error(i))
358 }).1;
359
360 if ch == '\n' {
361 eat(&mut chars);
362 } else if ch == '\r' {
363 chars.next();
364 let ch = chars.peek().unwrap_or_else(|| {
365 panic!("{}", error(i))
366 }).1;
367
368 if ch != '\n' {
369 panic!("lexer accepted bare CR");
370 }
371 eat(&mut chars);
372 } else {
373 // otherwise, a normal escape
374 let (c, n) = char_lit(&lit[i..]);
375 for _ in 0..n - 1 { // we don't need to move past the first \
376 chars.next();
377 }
378 res.push(c);
379 }
380 },
381 '\r' => {
382 let ch = chars.peek().unwrap_or_else(|| {
383 panic!("{}", error(i))
384 }).1;
385
386 if ch != '\n' {
387 panic!("lexer accepted bare CR");
388 }
389 chars.next();
390 res.push('\n');
391 }
392 c => res.push(c),
393 }
394 },
395 None => break
396 }
397 }
398
399 res.shrink_to_fit(); // probably not going to do anything, unless there was an escape.
400 debug!("parse_str_lit: returning {}", res);
401 res
402 }
403
404 /// Parse a string representing a raw string literal into its final form. The
405 /// only operation this does is convert embedded CRLF into a single LF.
406 pub fn raw_str_lit(lit: &str) -> String {
407 debug!("raw_str_lit: given {}", lit.escape_default());
408 let mut res = String::with_capacity(lit.len());
409
410 // FIXME #8372: This could be a for-loop if it didn't borrow the iterator
411 let mut chars = lit.chars().peekable();
412 loop {
413 match chars.next() {
414 Some(c) => {
415 if c == '\r' {
416 if *chars.peek().unwrap() != '\n' {
417 panic!("lexer accepted bare CR");
418 }
419 chars.next();
420 res.push('\n');
421 } else {
422 res.push(c);
423 }
424 },
425 None => break
426 }
427 }
428
429 res.shrink_to_fit();
430 res
431 }
432
433 // check if `s` looks like i32 or u1234 etc.
434 fn looks_like_width_suffix(first_chars: &[char], s: &str) -> bool {
435 s.len() > 1 &&
436 first_chars.contains(&char_at(s, 0)) &&
437 s[1..].chars().all(|c| '0' <= c && c <= '9')
438 }
439
440 fn filtered_float_lit(data: token::InternedString, suffix: Option<&str>,
441 sd: &SpanHandler, sp: Span) -> ast::Lit_ {
442 debug!("filtered_float_lit: {}, {:?}", data, suffix);
443 match suffix.as_ref().map(|s| &**s) {
444 Some("f32") => ast::LitFloat(data, ast::TyF32),
445 Some("f64") => ast::LitFloat(data, ast::TyF64),
446 Some(suf) => {
447 if suf.len() >= 2 && looks_like_width_suffix(&['f'], suf) {
448 // if it looks like a width, lets try to be helpful.
449 sd.span_err(sp, &*format!("invalid width `{}` for float literal", &suf[1..]));
450 sd.fileline_help(sp, "valid widths are 32 and 64");
451 } else {
452 sd.span_err(sp, &*format!("invalid suffix `{}` for float literal", suf));
453 sd.fileline_help(sp, "valid suffixes are `f32` and `f64`");
454 }
455
456 ast::LitFloatUnsuffixed(data)
457 }
458 None => ast::LitFloatUnsuffixed(data)
459 }
460 }
461 pub fn float_lit(s: &str, suffix: Option<InternedString>,
462 sd: &SpanHandler, sp: Span) -> ast::Lit_ {
463 debug!("float_lit: {:?}, {:?}", s, suffix);
464 // FIXME #2252: bounds checking float literals is deferred until trans
465 let s = s.chars().filter(|&c| c != '_').collect::<String>();
466 let data = token::intern_and_get_ident(&s);
467 filtered_float_lit(data, suffix.as_ref().map(|s| &**s), sd, sp)
468 }
469
470 /// Parse a string representing a byte literal into its final form. Similar to `char_lit`
471 pub fn byte_lit(lit: &str) -> (u8, usize) {
472 let err = |i| format!("lexer accepted invalid byte literal {} step {}", lit, i);
473
474 if lit.len() == 1 {
475 (lit.as_bytes()[0], 1)
476 } else {
477 assert!(lit.as_bytes()[0] == b'\\', err(0));
478 let b = match lit.as_bytes()[1] {
479 b'"' => b'"',
480 b'n' => b'\n',
481 b'r' => b'\r',
482 b't' => b'\t',
483 b'\\' => b'\\',
484 b'\'' => b'\'',
485 b'0' => b'\0',
486 _ => {
487 match u64::from_str_radix(&lit[2..4], 16).ok() {
488 Some(c) =>
489 if c > 0xFF {
490 panic!(err(2))
491 } else {
492 return (c as u8, 4)
493 },
494 None => panic!(err(3))
495 }
496 }
497 };
498 return (b, 2);
499 }
500 }
501
502 pub fn byte_str_lit(lit: &str) -> Rc<Vec<u8>> {
503 let mut res = Vec::with_capacity(lit.len());
504
505 // FIXME #8372: This could be a for-loop if it didn't borrow the iterator
506 let error = |i| format!("lexer should have rejected {} at {}", lit, i);
507
508 /// Eat everything up to a non-whitespace
509 fn eat<'a, I: Iterator<Item=(usize, u8)>>(it: &mut iter::Peekable<I>) {
510 loop {
511 match it.peek().map(|x| x.1) {
512 Some(b' ') | Some(b'\n') | Some(b'\r') | Some(b'\t') => {
513 it.next();
514 },
515 _ => { break; }
516 }
517 }
518 }
519
520 // byte string literals *must* be ASCII, but the escapes don't have to be
521 let mut chars = lit.bytes().enumerate().peekable();
522 loop {
523 match chars.next() {
524 Some((i, b'\\')) => {
525 let em = error(i);
526 match chars.peek().expect(&em).1 {
527 b'\n' => eat(&mut chars),
528 b'\r' => {
529 chars.next();
530 if chars.peek().expect(&em).1 != b'\n' {
531 panic!("lexer accepted bare CR");
532 }
533 eat(&mut chars);
534 }
535 _ => {
536 // otherwise, a normal escape
537 let (c, n) = byte_lit(&lit[i..]);
538 // we don't need to move past the first \
539 for _ in 0..n - 1 {
540 chars.next();
541 }
542 res.push(c);
543 }
544 }
545 },
546 Some((i, b'\r')) => {
547 let em = error(i);
548 if chars.peek().expect(&em).1 != b'\n' {
549 panic!("lexer accepted bare CR");
550 }
551 chars.next();
552 res.push(b'\n');
553 }
554 Some((_, c)) => res.push(c),
555 None => break,
556 }
557 }
558
559 Rc::new(res)
560 }
561
562 pub fn integer_lit(s: &str,
563 suffix: Option<InternedString>,
564 sd: &SpanHandler,
565 sp: Span)
566 -> ast::Lit_ {
567 // s can only be ascii, byte indexing is fine
568
569 let s2 = s.chars().filter(|&c| c != '_').collect::<String>();
570 let mut s = &s2[..];
571
572 debug!("integer_lit: {}, {:?}", s, suffix);
573
574 let mut base = 10;
575 let orig = s;
576 let mut ty = ast::UnsuffixedIntLit(ast::Plus);
577
578 if char_at(s, 0) == '0' && s.len() > 1 {
579 match char_at(s, 1) {
580 'x' => base = 16,
581 'o' => base = 8,
582 'b' => base = 2,
583 _ => { }
584 }
585 }
586
587 // 1f64 and 2f32 etc. are valid float literals.
588 if let Some(ref suf) = suffix {
589 if looks_like_width_suffix(&['f'], suf) {
590 match base {
591 16 => sd.span_err(sp, "hexadecimal float literal is not supported"),
592 8 => sd.span_err(sp, "octal float literal is not supported"),
593 2 => sd.span_err(sp, "binary float literal is not supported"),
594 _ => ()
595 }
596 let ident = token::intern_and_get_ident(&*s);
597 return filtered_float_lit(ident, Some(&**suf), sd, sp)
598 }
599 }
600
601 if base != 10 {
602 s = &s[2..];
603 }
604
605 if let Some(ref suf) = suffix {
606 if suf.is_empty() { sd.span_bug(sp, "found empty literal suffix in Some")}
607 ty = match &**suf {
608 "isize" => ast::SignedIntLit(ast::TyIs, ast::Plus),
609 "i8" => ast::SignedIntLit(ast::TyI8, ast::Plus),
610 "i16" => ast::SignedIntLit(ast::TyI16, ast::Plus),
611 "i32" => ast::SignedIntLit(ast::TyI32, ast::Plus),
612 "i64" => ast::SignedIntLit(ast::TyI64, ast::Plus),
613 "usize" => ast::UnsignedIntLit(ast::TyUs),
614 "u8" => ast::UnsignedIntLit(ast::TyU8),
615 "u16" => ast::UnsignedIntLit(ast::TyU16),
616 "u32" => ast::UnsignedIntLit(ast::TyU32),
617 "u64" => ast::UnsignedIntLit(ast::TyU64),
618 _ => {
619 // i<digits> and u<digits> look like widths, so lets
620 // give an error message along those lines
621 if looks_like_width_suffix(&['i', 'u'], suf) {
622 sd.span_err(sp, &*format!("invalid width `{}` for integer literal",
623 &suf[1..]));
624 sd.fileline_help(sp, "valid widths are 8, 16, 32 and 64");
625 } else {
626 sd.span_err(sp, &*format!("invalid suffix `{}` for numeric literal", suf));
627 sd.fileline_help(sp, "the suffix must be one of the integral types \
628 (`u32`, `isize`, etc)");
629 }
630
631 ty
632 }
633 }
634 }
635
636 debug!("integer_lit: the type is {:?}, base {:?}, the new string is {:?}, the original \
637 string was {:?}, the original suffix was {:?}", ty, base, s, orig, suffix);
638
639 let res = match u64::from_str_radix(s, base).ok() {
640 Some(r) => r,
641 None => {
642 // small bases are lexed as if they were base 10, e.g, the string
643 // might be `0b10201`. This will cause the conversion above to fail,
644 // but these cases have errors in the lexer: we don't want to emit
645 // two errors, and we especially don't want to emit this error since
646 // it isn't necessarily true.
647 let already_errored = base < 10 &&
648 s.chars().any(|c| c.to_digit(10).map_or(false, |d| d >= base));
649
650 if !already_errored {
651 sd.span_err(sp, "int literal is too large");
652 }
653 0
654 }
655 };
656
657 // adjust the sign
658 let sign = ast::Sign::new(res);
659 match ty {
660 ast::SignedIntLit(t, _) => ast::LitInt(res, ast::SignedIntLit(t, sign)),
661 ast::UnsuffixedIntLit(_) => ast::LitInt(res, ast::UnsuffixedIntLit(sign)),
662 us@ast::UnsignedIntLit(_) => ast::LitInt(res, us)
663 }
664 }
665
666 #[cfg(test)]
667 mod tests {
668 use super::*;
669 use std::rc::Rc;
670 use codemap::{Span, BytePos, Pos, Spanned, NO_EXPANSION};
671 use owned_slice::OwnedSlice;
672 use ast;
673 use abi;
674 use attr::{first_attr_value_str_by_name, AttrMetaMethods};
675 use parse;
676 use parse::parser::Parser;
677 use parse::token::{str_to_ident};
678 use print::pprust::item_to_string;
679 use ptr::P;
680 use util::parser_testing::{string_to_tts, string_to_parser};
681 use util::parser_testing::{string_to_expr, string_to_item, string_to_stmt};
682
683 // produce a codemap::span
684 fn sp(a: u32, b: u32) -> Span {
685 Span {lo: BytePos(a), hi: BytePos(b), expn_id: NO_EXPANSION}
686 }
687
688 #[test] fn path_exprs_1() {
689 assert!(string_to_expr("a".to_string()) ==
690 P(ast::Expr{
691 id: ast::DUMMY_NODE_ID,
692 node: ast::ExprPath(None, ast::Path {
693 span: sp(0, 1),
694 global: false,
695 segments: vec!(
696 ast::PathSegment {
697 identifier: str_to_ident("a"),
698 parameters: ast::PathParameters::none(),
699 }
700 ),
701 }),
702 span: sp(0, 1)
703 }))
704 }
705
706 #[test] fn path_exprs_2 () {
707 assert!(string_to_expr("::a::b".to_string()) ==
708 P(ast::Expr {
709 id: ast::DUMMY_NODE_ID,
710 node: ast::ExprPath(None, ast::Path {
711 span: sp(0, 6),
712 global: true,
713 segments: vec!(
714 ast::PathSegment {
715 identifier: str_to_ident("a"),
716 parameters: ast::PathParameters::none(),
717 },
718 ast::PathSegment {
719 identifier: str_to_ident("b"),
720 parameters: ast::PathParameters::none(),
721 }
722 )
723 }),
724 span: sp(0, 6)
725 }))
726 }
727
728 #[should_panic]
729 #[test] fn bad_path_expr_1() {
730 string_to_expr("::abc::def::return".to_string());
731 }
732
733 // check the token-tree-ization of macros
734 #[test]
735 fn string_to_tts_macro () {
736 let tts = string_to_tts("macro_rules! zip (($a)=>($a))".to_string());
737 let tts: &[ast::TokenTree] = &tts[..];
738
739 match (tts.len(), tts.get(0), tts.get(1), tts.get(2), tts.get(3)) {
740 (
741 4,
742 Some(&ast::TtToken(_, token::Ident(name_macro_rules, token::Plain))),
743 Some(&ast::TtToken(_, token::Not)),
744 Some(&ast::TtToken(_, token::Ident(name_zip, token::Plain))),
745 Some(&ast::TtDelimited(_, ref macro_delimed)),
746 )
747 if name_macro_rules.name.as_str() == "macro_rules"
748 && name_zip.name.as_str() == "zip" => {
749 let tts = &macro_delimed.tts[..];
750 match (tts.len(), tts.get(0), tts.get(1), tts.get(2)) {
751 (
752 3,
753 Some(&ast::TtDelimited(_, ref first_delimed)),
754 Some(&ast::TtToken(_, token::FatArrow)),
755 Some(&ast::TtDelimited(_, ref second_delimed)),
756 )
757 if macro_delimed.delim == token::Paren => {
758 let tts = &first_delimed.tts[..];
759 match (tts.len(), tts.get(0), tts.get(1)) {
760 (
761 2,
762 Some(&ast::TtToken(_, token::Dollar)),
763 Some(&ast::TtToken(_, token::Ident(ident, token::Plain))),
764 )
765 if first_delimed.delim == token::Paren
766 && ident.name.as_str() == "a" => {},
767 _ => panic!("value 3: {:?}", **first_delimed),
768 }
769 let tts = &second_delimed.tts[..];
770 match (tts.len(), tts.get(0), tts.get(1)) {
771 (
772 2,
773 Some(&ast::TtToken(_, token::Dollar)),
774 Some(&ast::TtToken(_, token::Ident(ident, token::Plain))),
775 )
776 if second_delimed.delim == token::Paren
777 && ident.name.as_str() == "a" => {},
778 _ => panic!("value 4: {:?}", **second_delimed),
779 }
780 },
781 _ => panic!("value 2: {:?}", **macro_delimed),
782 }
783 },
784 _ => panic!("value: {:?}",tts),
785 }
786 }
787
788 #[test]
789 fn string_to_tts_1() {
790 let tts = string_to_tts("fn a (b : i32) { b; }".to_string());
791
792 let expected = vec![
793 ast::TtToken(sp(0, 2),
794 token::Ident(str_to_ident("fn"),
795 token::IdentStyle::Plain)),
796 ast::TtToken(sp(3, 4),
797 token::Ident(str_to_ident("a"),
798 token::IdentStyle::Plain)),
799 ast::TtDelimited(
800 sp(5, 14),
801 Rc::new(ast::Delimited {
802 delim: token::DelimToken::Paren,
803 open_span: sp(5, 6),
804 tts: vec![
805 ast::TtToken(sp(6, 7),
806 token::Ident(str_to_ident("b"),
807 token::IdentStyle::Plain)),
808 ast::TtToken(sp(8, 9),
809 token::Colon),
810 ast::TtToken(sp(10, 13),
811 token::Ident(str_to_ident("i32"),
812 token::IdentStyle::Plain)),
813 ],
814 close_span: sp(13, 14),
815 })),
816 ast::TtDelimited(
817 sp(15, 21),
818 Rc::new(ast::Delimited {
819 delim: token::DelimToken::Brace,
820 open_span: sp(15, 16),
821 tts: vec![
822 ast::TtToken(sp(17, 18),
823 token::Ident(str_to_ident("b"),
824 token::IdentStyle::Plain)),
825 ast::TtToken(sp(18, 19),
826 token::Semi)
827 ],
828 close_span: sp(20, 21),
829 }))
830 ];
831
832 assert_eq!(tts, expected);
833 }
834
835 #[test] fn ret_expr() {
836 assert!(string_to_expr("return d".to_string()) ==
837 P(ast::Expr{
838 id: ast::DUMMY_NODE_ID,
839 node:ast::ExprRet(Some(P(ast::Expr{
840 id: ast::DUMMY_NODE_ID,
841 node:ast::ExprPath(None, ast::Path{
842 span: sp(7, 8),
843 global: false,
844 segments: vec!(
845 ast::PathSegment {
846 identifier: str_to_ident("d"),
847 parameters: ast::PathParameters::none(),
848 }
849 ),
850 }),
851 span:sp(7,8)
852 }))),
853 span:sp(0,8)
854 }))
855 }
856
857 #[test] fn parse_stmt_1 () {
858 assert!(string_to_stmt("b;".to_string()) ==
859 P(Spanned{
860 node: ast::StmtExpr(P(ast::Expr {
861 id: ast::DUMMY_NODE_ID,
862 node: ast::ExprPath(None, ast::Path {
863 span:sp(0,1),
864 global:false,
865 segments: vec!(
866 ast::PathSegment {
867 identifier: str_to_ident("b"),
868 parameters: ast::PathParameters::none(),
869 }
870 ),
871 }),
872 span: sp(0,1)}),
873 ast::DUMMY_NODE_ID),
874 span: sp(0,1)}))
875
876 }
877
878 fn parser_done(p: Parser){
879 assert_eq!(p.token.clone(), token::Eof);
880 }
881
882 #[test] fn parse_ident_pat () {
883 let sess = ParseSess::new();
884 let mut parser = string_to_parser(&sess, "b".to_string());
885 assert!(panictry!(parser.parse_pat_nopanic())
886 == P(ast::Pat{
887 id: ast::DUMMY_NODE_ID,
888 node: ast::PatIdent(ast::BindByValue(ast::MutImmutable),
889 Spanned{ span:sp(0, 1),
890 node: str_to_ident("b")
891 },
892 None),
893 span: sp(0,1)}));
894 parser_done(parser);
895 }
896
897 // check the contents of the tt manually:
898 #[test] fn parse_fundecl () {
899 // this test depends on the intern order of "fn" and "i32"
900 assert_eq!(string_to_item("fn a (b : i32) { b; }".to_string()),
901 Some(
902 P(ast::Item{ident:str_to_ident("a"),
903 attrs:Vec::new(),
904 id: ast::DUMMY_NODE_ID,
905 node: ast::ItemFn(P(ast::FnDecl {
906 inputs: vec!(ast::Arg{
907 ty: P(ast::Ty{id: ast::DUMMY_NODE_ID,
908 node: ast::TyPath(None, ast::Path{
909 span:sp(10,13),
910 global:false,
911 segments: vec!(
912 ast::PathSegment {
913 identifier:
914 str_to_ident("i32"),
915 parameters: ast::PathParameters::none(),
916 }
917 ),
918 }),
919 span:sp(10,13)
920 }),
921 pat: P(ast::Pat {
922 id: ast::DUMMY_NODE_ID,
923 node: ast::PatIdent(
924 ast::BindByValue(ast::MutImmutable),
925 Spanned{
926 span: sp(6,7),
927 node: str_to_ident("b")},
928 None
929 ),
930 span: sp(6,7)
931 }),
932 id: ast::DUMMY_NODE_ID
933 }),
934 output: ast::DefaultReturn(sp(15, 15)),
935 variadic: false
936 }),
937 ast::Unsafety::Normal,
938 ast::Constness::NotConst,
939 abi::Rust,
940 ast::Generics{ // no idea on either of these:
941 lifetimes: Vec::new(),
942 ty_params: OwnedSlice::empty(),
943 where_clause: ast::WhereClause {
944 id: ast::DUMMY_NODE_ID,
945 predicates: Vec::new(),
946 }
947 },
948 P(ast::Block {
949 stmts: vec!(P(Spanned{
950 node: ast::StmtSemi(P(ast::Expr{
951 id: ast::DUMMY_NODE_ID,
952 node: ast::ExprPath(None,
953 ast::Path{
954 span:sp(17,18),
955 global:false,
956 segments: vec!(
957 ast::PathSegment {
958 identifier:
959 str_to_ident(
960 "b"),
961 parameters:
962 ast::PathParameters::none(),
963 }
964 ),
965 }),
966 span: sp(17,18)}),
967 ast::DUMMY_NODE_ID),
968 span: sp(17,19)})),
969 expr: None,
970 id: ast::DUMMY_NODE_ID,
971 rules: ast::DefaultBlock, // no idea
972 span: sp(15,21),
973 })),
974 vis: ast::Inherited,
975 span: sp(0,21)})));
976 }
977
978 #[test] fn parse_use() {
979 let use_s = "use foo::bar::baz;";
980 let vitem = string_to_item(use_s.to_string()).unwrap();
981 let vitem_s = item_to_string(&*vitem);
982 assert_eq!(&vitem_s[..], use_s);
983
984 let use_s = "use foo::bar as baz;";
985 let vitem = string_to_item(use_s.to_string()).unwrap();
986 let vitem_s = item_to_string(&*vitem);
987 assert_eq!(&vitem_s[..], use_s);
988 }
989
990 #[test] fn parse_extern_crate() {
991 let ex_s = "extern crate foo;";
992 let vitem = string_to_item(ex_s.to_string()).unwrap();
993 let vitem_s = item_to_string(&*vitem);
994 assert_eq!(&vitem_s[..], ex_s);
995
996 let ex_s = "extern crate foo as bar;";
997 let vitem = string_to_item(ex_s.to_string()).unwrap();
998 let vitem_s = item_to_string(&*vitem);
999 assert_eq!(&vitem_s[..], ex_s);
1000 }
1001
1002 fn get_spans_of_pat_idents(src: &str) -> Vec<Span> {
1003 let item = string_to_item(src.to_string()).unwrap();
1004
1005 struct PatIdentVisitor {
1006 spans: Vec<Span>
1007 }
1008 impl<'v> ::visit::Visitor<'v> for PatIdentVisitor {
1009 fn visit_pat(&mut self, p: &'v ast::Pat) {
1010 match p.node {
1011 ast::PatIdent(_ , ref spannedident, _) => {
1012 self.spans.push(spannedident.span.clone());
1013 }
1014 _ => {
1015 ::visit::walk_pat(self, p);
1016 }
1017 }
1018 }
1019 }
1020 let mut v = PatIdentVisitor { spans: Vec::new() };
1021 ::visit::walk_item(&mut v, &*item);
1022 return v.spans;
1023 }
1024
1025 #[test] fn span_of_self_arg_pat_idents_are_correct() {
1026
1027 let srcs = ["impl z { fn a (&self, &myarg: i32) {} }",
1028 "impl z { fn a (&mut self, &myarg: i32) {} }",
1029 "impl z { fn a (&'a self, &myarg: i32) {} }",
1030 "impl z { fn a (self, &myarg: i32) {} }",
1031 "impl z { fn a (self: Foo, &myarg: i32) {} }",
1032 ];
1033
1034 for &src in &srcs {
1035 let spans = get_spans_of_pat_idents(src);
1036 let Span{ lo, hi, .. } = spans[0];
1037 assert!("self" == &src[lo.to_usize()..hi.to_usize()],
1038 "\"{}\" != \"self\". src=\"{}\"",
1039 &src[lo.to_usize()..hi.to_usize()], src)
1040 }
1041 }
1042
1043 #[test] fn parse_exprs () {
1044 // just make sure that they parse....
1045 string_to_expr("3 + 4".to_string());
1046 string_to_expr("a::z.froob(b,&(987+3))".to_string());
1047 }
1048
1049 #[test] fn attrs_fix_bug () {
1050 string_to_item("pub fn mk_file_writer(path: &Path, flags: &[FileFlag])
1051 -> Result<Box<Writer>, String> {
1052 #[cfg(windows)]
1053 fn wb() -> c_int {
1054 (O_WRONLY | libc::consts::os::extra::O_BINARY) as c_int
1055 }
1056
1057 #[cfg(unix)]
1058 fn wb() -> c_int { O_WRONLY as c_int }
1059
1060 let mut fflags: c_int = wb();
1061 }".to_string());
1062 }
1063
1064 #[test] fn crlf_doc_comments() {
1065 let sess = ParseSess::new();
1066
1067 let name = "<source>".to_string();
1068 let source = "/// doc comment\r\nfn foo() {}".to_string();
1069 let item = parse_item_from_source_str(name.clone(), source, Vec::new(), &sess).unwrap();
1070 let doc = first_attr_value_str_by_name(&item.attrs, "doc").unwrap();
1071 assert_eq!(&doc[..], "/// doc comment");
1072
1073 let source = "/// doc comment\r\n/// line 2\r\nfn foo() {}".to_string();
1074 let item = parse_item_from_source_str(name.clone(), source, Vec::new(), &sess).unwrap();
1075 let docs = item.attrs.iter().filter(|a| &*a.name() == "doc")
1076 .map(|a| a.value_str().unwrap().to_string()).collect::<Vec<_>>();
1077 let b: &[_] = &["/// doc comment".to_string(), "/// line 2".to_string()];
1078 assert_eq!(&docs[..], b);
1079
1080 let source = "/** doc comment\r\n * with CRLF */\r\nfn foo() {}".to_string();
1081 let item = parse_item_from_source_str(name, source, Vec::new(), &sess).unwrap();
1082 let doc = first_attr_value_str_by_name(&item.attrs, "doc").unwrap();
1083 assert_eq!(&doc[..], "/** doc comment\n * with CRLF */");
1084 }
1085
1086 #[test]
1087 fn ttdelim_span() {
1088 let sess = ParseSess::new();
1089 let expr = parse::parse_expr_from_source_str("foo".to_string(),
1090 "foo!( fn main() { body } )".to_string(), vec![], &sess);
1091
1092 let tts = match expr.node {
1093 ast::ExprMac(ref mac) => mac.node.tts.clone(),
1094 _ => panic!("not a macro"),
1095 };
1096
1097 let span = tts.iter().rev().next().unwrap().get_span();
1098
1099 match sess.codemap().span_to_snippet(span) {
1100 Ok(s) => assert_eq!(&s[..], "{ body }"),
1101 Err(_) => panic!("could not get snippet"),
1102 }
1103 }
1104 }