src/librustc_parse/lexer/tokentrees.rs

   1 use super::{StringReader, UnmatchedBrace};
   2
   3 use rustc_ast_pretty::pprust::token_to_string;
   4 use rustc_data_structures::fx::FxHashMap;
   5 use rustc_errors::PResult;
   6 use rustc_span::Span;
   7 use syntax::token::{self, Token};
   8 use syntax::tokenstream::{
   9     DelimSpan,
  10     IsJoint::{self, *},
  11     TokenStream, TokenTree, TreeAndJoint,
  12 };
  13
  14 impl<'a> StringReader<'a> {
  15     crate fn into_token_trees(self) -> (PResult<'a, TokenStream>, Vec<UnmatchedBrace>) {
  16         let mut tt_reader = TokenTreesReader {
  17             string_reader: self,
  18             token: Token::dummy(),
  19             joint_to_prev: Joint,
  20             open_braces: Vec::new(),
  21             unmatched_braces: Vec::new(),
  22             matching_delim_spans: Vec::new(),
  23             last_unclosed_found_span: None,
  24             last_delim_empty_block_spans: FxHashMap::default(),
  25         };
  26         let res = tt_reader.parse_all_token_trees();
  27         (res, tt_reader.unmatched_braces)
  28     }
  29 }
  30
  31 struct TokenTreesReader<'a> {
  32     string_reader: StringReader<'a>,
  33     token: Token,
  34     joint_to_prev: IsJoint,
  35     /// Stack of open delimiters and their spans. Used for error message.
  36     open_braces: Vec<(token::DelimToken, Span)>,
  37     unmatched_braces: Vec<UnmatchedBrace>,
  38     /// The type and spans for all braces
  39     ///
  40     /// Used only for error recovery when arriving to EOF with mismatched braces.
  41     matching_delim_spans: Vec<(token::DelimToken, Span, Span)>,
  42     last_unclosed_found_span: Option<Span>,
  43     last_delim_empty_block_spans: FxHashMap<token::DelimToken, Span>,
  44 }
  45
  46 impl<'a> TokenTreesReader<'a> {
  47     // Parse a stream of tokens into a list of `TokenTree`s, up to an `Eof`.
  48     fn parse_all_token_trees(&mut self) -> PResult<'a, TokenStream> {
  49         let mut buf = TokenStreamBuilder::default();
  50
  51         self.real_token();
  52         while self.token != token::Eof {
  53             buf.push(self.parse_token_tree()?);
  54         }
  55
  56         Ok(buf.into_token_stream())
  57     }
  58
  59     // Parse a stream of tokens into a list of `TokenTree`s, up to a `CloseDelim`.
  60     fn parse_token_trees_until_close_delim(&mut self) -> TokenStream {
  61         let mut buf = TokenStreamBuilder::default();
  62         loop {
  63             if let token::CloseDelim(..) = self.token.kind {
  64                 return buf.into_token_stream();
  65             }
  66
  67             match self.parse_token_tree() {
  68                 Ok(tree) => buf.push(tree),
  69                 Err(mut e) => {
  70                     e.emit();
  71                     return buf.into_token_stream();
  72                 }
  73             }
  74         }
  75     }
  76
  77     fn parse_token_tree(&mut self) -> PResult<'a, TreeAndJoint> {
  78         let sm = self.string_reader.sess.source_map();
  79         match self.token.kind {
  80             token::Eof => {
  81                 let msg = "this file contains an unclosed delimiter";
  82                 let mut err =
  83                     self.string_reader.sess.span_diagnostic.struct_span_err(self.token.span, msg);
  84                 for &(_, sp) in &self.open_braces {
  85                     err.span_label(sp, "unclosed delimiter");
  86                     self.unmatched_braces.push(UnmatchedBrace {
  87                         expected_delim: token::DelimToken::Brace,
  88                         found_delim: None,
  89                         found_span: self.token.span,
  90                         unclosed_span: Some(sp),
  91                         candidate_span: None,
  92                     });
  93                 }
  94
  95                 if let Some((delim, _)) = self.open_braces.last() {
  96                     if let Some((_, open_sp, close_sp)) =
  97                         self.matching_delim_spans.iter().find(|(d, open_sp, close_sp)| {
  98                             if let Some(close_padding) = sm.span_to_margin(*close_sp) {
  99                                 if let Some(open_padding) = sm.span_to_margin(*open_sp) {
 100                                     return delim == d && close_padding != open_padding;
 101                                 }
 102                             }
 103                             false
 104                         })
 105                     // these are in reverse order as they get inserted on close, but
 106                     {
 107                         // we want the last open/first close
 108                         err.span_label(*open_sp, "this delimiter might not be properly closed...");
 109                         err.span_label(
 110                             *close_sp,
 111                             "...as it matches this but it has different indentation",
 112                         );
 113                     }
 114                 }
 115                 Err(err)
 116             }
 117             token::OpenDelim(delim) => {
 118                 // The span for beginning of the delimited section
 119                 let pre_span = self.token.span;
 120
 121                 // Parse the open delimiter.
 122                 self.open_braces.push((delim, self.token.span));
 123                 self.real_token();
 124
 125                 // Parse the token trees within the delimiters.
 126                 // We stop at any delimiter so we can try to recover if the user
 127                 // uses an incorrect delimiter.
 128                 let tts = self.parse_token_trees_until_close_delim();
 129
 130                 // Expand to cover the entire delimited token tree
 131                 let delim_span = DelimSpan::from_pair(pre_span, self.token.span);
 132
 133                 match self.token.kind {
 134                     // Correct delimiter.
 135                     token::CloseDelim(d) if d == delim => {
 136                         let (open_brace, open_brace_span) = self.open_braces.pop().unwrap();
 137                         let close_brace_span = self.token.span;
 138
 139                         if tts.is_empty() {
 140                             let empty_block_span = open_brace_span.to(close_brace_span);
 141                             self.last_delim_empty_block_spans.insert(delim, empty_block_span);
 142                         }
 143
 144                         if self.open_braces.len() == 0 {
 145                             // Clear up these spans to avoid suggesting them as we've found
 146                             // properly matched delimiters so far for an entire block.
 147                             self.matching_delim_spans.clear();
 148                         } else {
 149                             self.matching_delim_spans.push((
 150                                 open_brace,
 151                                 open_brace_span,
 152                                 close_brace_span,
 153                             ));
 154                         }
 155                         // Parse the closing delimiter.
 156                         self.real_token();
 157                     }
 158                     // Incorrect delimiter.
 159                     token::CloseDelim(other) => {
 160                         let mut unclosed_delimiter = None;
 161                         let mut candidate = None;
 162                         if self.last_unclosed_found_span != Some(self.token.span) {
 163                             // do not complain about the same unclosed delimiter multiple times
 164                             self.last_unclosed_found_span = Some(self.token.span);
 165                             // This is a conservative error: only report the last unclosed
 166                             // delimiter. The previous unclosed delimiters could actually be
 167                             // closed! The parser just hasn't gotten to them yet.
 168                             if let Some(&(_, sp)) = self.open_braces.last() {
 169                                 unclosed_delimiter = Some(sp);
 170                             };
 171                             if let Some(current_padding) = sm.span_to_margin(self.token.span) {
 172                                 for (brace, brace_span) in &self.open_braces {
 173                                     if let Some(padding) = sm.span_to_margin(*brace_span) {
 174                                         // high likelihood of these two corresponding
 175                                         if current_padding == padding && brace == &other {
 176                                             candidate = Some(*brace_span);
 177                                         }
 178                                     }
 179                                 }
 180                             }
 181                             let (tok, _) = self.open_braces.pop().unwrap();
 182                             self.unmatched_braces.push(UnmatchedBrace {
 183                                 expected_delim: tok,
 184                                 found_delim: Some(other),
 185                                 found_span: self.token.span,
 186                                 unclosed_span: unclosed_delimiter,
 187                                 candidate_span: candidate,
 188                             });
 189                         } else {
 190                             self.open_braces.pop();
 191                         }
 192
 193                         // If the incorrect delimiter matches an earlier opening
 194                         // delimiter, then don't consume it (it can be used to
 195                         // close the earlier one). Otherwise, consume it.
 196                         // E.g., we try to recover from:
 197                         // fn foo() {
 198                         //     bar(baz(
 199                         // }  // Incorrect delimiter but matches the earlier `{`
 200                         if !self.open_braces.iter().any(|&(b, _)| b == other) {
 201                             self.real_token();
 202                         }
 203                     }
 204                     token::Eof => {
 205                         // Silently recover, the EOF token will be seen again
 206                         // and an error emitted then. Thus we don't pop from
 207                         // self.open_braces here.
 208                     }
 209                     _ => {}
 210                 }
 211
 212                 Ok(TokenTree::Delimited(delim_span, delim, tts.into()).into())
 213             }
 214             token::CloseDelim(delim) => {
 215                 // An unexpected closing delimiter (i.e., there is no
 216                 // matching opening delimiter).
 217                 let token_str = token_to_string(&self.token);
 218                 let msg = format!("unexpected closing delimiter: `{}`", token_str);
 219                 let mut err =
 220                     self.string_reader.sess.span_diagnostic.struct_span_err(self.token.span, &msg);
 221
 222                 if let Some(span) = self.last_delim_empty_block_spans.remove(&delim) {
 223                     err.span_label(
 224                         span,
 225                         "this block is empty, you might have not meant to close it",
 226                     );
 227                 }
 228                 err.span_label(self.token.span, "unexpected closing delimiter");
 229                 Err(err)
 230             }
 231             _ => {
 232                 let tt = TokenTree::Token(self.token.take());
 233                 self.real_token();
 234                 let is_joint = self.joint_to_prev == Joint && self.token.is_op();
 235                 Ok((tt, if is_joint { Joint } else { NonJoint }))
 236             }
 237         }
 238     }
 239
 240     fn real_token(&mut self) {
 241         self.joint_to_prev = Joint;
 242         loop {
 243             let token = self.string_reader.next_token();
 244             match token.kind {
 245                 token::Whitespace | token::Comment | token::Shebang(_) | token::Unknown(_) => {
 246                     self.joint_to_prev = NonJoint;
 247                 }
 248                 _ => {
 249                     self.token = token;
 250                     return;
 251                 }
 252             }
 253         }
 254     }
 255 }
 256
 257 #[derive(Default)]
 258 struct TokenStreamBuilder {
 259     buf: Vec<TreeAndJoint>,
 260 }
 261
 262 impl TokenStreamBuilder {
 263     fn push(&mut self, (tree, joint): TreeAndJoint) {
 264         if let Some((TokenTree::Token(prev_token), Joint)) = self.buf.last() {
 265             if let TokenTree::Token(token) = &tree {
 266                 if let Some(glued) = prev_token.glue(token) {
 267                     self.buf.pop();
 268                     self.buf.push((TokenTree::Token(glued), joint));
 269                     return;
 270                 }
 271             }
 272         }
 273         self.buf.push((tree, joint))
 274     }
 275
 276     fn into_token_stream(self) -> TokenStream {
 277         TokenStream::new(self.buf)
 278     }
 279 }