Auto merge of #57944 - estebank:unclosed-delim-the-quickening, r=<try>

Deduplicate mismatched delimiter errors

Delay unmatched delimiter errors until after the parser has run to deduplicate them when parsing and attempt recovering intelligently.

Second attempt at #54029, follow up to #53949. Fix #31528.

Author: bors

src/librustc_errors/emitter.rs

@@ -672,8 +672,8 @@ impl EmitterWriter {
         //   | |  something about `foo`
         //   | something about `fn foo()`
         annotations_position.sort_by(|a, b| {
-            // Decreasing order
-            a.1.len().cmp(&b.1.len()).reverse()
+            // Decreasing order. When `a` and `b` are the same length, prefer `Primary`.
+            (a.1.len(), !a.1.is_primary).cmp(&(b.1.len(), !b.1.is_primary)).reverse()
         });
 
         // Write the underlines.

src/librustc_metadata/cstore_impl.rs

@@ -29,6 +29,7 @@ use syntax::attr;
 use syntax::source_map;
 use syntax::edition::Edition;
 use syntax::parse::source_file_to_stream;
+use syntax::parse::parser::emit_unclosed_delims;
 use syntax::symbol::Symbol;
 use syntax_pos::{Span, NO_EXPANSION, FileName};
 use rustc_data_structures::bit_set::BitSet;
@@ -436,7 +437,8 @@ impl cstore::CStore {
 
         let source_file = sess.parse_sess.source_map().new_source_file(source_name, def.body);
         let local_span = Span::new(source_file.start_pos, source_file.end_pos, NO_EXPANSION);
-        let body = source_file_to_stream(&sess.parse_sess, source_file, None);
+        let (body, errors) = source_file_to_stream(&sess.parse_sess, source_file, None);
+        emit_unclosed_delims(&errors, &sess.diagnostic());
 
         // Mark the attrs as used
         let attrs = data.get_item_attrs(id.index, sess);

src/libsyntax/parse/lexer/mod.rs

@@ -33,6 +33,15 @@ impl Default for TokenAndSpan {
     }
 }
 
+#[derive(Clone, Debug)]
+pub struct UnmatchedBrace {
+    pub expected_delim: token::DelimToken,
+    pub found_delim: token::DelimToken,
+    pub found_span: Span,
+    pub unclosed_span: Option<Span>,
+    pub candidate_span: Option<Span>,
+}
+
 pub struct StringReader<'a> {
     pub sess: &'a ParseSess,
     /// The absolute offset within the source_map of the next character to read
@@ -58,6 +67,7 @@ pub struct StringReader<'a> {
     span_src_raw: Span,
     /// Stack of open delimiters and their spans. Used for error message.
     open_braces: Vec<(token::DelimToken, Span)>,
+    crate unmatched_braces: Vec<UnmatchedBrace>,
     /// The type and spans for all braces
     ///
     /// Used only for error recovery when arriving to EOF with mismatched braces.
@@ -222,6 +232,7 @@ impl<'a> StringReader<'a> {
             span: syntax_pos::DUMMY_SP,
             span_src_raw: syntax_pos::DUMMY_SP,
             open_braces: Vec::new(),
+            unmatched_braces: Vec::new(),
             matching_delim_spans: Vec::new(),
             override_span,
             last_unclosed_found_span: None,

src/libsyntax/parse/lexer/tokentrees.rs

@@ -1,5 +1,5 @@
 use crate::print::pprust::token_to_string;
-use crate::parse::lexer::StringReader;
+use crate::parse::lexer::{StringReader, UnmatchedBrace};
 use crate::parse::{token, PResult};
 use crate::tokenstream::{DelimSpan, IsJoint::*, TokenStream, TokenTree, TreeAndJoint};
 
@@ -101,38 +101,38 @@ impl<'a> StringReader<'a> {
                     }
                     // Incorrect delimiter.
                     token::CloseDelim(other) => {
-                        let token_str = token_to_string(&self.token);
+                        let mut unclosed_delimiter = None;
+                        let mut candidate = None;
                         if self.last_unclosed_found_span != Some(self.span) {
                             // do not complain about the same unclosed delimiter multiple times
                             self.last_unclosed_found_span = Some(self.span);
-                            let msg = format!("incorrect close delimiter: `{}`", token_str);
-                            let mut err = self.sess.span_diagnostic.struct_span_err(
-                                self.span,
-                                &msg,
-                            );
-                            err.span_label(self.span, "incorrect close delimiter");
                             // This is a conservative error: only report the last unclosed
                             // delimiter. The previous unclosed delimiters could actually be
                             // closed! The parser just hasn't gotten to them yet.
                             if let Some(&(_, sp)) = self.open_braces.last() {
-                                err.span_label(sp, "un-closed delimiter");
+                                unclosed_delimiter = Some(sp);
                             };
                             if let Some(current_padding) = sm.span_to_margin(self.span) {
                                 for (brace, brace_span) in &self.open_braces {
                                     if let Some(padding) = sm.span_to_margin(*brace_span) {
                                         // high likelihood of these two corresponding
                                         if current_padding == padding && brace == &other {
-                                            err.span_label(
-                                                *brace_span,
-                                                "close delimiter possibly meant for this",
-                                            );
+                                            candidate = Some(*brace_span);
                                         }
                                     }
                                 }
                             }
-                            err.emit();
+                            let (tok, _) = self.open_braces.pop().unwrap();
+                            self.unmatched_braces.push(UnmatchedBrace {
+                                expected_delim: tok,
+                                found_delim: other,
+                                found_span: self.span,
+                                unclosed_span: unclosed_delimiter,
+                                candidate_span: candidate,
+                            });
+                        } else {
+                            self.open_braces.pop();
                         }
-                        self.open_braces.pop().unwrap();
 
                         // If the incorrect delimiter matches an earlier opening
                         // delimiter, then don't consume it (it can be used to

src/libsyntax/parse/mod.rs

@@ -9,6 +9,7 @@ use crate::parse::parser::Parser;
 use crate::symbol::Symbol;
 use crate::tokenstream::{TokenStream, TokenTree};
 use crate::diagnostics::plugin::ErrorMap;
+use crate::print::pprust::token_to_string;
 
 use rustc_data_structures::sync::{Lrc, Lock};
 use syntax_pos::{Span, SourceFile, FileName, MultiSpan};
@@ -136,15 +137,17 @@ pub fn parse_crate_attrs_from_source_str(name: FileName, source: String, sess: &
     new_parser_from_source_str(sess, name, source).parse_inner_attributes()
 }
 
-pub fn parse_stream_from_source_str(name: FileName, source: String, sess: &ParseSess,
-                                    override_span: Option<Span>)
-                                    -> TokenStream {
+pub fn parse_stream_from_source_str(
+    name: FileName,
+    source: String,
+    sess: &ParseSess,
+    override_span: Option<Span>,
+) -> (TokenStream, Vec<lexer::UnmatchedBrace>) {
     source_file_to_stream(sess, sess.source_map().new_source_file(name, source), override_span)
 }
 
 /// Create a new parser from a source string
-pub fn new_parser_from_source_str(sess: &ParseSess, name: FileName, source: String)
-                                      -> Parser<'_> {
+pub fn new_parser_from_source_str(sess: &ParseSess, name: FileName, source: String) -> Parser<'_> {
     panictry_buffer!(&sess.span_diagnostic, maybe_new_parser_from_source_str(sess, name, source))
 }
 
@@ -195,12 +198,14 @@ fn source_file_to_parser(sess: &ParseSess, source_file: Lrc<SourceFile>) -> Pars
 
 /// Given a source_file and config, return a parser. Returns any buffered errors from lexing the
 /// initial token stream.
-fn maybe_source_file_to_parser(sess: &ParseSess, source_file: Lrc<SourceFile>)
-    -> Result<Parser<'_>, Vec<Diagnostic>>
-{
+fn maybe_source_file_to_parser(
+    sess: &ParseSess,
+    source_file: Lrc<SourceFile>,
+) -> Result<Parser<'_>, Vec<Diagnostic>> {
     let end_pos = source_file.end_pos;
-    let mut parser = stream_to_parser(sess, maybe_file_to_stream(sess, source_file, None)?);
-
+    let (stream, unclosed_delims) = maybe_file_to_stream(sess, source_file, None)?;
+    let mut parser = stream_to_parser(sess, stream);
+    parser.unclosed_delims = unclosed_delims;
     if parser.token == token::Eof && parser.span.is_dummy() {
         parser.span = Span::new(end_pos, end_pos, parser.span.ctxt());
     }
@@ -247,25 +252,44 @@ fn file_to_source_file(sess: &ParseSess, path: &Path, spanopt: Option<Span>)
 }
 
 /// Given a source_file, produce a sequence of token-trees
-pub fn source_file_to_stream(sess: &ParseSess,
-                             source_file: Lrc<SourceFile>,
-                             override_span: Option<Span>) -> TokenStream {
+pub fn source_file_to_stream(
+    sess: &ParseSess,
+    source_file: Lrc<SourceFile>,
+    override_span: Option<Span>,
+) -> (TokenStream, Vec<lexer::UnmatchedBrace>) {
     panictry_buffer!(&sess.span_diagnostic, maybe_file_to_stream(sess, source_file, override_span))
 }
 
 /// Given a source file, produce a sequence of token-trees. Returns any buffered errors from
 /// parsing the token tream.
-pub fn maybe_file_to_stream(sess: &ParseSess,
-                            source_file: Lrc<SourceFile>,
-                            override_span: Option<Span>) -> Result<TokenStream, Vec<Diagnostic>> {
+pub fn maybe_file_to_stream(
+    sess: &ParseSess,
+    source_file: Lrc<SourceFile>,
+    override_span: Option<Span>,
+) -> Result<(TokenStream, Vec<lexer::UnmatchedBrace>), Vec<Diagnostic>> {
     let mut srdr = lexer::StringReader::new_or_buffered_errs(sess, source_file, override_span)?;
     srdr.real_token();
 
     match srdr.parse_all_token_trees() {
-        Ok(stream) => Ok(stream),
+        Ok(stream) => Ok((stream, srdr.unmatched_braces)),
         Err(err) => {
             let mut buffer = Vec::with_capacity(1);
             err.buffer(&mut buffer);
+            // Not using `emit_unclosed_delims` to use `db.buffer`
+            for unmatched in srdr.unmatched_braces {
+                let mut db = sess.span_diagnostic.struct_span_err(unmatched.found_span, &format!(
+                    "incorrect close delimiter: `{}`",
+                    token_to_string(&token::Token::CloseDelim(unmatched.found_delim)),
+                ));
+                db.span_label(unmatched.found_span, "incorrect close delimiter");
+                if let Some(sp) = unmatched.candidate_span {
+                    db.span_label(sp, "close delimiter possibly meant for this");
+                }
+                if let Some(sp) = unmatched.unclosed_span {
+                    db.span_label(sp, "un-closed delimiter");
+                }
+                db.buffer(&mut buffer);
+            }
             Err(buffer)
         }
     }