Auto merge of #46653 - estebank:str-as-ch, r=petrochenkov

When attempting to write str with single quote suggest double quotes

Fix #26101.

Author: bors

src/Cargo.lock

@@ -12,3 +12,4 @@ crate-type = ["dylib"]
 serialize = { path = "../libserialize" }
 syntax_pos = { path = "../libsyntax_pos" }
 rustc_data_structures = { path = "../librustc_data_structures" }
+unicode-width = "0.1.4"

src/librustc_errors/Cargo.toml

@@ -23,6 +23,7 @@ use std::rc::Rc;
 use term;
 use std::collections::HashMap;
 use std::cmp::min;
+use unicode_width;
 
 /// Emitter trait for emitting errors.
 pub trait Emitter {
@@ -1182,7 +1183,10 @@ impl EmitterWriter {
                 if show_underline {
                     draw_col_separator(&mut buffer, row_num, max_line_num_len + 1);
                     let start = parts[0].snippet.len() - parts[0].snippet.trim_left().len();
-                    let sub_len = parts[0].snippet.trim().len();
+                    // account for substitutions containing unicode characters
+                    let sub_len = parts[0].snippet.trim().chars().fold(0, |acc, ch| {
+                        acc + unicode_width::UnicodeWidthChar::width(ch).unwrap_or(0)
+                    });
                     let underline_start = span_start_pos.col.0 + start;
                     let underline_end = span_start_pos.col.0 + start + sub_len;
                     for p in underline_start..underline_end {

src/librustc_errors/emitter.rs

@@ -26,6 +26,7 @@ extern crate libc;
 extern crate rustc_data_structures;
 extern crate serialize as rustc_serialize;
 extern crate syntax_pos;
+extern crate unicode_width;
 
 pub use emitter::ColorConfig;
 

src/librustc_errors/lib.rs

@@ -1306,8 +1306,34 @@ impl<'a> StringReader<'a> {
                                                    '\'');
 
                 if !self.ch_is('\'') {
+                    let pos = self.pos;
+                    loop {
+                        self.bump();
+                        if self.ch_is('\'') {
+                            let start = self.byte_offset(start).to_usize();
+                            let end = self.byte_offset(self.pos).to_usize();
+                            self.bump();
+                            let span = self.mk_sp(start_with_quote, self.pos);
+                            self.sess.span_diagnostic
+                                .struct_span_err(span,
+                                                 "character literal may only contain one codepoint")
+                                .span_suggestion(span,
+                                                 "if you meant to write a `str` literal, \
+                                                  use double quotes",
+                                                 format!("\"{}\"",
+                                                         &self.source_text[start..end]))
+                                .emit();
+                            return Ok(token::Literal(token::Str_(Symbol::intern("??")), None))
+                        }
+                        if self.ch_is('\n') || self.is_eof() || self.ch_is('/') {
+                            // Only attempt to infer single line string literals. If we encounter
+                            // a slash, bail out in order to avoid nonsensical suggestion when
+                            // involving comments.
+                            break;
+                        }
+                    }
                     panic!(self.fatal_span_verbose(
-                           start_with_quote, self.pos,
+                           start_with_quote, pos,
                            String::from("character literal may only contain one codepoint")));
                 }
 

src/libsyntax/parse/lexer/mod.rs

@@ -12,5 +12,5 @@
 
 // This test needs to the last one appearing in this file as it kills the parser
 static c: char =
-    '●●' //~ ERROR: character literal may only contain one codepoint: '●
+    '●●' //~ ERROR: character literal may only contain one codepoint
 ;

src/test/parse-fail/lex-bad-char-literals-3.rs

@@ -12,5 +12,5 @@
 //
 // This test needs to the last one appearing in this file as it kills the parser
 static c: char =
-    '\x10\x10'  //~ ERROR: character literal may only contain one codepoint: '\x10
+    '\x10\x10'  //~ ERROR: character literal may only contain one codepoint
 ;

src/test/parse-fail/lex-bad-char-literals-5.rs

@@ -0,0 +1,14 @@
+// Copyright 2017 The Rust Project Developers. See the COPYRIGHT
+// file at the top-level directory of this distribution and at
+// http://rust-lang.org/COPYRIGHT.
+//
+// Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or
+// http://www.apache.org/licenses/LICENSE-2.0> or the MIT license
+// <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your
+// option. This file may not be copied, modified, or distributed
+// except according to those terms.
+
+fn main() {
+    println!('●●');
+    //~^ ERROR character literal may only contain one codepoint
+}

src/test/ui/suggestions/str-as-char.rs

@@ -0,0 +1,12 @@
+error: character literal may only contain one codepoint
+  --> $DIR/str-as-char.rs:12:14
+   |
+12 |     println!('●●');
+   |              ^^^^
+help: if you meant to write a `str` literal, use double quotes
+   |
+12 |     println!("●●");
+   |              ^^^^
+
+error: aborting due to previous error
+