Account for unicode confusables on typoed identifiers

Cargo.lock

@@ -3871,6 +3871,7 @@ dependencies = [
  "syntax",
  "syntax_expand",
  "syntax_pos",
+ "unicode_skeleton",
 ]
 
 [[package]]
@@ -4471,6 +4472,7 @@ dependencies = [
  "serialize",
  "smallvec 1.0.0",
  "syntax_pos",
+ "unicode_skeleton",
 ]
 
 [[package]]
@@ -5045,6 +5047,15 @@ version = "0.1.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "39ec24b3121d976906ece63c9daad25b85969647682eee313cb5779fdd69e14e"
 
+[[package]]
+name = "unicode_skeleton"
+version = "0.1.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "66bd74db2c088d393d1fbf83db2cd5663137640f072d128287dd53c882a0f412"
+dependencies = [
+ "unicode-normalization",
+]
+
 [[package]]
 name = "unstable-book-gen"
 version = "0.1.0"

src/librustc_resolve/Cargo.toml

@@ -23,3 +23,4 @@ rustc_data_structures = { path = "../librustc_data_structures" }
 rustc_metadata = { path = "../librustc_metadata" }
 rustc_error_codes = { path = "../librustc_error_codes" }
 smallvec = { version = "1.0", features = ["union", "may_dangle"] }
+unicode_skeleton = "0.1.1"

src/librustc_resolve/diagnostics.rs

@@ -488,9 +488,11 @@ impl<'a> Resolver<'a> {
             &ident.as_str(),
             None,
         ) {
-            Some(found) if found != ident.name => suggestions
-                .into_iter()
-                .find(|suggestion| suggestion.candidate == found),
+            Some(found) if found != ident.name => suggestions.into_iter()
+                .find(|suggestion| unicode_skeleton::confusable(
+                    suggestion.candidate.as_str().chars(),
+                    found.as_str().chars(),
+                )),
             _ => None,
         }
     }

src/libsyntax/Cargo.toml

@@ -23,3 +23,4 @@ rustc_lexer = { path = "../librustc_lexer" }
 rustc_macros = { path = "../librustc_macros" }
 smallvec = { version = "1.0", features = ["union", "may_dangle"] }
 rustc_error_codes = { path = "../librustc_error_codes" }
+unicode_skeleton = "0.1.1"

src/libsyntax/util/lev_distance.rs

@@ -49,34 +49,45 @@ pub fn find_best_match_for_name<'a, T>(iter_names: T,
     where T: Iterator<Item = &'a Symbol> {
     let max_dist = dist.map_or_else(|| cmp::max(lookup.len(), 3) / 3, |d| d);
 
-    let (case_insensitive_match, levenstein_match) = iter_names
-    .filter_map(|&name| {
-        let dist = lev_distance(lookup, &name.as_str());
-        if dist <= max_dist {
-            Some((name, dist))
-        } else {
-            None
-        }
-    })
-    // Here we are collecting the next structure:
-    // (case_insensitive_match, (levenstein_match, levenstein_distance))
-    .fold((None, None), |result, (candidate, dist)| {
-        (
-            if candidate.as_str().to_uppercase() == lookup.to_uppercase() {
-                Some(candidate)
+    let (case_insensitive_match, levenstein_match, is_confusable) = iter_names
+        .filter_map(|&name| {
+            let dist = lev_distance(lookup, &name.as_str());
+            let is_confusable = unicode_skeleton::confusable(lookup, name.as_str().chars());
+            if dist <= max_dist || is_confusable {
+                Some((name, dist, is_confusable))
             } else {
-                result.0
-            },
-            match result.1 {
-                None => Some((candidate, dist)),
-                Some((c, d)) => Some(if dist < d { (candidate, dist) } else { (c, d) })
+                None
             }
-        )
-    });
+        })
+        // Here we are collecting the next structure:
+        // (case_insensitive_match, (levenstein_match, levenstein_distance))
+        .fold((None, None, None), |result, (candidate, dist, is_confusable)| {
+            (
+                if candidate.as_str().to_uppercase() == lookup.to_uppercase() {
+                    Some(candidate)
+                } else {
+                    result.0
+                },
+                match result.1 {
+                    None => Some((candidate, dist)),
+                    Some((c, d)) => Some(if dist < d { (candidate, dist) } else { (c, d) }),
+                },
+                match (is_confusable, result.2) {
+                    (false, Some((c, d))) => Some((c, d)),
+                    (false, None) => None,
+                    (true, None) => Some((candidate, dist)),
+                    (true, Some((c, d))) => Some(if dist < d { (candidate, dist) } else { (c, d) }),
+                }
+            )
+        });
 
     if let Some(candidate) = case_insensitive_match {
         Some(candidate) // exact case insensitive match has a higher priority
+    } else if let Some((candidate, _)) = is_confusable {
+        Some(candidate)
+    } else if let Some((candidate, _)) = levenstein_match {
+        Some(candidate)
     } else {
-        if let Some((candidate, _)) = levenstein_match { Some(candidate) } else { None }
+        None
     }
 }

src/test/ui/suggestions/unicode-confusable-typo.fixed

@@ -0,0 +1,11 @@
+// run-rustfix
+#![feature(non_ascii_idents)]
+
+struct ℝ𝓊𝓈𝓉;
+
+fn main() {
+    let ü = ℝ𝓊𝓈𝓉;
+    //~^ ERROR cannot find value `Rust` in this scope
+    let _ = ü;
+    //~^ ERROR cannot find value `u` in this scope
+}

src/test/ui/suggestions/unicode-confusable-typo.rs

@@ -0,0 +1,11 @@
+// run-rustfix
+#![feature(non_ascii_idents)]
+
+struct ℝ𝓊𝓈𝓉;
+
+fn main() {
+    let ü = Rust;
+    //~^ ERROR cannot find value `Rust` in this scope
+    let _ = u;
+    //~^ ERROR cannot find value `u` in this scope
+}

src/test/ui/suggestions/unicode-confusable-typo.stderr

@@ -0,0 +1,18 @@
+error[E0425]: cannot find value `Rust` in this scope
+  --> $DIR/unicode-confusable-typo.rs:7:13
+   |
+LL | struct ℝ𝓊𝓈𝓉;
+   | ------------ similarly named unit struct `ℝ𝓊𝓈𝓉` defined here
+...
+LL |     let ü = Rust;
+   |             ^^^^ help: a unit struct with a similar name exists: `ℝ𝓊𝓈𝓉`
+
+error[E0425]: cannot find value `u` in this scope
+  --> $DIR/unicode-confusable-typo.rs:9:13
+   |
+LL |     let _ = u;
+   |             ^ help: a local variable with a similar name exists: `ü`
+
+error: aborting due to 2 previous errors
+
+For more information about this error, try `rustc --explain E0425`.