Improve handling of unicode characters (#3618)

Author: scampi

src/chains.rs

@@ -557,7 +557,10 @@ impl<'a> ChainFormatterShared<'a> {
         let almost_total = if extendable {
             prev_last_line_width
         } else {
-            self.rewrites.iter().map(String::len).sum()
+            self.rewrites
+                .iter()
+                .map(|rw| utils::unicode_str_width(&rw))
+                .sum()
         } + last.tries;
         let one_line_budget = if self.child_count == 1 {
             shape.width

src/expr.rs

@@ -33,7 +33,7 @@ use crate::types::{rewrite_path, PathContext};
 use crate::utils::{
     colon_spaces, contains_skip, count_newlines, first_line_ends_with, inner_attributes,
     last_line_extendable, last_line_width, mk_sp, outer_attributes, ptr_vec_to_ref_vec,
-    semicolon_for_expr, semicolon_for_stmt, wrap_str,
+    semicolon_for_expr, semicolon_for_stmt, unicode_str_width, wrap_str,
 };
 use crate::vertical::rewrite_with_alignment;
 use crate::visitor::FmtVisitor;
@@ -1973,7 +1973,9 @@ fn choose_rhs<R: Rewrite>(
     rhs_tactics: RhsTactics,
 ) -> Option<String> {
     match orig_rhs {
-        Some(ref new_str) if !new_str.contains('\n') && new_str.len() <= shape.width => {
+        Some(ref new_str)
+            if !new_str.contains('\n') && unicode_str_width(new_str) <= shape.width =>
+        {
             Some(format!(" {}", new_str))
         }
         _ => {

tests/source/unicode.rs

@@ -0,0 +1,25 @@
+// rustfmt-wrap_comments: true
+
+fn foo() {
+    let s = "this line goes to 100: ͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶ";
+    let s     =     42;
+
+    // a comment of length 80, with the starting sigil: ҘҘҘҘҘҘҘҘҘҘ ҘҘҘҘҘҘҘҘҘҘҘҘҘҘ
+    let s     =     42;
+}
+
+pub fn bar(config: &Config) {
+    let csv = RefCell::new(create_csv(config, "foo"));
+    {
+        let mut csv = csv.borrow_mut();
+        for (i1, i2, i3) in iproduct!(0..2, 0..3, 0..3) {
+            csv.write_field(format!("γ[{}.{}.{}]", i1, i2, i3))
+                .unwrap();
+            csv.write_field(format!("d[{}.{}.{}]", i1, i2, i3))
+                .unwrap();
+            csv.write_field(format!("i[{}.{}.{}]", i1, i2, i3))
+                .unwrap();
+        }
+        csv.write_record(None::<&[u8]>).unwrap();
+    }
+}

tests/target/unicode.rs

@@ -0,0 +1,22 @@
+// rustfmt-wrap_comments: true
+
+fn foo() {
+    let s = "this line goes to 100: ͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶͶ";
+    let s = 42;
+
+    // a comment of length 80, with the starting sigil: ҘҘҘҘҘҘҘҘҘҘ ҘҘҘҘҘҘҘҘҘҘҘҘҘҘ
+    let s = 42;
+}
+
+pub fn bar(config: &Config) {
+    let csv = RefCell::new(create_csv(config, "foo"));
+    {
+        let mut csv = csv.borrow_mut();
+        for (i1, i2, i3) in iproduct!(0..2, 0..3, 0..3) {
+            csv.write_field(format!("γ[{}.{}.{}]", i1, i2, i3)).unwrap();
+            csv.write_field(format!("d[{}.{}.{}]", i1, i2, i3)).unwrap();
+            csv.write_field(format!("i[{}.{}.{}]", i1, i2, i3)).unwrap();
+        }
+        csv.write_record(None::<&[u8]>).unwrap();
+    }
+}