Merge pull request #334 from Muscraft/fix-annotating-leading-whitespace

fix: Properly handle annotating leading whitespace

Author: epage

src/renderer/margin.rs

@@ -71,7 +71,12 @@ impl Margin {
 
         if self.computed_right - self.computed_left > self.term_width {
             // Trimming only whitespace isn't enough, let's get craftier.
-            if self.label_right - self.whitespace_left <= self.term_width {
+            if self.label_right.saturating_sub(self.whitespace_left) <= self.term_width
+                // Trimming whitespace when the right-most label is somewhrere
+                // within it would result in the label pointing to the wrong
+                // place
+                && self.label_right >= self.whitespace_left
+            {
                 // Attempt to fit the code window only trimming whitespace.
                 self.computed_left = self.whitespace_left;
                 self.computed_right = self.computed_left + self.term_width;

src/renderer/render.rs

@@ -622,11 +622,6 @@ fn render_snippet_annotations(
     // Get the left-side margin to remove it
     let mut whitespace_margin = usize::MAX;
     for line_info in annotated_lines {
-        // Whitespace can only be removed (aka considered leading)
-        // if the lexer considers it whitespace.
-        // non-rustc_lexer::is_whitespace() chars are reported as an
-        // error (ex. no-break-spaces \u{a0}), and thus can't be considered
-        // for removal during error reporting.
         let leading_whitespace = line_info
             .line
             .chars()

tests/rustc_tests.rs

@@ -5646,3 +5646,55 @@ LL │ static ROOK_ATTACKS_TABLE: () = {
     let renderer = renderer.decor_style(DecorStyle::Unicode);
     assert_data_eq!(renderer.render(input), expected_unicode);
 }
+
+#[test]
+fn emitter_overflow_bad_whitespace() {
+    // tests/ui/errors/emitter-overflow-bad-whitespace.rs
+    let source = r#"                                         fn main() {              return;              }
+"#;
+    let title_0 = "Unicode character ' ' (No-Break Space) looks like ' ' (Space), but it is not";
+
+    let report = &[
+        Group::with_title(Level::ERROR.primary_title("unknown start of token: \u{a0}")).element(
+            Snippet::source(source)
+                .path("$DIR/emitter-overflow-bad-whitespace.rs")
+                .line_start(10)
+                .annotation(AnnotationKind::Primary.span(0..2)),
+        ),
+        Group::with_title(Level::HELP.secondary_title(title_0)).element(
+            Snippet::source(source)
+                .path("$DIR/emitter-overflow-bad-whitespace.rs")
+                .line_start(10)
+                .patch(Patch::new(0..2, " ")),
+        ),
+    ];
+    let expected_ascii = str![[r#"
+error: unknown start of token:  
+  --> $DIR/emitter-overflow-bad-whitespace.rs:10:1
+   |
+10 |     ...
+   | ^
+   |
+help: Unicode character ' ' (No-Break Space) looks like ' ' (Space), but it is not
+   |
+10 |                                          fn main() {              return;              }
+   | +
+"#]];
+    let renderer_ascii = Renderer::plain().term_width(1);
+    assert_data_eq!(renderer_ascii.render(report), expected_ascii);
+
+    let expected_unicode = str![[r#"
+error: unknown start of token:  
+   ╭▸ $DIR/emitter-overflow-bad-whitespace.rs:10:1
+   │
+10 │       …
+   │ ━
+   ╰╴
+help: Unicode character ' ' (No-Break Space) looks like ' ' (Space), but it is not
+   ╭╴
+10 │                                          fn main() {              return;              }
+   ╰╴+
+"#]];
+    let renderer_unicode = renderer_ascii.decor_style(DecorStyle::Unicode);
+    assert_data_eq!(renderer_unicode.render(report), expected_unicode);
+}