Auto merge of rust-lang#128200 - estebank:normalize-whitespace, r=<try>

[perf] Change output normalization logic to be linear against size of output

I believe the previous code was accidentally quadratic. Let's perf it.

Author: bors

compiler/rustc_errors/src/emitter.rs

@@ -2557,62 +2557,60 @@ fn num_decimal_digits(num: usize) -> usize {
     MAX_DIGITS
 }
 
-// We replace some characters so the CLI output is always consistent and underlines aligned.
-// Keep the following list in sync with `rustc_span::char_width`.
-const OUTPUT_REPLACEMENTS: &[(char, &str)] = &[
-    ('\t', "    "),    // We do our own tab replacement
-    ('\u{200D}', ""), // Replace ZWJ with nothing for consistent terminal output of grapheme clusters.
-    ('\u{202A}', "�"), // The following unicode text flow control characters are inconsistently
-    ('\u{202B}', "�"), // supported across CLIs and can cause confusion due to the bytes on disk
-    ('\u{202D}', "�"), // not corresponding to the visible source code, so we replace them always.
-    ('\u{202E}', "�"),
-    ('\u{2066}', "�"),
-    ('\u{2067}', "�"),
-    ('\u{2068}', "�"),
-    ('\u{202C}', "�"),
-    ('\u{2069}', "�"),
-    // In terminals without Unicode support the following will be garbled, but in *all* terminals
-    // the underlying codepoint will be as well. We could gate this replacement behind a "unicode
-    // support" gate.
-    ('\u{0000}', "␀"),
-    ('\u{0001}', "␁"),
-    ('\u{0002}', "␂"),
-    ('\u{0003}', "␃"),
-    ('\u{0004}', "␄"),
-    ('\u{0005}', "␅"),
-    ('\u{0006}', "␆"),
-    ('\u{0007}', "␇"),
-    ('\u{0008}', "␈"),
-    ('\u{000B}', "␋"),
-    ('\u{000C}', "␌"),
-    ('\u{000D}', "␍"),
-    ('\u{000E}', "␎"),
-    ('\u{000F}', "␏"),
-    ('\u{0010}', "␐"),
-    ('\u{0011}', "␑"),
-    ('\u{0012}', "␒"),
-    ('\u{0013}', "␓"),
-    ('\u{0014}', "␔"),
-    ('\u{0015}', "␕"),
-    ('\u{0016}', "␖"),
-    ('\u{0017}', "␗"),
-    ('\u{0018}', "␘"),
-    ('\u{0019}', "␙"),
-    ('\u{001A}', "␚"),
-    ('\u{001B}', "␛"),
-    ('\u{001C}', "␜"),
-    ('\u{001D}', "␝"),
-    ('\u{001E}', "␞"),
-    ('\u{001F}', "␟"),
-    ('\u{007F}', "␡"),
-];
-
 fn normalize_whitespace(str: &str) -> String {
-    let mut s = str.to_string();
-    for (c, replacement) in OUTPUT_REPLACEMENTS {
-        s = s.replace(*c, replacement);
-    }
-    s
+    // We replace some characters so the CLI output is always consistent and underlines aligned.
+    // Keep the following list in sync with `rustc_span::char_width`.
+    let output_replacements = FxHashMap::from_iter([
+        ('\t', "    "),    // We do our own tab replacement
+        ('\u{200D}', ""),  // Replace ZWJ for consistent terminal output of grapheme clusters.
+        ('\u{202A}', "�"), // The following unicode text flow control characters are inconsistently
+        ('\u{202B}', "�"), // supported across CLIs and can cause confusion due to the bytes on disk
+        ('\u{202D}', "�"), // not corresponding to the visible source code, so we replace them always.
+        ('\u{202E}', "�"),
+        ('\u{2066}', "�"),
+        ('\u{2067}', "�"),
+        ('\u{2068}', "�"),
+        ('\u{202C}', "�"),
+        ('\u{2069}', "�"),
+        // In terminals without Unicode support the following will be garbled, but in *all* terminals
+        // the underlying codepoint will be as well. We could gate this replacement behind a "unicode
+        // support" gate.
+        ('\u{0000}', "␀"),
+        ('\u{0001}', "␁"),
+        ('\u{0002}', "␂"),
+        ('\u{0003}', "␃"),
+        ('\u{0004}', "␄"),
+        ('\u{0005}', "␅"),
+        ('\u{0006}', "␆"),
+        ('\u{0007}', "␇"),
+        ('\u{0008}', "␈"),
+        ('\u{000B}', "␋"),
+        ('\u{000C}', "␌"),
+        ('\u{000D}', "␍"),
+        ('\u{000E}', "␎"),
+        ('\u{000F}', "␏"),
+        ('\u{0010}', "␐"),
+        ('\u{0011}', "␑"),
+        ('\u{0012}', "␒"),
+        ('\u{0013}', "␓"),
+        ('\u{0014}', "␔"),
+        ('\u{0015}', "␕"),
+        ('\u{0016}', "␖"),
+        ('\u{0017}', "␗"),
+        ('\u{0018}', "␘"),
+        ('\u{0019}', "␙"),
+        ('\u{001A}', "␚"),
+        ('\u{001B}', "␛"),
+        ('\u{001C}', "␜"),
+        ('\u{001D}', "␝"),
+        ('\u{001E}', "␞"),
+        ('\u{001F}', "␟"),
+        ('\u{007F}', "␡"),
+    ]);
+
+    str.chars()
+        .map(|c| output_replacements.get(&c).map_or(c.to_string(), |s| s.to_string()))
+        .collect()
 }
 
 fn draw_col_separator(buffer: &mut StyledBuffer, line: usize, col: usize) {