fix(cpp) improve number literals in C++ (#3847)

highlightjs · Nov 8, 2023 · fdd24e9 · fdd24e9
1 parent 5bb7114
commit fdd24e9
Show file tree

Hide file tree

Showing 4 changed files with 148 additions and 23 deletions.
diff --git a/CHANGES.md b/CHANGES.md
@@ -2,6 +2,7 @@
 
 Core Grammars:
 
+- fix(cpp) not all kinds of number literals are highlighted correctly [Lê Duy Quang][]
 - fix(css) fix overly greedy pseudo class matching [Bradley Mackey][]
 - enh(arcade) updated to ArcGIS Arcade version 1.24 [Kristian Ekenes][]
 - fix(rust) fix escaped double quotes in string  [Mohamed Ali][]
@@ -17,6 +18,7 @@ Developer Tool:
 
 - enh(tools): order CSS options picklist [David Schach][]
 
+[Lê Duy Quang]: https://github.com/leduyquang753
 [Mohamed Ali]: https://github.com/MohamedAli00949
 [Bradley Mackey]: https://github.com/bradleymackey
 [Kristian Ekenes]: https://github.com/ekenes

diff --git a/src/languages/cpp.js b/src/languages/cpp.js
@@ -52,9 +52,44 @@ export default function(hljs) {
   const NUMBERS = {
     className: 'number',
     variants: [
-      { begin: '\\b(0b[01\']+)' },
-      { begin: '(-?)\\b([\\d\']+(\\.[\\d\']*)?|\\.[\\d\']+)((ll|LL|l|L)(u|U)?|(u|U)(ll|LL|l|L)?|f|F|b|B)' },
-      { begin: '(-?)(\\b0[xX][a-fA-F0-9\']+|(\\b[\\d\']+(\\.[\\d\']*)?|\\.[\\d\']+)([eE][-+]?[\\d\']+)?)' }
+      // Floating-point literal.
+      { begin:
+        "[+-]?(?:" // Leading sign.
+          // Decimal.
+          + "(?:"
+            +"[0-9](?:'?[0-9])*\\.(?:[0-9](?:'?[0-9])*)?"
+            + "|\\.[0-9](?:'?[0-9])*"
+          + ")(?:[Ee][+-]?[0-9](?:'?[0-9])*)?"
+          + "|[0-9](?:'?[0-9])*[Ee][+-]?[0-9](?:'?[0-9])*"
+          // Hexadecimal.
+          + "|0[Xx](?:"
+            +"[0-9A-Fa-f](?:'?[0-9A-Fa-f])*(?:\\.(?:[0-9A-Fa-f](?:'?[0-9A-Fa-f])*)?)?"
+            + "|\\.[0-9A-Fa-f](?:'?[0-9A-Fa-f])*"
+          + ")[Pp][+-]?[0-9](?:'?[0-9])*"
+        + ")(?:" // Literal suffixes.
+          + "[Ff](?:16|32|64|128)?"
+          + "|(BF|bf)16"
+          + "|[Ll]"
+          + "|" // Literal suffix is optional.
+        + ")"
+      },
+      // Integer literal.
+      { begin:
+        "[+-]?\\b(?:" // Leading sign.
+          + "0[Bb][01](?:'?[01])*" // Binary.
+          + "|0[Xx][0-9A-Fa-f](?:'?[0-9A-Fa-f])*" // Hexadecimal.
+          + "|0(?:'?[0-7])*" // Octal or just a lone zero.
+          + "|[1-9](?:'?[0-9])*" // Decimal.
+        + ")(?:" // Literal suffixes.
+          + "[Uu](?:LL?|ll?)"
+          + "|[Uu][Zz]?"
+          + "|(?:LL?|ll?)[Uu]?"
+          + "|[Zz][Uu]"
+          + "|" // Literal suffix is optional.
+        + ")"
+        // Note: there are user-defined literal suffixes too, but perhaps having the custom suffix not part of the
+        // literal highlight actually makes it stand out more.
+      }
     ],
     relevance: 0
   };

diff --git a/test/markup/cpp/number-literals.expect.txt b/test/markup/cpp/number-literals.expect.txt
@@ -1,11 +1,55 @@
-<span class="hljs-comment">/* digit separators */</span>
-<span class="hljs-type">int</span> number = <span class="hljs-number">2&#x27;555&#x27;555&#x27;555</span>; <span class="hljs-comment">// digit separators</span>
-<span class="hljs-type">float</span> exponentFloat = <span class="hljs-number">.123&#x27;456e3&#x27;000</span>; <span class="hljs-comment">// digit separators in floats</span>
-<span class="hljs-type">float</span> suffixed = <span class="hljs-number">3.000&#x27;001&#x27;234f</span> <span class="hljs-comment">// digit separators in suffixed numbers</span>
-<span class="hljs-type">char</span> word[] = { <span class="hljs-string">&#x27;3&#x27;</span>, <span class="hljs-string">&#x27;\0&#x27;</span> }; <span class="hljs-comment">// make sure digit separators don&#x27;t mess up chars</span>
-<span class="hljs-type">float</span> negative = <span class="hljs-number">-123.0f</span>; <span class="hljs-comment">// negative floating point numbers</span>
+<span class="hljs-comment">/* Floating-point literals. */</span>
+<span class="hljs-comment">// Decimal.</span>
+<span class="hljs-number">1.</span>
+<span class="hljs-number">+12.</span>
+<span class="hljs-number">-1&#x27;234.</span>
+<span class="hljs-number">1.2</span>
+<span class="hljs-number">1.234&#x27;5</span>
+<span class="hljs-number">1&#x27;234.567&#x27;8</span>
+<span class="hljs-number">1&#x27;234.567&#x27;8e-9</span>
+<span class="hljs-number">1&#x27;234.567&#x27;8E9&#x27;0</span>
+<span class="hljs-number">1234e56</span>
+<span class="hljs-number">1&#x27;234e5&#x27;6</span>
+<span class="hljs-comment">// Hexadecimal.</span>
+<span class="hljs-number">0x1p2</span>
+<span class="hljs-number">+0x1.p2</span>
+<span class="hljs-number">-0X1A.P2</span>
+<span class="hljs-number">0x1A&#x27;2B3C.p4</span>
+<span class="hljs-number">0x1.Ap2</span>
+<span class="hljs-number">0x1A.2B3C&#x27;4Dp+5</span>
+<span class="hljs-number">0x1A&#x27;2b3C.4D5e&#x27;6Fp7&#x27;8</span>
+<span class="hljs-comment">// Literal suffixes.</span>
+<span class="hljs-number">1.F</span> <span class="hljs-number">1.F16</span> <span class="hljs-number">1.F32</span> <span class="hljs-number">1.f64</span> <span class="hljs-number">1.f128</span>
+<span class="hljs-number">2.BF16</span> <span class="hljs-number">2.bf16</span>
+<span class="hljs-number">0X3&#x27;4.P5L</span> <span class="hljs-number">0x3&#x27;4.p5l</span>
 
-<span class="hljs-comment">/* literal suffixes: these are to be highlighted as numbers */</span>
-<span class="hljs-number">90ll</span> <span class="hljs-number">90LL</span> <span class="hljs-number">90u</span> <span class="hljs-number">90U</span> <span class="hljs-number">90l</span> <span class="hljs-number">90L</span>
-<span class="hljs-number">90ul</span> <span class="hljs-number">90ul</span> <span class="hljs-number">90Ul</span> <span class="hljs-number">90UL</span> <span class="hljs-number">90lu</span> <span class="hljs-number">90lU</span> <span class="hljs-number">90Lu</span> <span class="hljs-number">90LU</span>
-<span class="hljs-number">90ull</span> <span class="hljs-number">90uLL</span> <span class="hljs-number">90Ull</span> <span class="hljs-number">90ULL</span> <span class="hljs-number">90llu</span> <span class="hljs-number">90llU</span> <span class="hljs-number">90LLu</span> <span class="hljs-number">90LLU</span>
+<span class="hljs-comment">/* Integer literals. */</span>
+<span class="hljs-comment">// Binary.</span>
+<span class="hljs-number">+0b1</span>
+<span class="hljs-number">0B01</span>
+<span class="hljs-number">-0b10&#x27;1011</span>
+<span class="hljs-comment">// Hexadecimal.</span>
+<span class="hljs-number">+0x1</span>
+<span class="hljs-number">0X1A</span>
+<span class="hljs-number">-0x1A&#x27;2b3C</span>
+<span class="hljs-comment">// Octal.</span>
+<span class="hljs-number">+01</span>
+<span class="hljs-number">012</span>
+<span class="hljs-number">-0123&#x27;4567</span>
+<span class="hljs-comment">// Decimal.</span>
+<span class="hljs-number">0</span>
+<span class="hljs-number">+1</span>
+<span class="hljs-number">12</span>
+<span class="hljs-number">-123&#x27;456&#x27;789</span>
+<span class="hljs-comment">// Literal suffixes.</span>
+<span class="hljs-number">0B1U</span> <span class="hljs-number">0b1u</span>
+<span class="hljs-number">0X2L</span> <span class="hljs-number">0x2l</span>
+<span class="hljs-number">03LL</span> <span class="hljs-number">03ll</span>
+<span class="hljs-number">4UL</span> <span class="hljs-number">4Ul</span> <span class="hljs-number">4uL</span> <span class="hljs-number">4ul</span>
+<span class="hljs-number">5LU</span> <span class="hljs-number">5Lu</span> <span class="hljs-number">5lU</span> <span class="hljs-number">5lu</span>
+<span class="hljs-number">6ULL</span> <span class="hljs-number">6Ull</span> <span class="hljs-number">6uLL</span> <span class="hljs-number">6ull</span>
+<span class="hljs-number">7LLU</span> <span class="hljs-number">7LLu</span> <span class="hljs-number">7llU</span> <span class="hljs-number">7llu</span>
+<span class="hljs-number">8ZU</span> <span class="hljs-number">8Zu</span> <span class="hljs-number">8zU</span> <span class="hljs-number">8zu</span>
+<span class="hljs-number">9&#x27;0UZ</span> <span class="hljs-number">9&#x27;0Uz</span> <span class="hljs-number">9&#x27;0uZ</span> <span class="hljs-number">9&#x27;0uz</span>
+
+<span class="hljs-type">char</span> word[] = { <span class="hljs-string">&#x27;3&#x27;</span>, <span class="hljs-string">&#x27;\0&#x27;</span> }; <span class="hljs-comment">// Make sure digit separators don&#x27;t mess up chars.</span>
diff --git a/test/markup/cpp/number-literals.txt b/test/markup/cpp/number-literals.txt
@@ -1,11 +1,55 @@
-/* digit separators */
-int number = 2'555'555'555; // digit separators
-float exponentFloat = .123'456e3'000; // digit separators in floats
-float suffixed = 3.000'001'234f // digit separators in suffixed numbers
-char word[] = { '3', '\0' }; // make sure digit separators don't mess up chars
-float negative = -123.0f; // negative floating point numbers
+/* Floating-point literals. */
+// Decimal.
+1.
++12.
+-1'234.
+1.2
+1.234'5
+1'234.567'8
+1'234.567'8e-9
+1'234.567'8E9'0
+1234e56
+1'234e5'6
+// Hexadecimal.
+0x1p2
++0x1.p2
+-0X1A.P2
+0x1A'2B3C.p4
+0x1.Ap2
+0x1A.2B3C'4Dp+5
+0x1A'2b3C.4D5e'6Fp7'8
+// Literal suffixes.
+1.F 1.F16 1.F32 1.f64 1.f128
+2.BF16 2.bf16
+0X3'4.P5L 0x3'4.p5l
 
-/* literal suffixes: these are to be highlighted as numbers */
-90ll 90LL 90u 90U 90l 90L
-90ul 90ul 90Ul 90UL 90lu 90lU 90Lu 90LU
-90ull 90uLL 90Ull 90ULL 90llu 90llU 90LLu 90LLU
+/* Integer literals. */
+// Binary.
++0b1
+0B01
+-0b10'1011
+// Hexadecimal.
++0x1
+0X1A
+-0x1A'2b3C
+// Octal.
++01
+012
+-0123'4567
+// Decimal.
+0
++1
+12
+-123'456'789
+// Literal suffixes.
+0B1U 0b1u
+0X2L 0x2l
+03LL 03ll
+4UL 4Ul 4uL 4ul
+5LU 5Lu 5lU 5lu
+6ULL 6Ull 6uLL 6ull
+7LLU 7LLu 7llU 7llu
+8ZU 8Zu 8zU 8zu
+9'0UZ 9'0Uz 9'0uZ 9'0uz
+
+char word[] = { '3', '\0' }; // Make sure digit separators don't mess up chars.