From d5aa057b348c3698db3457314a98bf6aafb723f5 Mon Sep 17 00:00:00 2001
From: flofriday <flohacksfriday@gmail.com>
Date: Mon, 25 Aug 2025 19:01:05 +0200
Subject: [PATCH] fix(kotlin) number literals

Previously kotlin resused the java number parsing but there are
significant differnces in how the two languages express numbers
causing false-negatives and false-positives.
---
 CHANGES.md                            |  1 +
 src/languages/kotlin.js               | 31 ++++++++++++++----
 test/markup/kotlin/numbers.expect.txt | 45 +++++++++++++++++++++++++++
 test/markup/kotlin/numbers.txt        | 45 +++++++++++++++++++++++++++
 4 files changed, 116 insertions(+), 6 deletions(-)
 create mode 100644 test/markup/kotlin/numbers.expect.txt
 create mode 100644 test/markup/kotlin/numbers.txt

diff --git a/CHANGES.md b/CHANGES.md
index b1413c29a5..057f74d9a2 100644
--- a/CHANGES.md
+++ b/CHANGES.md
@@ -26,6 +26,7 @@ Core Grammars:
 - enh(json) add json5 support [Kerry Shetline][]
 - fix(css) `unicode-range` parsing, issue #4253 [Kerry Shetline][]
 - fix(csharp) Support digit separators [te-ing][]
+- fix(kotlin) Improve number literal parsing[Florian Freitag][]
 
 Documentation:
 
diff --git a/src/languages/kotlin.js b/src/languages/kotlin.js
index c31f5a9d49..c63be9b7d5 100644
--- a/src/languages/kotlin.js
+++ b/src/languages/kotlin.js
@@ -6,8 +6,6 @@
  Category: common
  */
 
-import { NUMERIC } from "./lib/java.js";
-
 export default function(hljs) {
   const KEYWORDS = {
     keyword:
@@ -100,10 +98,31 @@ export default function(hljs) {
     ]
   };
 
-  // https://kotlinlang.org/docs/reference/whatsnew11.html#underscores-in-numeric-literals
-  // According to the doc above, the number mode of kotlin is the same as java 8,
-  // so the code below is copied from java.js
-  const KOTLIN_NUMBER_MODE = NUMERIC;
+  const decimalDigits = '[0-9](_*[0-9])*';
+  const frac = `\\.(${decimalDigits})`;
+  const KOTLIN_NUMBER_MODE = {
+    className: 'number',
+    variants: [
+      // DecimalFloatingPointLiteral
+      // including ExponentPart
+      { begin: `(\\b(${decimalDigits})((${frac})|\\.)?|(${frac}))` +
+        `[eE][+-]?(${decimalDigits})[fF]?\\b` },
+      // excluding ExponentPart
+      { begin: `\\b(${decimalDigits})((${frac})[fF]?\\b|\\.([fF]\\b)?)` },
+      { begin: `(${frac})[fF]?\\b` },
+      { begin: `\\b(${decimalDigits})[fF]\\b` },
+  
+      // DecimalIntegerLiteral
+      { begin: '\\b(0|[1-9](_*[0-9])*)[uU]?L?\\b' },
+  
+      // HexIntegerLiteral
+      { begin: `\\b0[xX]([0-9a-fA-F](_*[0-9a-fA-F])*)[uU]?L?\\b` },
+  
+      // BinaryIntegerLiteral
+      { begin: '\\b0[bB][01](_*[01])*[uU]?L?\\b' },
+    ],
+    relevance: 0
+  };
   const KOTLIN_NESTED_COMMENT = hljs.COMMENT(
     '/\\*', '\\*/',
     { contains: [ hljs.C_BLOCK_COMMENT_MODE ] }
diff --git a/test/markup/kotlin/numbers.expect.txt b/test/markup/kotlin/numbers.expect.txt
new file mode 100644
index 0000000000..6036cc062b
--- /dev/null
+++ b/test/markup/kotlin/numbers.expect.txt
@@ -0,0 +1,45 @@
+<span class="hljs-comment">// Postive cases</span>
+<span class="hljs-number">1</span>
+<span class="hljs-number">123</span>
+<span class="hljs-number">123L</span>
+<span class="hljs-number">123f</span>
+<span class="hljs-number">123F</span>
+<span class="hljs-number">123u</span>
+<span class="hljs-number">123U</span>
+<span class="hljs-number">123UL</span>
+<span class="hljs-number">123uL</span>
+<span class="hljs-number">1.23</span>
+<span class="hljs-number">1.23f</span>
+<span class="hljs-number">1.23F</span>
+<span class="hljs-number">.123</span>
+<span class="hljs-number">.123f</span>
+<span class="hljs-number">.123F</span>
+<span class="hljs-number">0b10101</span>
+<span class="hljs-number">0B10101</span>
+<span class="hljs-number">0b10101u</span>
+<span class="hljs-number">0B10101u</span>
+<span class="hljs-number">0b10101U</span>
+<span class="hljs-number">0B10101U</span>
+<span class="hljs-number">0b10101L</span>
+<span class="hljs-number">0B10101L</span>
+<span class="hljs-number">0b10101uL</span>
+<span class="hljs-number">0B10101uL</span>
+<span class="hljs-number">0b10101UL</span>
+<span class="hljs-number">0B10101UL</span>
+<span class="hljs-number">0xff8842</span>
+<span class="hljs-number">0Xff8842</span>
+<span class="hljs-number">0xff8842u</span>
+<span class="hljs-number">0Xff8842u</span>
+<span class="hljs-number">0xff8842U</span>
+<span class="hljs-number">0Xff8842U</span>
+<span class="hljs-number">0xff8842L</span>
+<span class="hljs-number">0Xff8842L</span>
+<span class="hljs-number">0xff8842uL</span>
+<span class="hljs-number">0Xff8842uL</span>
+<span class="hljs-number">0xff8842UL</span>
+<span class="hljs-number">0Xff8842UL</span>
+
+<span class="hljs-comment">// Negative cases</span>
+123l
+123d
+123D
diff --git a/test/markup/kotlin/numbers.txt b/test/markup/kotlin/numbers.txt
new file mode 100644
index 0000000000..75432e9e40
--- /dev/null
+++ b/test/markup/kotlin/numbers.txt
@@ -0,0 +1,45 @@
+// Postive cases
+1
+123
+123L
+123f
+123F
+123u
+123U
+123UL
+123uL
+1.23
+1.23f
+1.23F
+.123
+.123f
+.123F
+0b10101
+0B10101
+0b10101u
+0B10101u
+0b10101U
+0B10101U
+0b10101L
+0B10101L
+0b10101uL
+0B10101uL
+0b10101UL
+0B10101UL
+0xff8842
+0Xff8842
+0xff8842u
+0Xff8842u
+0xff8842U
+0Xff8842U
+0xff8842L
+0Xff8842L
+0xff8842uL
+0Xff8842uL
+0xff8842UL
+0Xff8842UL
+
+// Negative cases
+123l
+123d
+123D