enh(haskell) Restucture Haskell number parsing

Restructured Haskell's number parsing, taking a cue from Swift's implementation. As a happy coincidence, this patch adds support for Haskell's `HexFloatLiterals` language extension.
highlightjs · Apr 19, 2021 · afa912b · afa912b
1 parent c55ad76
commit afa912b
Show file tree

Hide file tree

Showing 4 changed files with 160 additions and 13 deletions.
diff --git a/CHANGES.md b/CHANGES.md
@@ -80,6 +80,7 @@ Grammars:
 - enh(javascript/typescript) highlight object properties (#3072) [Josh Goebel][]
 - enh(haskell) add support for BinaryLiterals (#3150) [Martijn Bastiaan][]
 - enh(haskell) add support for NumericUnderscores (#3150) [Martijn Bastiaan][]
+- enh(haskell) add support for HexFloatLiterals (#3150) [Martijn Bastiaan][]
 
 New Languages:
 

diff --git a/src/languages/haskell.js b/src/languages/haskell.js
@@ -62,6 +62,42 @@ export default function(hljs) {
     contains: LIST.contains
   };
 
+  /* See:
+
+     - https://www.haskell.org/onlinereport/lexemes.html
+     - https://downloads.haskell.org/ghc/9.0.1/docs/html/users_guide/exts/binary_literals.html
+     - https://downloads.haskell.org/ghc/9.0.1/docs/html/users_guide/exts/numeric_underscores.html
+     - https://downloads.haskell.org/ghc/9.0.1/docs/html/users_guide/exts/hex_float_literals.html
+
+  */
+  const decimalDigits = '([0-9]_*)+';
+  const hexDigits = '([0-9a-fA-F]_*)+';
+  const binaryDigits = '([01]_*)+';
+  const octalDigits = '([0-7]_*)+';
+
+  const NUMBER = {
+    className: 'number',
+    relevance: 0,
+    variants: [
+      // decimal floating-point-literal (subsumes decimal-literal)
+      {
+        match: `(-?)\\b(${decimalDigits})(\\.(${decimalDigits}))?` + `([eE][+-]?(${decimalDigits}))?\\b`
+      },
+      // hexadecimal floating-point-literal (subsumes hexadecimal-literal)
+      {
+        match: `(-?)\\b0[xX]_*(${hexDigits})(\\.(${hexDigits}))?` + `([pP][+-]?(${decimalDigits}))?\\b`
+      },
+      // octal-literal
+      {
+        match: `(-?)\\b0[oO](${octalDigits})\\b`
+      },
+      // binary-literal
+      {
+        match: `(-?)\\b0[bB](${binaryDigits})\\b`
+      }
+    ]
+  };
+
   return {
     name: 'Haskell',
     aliases: ['hs'],
@@ -154,13 +190,7 @@ export default function(hljs) {
       PREPROCESSOR,
 
       // Literals and names.
-      hljs.inherit(hljs.BINARY_NUMBER_MODE, {
-        begin: '\\b(0b[01_]+)'
-      }),
-
-      hljs.inherit(hljs.C_NUMBER_MODE, {
-        begin: '(-?)(\\b0[xX][a-fA-F0-9_]+|(\\b(\\d|_)+(\\.(\\d|_)*)?|\\.(\\d|_)+)([eE][-+]?(\\d|_)+)?)'
-      }),
+      NUMBER,
 
       // TODO: characters.
       hljs.QUOTE_STRING_MODE,

diff --git a/test/markup/haskell/numbers.expect.txt b/test/markup/haskell/numbers.expect.txt
@@ -1,3 +1,61 @@
-<span class="hljs-number">0b001_001</span>
-<span class="hljs-number">0x000_0FF</span>
-<span class="hljs-number">132_15</span>
+<span class="hljs-comment">-- decimal</span>
+<span class="hljs-title">million</span>    = <span class="hljs-number">1_000_000</span>
+<span class="hljs-title">billion</span>    = <span class="hljs-number">1_000_000_000</span>
+<span class="hljs-title">lightspeed</span> = <span class="hljs-number">299_792_458</span>
+<span class="hljs-title">version</span>    = <span class="hljs-number">8_04_1</span>
+<span class="hljs-title">date</span>       = <span class="hljs-number">2017_12_31</span>
+<span class="hljs-title">the_answer</span> = <span class="hljs-number">42</span>
+
+<span class="hljs-comment">-- hexadecimal</span>
+<span class="hljs-title">red_mask</span> = <span class="hljs-number">0xff_00_00</span>
+<span class="hljs-title">size1G</span>   = <span class="hljs-number">0x3fff_fff</span>
+<span class="hljs-title">bigX</span>     = <span class="hljs-number">0Xffff</span>
+
+<span class="hljs-comment">-- octal</span>
+<span class="hljs-title">world</span> = <span class="hljs-number">0o777</span>
+<span class="hljs-title">me</span>    = <span class="hljs-number">0o700</span>
+<span class="hljs-title">bigO</span>  = <span class="hljs-number">0O700</span>
+
+<span class="hljs-comment">-- binary</span>
+<span class="hljs-title">bit8th</span>   = <span class="hljs-number">0b01_0000_0000</span>
+<span class="hljs-title">packbits</span> = <span class="hljs-number">0b1_11_01_0000_0_111</span>
+<span class="hljs-title">bigbits</span>  = <span class="hljs-number">0b1100_1011__1110_1111__0101_0011</span>
+<span class="hljs-title">bigB</span>     = <span class="hljs-number">0B10</span>
+
+<span class="hljs-comment">-- float</span>
+<span class="hljs-title">pi</span>       = <span class="hljs-number">3.141_592_653_589_793</span>
+<span class="hljs-title">faraday</span>  = <span class="hljs-number">96_485.332_89</span>
+<span class="hljs-title">avogadro</span> = <span class="hljs-number">6.022_140_857e+23</span>
+
+<span class="hljs-comment">-- negative numbers</span>
+<span class="hljs-title">n0</span> = <span class="hljs-number">-1</span>
+<span class="hljs-title">n1</span> = <span class="hljs-number">-1.0</span>
+<span class="hljs-title">n2</span> = <span class="hljs-number">-0xFF0</span>
+<span class="hljs-title">n3</span> = <span class="hljs-number">-0o123</span>
+<span class="hljs-title">n4</span> = <span class="hljs-number">-0b001</span>
+
+<span class="hljs-comment">-- Numeric underscores</span>
+<span class="hljs-title">x0</span> = <span class="hljs-number">1_000_000</span>
+<span class="hljs-title">x1</span> = <span class="hljs-number">1__000000</span>
+
+<span class="hljs-title">e0</span> = <span class="hljs-number">0.0001</span>
+<span class="hljs-title">e1</span> = <span class="hljs-number">0.000_1</span>
+
+<span class="hljs-title">f0</span> = <span class="hljs-number">1e+23</span>
+<span class="hljs-title">f1</span> = <span class="hljs-number">1_e+23</span>
+<span class="hljs-title">f2</span> = <span class="hljs-number">1__e+23</span>
+
+<span class="hljs-title">g0</span> = <span class="hljs-number">1e+23</span>
+
+<span class="hljs-title">h0</span> = <span class="hljs-number">0xffff</span>
+<span class="hljs-title">h1</span> = <span class="hljs-number">0xff_ff</span>
+<span class="hljs-title">h2</span> = <span class="hljs-number">0x_ffff</span>
+<span class="hljs-title">h3</span> = <span class="hljs-number">0x__ffff</span>
+
+<span class="hljs-comment">-- hex float</span>
+<span class="hljs-title">hf0</span> = <span class="hljs-number">0x0.1</span>
+<span class="hljs-title">hf1</span> = <span class="hljs-number">0x0.01</span>
+<span class="hljs-title">hf2</span> = <span class="hljs-number">0xF.FF</span>
+<span class="hljs-title">hf3</span> = <span class="hljs-number">0x0.1p41</span>
+<span class="hljs-title">hf4</span> = <span class="hljs-number">0x0.1p-4</span>
+<span class="hljs-title">hf5</span> = <span class="hljs-number">0x0.1p12</span>
diff --git a/test/markup/haskell/numbers.txt b/test/markup/haskell/numbers.txt
@@ -1,3 +1,61 @@
-0b001_001
-0x000_0FF
-132_15
+-- decimal
+million    = 1_000_000
+billion    = 1_000_000_000
+lightspeed = 299_792_458
+version    = 8_04_1
+date       = 2017_12_31
+the_answer = 42
+
+-- hexadecimal
+red_mask = 0xff_00_00
+size1G   = 0x3fff_fff
+bigX     = 0Xffff
+
+-- octal
+world = 0o777
+me    = 0o700
+bigO  = 0O700
+
+-- binary
+bit8th   = 0b01_0000_0000
+packbits = 0b1_11_01_0000_0_111
+bigbits  = 0b1100_1011__1110_1111__0101_0011
+bigB     = 0B10
+
+-- float
+pi       = 3.141_592_653_589_793
+faraday  = 96_485.332_89
+avogadro = 6.022_140_857e+23
+
+-- negative numbers
+n0 = -1
+n1 = -1.0
+n2 = -0xFF0
+n3 = -0o123
+n4 = -0b001
+
+-- Numeric underscores
+x0 = 1_000_000
+x1 = 1__000000
+
+e0 = 0.0001
+e1 = 0.000_1
+
+f0 = 1e+23
+f1 = 1_e+23
+f2 = 1__e+23
+
+g0 = 1e+23
+
+h0 = 0xffff
+h1 = 0xff_ff
+h2 = 0x_ffff
+h3 = 0x__ffff
+
+-- hex float
+hf0 = 0x0.1
+hf1 = 0x0.01
+hf2 = 0xF.FF
+hf3 = 0x0.1p41
+hf4 = 0x0.1p-4
+hf5 = 0x0.1p12