enh(llvm) Improve number support, add char.escape (#3471)

- Numbers are allowed to have a leading + sign
- Hex floats can have a format specified after the 'x'
- Strings cannot escape quotes with backslashes, so a '\"' ends a string
- Add support for `\00` style `char.escape` inside strings

Author: Flakebi

CHANGES.md

@@ -7,11 +7,13 @@ Grammars:
 - fix(angelscript) incomplete int8, int16, int32, int64 highlighting [Melissa Geels][]
 - enh(ts) modify TypeScript-specific keywords and types list [anydonym][]
 - fix(brainfuck) fix highlighting of initial ++/-- [Christina Hanson][]
+- fix(llvm) escaping in strings and number formats [Flakebi][]
 
 [Wojciech Kania]: https://github.com/wkania
 [Melissa Geels]: https://github.com/codecat
 [anydonym]: https://github.com/anydonym
 [Christina Hanson]: https://github.com/LyricLy
+[Flakebi]: https://github.com/Flakebi
 
 ## Version 11.4.0
 

src/languages/llvm.js

@@ -28,8 +28,8 @@ export default function(hljs) {
   const NUMBER = {
     className: 'number',
     variants: [
-        { begin: /0[xX][a-fA-F0-9]+/ },
-        { begin: /-?\d+(?:[.]\d+)?(?:[eE][-+]?\d+(?:[.]\d+)?)?/ }
+        { begin: /[su]?0[xX][KMLHR]?[a-fA-F0-9]+/ },
+        { begin: /[-+]?\d+(?:[.]\d+)?(?:[eE][-+]?\d+(?:[.]\d+)?)?/ }
     ],
     relevance: 0
   };
@@ -109,12 +109,15 @@ export default function(hljs) {
       // another language than an actual comment
       hljs.COMMENT(/;\s*$/, null, { relevance: 0 }),
       hljs.COMMENT(/;/, /$/),
-      hljs.QUOTE_STRING_MODE,
       {
         className: 'string',
-        variants: [
-          // Double-quoted string
-          { begin: /"/, end: /[^\\]"/ },
+        begin: /"/,
+        end: /"/,
+        contains: [
+          {
+            className: 'char.escape',
+            match: /\\\d\d/
+          }
         ]
       },
       FUNCTION,

test/markup/llvm/default.expect.txt

@@ -6,10 +6,10 @@
 <span class="hljs-variable">%struct._IO_marker</span> <span class="hljs-operator">=</span> <span class="hljs-keyword">type</span> { <span class="hljs-variable">%struct._IO_marker</span>*<span class="hljs-punctuation">,</span> <span class="hljs-variable">%struct._IO_FILE</span>*<span class="hljs-punctuation">,</span> <span class="hljs-type">i32</span> }
 <span class="hljs-variable">%struct.what</span> <span class="hljs-operator">=</span> <span class="hljs-keyword">type</span> { <span class="hljs-type">i8</span><span class="hljs-punctuation">,</span> <span class="hljs-type">i16</span> }
 
-<span class="hljs-title">@.str</span> <span class="hljs-operator">=</span> <span class="hljs-keyword">private</span> <span class="hljs-keyword">unnamed_addr</span> <span class="hljs-keyword">constant</span> [<span class="hljs-number">6</span> <span class="hljs-keyword">x</span> <span class="hljs-type">i8</span>] <span class="hljs-keyword">c</span><span class="hljs-string">&quot;foo()\00&quot;</span><span class="hljs-punctuation">,</span> <span class="hljs-keyword">align</span> <span class="hljs-number">1</span>
+<span class="hljs-title">@.str</span> <span class="hljs-operator">=</span> <span class="hljs-keyword">private</span> <span class="hljs-keyword">unnamed_addr</span> <span class="hljs-keyword">constant</span> [<span class="hljs-number">6</span> <span class="hljs-keyword">x</span> <span class="hljs-type">i8</span>] <span class="hljs-keyword">c</span><span class="hljs-string">&quot;foo()<span class="hljs-char escape_">\00</span>&quot;</span><span class="hljs-punctuation">,</span> <span class="hljs-keyword">align</span> <span class="hljs-number">1</span>
 <span class="hljs-title">@e_long</span> <span class="hljs-operator">=</span> <span class="hljs-keyword">common</span> <span class="hljs-keyword">global</span> <span class="hljs-type">i64</span> <span class="hljs-number">0</span><span class="hljs-punctuation">,</span> <span class="hljs-keyword">align</span> <span class="hljs-number">8</span>
 <span class="hljs-title">@g_double</span> <span class="hljs-operator">=</span> <span class="hljs-keyword">common</span> <span class="hljs-keyword">global</span> <span class="hljs-keyword">double</span> <span class="hljs-number">0.000000e+00</span><span class="hljs-punctuation">,</span> <span class="hljs-keyword">align</span> <span class="hljs-number">8</span>
-<span class="hljs-title">@.str.1</span> <span class="hljs-operator">=</span> <span class="hljs-keyword">private</span> <span class="hljs-keyword">unnamed_addr</span> <span class="hljs-keyword">constant</span> [<span class="hljs-number">7</span> <span class="hljs-keyword">x</span> <span class="hljs-type">i8</span>] <span class="hljs-keyword">c</span><span class="hljs-string">&quot;oooooh\00&quot;</span><span class="hljs-punctuation">,</span> <span class="hljs-keyword">align</span> <span class="hljs-number">1</span>
+<span class="hljs-title">@.str.1</span> <span class="hljs-operator">=</span> <span class="hljs-keyword">private</span> <span class="hljs-keyword">unnamed_addr</span> <span class="hljs-keyword">constant</span> [<span class="hljs-number">7</span> <span class="hljs-keyword">x</span> <span class="hljs-type">i8</span>] <span class="hljs-keyword">c</span><span class="hljs-string">&quot;oooooh<span class="hljs-char escape_">\00</span>&quot;</span><span class="hljs-punctuation">,</span> <span class="hljs-keyword">align</span> <span class="hljs-number">1</span>
 <span class="hljs-title">@func_ptr</span> <span class="hljs-operator">=</span> <span class="hljs-keyword">common</span> <span class="hljs-keyword">global</span> <span class="hljs-type">i32</span> (...)* <span class="hljs-keyword">null</span><span class="hljs-punctuation">,</span> <span class="hljs-keyword">align</span> <span class="hljs-number">8</span>
 <span class="hljs-title">@stderr</span> <span class="hljs-operator">=</span> <span class="hljs-keyword">external</span> <span class="hljs-keyword">global</span> <span class="hljs-variable">%struct._IO_FILE</span>*<span class="hljs-punctuation">,</span> <span class="hljs-keyword">align</span> <span class="hljs-number">8</span>
 

test/markup/llvm/numbers.expect.txt

@@ -0,0 +1,11 @@
+<span class="hljs-keyword">define</span> <span class="hljs-type">i32</span> <span class="hljs-title">@signed_int</span>() {
+  <span class="hljs-keyword">ret</span> <span class="hljs-type">i32</span> <span class="hljs-number">s0x12a</span>
+}
+
+<span class="hljs-keyword">define</span> float <span class="hljs-title">@signed_float</span>() {
+  <span class="hljs-keyword">ret</span> float <span class="hljs-number">+1.0</span>
+}
+
+<span class="hljs-keyword">define</span> float <span class="hljs-title">@float_format</span>() {
+  <span class="hljs-keyword">ret</span> float <span class="hljs-number">0xK12a</span>
+}

test/markup/llvm/numbers.txt

@@ -0,0 +1,11 @@
+define i32 @signed_int() {
+  ret i32 s0x12a
+}
+
+define float @signed_float() {
+  ret float +1.0
+}
+
+define float @float_format() {
+  ret float 0xK12a
+}

test/markup/llvm/string.expect.txt

@@ -0,0 +1,8 @@
+<span class="hljs-comment">; Backslashes do not escape quotes, this is a legal string</span>
+<span class="hljs-keyword">define</span> void @<span class="hljs-string">&quot;C:\&quot;</span>() {
+  <span class="hljs-keyword">ret</span> void
+}
+
+<span class="hljs-keyword">define</span> void @<span class="hljs-string">&quot;escape_code_<span class="hljs-char escape_">\01</span>23<span class="hljs-char escape_">\04</span>end&quot;</span>() {
+  <span class="hljs-keyword">ret</span> void
+}

test/markup/llvm/string.txt

@@ -0,0 +1,8 @@
+; Backslashes do not escape quotes, this is a legal string
+define void @"C:\"() {
+  ret void
+}
+
+define void @"escape_code_\0123\04end"() {
+  ret void
+}