(parser) Adds SHEBANG mode

highlightjs · Apr 29, 2020 · a23f19e · a23f19e
1 parent b984c7b
commit a23f19e
Show file tree

Hide file tree

Showing 9 changed files with 51 additions and 27 deletions.
diff --git a/CHANGES.md b/CHANGES.md
@@ -2,6 +2,7 @@
 
 Parser Engine:
 
+- (parser) Adds SHEBANG utility mode [Josh Goebel][]
 - (enh) Added `on:begin` callback for modes (#2261) [Josh Goebel][]
 - (enh) Added `on:end` callback for modes (#2261) [Josh Goebel][]
 - (enh) Added ability to programatically ignore begin and end matches (#2261) [Josh Goebel][]

diff --git a/src/languages/bash.js b/src/languages/bash.js
@@ -55,11 +55,21 @@ export default function(hljs) {
       VAR
     ]
   };
-  const SHEBANG = {
-    className: 'meta',
-    begin: /^#![^\n]+sh\s*$/,
+  const SH_LIKE_SHELLS = [
+    "fish",
+    "bash",
+    "zsh",
+    "sh",
+    "csh",
+    "ksh",
+    "tcsh",
+    "dash",
+    "scsh",
+  ];
+  const KNOWN_SHEBANG = hljs.SHEBANG({
+    binary: `(${SH_LIKE_SHELLS.join("|")})`,
     relevance: 10
-  };
+  });
   const FUNCTION = {
     className: 'function',
     begin: /\w[\w\d_]*\s*\(\s*\)\s*\{/,
@@ -98,7 +108,8 @@ export default function(hljs) {
         '-ne -eq -lt -gt -f -d -e -s -l -a' // relevance booster
     },
     contains: [
-      SHEBANG,
+      KNOWN_SHEBANG, // to catch known shells and boost relevancy
+      hljs.SHEBANG(), // to catch unknown shells but still highlight the shebang
       FUNCTION,
       ARITHMETIC,
       hljs.HASH_COMMENT_MODE,

diff --git a/src/languages/hy.js b/src/languages/hy.js
@@ -45,11 +45,6 @@ export default function(hljs) {
   var SYMBOL_RE = '[' + SYMBOLSTART + '][' + SYMBOLSTART + '0-9/;:]*';
   var SIMPLE_NUMBER_RE = '[-+]?\\d+(\\.\\d+)?';
 
-  var SHEBANG = {
-    className: 'meta',
-    begin: '^#!', end: '$'
-  };
-
   var SYMBOL = {
     begin: SYMBOL_RE,
     relevance: 0
@@ -105,6 +100,6 @@ export default function(hljs) {
     name: 'Hy',
     aliases: ['hylang'],
     illegal: /\S/,
-    contains: [SHEBANG, LIST, STRING, HINT, HINT_COL, COMMENT, KEY, COLLECTION, NUMBER, LITERAL]
+    contains: [hljs.SHEBANG(), LIST, STRING, HINT, HINT_COL, COMMENT, KEY, COLLECTION, NUMBER, LITERAL]
   }
 }
diff --git a/src/languages/javascript.js b/src/languages/javascript.js
@@ -97,15 +97,15 @@ export default function(hljs) {
     aliases: ['js', 'jsx', 'mjs', 'cjs'],
     keywords: KEYWORDS,
     contains: [
+      hljs.SHEBANG({
+        binary: "node",
+        relevance: 5
+      }),
       {
         className: 'meta',
         relevance: 10,
         begin: /^\s*['"]use (strict|asm)['"]/
       },
-      {
-        className: 'meta',
-        begin: /^#!/, end: /$/
-      },
       hljs.APOS_STRING_MODE,
       hljs.QUOTE_STRING_MODE,
       HTML_TEMPLATE,

diff --git a/src/languages/lisp.js b/src/languages/lisp.js
@@ -9,10 +9,6 @@ export default function(hljs) {
   var LISP_IDENT_RE = '[a-zA-Z_\\-\\+\\*\\/\\<\\=\\>\\&\\#][a-zA-Z0-9_\\-\\+\\*\\/\\<\\=\\>\\&\\#!]*';
   var MEC_RE = '\\|[^]*?\\|';
   var LISP_SIMPLE_NUMBER_RE = '(\\-|\\+)?\\d+(\\.\\d+|\\/\\d+)?((d|e|f|l|s|D|E|F|L|S)(\\+|\\-)?\\d+)?';
-  var SHEBANG = {
-    className: 'meta',
-    begin: '^#!', end: '$'
-  };
   var LITERAL = {
     className: 'literal',
     begin: '\\b(t{1}|nil)\\b'
@@ -97,7 +93,7 @@ export default function(hljs) {
     illegal: /\S/,
     contains: [
       NUMBER,
-      SHEBANG,
+      hljs.SHEBANG(),
       LITERAL,
       STRING,
       COMMENT,

diff --git a/src/languages/scheme.js b/src/languages/scheme.js
@@ -50,12 +50,6 @@ export default function(hljs) {
       'with-input-from-file with-output-to-file write write-char zero?'
   };
 
-  var SHEBANG = {
-    className: 'meta',
-    begin: '^#!',
-    end: '$'
-  };
-
   var LITERAL = {
     className: 'literal',
     begin: '(#t|#f|#\\\\' + SCHEME_IDENT_RE + '|#\\\\.)'
@@ -150,6 +144,6 @@ export default function(hljs) {
   return {
     name: 'Scheme',
     illegal: /\S/,
-    contains: [SHEBANG, NUMBER, STRING, QUOTED_IDENT, QUOTED_LIST, LIST].concat(COMMENT_MODES)
+    contains: [hljs.SHEBANG(), NUMBER, STRING, QUOTED_IDENT, QUOTED_LIST, LIST].concat(COMMENT_MODES)
   };
 }
diff --git a/src/languages/typescript.js b/src/languages/typescript.js
@@ -127,6 +127,7 @@ export default function(hljs) {
     aliases: ['ts'],
     keywords: KEYWORDS,
     contains: [
+      hljs.SHEBANG(),
       {
         className: 'meta',
         begin: /^\s*['"]use strict['"]/

diff --git a/src/lib/modes.js b/src/lib/modes.js
@@ -1,4 +1,5 @@
 import { inherit } from './utils';
+import * as regex from './regex';
 
 // Common regexps
 export const IDENT_RE = '[a-zA-Z]\\w*';
@@ -8,6 +9,26 @@ export const C_NUMBER_RE = '(-?)(\\b0[xX][a-fA-F0-9]+|(\\b\\d+(\\.\\d*)?|\\.\\d+
 export const BINARY_NUMBER_RE = '\\b(0b[01]+)'; // 0b...
 export const RE_STARTERS_RE = '!|!=|!==|%|%=|&|&&|&=|\\*|\\*=|\\+|\\+=|,|-|-=|/=|/|:|;|<<|<<=|<=|<|===|==|=|>>>=|>>=|>=|>>>|>>|>|\\?|\\[|\\{|\\(|\\^|\\^=|\\||\\|=|\\|\\||~';
 
+export const SHEBANG = (opts = {}) => {
+  const beginShebang = /^#![ ]*\//;
+  if (opts.binary) {
+    opts.begin = regex.concat(
+      beginShebang,
+      /.*\b/,
+      opts.binary,
+      /\b.*/);
+  }
+  return inherit({
+    className: 'meta',
+    begin: beginShebang,
+    end: /$/,
+    relevance: 0,
+    "on:begin": (m, resp) => {
+      if (m.index !== 0) resp.ignoreMatch();
+    }
+  }, opts);
+};
+
 // Common modes
 export const BACKSLASH_ESCAPE = {
   begin: '\\\\[\\s\\S]', relevance: 0

diff --git a/src/lib/regex.js b/src/lib/regex.js
@@ -8,6 +8,11 @@ export function source(re) {
   return (re && re.source) || re;
 }
 
+export function concat(...args) {
+  const joined = args.map((x) => source(x)).join("");
+  return joined;
+}
+
 export function countMatchGroups(re) {
   return (new RegExp(re.toString() + '|')).exec('').length - 1;
 }