Merge branch 'master' into expression-variants

highlightjs · Oct 15, 2019 · 640db25 · 640db25
2 parents 6ebaa5f + e859d30
commit 640db25
Show file tree

Hide file tree

Showing 180 changed files with 491 additions and 122 deletions.
diff --git a/AUTHORS.en.txt b/AUTHORS.en.txt
@@ -279,3 +279,5 @@ Contributors:
 - Taif Alimov <inzeppelin@gmail.com>
 - Yuri Mazursky <mail@colomolome.com>
 - Carl Baxter <carl@cbax.tech>
+- Thomas Reichel <tom.p.reichel@gmail.com>
+
diff --git a/CHANGES.md b/CHANGES.md
@@ -6,13 +6,23 @@ New styles:
 - *Night Owl* by [Carl Baxter][]
 
 Improvements:
+- blacklist super-common keywords from having relevance (#2179)
+- fix(swift): support for `@dynamicMemberLookup` and `@propertyWrapper` (#2202)
+- fix(typescript): constructor in declaration doesn't break highlighting
+- fix(typescript): only match function keyword as a separate identifier (#2191)
+- feature(arduino) make arduino a super-set of cpp grammar
 - fix(javascript): fix object attributes immediately following line comments
 - fix(xml): remove `vbscript` as potential script tag subLanguage
 - fix(Elixir): improve regex for numbers
 - fix(YAML): improve matching for keys
 - fix(Pony): improve regex for numbers
+- fix(handlebars): add support for raw-blocks and triple-mustaches (#2175)
 - JSON: support for comments in JSON (#2016)
 - fix(cpp): improve string literal matching
+- fix(highlight.js): omit empty span-tags in the output (#2182)
+- fix(Go): improve function declaration matching
+- fix(python): added support for f-string literal curly braces (#2195)
+- fix(cpp): add `future` built-in (#1610)
 
 [Carl Baxter]: https://github.com/cdbax
 

diff --git a/src/highlight.js b/src/highlight.js
@@ -41,7 +41,7 @@ https://highlightjs.org/
       languagePrefixRe = /\blang(?:uage)?-([\w-]+)\b/i,
       fixMarkupRe      = /((^(<[^>]+>|\t|)+|(?:\n)))/gm;
 
-  // The object will be assigned by the build tool. It used to synchronize API 
+  // The object will be assigned by the build tool. It used to synchronize API
   // of external language files with minified version of the highlight.js library.
   var API_REPLACES;
 
@@ -56,6 +56,9 @@ https://highlightjs.org/
     languages: undefined
   };
 
+  // keywords that should have no default relevance value
+  var COMMON_KEYWORDS = 'of and for in not or if then'.split(' ')
+
 
   /* Utility functions */
 
@@ -238,6 +241,44 @@ https://highlightjs.org/
     }
   }
 
+  function compileKeywords(rawKeywords, case_insensitive) {
+      var compiled_keywords = {};
+
+      if (typeof rawKeywords === 'string') { // string
+        splitAndCompile('keyword', rawKeywords);
+      } else {
+        objectKeys(rawKeywords).forEach(function (className) {
+          splitAndCompile(className, rawKeywords[className]);
+        });
+      }
+    return compiled_keywords;
+
+    // ---
+
+    function splitAndCompile(className, str) {
+      if (case_insensitive) {
+        str = str.toLowerCase();
+      }
+      str.split(' ').forEach(function(keyword) {
+        var pair = keyword.split('|');
+        compiled_keywords[pair[0]] = [className, scoreForKeyword(pair[0], pair[1])];
+      });
+    };
+  }
+
+  function scoreForKeyword(keyword, providedScore) {
+    // manual scores always win over common keywords
+    // so you can force a score of 1 if you really insist
+    if (providedScore)
+      return Number(providedScore)
+
+    return commonKeyword(keyword) ? 0 : 1;
+  }
+
+  function commonKeyword(word) {
+    return COMMON_KEYWORDS.indexOf(word.toLowerCase()) != -1
+  }
+
   function compileLanguage(language) {
 
     function reStr(re) {
@@ -298,28 +339,9 @@ https://highlightjs.org/
       mode.compiled = true;
 
       mode.keywords = mode.keywords || mode.beginKeywords;
-      if (mode.keywords) {
-        var compiled_keywords = {};
-
-        var flatten = function(className, str) {
-          if (language.case_insensitive) {
-            str = str.toLowerCase();
-          }
-          str.split(' ').forEach(function(kw) {
-            var pair = kw.split('|');
-            compiled_keywords[pair[0]] = [className, pair[1] ? Number(pair[1]) : 1];
-          });
-        };
+      if (mode.keywords)
+        mode.keywords = compileKeywords(mode.keywords, language.case_insensitive)
 
-        if (typeof mode.keywords === 'string') { // string
-          flatten('keyword', mode.keywords);
-        } else {
-          objectKeys(mode.keywords).forEach(function (className) {
-            flatten(className, mode.keywords[className]);
-          });
-        }
-        mode.keywords = compiled_keywords;
-      }
       mode.lexemesRe = langRe(mode.lexemes || /\w+/, true);
 
       if (parent) {
@@ -364,7 +386,7 @@ https://highlightjs.org/
         .filter(Boolean);
       mode.terminators = terminators.length ? langRe(joinRe(terminators, '|'), true) : {exec: function(/*s*/) {return null;}};
     }
-    
+
     compileMode(language);
   }
 
@@ -751,6 +773,8 @@ https://highlightjs.org/
   function registerLanguage(name, language) {
     var lang = languages[name] = language(hljs);
     restoreLanguageApi(lang);
+    lang.rawDefinition = language.bind(null,hljs);
+
     if (lang.aliases) {
       lang.aliases.forEach(function(alias) {aliases[alias] = name;});
     }

diff --git a/src/languages/abnf.js b/src/languages/abnf.js
@@ -1,6 +1,7 @@
 /*
 Language: Augmented Backus-Naur Form
 Author: Alex McKibben <alex@nullscope.net>
+Website: https://tools.ietf.org/html/rfc5234
 */
 
 function(hljs) {

diff --git a/src/languages/accesslog.js b/src/languages/accesslog.js
@@ -2,6 +2,7 @@
  Language: Access log
  Author: Oleg Efimov <efimovov@gmail.com>
  Description: Apache/Nginx Access Logs
+ Website: https://httpd.apache.org/docs/2.4/logs.html#accesslog
  */
 
 function(hljs) {

diff --git a/src/languages/angelscript.js b/src/languages/angelscript.js
@@ -2,6 +2,7 @@
 Language: AngelScript
 Author: Melissa Geels <melissa@nimble.tools>
 Category: scripting
+Website: https://www.angelcode.com/angelscript/
 */
 
 function(hljs) {

diff --git a/src/languages/apache.js b/src/languages/apache.js
@@ -2,7 +2,7 @@
 Language: Apache
 Author: Ruslan Keba <rukeba@gmail.com>
 Contributors: Ivan Sagalaev <maniac@softwaremaniacs.org>
-Website: http://rukeba.com/
+Website: https://httpd.apache.org
 Description: language definition for Apache configuration files (httpd.conf & .htaccess)
 Category: common, config
 */

diff --git a/src/languages/applescript.js b/src/languages/applescript.js
@@ -2,6 +2,7 @@
 Language: AppleScript
 Authors: Nathan Grigg <nathan@nathanamy.org>, Dr. Drang <drdrang@gmail.com>
 Category: scripting
+Website: https://developer.apple.com/library/archive/documentation/AppleScript/Conceptual/AppleScriptLangGuide/introduction/ASLR_intro.html
 */
 
 function(hljs) {

diff --git a/src/languages/arcade.js b/src/languages/arcade.js
@@ -2,6 +2,7 @@
  Language: ArcGIS Arcade
  Category: scripting
  Author: John Foster <jfoster@esri.com>
+ Website: https://developers.arcgis.com/arcade/
  Description: ArcGIS Arcade is an expression language used in many Esri ArcGIS products such as Pro, Online, Server, Runtime, JavaScript, and Python
 */
 function(hljs) {

diff --git a/src/languages/arduino.js b/src/languages/arduino.js
@@ -3,17 +3,16 @@ Language: Arduino
 Author: Stefania Mellai <s.mellai@arduino.cc>
 Description: The Arduino® Language is a superset of C++. This rules are designed to highlight the Arduino® source code. For info about language see http://www.arduino.cc.
 Requires: cpp.js
+Website: https://www.arduino.cc
 */
 
 function(hljs) {
-  var CPP = hljs.getLanguage('cpp').exports;
-	return {
-    keywords: {
+
+	ARDUINO_KW = {
       keyword:
-        'boolean byte word string String array ' + CPP.keywords.keyword,
+        'boolean byte word String',
       built_in:
-        'setup loop while catch for if do goto try switch case else ' +
-        'default break continue return ' +
+        'setup loop' +
         'KeyboardController MouseController SoftwareSerial ' +
         'EthernetServer EthernetClient LiquidCrystal ' +
         'RobotControl GSMVoiceCall EthernetUDP EsploraTFT ' +
@@ -93,14 +92,15 @@ function(hljs) {
         'SET_PIN_MODE INTERNAL2V56 SYSTEM_RESET LED_BUILTIN ' +
         'INTERNAL1V1 SYSEX_START INTERNAL EXTERNAL ' +
         'DEFAULT OUTPUT INPUT HIGH LOW'
-    },
-    contains: [
-      CPP.preprocessor,
-      hljs.C_LINE_COMMENT_MODE,
-      hljs.C_BLOCK_COMMENT_MODE,
-      hljs.APOS_STRING_MODE,
-      hljs.QUOTE_STRING_MODE,
-      hljs.C_NUMBER_MODE
-    ]
   };
+
+  var ARDUINO = hljs.getLanguage('cpp').rawDefinition();
+
+  var kws = ARDUINO.keywords;
+
+  kws.keyword += ' ' + ARDUINO_KW.keyword;
+  kws.literal += ' ' + ARDUINO_KW.literal;
+  kws.built_in += ' ' + ARDUINO_KW.built_in;
+
+  return ARDUINO;
 }
diff --git a/src/languages/asciidoc.js b/src/languages/asciidoc.js
@@ -2,7 +2,7 @@
 Language: AsciiDoc
 Requires: xml.js
 Author: Dan Allen <dan.j.allen@gmail.com>
-Website: http://google.com/profiles/dan.j.allen
+Website: http://asciidoc.org
 Description: A semantic, text-based document format that can be exported to HTML, DocBook and other backends.
 Category: markup
 */

diff --git a/src/languages/aspectj.js b/src/languages/aspectj.js
@@ -1,6 +1,7 @@
 /*
 Language: AspectJ
 Author: Hakan Ozler <ozler.hakan@gmail.com>
+Website: https://www.eclipse.org/aspectj/
 Description: Syntax Highlighting for the AspectJ Language which is a general-purpose aspect-oriented extension to the Java programming language.
  */
 function (hljs) {

diff --git a/src/languages/avrasm.js b/src/languages/avrasm.js
@@ -2,6 +2,7 @@
 Language: AVR Assembler
 Author: Vladimir Ermakov <vooon341@gmail.com>
 Category: assembler
+Website: https://www.microchip.com/webdoc/avrassembler/avrassembler.wb_instruction_list.html
 */
 
 function(hljs) {
@@ -58,7 +59,7 @@ function(hljs) {
       },
       {className: 'symbol',  begin: '^[A-Za-z0-9_.$]+:'},
       {className: 'meta', begin: '#', end: '$'},
-      {  // подстановка в «.macro»
+      {  // substitution within a macro
         className: 'subst',
         begin: '@[0-9]+'
       }

diff --git a/src/languages/awk.js b/src/languages/awk.js
@@ -1,7 +1,7 @@
 /*
 Language: Awk
 Author: Matthew Daly <matthewbdaly@gmail.com>
-Website: http://matthewdaly.co.uk/
+Website: https://www.gnu.org/software/gawk/manual/gawk.html
 Description: language definition for Awk scripts
 */
 

diff --git a/src/languages/axapta.js b/src/languages/axapta.js
@@ -1,6 +1,7 @@
 /*
-Language: Axapta
+Language: Microsoft Axapta (now Dynamics 365)
 Author: Dmitri Roudakov <dmitri@roudakov.ru>
+Website: https://dynamics.microsoft.com/en-us/ax-overview/
 Category: enterprise
 */
 

diff --git a/src/languages/bash.js b/src/languages/bash.js
@@ -2,6 +2,7 @@
 Language: Bash
 Author: vah <vahtenberg@gmail.com>
 Contributrors: Benjamin Pannell <contact@sierrasoftworks.com>
+Website: https://www.gnu.org/software/bash/
 Category: common
 */
 

diff --git a/src/languages/basic.js b/src/languages/basic.js
@@ -2,6 +2,7 @@
 Language: Basic
 Author: Raphaël Assénat <raph@raphnet.net>
 Description: Based on the BASIC reference from the Tandy 1000 guide
+Website: https://en.wikipedia.org/wiki/Tandy_1000
 */
 function(hljs) {
   return {
@@ -15,7 +16,7 @@ function(hljs) {
           'CLEAR CLOSE CLS COLOR COM COMMON CONT COS CSNG CSRLIN CVD CVI CVS DATA DATE$ ' +
           'DEFDBL DEFINT DEFSNG DEFSTR DEF|0 SEG USR DELETE DIM DRAW EDIT END ENVIRON ENVIRON$ ' +
           'EOF EQV ERASE ERDEV ERDEV$ ERL ERR ERROR EXP FIELD FILES FIX FOR|0 FRE GET GOSUB|10 GOTO ' +
-          'HEX$ IF|0 THEN ELSE|0 INKEY$ INP INPUT INPUT# INPUT$ INSTR IMP INT IOCTL IOCTL$ KEY ON ' +
+          'HEX$ IF THEN ELSE|0 INKEY$ INP INPUT INPUT# INPUT$ INSTR IMP INT IOCTL IOCTL$ KEY ON ' +
           'OFF LIST KILL LEFT$ LEN LET LINE LLIST LOAD LOC LOCATE LOF LOG LPRINT USING LSET ' +
           'MERGE MID$ MKDIR MKD$ MKI$ MKS$ MOD NAME NEW NEXT NOISE NOT OCT$ ON OR PEN PLAY STRIG OPEN OPTION ' +
           'BASE OUT PAINT PALETTE PCOPY PEEK PMAP POINT POKE POS PRINT PRINT] PSET PRESET ' +

diff --git a/src/languages/bnf.js b/src/languages/bnf.js
@@ -1,5 +1,6 @@
 /*
 Language: Backus–Naur Form
+Website: https://en.wikipedia.org/wiki/Backus–Naur_form
 Author: Oleg Efimov <efimovov@gmail.com>
 */
 

diff --git a/src/languages/brainfuck.js b/src/languages/brainfuck.js
@@ -1,6 +1,7 @@
 /*
 Language: Brainfuck
 Author: Evgeny Stepanischev <imbolk@gmail.com>
+Website: https://esolangs.org/wiki/Brainfuck
 */
 
 function(hljs){

diff --git a/src/languages/cal.js b/src/languages/cal.js
@@ -2,6 +2,7 @@
 Language: C/AL
 Author: Kenneth Fuglsang Christensen <kfuglsang@gmail.com>
 Description: Provides highlighting of Microsoft Dynamics NAV C/AL code files
+Website: https://docs.microsoft.com/en-us/dynamics-nav/programming-in-c-al
 */
 
 function(hljs) {

diff --git a/src/languages/capnproto.js b/src/languages/capnproto.js
@@ -2,6 +2,7 @@
 Language: Cap’n Proto
 Author: Oleg Efimov <efimovov@gmail.com>
 Description: Cap’n Proto message definition format
+Website: https://capnproto.org/capnp-tool.html
 Category: protocols
 */
 

diff --git a/src/languages/ceylon.js b/src/languages/ceylon.js
@@ -1,6 +1,7 @@
 /*
 Language: Ceylon
 Author: Lucas Werkmeister <mail@lucaswerkmeister.de>
+Website: https://ceylon-lang.org
 */
 function(hljs) {
   // 2.3. Identifiers and keywords

diff --git a/src/languages/clojure-repl.js b/src/languages/clojure-repl.js
@@ -3,6 +3,7 @@ Language: Clojure REPL
 Description: Clojure REPL sessions
 Author: Ivan Sagalaev <maniac@softwaremaniacs.org>
 Requires: clojure.js
+Website: https://clojure.org
 Category: lisp
 */
 

diff --git a/src/languages/clojure.js b/src/languages/clojure.js
@@ -3,6 +3,7 @@ Language: Clojure
 Description: Clojure syntax (based on lisp.js)
 Author: mfornos
 Contributors: Martin Clausen <martin.clausene@gmail.com>
+Website: https://clojure.org
 Category: lisp
 */
 

diff --git a/src/languages/cmake.js b/src/languages/cmake.js
@@ -2,7 +2,7 @@
 Language: CMake
 Description: CMake is an open-source cross-platform system for build automation.
 Author: Igor Kalnitsky <igor@kalnitsky.org>
-Website: http://kalnitsky.org/
+Website: https://cmake.org
 */
 
 function(hljs) {

diff --git a/src/languages/coffeescript.js b/src/languages/coffeescript.js
@@ -4,6 +4,7 @@ Author: Dmytrii Nagirniak <dnagir@gmail.com>
 Contributors: Oleg Efimov <efimovov@gmail.com>, Cédric Néhémie <cedric.nehemie@gmail.com>
 Description: CoffeeScript is a programming language that transcompiles to JavaScript. For info about language see http://coffeescript.org/
 Category: common, scripting
+Website: https://coffeescript.org
 */
 
 function(hljs) {