Implement rudimentary JSX lexer based on https://github.com/fcurella/jsx-lexer/blob/master/jsx/lexer.py

Fixes #111.

lexers/javascript.go

@@ -4,16 +4,7 @@ import (
 	. "github.com/alecthomas/chroma" // nolint
 )
 
-// Javascript lexer.
-var Javascript = Register(MustNewLexer(
-	&Config{
-		Name:      "JavaScript",
-		Aliases:   []string{"js", "javascript"},
-		Filenames: []string{"*.js", "*.jsm"},
-		MimeTypes: []string{"application/javascript", "application/x-javascript", "text/x-javascript", "text/javascript"},
-		DotAll:    true,
-	},
-	Rules{
+var JavascriptRules = Rules{
 	"commentsandwhitespace": {
 		{`\s+`, Text, nil},
 		{`<!--`, Comment, nil},
@@ -64,5 +55,16 @@ var Javascript = Register(MustNewLexer(
 		{`\}`, LiteralStringInterpol, Pop(1)},
 		Include("root"),
 	},
+}
+
+// Javascript lexer.
+var Javascript = Register(MustNewLexer(
+	&Config{
+		Name:      "JavaScript",
+		Aliases:   []string{"js", "javascript"},
+		Filenames: []string{"*.js", "*.jsm"},
+		MimeTypes: []string{"application/javascript", "application/x-javascript", "text/x-javascript", "text/javascript"},
+		DotAll:    true,
 	},
+	JavascriptRules,
 ))

lexers/jsx.go

@@ -0,0 +1,41 @@
+package lexers
+
+import (
+	. "github.com/alecthomas/chroma" // nolint
+)
+
+var JSXRules = func() Rules {
+	rules := JavascriptRules.Clone()
+	rules["jsx"] = []Rule{
+		{`(<)([\w_\-]+)`, ByGroups(Punctuation, NameTag), Push("tag")},
+		{`(<)(/)(\s*)([\w_\-]+)(\s*)(>)`, ByGroups(Punctuation, Punctuation, Text, NameTag, Text, Punctuation), nil},
+	}
+	rules["tag"] = []Rule{
+		{`\s+`, Text, nil},
+		{`([\w]+\s*)(=)(\s*)`, ByGroups(NameAttribute, Operator, Text), Push("attr")},
+		{`[{}]+`, Punctuation, nil},
+		{`[\w\.]+`, NameAttribute, nil},
+		{`(/?)(\s*)(>)`, ByGroups(Punctuation, Text, Punctuation), Pop(1)},
+	}
+	rules["attr"] = []Rule{
+		{`\s+`, Text, nil},
+		{`".*?"`, String, Pop(1)},
+		{`'.*?'`, String, Pop(1)},
+		{`[^\s>]+`, String, Pop(1)},
+	}
+
+	rules["root"] = append([]Rule{Include("jsx")}, rules["root"]...)
+	return rules
+}()
+
+// JSX lexer.
+var JSX = Register(MustNewLexer(
+	&Config{
+		Name:      "JSX",
+		Aliases:   []string{"react"},
+		Filenames: []string{"*.jsx", "*.react"},
+		MimeTypes: []string{"text/jsx", "text/typescript-jsx"},
+		DotAll:    true,
+	},
+	JSXRules,
+))

lexers/testdata/jsx.actual

@@ -0,0 +1,10 @@
+import React from 'react';
+import ReactDOM from 'react-dom';
+import App from './component/App';
+import './index.css';
+import 'github-fork-ribbon-css/gh-fork-ribbon.css';
+
+ReactDOM.render(
+  <App />,
+  document.getElementById('root')
+);

lexers/testdata/jsx.expected

@@ -0,0 +1,60 @@
+[
+  {"type":"Keyword","value":"import"},
+  {"type":"Text","value":" "},
+  {"type":"NameOther","value":"React"},
+  {"type":"Text","value":" "},
+  {"type":"NameOther","value":"from"},
+  {"type":"Text","value":" "},
+  {"type":"LiteralStringSingle","value":"'react'"},
+  {"type":"Punctuation","value":";"},
+  {"type":"Text","value":"\n"},
+  {"type":"Keyword","value":"import"},
+  {"type":"Text","value":" "},
+  {"type":"NameOther","value":"ReactDOM"},
+  {"type":"Text","value":" "},
+  {"type":"NameOther","value":"from"},
+  {"type":"Text","value":" "},
+  {"type":"LiteralStringSingle","value":"'react-dom'"},
+  {"type":"Punctuation","value":";"},
+  {"type":"Text","value":"\n"},
+  {"type":"Keyword","value":"import"},
+  {"type":"Text","value":" "},
+  {"type":"NameOther","value":"App"},
+  {"type":"Text","value":" "},
+  {"type":"NameOther","value":"from"},
+  {"type":"Text","value":" "},
+  {"type":"LiteralStringSingle","value":"'./component/App'"},
+  {"type":"Punctuation","value":";"},
+  {"type":"Text","value":"\n"},
+  {"type":"Keyword","value":"import"},
+  {"type":"Text","value":" "},
+  {"type":"LiteralStringSingle","value":"'./index.css'"},
+  {"type":"Punctuation","value":";"},
+  {"type":"Text","value":"\n"},
+  {"type":"Keyword","value":"import"},
+  {"type":"Text","value":" "},
+  {"type":"LiteralStringSingle","value":"'github-fork-ribbon-css/gh-fork-ribbon.css'"},
+  {"type":"Punctuation","value":";"},
+  {"type":"Text","value":"\n\n"},
+  {"type":"NameOther","value":"ReactDOM"},
+  {"type":"Punctuation","value":"."},
+  {"type":"NameOther","value":"render"},
+  {"type":"Punctuation","value":"("},
+  {"type":"Text","value":"\n  "},
+  {"type":"Punctuation","value":"\u003c"},
+  {"type":"NameTag","value":"App"},
+  {"type":"Text","value":" "},
+  {"type":"Punctuation","value":"/"},
+  {"type":"Text","value":""},
+  {"type":"Punctuation","value":"\u003e,"},
+  {"type":"Text","value":"\n  "},
+  {"type":"NameBuiltin","value":"document"},
+  {"type":"Punctuation","value":"."},
+  {"type":"NameOther","value":"getElementById"},
+  {"type":"Punctuation","value":"("},
+  {"type":"LiteralStringSingle","value":"'root'"},
+  {"type":"Punctuation","value":")"},
+  {"type":"Text","value":"\n"},
+  {"type":"Punctuation","value":");"},
+  {"type":"Text","value":"\n"}
+]

regexp.go

@@ -87,6 +87,15 @@ func Tokenise(lexer Lexer, options *TokeniseOptions, text string) ([]*Token, err
 // Rules maps from state to a sequence of Rules.
 type Rules map[string][]Rule
 
+func (r Rules) Clone() Rules {
+	out := map[string][]Rule{}
+	for key, rules := range r {
+		out[key] = make([]Rule, len(rules))
+		copy(out[key], rules)
+	}
+	return out
+}
+
 // MustNewLexer creates a new Lexer or panics.
 func MustNewLexer(config *Config, rules Rules) *RegexLexer {
 	lexer, err := NewLexer(config, rules)