0
|
1 /* Tokenizer for JavaScript code */
|
|
2
|
|
3 var tokenizeJavaScript = (function() {
|
|
4 // Advance the stream until the given character (not preceded by a
|
|
5 // backslash) is encountered, or the end of the line is reached.
|
|
6 function nextUntilUnescaped(source, end) {
|
|
7 var escaped = false;
|
|
8 while (!source.endOfLine()) {
|
|
9 var next = source.next();
|
|
10 if (next == end && !escaped)
|
|
11 return false;
|
|
12 escaped = !escaped && next == "\\";
|
|
13 }
|
|
14 return escaped;
|
|
15 }
|
|
16
|
|
17 // A map of JavaScript's keywords. The a/b/c keyword distinction is
|
|
18 // very rough, but it gives the parser enough information to parse
|
|
19 // correct code correctly (we don't care that much how we parse
|
|
20 // incorrect code). The style information included in these objects
|
|
21 // is used by the highlighter to pick the correct CSS style for a
|
|
22 // token.
|
|
23 var keywords = function(){
|
|
24 function result(type, style){
|
|
25 return {type: type, style: "js-" + style};
|
|
26 }
|
|
27 // keywords that take a parenthised expression, and then a
|
|
28 // statement (if)
|
|
29 var keywordA = result("keyword a", "keyword");
|
|
30 // keywords that take just a statement (else)
|
|
31 var keywordB = result("keyword b", "keyword");
|
|
32 // keywords that optionally take an expression, and form a
|
|
33 // statement (return)
|
|
34 var keywordC = result("keyword c", "keyword");
|
|
35 var operator = result("operator", "keyword");
|
|
36 var atom = result("atom", "atom");
|
|
37 return {
|
|
38 "if": keywordA, "while": keywordA, "with": keywordA,
|
|
39 "else": keywordB, "do": keywordB, "try": keywordB, "finally": keywordB,
|
|
40 "return": keywordC, "break": keywordC, "continue": keywordC, "new": keywordC, "delete": keywordC, "throw": keywordC,
|
|
41 "in": operator, "typeof": operator, "instanceof": operator,
|
|
42 "var": result("var", "keyword"), "function": result("function", "keyword"), "catch": result("catch", "keyword"),
|
|
43 "for": result("for", "keyword"), "switch": result("switch", "keyword"),
|
|
44 "case": result("case", "keyword"), "default": result("default", "keyword"),
|
|
45 "true": atom, "false": atom, "null": atom, "undefined": atom, "NaN": atom, "Infinity": atom
|
|
46 };
|
|
47 }();
|
|
48
|
|
49 // Some helper regexps
|
|
50 var isOperatorChar = /[+\-*&%=<>!?|]/;
|
|
51 var isHexDigit = /[0-9A-Fa-f]/;
|
|
52 var isWordChar = /[\w\$_]/;
|
|
53
|
|
54 // Wrapper around jsToken that helps maintain parser state (whether
|
|
55 // we are inside of a multi-line comment and whether the next token
|
|
56 // could be a regular expression).
|
|
57 function jsTokenState(inside, regexp) {
|
|
58 return function(source, setState) {
|
|
59 var newInside = inside;
|
|
60 var type = jsToken(inside, regexp, source, function(c) {newInside = c;});
|
|
61 var newRegexp = type.type == "operator" || type.type == "keyword c" || type.type.match(/^[\[{}\(,;:]$/);
|
|
62 if (newRegexp != regexp || newInside != inside)
|
|
63 setState(jsTokenState(newInside, newRegexp));
|
|
64 return type;
|
|
65 };
|
|
66 }
|
|
67
|
|
68 // The token reader, intended to be used by the tokenizer from
|
|
69 // tokenize.js (through jsTokenState). Advances the source stream
|
|
70 // over a token, and returns an object containing the type and style
|
|
71 // of that token.
|
|
72 function jsToken(inside, regexp, source, setInside) {
|
|
73 function readHexNumber(){
|
|
74 source.next(); // skip the 'x'
|
|
75 source.nextWhileMatches(isHexDigit);
|
|
76 return {type: "number", style: "js-atom"};
|
|
77 }
|
|
78
|
|
79 function readNumber() {
|
|
80 source.nextWhileMatches(/[0-9]/);
|
|
81 if (source.equals(".")){
|
|
82 source.next();
|
|
83 source.nextWhileMatches(/[0-9]/);
|
|
84 }
|
|
85 if (source.equals("e") || source.equals("E")){
|
|
86 source.next();
|
|
87 if (source.equals("-"))
|
|
88 source.next();
|
|
89 source.nextWhileMatches(/[0-9]/);
|
|
90 }
|
|
91 return {type: "number", style: "js-atom"};
|
|
92 }
|
|
93 // Read a word, look it up in keywords. If not found, it is a
|
|
94 // variable, otherwise it is a keyword of the type found.
|
|
95 function readWord() {
|
|
96 source.nextWhileMatches(isWordChar);
|
|
97 var word = source.get();
|
|
98 var known = keywords.hasOwnProperty(word) && keywords.propertyIsEnumerable(word) && keywords[word];
|
|
99 return known ? {type: known.type, style: known.style, content: word} :
|
|
100 {type: "variable", style: "js-variable", content: word};
|
|
101 }
|
|
102 function readRegexp() {
|
|
103 nextUntilUnescaped(source, "/");
|
|
104 source.nextWhileMatches(/[gimy]/); // 'y' is "sticky" option in Mozilla
|
|
105 return {type: "regexp", style: "js-string"};
|
|
106 }
|
|
107 // Mutli-line comments are tricky. We want to return the newlines
|
|
108 // embedded in them as regular newline tokens, and then continue
|
|
109 // returning a comment token for every line of the comment. So
|
|
110 // some state has to be saved (inside) to indicate whether we are
|
|
111 // inside a /* */ sequence.
|
|
112 function readMultilineComment(start){
|
|
113 var newInside = "/*";
|
|
114 var maybeEnd = (start == "*");
|
|
115 while (true) {
|
|
116 if (source.endOfLine())
|
|
117 break;
|
|
118 var next = source.next();
|
|
119 if (next == "/" && maybeEnd){
|
|
120 newInside = null;
|
|
121 break;
|
|
122 }
|
|
123 maybeEnd = (next == "*");
|
|
124 }
|
|
125 setInside(newInside);
|
|
126 return {type: "comment", style: "js-comment"};
|
|
127 }
|
|
128 function readOperator() {
|
|
129 source.nextWhileMatches(isOperatorChar);
|
|
130 return {type: "operator", style: "js-operator"};
|
|
131 }
|
|
132 function readString(quote) {
|
|
133 var endBackSlash = nextUntilUnescaped(source, quote);
|
|
134 setInside(endBackSlash ? quote : null);
|
|
135 return {type: "string", style: "js-string"};
|
|
136 }
|
|
137
|
|
138 // Fetch the next token. Dispatches on first character in the
|
|
139 // stream, or first two characters when the first is a slash.
|
|
140 if (inside == "\"" || inside == "'")
|
|
141 return readString(inside);
|
|
142 var ch = source.next();
|
|
143 if (inside == "/*")
|
|
144 return readMultilineComment(ch);
|
|
145 else if (ch == "\"" || ch == "'")
|
|
146 return readString(ch);
|
|
147 // with punctuation, the type of the token is the symbol itself
|
|
148 else if (/[\[\]{}\(\),;\:\.]/.test(ch))
|
|
149 return {type: ch, style: "js-punctuation"};
|
|
150 else if (ch == "0" && (source.equals("x") || source.equals("X")))
|
|
151 return readHexNumber();
|
|
152 else if (/[0-9]/.test(ch))
|
|
153 return readNumber();
|
|
154 else if (ch == "/"){
|
|
155 if (source.equals("*"))
|
|
156 { source.next(); return readMultilineComment(ch); }
|
|
157 else if (source.equals("/"))
|
|
158 { nextUntilUnescaped(source, null); return {type: "comment", style: "js-comment"};}
|
|
159 else if (regexp)
|
|
160 return readRegexp();
|
|
161 else
|
|
162 return readOperator();
|
|
163 }
|
|
164 else if (isOperatorChar.test(ch))
|
|
165 return readOperator();
|
|
166 else
|
|
167 return readWord();
|
|
168 }
|
|
169
|
|
170 // The external interface to the tokenizer.
|
|
171 return function(source, startState) {
|
|
172 return tokenizer(source, startState || jsTokenState(false, true));
|
|
173 };
|
|
174 })();
|