1 // A framework for simple tokenizers. Takes care of newlines and
2 // white-space, and of getting the text from the source stream into
3 // the token object. A state is a function of two arguments -- a
4 // string stream and a setState function. The second can be used to
5 // change the tokenizer's state, and can be ignored for stateless
6 // tokenizers. This function should advance the stream over a token
7 // and return a string or object containing information about the next
8 // token, or null to pass and have the (new) state be called to finish
9 // the token. When a string is given, it is wrapped in a {style, type}
10 // object. In the resulting object, the characters consumed are stored
11 // under the content property. Any whitespace following them is also
12 // automatically consumed, and added to the value property. (Thus,
13 // content is the actual meaningful part of the token, while value
14 // contains all the text it spans.)
16 function tokenizer(source, state) {
17 // Newlines are always a separate token.
18 function isWhiteSpace(ch) {
19 // The messy regexp is because IE's regexp matcher is of the
20 // opinion that non-breaking spaces are no whitespace.
21 return ch != "\n" && /^[\s\u00a0]*$/.test(ch);
27 take: function(type) {
28 if (typeof(type) == "string")
29 type = {style: type, type: type};
31 type.content = (type.content || "") + source.get();
32 if (!/\n$/.test(type.content))
33 source.nextWhile(isWhiteSpace);
34 type.value = type.content + source.get();
39 if (!source.more()) throw StopIteration;
42 if (source.equals("\n")) {
44 return this.take("whitespace");
47 if (source.applies(isWhiteSpace))
51 type = this.state(source, function(s) {tokenizer.state = s;});
53 return this.take(type);