1 | 'use strict';
|
2 |
|
3 | var htmlparser = require('htmlparser2');
|
4 | var isObject = require('isobject');
|
5 |
|
6 |
|
7 |
|
8 |
|
9 | var defaultOptions = {lowerCaseTags: false, lowerCaseAttributeNames: false};
|
10 |
|
11 | var defaultDirectives = [{name: '!doctype', start: '<', end: '>'}];
|
12 |
|
13 |
|
14 |
|
15 |
|
16 |
|
17 |
|
18 |
|
19 | function postHTMLParser(html, options) {
|
20 | var bufArray = [],
|
21 | results = [];
|
22 |
|
23 | bufArray.last = function() {
|
24 | return this[this.length - 1];
|
25 | };
|
26 |
|
27 | function parserDirective(name, data) {
|
28 | var directives = options.directives || defaultDirectives;
|
29 | var last = bufArray.last();
|
30 |
|
31 | for (var i = 0; i < directives.length; i++) {
|
32 | var directive = directives[i];
|
33 | var directiveText = directive.start + data + directive.end;
|
34 |
|
35 | if (name.toLowerCase() === directive.name) {
|
36 | if (!last) {
|
37 | results.push(directiveText);
|
38 | return;
|
39 | }
|
40 |
|
41 | last.content || (last.content = []);
|
42 | last.content.push(directiveText);
|
43 | }
|
44 | }
|
45 | }
|
46 |
|
47 | var parser = new htmlparser.Parser({
|
48 | onprocessinginstruction: parserDirective,
|
49 | oncomment: function(data) {
|
50 | var comment = '<!--' + data + '-->',
|
51 | last = bufArray.last();
|
52 |
|
53 | if (!last) {
|
54 | results.push(comment);
|
55 | return;
|
56 | }
|
57 |
|
58 | last.content || (last.content = []);
|
59 | last.content.push(comment);
|
60 | },
|
61 | onopentag: function(tag, attrs) {
|
62 | var buf = { tag: tag };
|
63 |
|
64 | if (Object.keys(attrs).length) {
|
65 | buf.attrs = attrs;
|
66 | }
|
67 |
|
68 | bufArray.push(buf);
|
69 | },
|
70 | onclosetag: function() {
|
71 | var buf = bufArray.pop();
|
72 |
|
73 | if (!bufArray.length) {
|
74 | results.push(buf);
|
75 | return;
|
76 | }
|
77 |
|
78 | var last = bufArray.last();
|
79 | if (!Array.isArray(last.content)) {
|
80 | last.content = [];
|
81 | }
|
82 |
|
83 | last.content.push(buf);
|
84 | },
|
85 | ontext: function(text) {
|
86 | var last = bufArray.last();
|
87 | if (!last) {
|
88 | results.push(text);
|
89 | return;
|
90 | }
|
91 |
|
92 | last.content || (last.content = []);
|
93 | last.content.push(text);
|
94 | }
|
95 | }, options || defaultOptions);
|
96 |
|
97 | parser.write(html);
|
98 | parser.end();
|
99 |
|
100 | return results;
|
101 | }
|
102 |
|
103 | function parserWrapper() {
|
104 | var option;
|
105 |
|
106 | function parser(html) {
|
107 | var opt = option || defaultOptions;
|
108 | return postHTMLParser(html, opt);
|
109 | }
|
110 |
|
111 | if (arguments.length === 1 && isObject(arguments[0])) {
|
112 | option = arguments[0];
|
113 | return parser;
|
114 | }
|
115 |
|
116 | option = arguments[1];
|
117 | return parser(arguments[0]);
|
118 | }
|
119 |
|
120 | module.exports = parserWrapper;
|
121 | module.exports.defaultOptions = defaultOptions;
|
122 | module.exports.defaultDirectives = defaultDirectives;
|