1 | var options = {};
|
2 |
|
3 | function setup(opt) {
|
4 | options = {
|
5 | 'attr-to-remove': [
|
6 | 'align',
|
7 | 'valign',
|
8 | 'bgcolor',
|
9 | 'color',
|
10 | 'width',
|
11 | 'height',
|
12 | 'border',
|
13 | 'cellpadding',
|
14 | 'cellspacing'
|
15 | ],
|
16 | 'block-tags': [
|
17 | 'div',
|
18 | 'p',
|
19 | 'table',
|
20 | 'tr',
|
21 | 'td',
|
22 | 'blockquote',
|
23 | 'hr'
|
24 | ],
|
25 | 'break-after-br': true,
|
26 | 'close-empty-tags': false,
|
27 | 'empty-tags': [
|
28 | 'br',
|
29 | 'hr',
|
30 | 'img'
|
31 | ],
|
32 | 'indent': ' ',
|
33 | 'pretty': true,
|
34 | 'remove-comments': false,
|
35 | 'tags-to-remove': [
|
36 | 'font'
|
37 | ]
|
38 | };
|
39 |
|
40 | if (!opt) {
|
41 | return;
|
42 | }
|
43 |
|
44 | options['attr-to-remove'] = opt['attr-to-remove'] || options['attr-to-remove'];
|
45 | options['block-tags'] = opt['block-tags'] || options['block-tags'];
|
46 | options['break-after-br'] = opt['break-after-br'] === false ? false : true;
|
47 | options['close-empty-tags'] = opt['close-empty-tags'] === true ? true : false;
|
48 | options['empty-tags'] = opt['empty-tags'] || options['empty-tags'];
|
49 | options['indent'] = opt['indent'] || options['indent'];
|
50 | options['pretty'] = opt['pretty'] === false ? false : true;
|
51 | options['remove-comments'] = opt['remove-comments'] === true ? true : false;
|
52 | options['tags-to-remove'] = opt['tags-to-remove'] || options['tags-to-remove'];
|
53 |
|
54 | if (opt['add-attr-to-remove']) {
|
55 | options['attr-to-remove'] = options['attr-to-remove'].concat(opt['add-attr-to-remove']);
|
56 | }
|
57 |
|
58 | if (opt['add-block-tags']) {
|
59 | options['block-tags'] = options['block-tags'].concat(opt['add-block-tags']);
|
60 | }
|
61 |
|
62 | if (opt['add-empty-tags']) {
|
63 | options['empty-tags'] = options['empty-tags'].concat(opt['add-empty-tags']);
|
64 | }
|
65 |
|
66 | if (opt['add-tags-to-remove']) {
|
67 | options['tags-to-remove'] = options['tags-to-remove'].concat(opt['add-tags-to-remove']);
|
68 | }
|
69 | }
|
70 |
|
71 | function replaceWhiteSpace(html) {
|
72 | return html.replace(/\s/g, ' ');
|
73 | }
|
74 |
|
75 | function removeExtraSpaces(html) {
|
76 | return html.replace(/ {2,}/g, ' ');
|
77 | }
|
78 |
|
79 | function closeEmptyTag(tag) {
|
80 | return tag.replace(/ ?\/?>/, '/>');
|
81 | }
|
82 |
|
83 | function removeTrailingSlash(tag) {
|
84 | return tag.replace(/ ?\/>/, '>');
|
85 | }
|
86 |
|
87 | function cleanAttributes(tag) {
|
88 | return tag.replace(/ (\w+)=['"].+?['"]/g, function (attribute, attributeName) {
|
89 | if (options['attr-to-remove'].indexOf(attributeName) > -1) {
|
90 | return '';
|
91 | }
|
92 |
|
93 | return attribute;
|
94 | });
|
95 | }
|
96 |
|
97 | function cleanTags(html) {
|
98 | return html.replace(/<\/?(\w+).*?>/g, function (tag, tagName) {
|
99 | tag = tag.toLowerCase();
|
100 | tagName = tagName.toLowerCase();
|
101 |
|
102 | if (options['tags-to-remove'].indexOf(tagName) > -1) {
|
103 | return '';
|
104 | }
|
105 |
|
106 | if (options['empty-tags'].indexOf(tagName) > -1) {
|
107 | if (options['close-empty-tags']) {
|
108 | tag = closeEmptyTag(tag);
|
109 | } else {
|
110 | tag = removeTrailingSlash(tag);
|
111 | }
|
112 | }
|
113 |
|
114 | tag = cleanAttributes(tag);
|
115 |
|
116 | return tag;
|
117 | });
|
118 | }
|
119 |
|
120 | function removeComments(html) {
|
121 | return html.replace(/<!--.*?-->/g, '');
|
122 | }
|
123 |
|
124 | function addLineBreaks(html) {
|
125 | return html.replace(/<\/?(\w+).*?>/g, function (tag, tagName) {
|
126 | if (options['block-tags'].indexOf(tagName) > -1) {
|
127 | return '\n' + tag + '\n';
|
128 | }
|
129 |
|
130 | if (tagName == 'br' && options['break-after-br']) {
|
131 | return tag + '\n';
|
132 | }
|
133 |
|
134 | return tag;
|
135 | });
|
136 | }
|
137 |
|
138 | function removeBlankLines(html) {
|
139 | return html.replace(/\s{2,}/g, '\n');
|
140 | }
|
141 |
|
142 | function indentLine(line, indentLevel) {
|
143 | var indent = '';
|
144 |
|
145 | for (var i = 0; i < indentLevel; i++) {
|
146 | indent += options['indent'];
|
147 | }
|
148 |
|
149 | return indent + line;
|
150 | }
|
151 |
|
152 | function indent(html) {
|
153 | var indentLevel = 0;
|
154 |
|
155 | return html.replace(/.*\n/g, function (line) {
|
156 | var match = line.match(/<\/?(\w+).*?>/);
|
157 |
|
158 | if (!match) {
|
159 | return indentLine(line, indentLevel);
|
160 | }
|
161 |
|
162 | var tag = match[0],
|
163 | tagName = match[1];
|
164 |
|
165 | if (options['block-tags'].indexOf(tagName) > -1) {
|
166 | if (tag.indexOf('</') === 0) {
|
167 | indentLevel--;
|
168 | line = indentLine(line, indentLevel);
|
169 | } else {
|
170 | line = indentLine(line, indentLevel);
|
171 | indentLevel++;
|
172 | }
|
173 |
|
174 | return line;
|
175 | }
|
176 |
|
177 | return indentLine(line, indentLevel);
|
178 | });
|
179 | }
|
180 |
|
181 | function clean(html, opt) {
|
182 | setup(opt);
|
183 |
|
184 | html = replaceWhiteSpace(html);
|
185 | html = removeExtraSpaces(html);
|
186 | html = cleanTags(html);
|
187 |
|
188 | if (options['remove-comments']) {
|
189 | html = removeComments(html);
|
190 | }
|
191 |
|
192 | if (options['pretty']) {
|
193 | html = addLineBreaks(html);
|
194 | html = removeBlankLines(html);
|
195 | html = indent(html);
|
196 | }
|
197 |
|
198 | return html.trim();
|
199 | }
|
200 |
|
201 | module.exports = {
|
202 | clean: clean
|
203 | };
|