1 | var q = require('q')
|
2 | var isCss = require('is-css')
|
3 | var isPresent = require('is-present')
|
4 | var isBlank = require('is-blank')
|
5 | var isUrl = require('is-url-superb')
|
6 | var request = require('requestretry')
|
7 | var cheerio = require('cheerio')
|
8 | var normalizeUrl = require('normalize-url')
|
9 | var stripHtmlComments = require('strip-html-comments')
|
10 | var stripWaybackToolbar = require('strip-wayback-toolbar')
|
11 | var resolveCssImportUrls = require('resolve-css-import-urls')
|
12 | var ua = require('ua-string')
|
13 |
|
14 | var getLinkContents = require('./utils/get-link-contents')
|
15 | var createLink = require('./utils/create-link')
|
16 |
|
17 | module.exports = function(url, options, html) {
|
18 | var deferred = q.defer()
|
19 | var options = options || {}
|
20 | options.headers = options.headers || {}
|
21 | options.headers['User-Agent'] = options.headers['User-Agent'] || ua
|
22 | options.timeout = options.timeout || 5000
|
23 | options.stripWayback = options.stripWayback || false
|
24 | options.gzip = true
|
25 |
|
26 | if (typeof url !== 'string' || isBlank(url) || !isUrl(url)) {
|
27 | throw new TypeError('get-css expected a url as a string')
|
28 | }
|
29 |
|
30 | url = normalizeUrl(url, { stripWWW: false })
|
31 | options.url = url
|
32 |
|
33 | if (options.ignoreCerts) {
|
34 | process.env.NODE_TLS_REJECT_UNAUTHORIZED = '0'
|
35 | }
|
36 |
|
37 | var status = {
|
38 | parsed: 0,
|
39 | total: 0
|
40 | }
|
41 |
|
42 | var result = {
|
43 | links: [],
|
44 | styles: [],
|
45 | css: ''
|
46 | }
|
47 |
|
48 | function handleResolve() {
|
49 | if (status.parsed >= status.total) {
|
50 | deferred.resolve(result)
|
51 | }
|
52 | }
|
53 |
|
54 | function parseHtml(html) {
|
55 | if (options.stripWayback) {
|
56 | html = stripWaybackToolbar(html)
|
57 | }
|
58 | var $ = cheerio.load(html)
|
59 | result.pageTitle = $('head > title').text()
|
60 | result.html = html
|
61 |
|
62 | $('[rel=stylesheet]').each(function() {
|
63 | var link = $(this).attr('href')
|
64 | if (isPresent(link)) {
|
65 | result.links.push(createLink(link, url))
|
66 | } else {
|
67 | result.styles.push(stripHtmlComments($(this).html()))
|
68 | }
|
69 | })
|
70 |
|
71 | $('style').each(function() {
|
72 | result.styles.push(stripHtmlComments($(this).html()))
|
73 | })
|
74 |
|
75 | status.total = result.links.length + result.styles.length
|
76 | if (!status.total) {
|
77 | handleResolve()
|
78 | }
|
79 |
|
80 | result.links.forEach(function(link) {
|
81 | getLinkContents(link.url, options)
|
82 | .then(function(css) {
|
83 | handleCssFromLink(link, css)
|
84 | })
|
85 | .catch(function(error) {
|
86 | link.error = error
|
87 | status.parsed++
|
88 | handleResolve()
|
89 | })
|
90 | })
|
91 |
|
92 | result.styles.forEach(function(css) {
|
93 | result.css += css
|
94 | status.parsed++
|
95 | handleResolve()
|
96 | })
|
97 | }
|
98 |
|
99 | function handleCssFromLink(link, css) {
|
100 | link.css += css
|
101 |
|
102 | parseCssForImports(link, css)
|
103 |
|
104 | status.parsed++
|
105 | handleResolve()
|
106 | }
|
107 |
|
108 |
|
109 | function parseCssForImports(link, css) {
|
110 | link.imports = resolveCssImportUrls(link.url, css)
|
111 | status.total += link.imports.length
|
112 | result.css += css
|
113 |
|
114 | link.imports.forEach(function(importUrl) {
|
115 | var importLink = createLink(importUrl, importUrl)
|
116 | result.links.push(importLink)
|
117 |
|
118 | getLinkContents(importLink.url, options)
|
119 | .then(function(css) {
|
120 | handleCssFromLink(importLink, css)
|
121 | })
|
122 | .catch(function(error) {
|
123 | link.error = error
|
124 | status.parsed++
|
125 | handleResolve()
|
126 | })
|
127 | })
|
128 | }
|
129 |
|
130 | function handleBody(body) {
|
131 | if (isCss(url)) {
|
132 | var link = createLink(url, url)
|
133 | result.links.push(link)
|
134 | handleCssFromLink(link, body)
|
135 | } else {
|
136 | parseHtml(body)
|
137 | }
|
138 | }
|
139 |
|
140 | if (html) {
|
141 | handleBody(html)
|
142 | } else {
|
143 | request(options, function(error, response, body) {
|
144 | if (error) {
|
145 | if (options.verbose) console.log('Error from ' + url + ' ' + error)
|
146 | deferred.reject(error)
|
147 | return
|
148 | }
|
149 |
|
150 | if (response && response.statusCode != 200) {
|
151 | if (options.verbose)
|
152 | console.log('Received a ' + response.statusCode + ' from: ' + url)
|
153 | deferred.reject({ url: url, statusCode: response.code })
|
154 | return
|
155 | }
|
156 |
|
157 | handleBody(body)
|
158 | })
|
159 | }
|
160 |
|
161 | return deferred.promise
|
162 | }
|