1 |
|
2 |
|
3 |
|
4 | exports = module.exports = Pickup
|
5 |
|
6 | var StringDecoder = require('string_decoder').StringDecoder
|
7 | var assert = require('assert')
|
8 | var attribute = require('./lib/attribute')
|
9 | var mappings = require('./lib/mappings')
|
10 | var os = require('os')
|
11 | var sax = require('sax')
|
12 | var stream = require('readable-stream')
|
13 | var util = require('util')
|
14 |
|
15 | function isString(obj) {
|
16 | return typeof obj === 'string'
|
17 | }
|
18 |
|
19 | function OpenHandlers (t) {
|
20 | this.channel = t.feedopen
|
21 | this.feed = t.feedopen
|
22 | this.item = t.entryopen
|
23 | this.entry = t.entryopen
|
24 | this.image = t.imageopen
|
25 | }
|
26 |
|
27 | function CloseHandlers (t) {
|
28 | this.channel = t.feedclose
|
29 | this.feed = t.feedclose
|
30 | this.item = t.entryclose
|
31 | this.entry = t.entryclose
|
32 | this.image = t.imageclose
|
33 | }
|
34 |
|
35 | function Opts (trim, normalize, position) {
|
36 | this.trim = trim
|
37 | this.normalize = normalize
|
38 | this.position = position
|
39 | }
|
40 | var saxOpts = new Opts(true, true, false)
|
41 |
|
42 | util.inherits(Pickup, stream.Transform)
|
43 | function Pickup (opts) {
|
44 | if (!(this instanceof Pickup)) return new Pickup(opts)
|
45 | stream.Transform.call(this, opts)
|
46 | if (!Pickup.openHandlers) {
|
47 | Pickup.openHandlers = new OpenHandlers(Pickup.prototype)
|
48 | Pickup.closeHandlers = new CloseHandlers(Pickup.prototype)
|
49 | }
|
50 |
|
51 | this.decoder = new StringDecoder('utf8')
|
52 | this.eventMode = opts && opts.eventMode
|
53 | this.map = null
|
54 | this.parser = sax.parser(true, saxOpts)
|
55 | this.state = new State()
|
56 |
|
57 | var me = this
|
58 | var parser = this.parser
|
59 | parser.ontext = function (t) {
|
60 | var current = me.current()
|
61 | var map = me.map
|
62 | var state = me.state
|
63 | var name = me.state.name
|
64 | if (!current || !map) return
|
65 | var key = state.image && name === 'url' ? 'image' : map[name]
|
66 | if (!key) return
|
67 | if (state.feed && key === 'link' && !!current.link) return
|
68 | var prop = current[key]
|
69 | var add = isString(prop) && isString(t) && t !== prop
|
70 | if (prop && add) {
|
71 | current[key] += t
|
72 | } else {
|
73 | current[key] = t
|
74 | }
|
75 | }
|
76 | parser.oncdata = function (d) {
|
77 | parser.ontext(d)
|
78 | }
|
79 | function handle (name, handlers) {
|
80 | if (handlers.hasOwnProperty(name)) {
|
81 | handlers[name].apply(me)
|
82 | }
|
83 | }
|
84 | parser.onopentag = function (node) {
|
85 | var name = node.name
|
86 | me.state.name = name
|
87 | me.map = mappings[name] || me.map
|
88 | handle(name, Pickup.openHandlers)
|
89 | var current = me.current()
|
90 | if (current) {
|
91 | var key = me.map[name]
|
92 | if (key) {
|
93 | var attributes = node.attributes
|
94 | var keys = Object.keys(attributes)
|
95 | if (keys.length) {
|
96 | var kv = attribute(key, attributes)
|
97 | if (kv) current[kv[0]] = kv[1]
|
98 | }
|
99 | }
|
100 | }
|
101 | }
|
102 | parser.onclosetag = function (name) {
|
103 | handle(name, Pickup.closeHandlers)
|
104 | me.state.name = null
|
105 | }
|
106 | }
|
107 |
|
108 | Pickup.prototype.current = function () {
|
109 | return this.state.entry || this.state.feed
|
110 | }
|
111 |
|
112 | Pickup.prototype.objectMode = function () {
|
113 | return this._readableState.objectMode
|
114 | }
|
115 |
|
116 | Pickup.prototype.feedopen = function () {
|
117 | this.state.feed = new Feed()
|
118 | }
|
119 |
|
120 | Pickup.prototype.entryopen = function () {
|
121 | this.state.entry = new Entry()
|
122 | }
|
123 |
|
124 | Pickup.prototype.imageopen = function () {
|
125 | this.state.image = true
|
126 | }
|
127 |
|
128 | Pickup.prototype.entryclose = function () {
|
129 | var entry = this.state.entry
|
130 | if (!this.eventMode) {
|
131 | if (this.objectMode()) {
|
132 | this.push(entry)
|
133 | } else {
|
134 | this.push(JSON.stringify(entry) + os.EOL)
|
135 | }
|
136 | } else {
|
137 | this.emit('entry', entry)
|
138 | }
|
139 | this.state.entry = null
|
140 | }
|
141 |
|
142 | Pickup.prototype.feedclose = function () {
|
143 | var feed = this.state.feed
|
144 | if (!this.eventMode) {
|
145 | if (this.objectMode()) {
|
146 | this.push(feed)
|
147 | } else {
|
148 | this.push(JSON.stringify(feed) + os.EOL)
|
149 | }
|
150 | } else {
|
151 | this.emit('feed', feed)
|
152 | }
|
153 | this.state.feed = null
|
154 | }
|
155 |
|
156 | Pickup.prototype.imageclose = function () {
|
157 | this.state.image = false
|
158 | }
|
159 |
|
160 | function free (parser) {
|
161 | ['ontext', 'oncdata', 'onopentag', 'onclosetag']
|
162 | .forEach(function (p) {
|
163 | delete parser[p]
|
164 | })
|
165 | }
|
166 |
|
167 | Pickup.prototype._flush = function (cb) {
|
168 | free(this.parser)
|
169 | this.parser.close()
|
170 | this.parser = null
|
171 | this.decoder = null
|
172 | this.map = null
|
173 | this.state.deinit()
|
174 | this.state = null
|
175 | cb()
|
176 | }
|
177 |
|
178 | Pickup.prototype._transform = function (chunk, enc, cb) {
|
179 | var str = this.decoder.write(chunk)
|
180 | var er = this.parser.write(str).error
|
181 | this.parser.error = null
|
182 | cb(er)
|
183 | }
|
184 |
|
185 | function Entry (
|
186 | author
|
187 | , enclosure
|
188 | , duration
|
189 | , id
|
190 | , image
|
191 | , link
|
192 | , subtitle
|
193 | , summary
|
194 | , title
|
195 | , updated) {
|
196 | this.author = author
|
197 | this.enclosure = enclosure
|
198 | this.duration = duration
|
199 | this.id = id
|
200 | this.image = image
|
201 | this.link = link
|
202 | this.subtitle = subtitle
|
203 | this.summary = summary
|
204 | this.title = title
|
205 | this.updated = updated
|
206 | }
|
207 |
|
208 | function Feed (
|
209 | author
|
210 | , copyright
|
211 | , id
|
212 | , image
|
213 | , language
|
214 | , link
|
215 | , payment
|
216 | , subtitle
|
217 | , summary
|
218 | , title
|
219 | , ttl
|
220 | , updated) {
|
221 | this.author = author
|
222 | this.copyright = copyright
|
223 | this.id = id
|
224 | this.image = image
|
225 | this.language = language
|
226 | this.link = link
|
227 | this.payment = payment
|
228 | this.subtitle = subtitle
|
229 | this.summary = summary
|
230 | this.title = title
|
231 | this.ttl = ttl
|
232 | this.updated = updated
|
233 | }
|
234 |
|
235 | function State (entry, feed, image, name) {
|
236 | this.entry = entry
|
237 | this.feed = feed
|
238 | this.image = image
|
239 | this.name = name
|
240 | }
|
241 |
|
242 | State.prototype.deinit = function () {
|
243 | this.entry = null
|
244 | this.feed = null
|
245 | this.image = false
|
246 | this.name = undefined
|
247 | }
|
248 |
|
249 | function extend (origin, add) {
|
250 | return util._extend(origin, add || Object.create(null)) }
|
251 | function entry (obj) {
|
252 | return extend(new Entry(), obj) }
|
253 | function feed (obj) {
|
254 | return extend(new Feed(), obj) }
|
255 |
|
256 | if (process.env.NODE_TEST) {
|
257 | exports.entry = entry
|
258 | exports.feed = feed
|
259 | exports.EVENTS = [
|
260 | 'data'
|
261 | , 'drain'
|
262 | , 'readable'
|
263 | , 'end'
|
264 | , 'entry'
|
265 | , 'error'
|
266 | , 'feed'
|
267 | , 'finish'
|
268 | ]
|
269 | }
|