1 | 'use strict';
|
2 | var sax = require('sax'),
|
3 | fs = require('fs'),
|
4 | util = require('util'),
|
5 | events = require('events')
|
6 | ;
|
7 | function within(arr1, arr2) {
|
8 | for (var i = 0; i < arr1.length; i++) {
|
9 | if (arr1.indexOf(i) !== arr2.indexOf(i)) {
|
10 | return false
|
11 | }
|
12 | }
|
13 | return true
|
14 | }
|
15 |
|
16 | function string2path(s)
|
17 | {
|
18 | if (typeof s === 'string') {
|
19 | var k = {}
|
20 | k.v = s.toLowerCase()
|
21 | k.a = k.v.split('/')
|
22 | if (k.a[0] === '') {
|
23 | k.a.shift()
|
24 | }
|
25 | return k
|
26 | }
|
27 | }
|
28 |
|
29 |
|
30 | function XMLSplitter(knife) {
|
31 |
|
32 | if (!(this instanceof XMLSplitter)) {
|
33 | return new XMLSplitter()
|
34 | }
|
35 |
|
36 | events.EventEmitter.call(this)
|
37 |
|
38 |
|
39 | var self = this
|
40 | var tagname = 'row'
|
41 |
|
42 |
|
43 | self.result = {}
|
44 | self.stack = []
|
45 | self.path = []
|
46 | self.cdata = ''
|
47 | self.knifes = []
|
48 | self.counter = 0
|
49 | self.tracker = false
|
50 | self.stream = require("sax").createStream(true, {trim:true})
|
51 |
|
52 |
|
53 | if (Array.isArray(knife)) {
|
54 | knife.forEach(function (item) {
|
55 | self.knifes.push(string2path(item))
|
56 | }
|
57 | )
|
58 | }
|
59 | else {
|
60 | self.knifes.push(string2path(knife))
|
61 | }
|
62 | if (self.knifes.length === 0) {
|
63 | throw new Error('Invalid Parameter')
|
64 | }
|
65 |
|
66 |
|
67 |
|
68 |
|
69 | self.stream.onerror = function (e) {
|
70 |
|
71 | }
|
72 | self.stream.onprocessinginstruction = function (pi) {
|
73 | }
|
74 | self.stream.ontext = function (v) {
|
75 | if (self.tracker === false) {
|
76 | return
|
77 | }
|
78 | self.cvalue('$t', v)
|
79 | }
|
80 | self.stream.oncomment = function (v) {
|
81 | if (self.tracker === false) {
|
82 | return
|
83 | }
|
84 | self.cvalue('$c', v)
|
85 | }
|
86 | self.stream.oncdata = function (v) {
|
87 | if (self.tracker === false) {
|
88 | return
|
89 | }
|
90 | self.cdata += v
|
91 | }
|
92 | self.stream.onopencdata = function () {
|
93 | if (self.tracker === false) {
|
94 | return
|
95 | }
|
96 | self.cdata = ''
|
97 | }
|
98 | self.stream.onclosecdata = function () {
|
99 | if (self.tracker === false) {
|
100 | return
|
101 | }
|
102 | self.cvalue('$cd', self.cdata)
|
103 | self.cdata = ''
|
104 | }
|
105 | self.stream.onopentag = function (node) {
|
106 | self.path.push(node.name.toLowerCase())
|
107 |
|
108 |
|
109 |
|
110 | var t1 = self.knifes.reduce(function (prev, cur) { return (prev === true || self.path.length >= cur.a.length) }, false)
|
111 | var t2 = self.knifes.reduce(function (prev, cur) { return (prev === true || within(cur, self.path)) }, false)
|
112 |
|
113 | if (t1 === false || t2 === false) {
|
114 | self.tracker = false
|
115 | return
|
116 | }
|
117 |
|
118 | self.tracker = true
|
119 |
|
120 | if (self.stack.length === 1 && node.name === tagname) {
|
121 | self.result = self.cattr(node.attributes);
|
122 | self.stack.push(this.result);
|
123 | }
|
124 | else {
|
125 | self.stack.push(self.cvalue(node.name, self.cattr(node.attributes)))
|
126 | }
|
127 | }
|
128 | self.stream.onclosetag = function (tag) {
|
129 | var l = self.stack.pop()
|
130 | var p = self.path.reduce(function (prev, cur) { return prev + '/' + cur }, '')
|
131 |
|
132 | if (self.knifes.some(function checkPath(x) { return (x.v === p) })) {
|
133 | self.emit('data', l, tag.replace(':', '$'), p)
|
134 | self.counter++
|
135 | self.stack = [{}]
|
136 | }
|
137 | self.path.pop()
|
138 | if (self.path.length === 0) {
|
139 | self.emit('end', self.counter)
|
140 | }
|
141 | }
|
142 | self.stream.onready = function () {
|
143 | var l = self.stack.pop()
|
144 | }
|
145 | self.stream.onend = function () {
|
146 | }
|
147 | }
|
148 | util.inherits(XMLSplitter, events.EventEmitter)
|
149 |
|
150 | XMLSplitter.prototype.parseString = function (string, encoding) {
|
151 | var self = this
|
152 | self.stream.end(string, encoding || 'utf8')
|
153 | }
|
154 |
|
155 | XMLSplitter.prototype.parseStream = function (stream) {
|
156 | var self = this;
|
157 | stream.pipe(self.stream);
|
158 | }
|
159 |
|
160 | XMLSplitter.prototype.cvalue = function (n, v) {
|
161 | var self = this
|
162 | n = n.replace(':', '$')
|
163 | var o = self.stack[self.stack.length - 1]
|
164 | if (o == undefined) {
|
165 | o = {}
|
166 | o[n] = v
|
167 | return o[n]
|
168 | }
|
169 | else if (o[n] == undefined) {
|
170 | o[n] = v
|
171 | return o[n]
|
172 | }
|
173 | else if (!Array.isArray(o[n])) {
|
174 | var x = o[n]
|
175 | o[n] = new Array(x, v)
|
176 | return o[n][1]
|
177 | }
|
178 | else {
|
179 | var i = o[n].push(v)
|
180 | return o[n][i - 1]
|
181 | }
|
182 | }
|
183 |
|
184 | XMLSplitter.prototype.cattr = function (o) {
|
185 | var self = this
|
186 | var r = {};
|
187 | for (var key in o) {
|
188 | if (o.hasOwnProperty(key) && o[key]) {
|
189 | r[key.replace(':', '$')] = o[key];
|
190 | }
|
191 | }
|
192 | return r;
|
193 | }
|
194 |
|
195 | module.exports = XMLSplitter
|