1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
/** |
4
|
|
|
* @package Import |
5
|
|
|
* @author Iurii Makukh <[email protected]> |
6
|
|
|
* @copyright Copyright (c) 2015, Iurii Makukh |
7
|
|
|
* @license https://www.gnu.org/licenses/gpl.html GNU/GPLv3 |
8
|
|
|
*/ |
9
|
|
|
|
10
|
|
|
namespace gplcart\modules\import\helpers; |
11
|
|
|
|
12
|
|
|
use Iterator; |
13
|
|
|
use UnexpectedValueException; |
14
|
|
|
|
15
|
|
|
/** |
16
|
|
|
* Provides methods to read CSV data |
17
|
|
|
*/ |
18
|
|
|
class Csv implements Iterator |
19
|
|
|
{ |
20
|
|
|
|
21
|
|
|
/** |
22
|
|
|
* File handler |
23
|
|
|
* @var resource |
24
|
|
|
*/ |
25
|
|
|
protected $handle; |
26
|
|
|
|
27
|
|
|
/** |
28
|
|
|
* Current CSV line |
29
|
|
|
* @var string |
30
|
|
|
*/ |
31
|
|
|
protected $line; |
32
|
|
|
|
33
|
|
|
/** |
34
|
|
|
* @var int |
35
|
|
|
*/ |
36
|
|
|
protected $key; |
37
|
|
|
|
38
|
|
|
/** |
39
|
|
|
* Current offset in bytes |
40
|
|
|
* @var integer |
41
|
|
|
*/ |
42
|
|
|
protected $position; |
43
|
|
|
|
44
|
|
|
/** |
45
|
|
|
* Total CSV file size in bytes |
46
|
|
|
* @var integer |
47
|
|
|
*/ |
48
|
|
|
protected $total; |
49
|
|
|
|
50
|
|
|
/** |
51
|
|
|
* Path to CSV file |
52
|
|
|
* @var string |
53
|
|
|
*/ |
54
|
|
|
protected $file; |
55
|
|
|
|
56
|
|
|
/** |
57
|
|
|
* Max number of rows to parse |
58
|
|
|
* @var integer |
59
|
|
|
*/ |
60
|
|
|
protected $limit; |
61
|
|
|
|
62
|
|
|
/** |
63
|
|
|
* Final offset in bytes |
64
|
|
|
* @var integer |
65
|
|
|
*/ |
66
|
|
|
protected $last = 0; |
67
|
|
|
|
68
|
|
|
/** |
69
|
|
|
* Starting offset in bytes |
70
|
|
|
* @var integer |
71
|
|
|
*/ |
72
|
|
|
protected $offset = 0; |
73
|
|
|
|
74
|
|
|
/** |
75
|
|
|
* Array of header names |
76
|
|
|
* @var array |
77
|
|
|
*/ |
78
|
|
|
protected $header = array(); |
79
|
|
|
|
80
|
|
|
/** |
81
|
|
|
* Skip or not first line |
82
|
|
|
* @var boolean |
83
|
|
|
*/ |
84
|
|
|
protected $skip_header = false; |
85
|
|
|
|
86
|
|
|
/** |
87
|
|
|
* CSV delimiter |
88
|
|
|
* @var string |
89
|
|
|
*/ |
90
|
|
|
protected $delimiter = ","; |
91
|
|
|
|
92
|
|
|
/** |
93
|
|
|
* Constructor |
94
|
|
|
*/ |
95
|
|
|
public function __construct() |
96
|
|
|
{ |
97
|
|
|
ini_set('auto_detect_line_endings', true); |
98
|
|
|
} |
99
|
|
|
|
100
|
|
|
/** |
101
|
|
|
* Closes file handler |
102
|
|
|
*/ |
103
|
|
|
public function __destruct() |
104
|
|
|
{ |
105
|
|
|
if (is_resource($this->handle)) { |
106
|
|
|
fclose($this->handle); |
107
|
|
|
} |
108
|
|
|
} |
109
|
|
|
|
110
|
|
|
/** |
111
|
|
|
* Sets file to parse |
112
|
|
|
* @param string $file |
113
|
|
|
* @param integer $filesize |
114
|
|
|
* @return $this |
115
|
|
|
*/ |
116
|
|
|
public function open($file, $filesize = null) |
117
|
|
|
{ |
118
|
|
|
$this->handle = fopen($file, 'r'); |
119
|
|
|
|
120
|
|
|
if (!is_resource($this->handle)) { |
121
|
|
|
throw new UnexpectedValueException('Failed to open CSV file'); |
122
|
|
|
} |
123
|
|
|
|
124
|
|
|
$this->file = $file; |
125
|
|
|
$this->total = isset($filesize) ? (int) $filesize : filesize($file); |
126
|
|
|
return $this; |
127
|
|
|
} |
128
|
|
|
|
129
|
|
|
/** |
130
|
|
|
* Sets max lines to parse |
131
|
|
|
* @param integer $limit |
132
|
|
|
* @return $this |
133
|
|
|
*/ |
134
|
|
|
public function setLimit($limit) |
135
|
|
|
{ |
136
|
|
|
$this->limit = (int) $limit; |
137
|
|
|
return $this; |
138
|
|
|
} |
139
|
|
|
|
140
|
|
|
/** |
141
|
|
|
* Sets separator between columns |
142
|
|
|
* @param string $character |
143
|
|
|
* @return $this |
144
|
|
|
*/ |
145
|
|
|
public function setDelimiter($character) |
146
|
|
|
{ |
147
|
|
|
$this->delimiter = $character; |
148
|
|
|
return $this; |
149
|
|
|
} |
150
|
|
|
|
151
|
|
|
/** |
152
|
|
|
* Get first line of CSV file |
153
|
|
|
* @return array |
154
|
|
|
*/ |
155
|
|
|
public function getHeader() |
156
|
|
|
{ |
157
|
|
|
$this->limit = 1; |
158
|
|
|
$header = $this->read(); |
159
|
|
|
|
160
|
|
|
return empty($header) ? array() : reset($header); |
161
|
|
|
} |
162
|
|
|
|
163
|
|
|
/** |
164
|
|
|
* Sets header (first line) |
165
|
|
|
* @param array $header |
166
|
|
|
* @return $this |
167
|
|
|
*/ |
168
|
|
|
public function setHeader(array $header) |
169
|
|
|
{ |
170
|
|
|
$this->header = $header; |
171
|
|
|
return $this; |
172
|
|
|
} |
173
|
|
|
|
174
|
|
|
/** |
175
|
|
|
* CSV reader |
176
|
|
|
* @return array |
177
|
|
|
*/ |
178
|
|
|
public function read() |
179
|
|
|
{ |
180
|
|
|
$rows = array(); |
181
|
|
|
|
182
|
|
|
$start = $this->offset ? $this->offset : $this->last; |
183
|
|
|
|
184
|
|
|
$this->key = 0; |
185
|
|
|
$this->last = 0; |
186
|
|
|
|
187
|
|
|
$parsed = 0; |
188
|
|
|
for ($this->rewind($start); $this->valid(); $this->next()) { |
189
|
|
|
|
190
|
|
|
$line = trim($this->current(), "\r\n"); |
191
|
|
|
|
192
|
|
|
if (empty($line)) { |
193
|
|
|
continue; |
194
|
|
|
} |
195
|
|
|
|
196
|
|
|
if ($this->skip_header) { |
197
|
|
|
$this->skip_header = false; |
198
|
|
|
continue; |
199
|
|
|
} |
200
|
|
|
|
201
|
|
|
$quoted = false; |
202
|
|
|
$current_index = 0; |
203
|
|
|
$current_field = ''; |
204
|
|
|
$fields = array(); |
205
|
|
|
|
206
|
|
|
while ($current_index <= strlen($line)) { |
207
|
|
|
if ($quoted) { |
208
|
|
|
$next_quote_index = strpos($line, '"', $current_index); |
209
|
|
|
|
210
|
|
|
if ($next_quote_index === false) { |
211
|
|
|
$current_field .= substr($line, $current_index); |
212
|
|
|
$this->next(); |
213
|
|
|
|
214
|
|
|
if (!$this->valid()) { |
215
|
|
|
$fields[] = $current_field; |
216
|
|
|
break; |
217
|
|
|
} |
218
|
|
|
|
219
|
|
|
$current_field .= "\n"; |
220
|
|
|
$line = trim($this->current(), "\r\n"); |
221
|
|
|
$current_index = 0; |
222
|
|
|
continue; |
223
|
|
|
} |
224
|
|
|
|
225
|
|
|
$current_field .= substr($line, $current_index, $next_quote_index - $current_index); |
226
|
|
|
|
227
|
|
|
if (isset($line[$next_quote_index + 1]) && $line[$next_quote_index + 1] === '"') { |
228
|
|
|
$current_field .= '"'; |
229
|
|
|
$current_index = $next_quote_index + 2; |
230
|
|
|
} else { |
231
|
|
|
$quoted = false; |
232
|
|
|
$current_index = $next_quote_index + 1; |
233
|
|
|
} |
234
|
|
|
} else { |
235
|
|
|
$next_quote_index = strpos($line, '"', $current_index); |
236
|
|
|
$next_delimiter_index = strpos($line, $this->delimiter, $current_index); |
237
|
|
|
|
238
|
|
|
if ($next_quote_index === false) { |
239
|
|
|
$next_index = $next_delimiter_index; |
240
|
|
|
} elseif ($next_delimiter_index === false) { |
241
|
|
|
$next_index = $next_quote_index; |
242
|
|
|
} else { |
243
|
|
|
$next_index = min($next_quote_index, $next_delimiter_index); |
244
|
|
|
} |
245
|
|
|
|
246
|
|
|
if ($next_index === false) { |
247
|
|
|
$current_field .= substr($line, $current_index); |
248
|
|
|
$fields[] = $current_field; |
249
|
|
|
break; |
250
|
|
|
} elseif ($line[$next_index] === $this->delimiter) { |
251
|
|
|
$length = ($next_index + strlen($this->delimiter) - 1) - $current_index; |
252
|
|
|
$current_field .= substr($line, $current_index, $length); |
253
|
|
|
$fields[] = $current_field; |
254
|
|
|
$current_field = ''; |
255
|
|
|
$current_index += $length + 1; |
256
|
|
|
} else { |
257
|
|
|
$quoted = true; |
258
|
|
|
$current_field .= substr($line, $current_index, $next_index - $current_index); |
259
|
|
|
$current_index = $next_index + 1; |
260
|
|
|
} |
261
|
|
|
} |
262
|
|
|
} |
263
|
|
|
|
264
|
|
|
if (!empty($this->header)) { |
265
|
|
|
$row = array(); |
266
|
|
|
foreach ($this->header as $key => $name) { |
267
|
|
|
$field = array_shift($fields); |
268
|
|
|
$row[$key] = isset($field) ? $field : ''; |
269
|
|
|
} |
270
|
|
|
} else { |
271
|
|
|
$row = $fields; |
272
|
|
|
} |
273
|
|
|
|
274
|
|
|
$rows[$this->key] = $row; |
275
|
|
|
|
276
|
|
|
$parsed++; |
277
|
|
|
|
278
|
|
|
if (!empty($this->limit) && $parsed >= $this->limit) { |
279
|
|
|
$this->last = $this->getPosition(); |
280
|
|
|
break; |
281
|
|
|
} |
282
|
|
|
|
283
|
|
|
$this->key++; |
284
|
|
|
} |
285
|
|
|
|
286
|
|
|
return $rows; |
287
|
|
|
} |
288
|
|
|
|
289
|
|
|
/** |
290
|
|
|
* Moves pointer to a certain position |
291
|
|
|
* @param integer $position |
292
|
|
|
*/ |
293
|
|
|
public function rewind($position = 0) |
294
|
|
|
{ |
295
|
|
|
if (isset($this->handle)) { |
296
|
|
|
fseek($this->handle, $position); |
297
|
|
|
$this->next(); |
298
|
|
|
} |
299
|
|
|
} |
300
|
|
|
|
301
|
|
|
/** |
302
|
|
|
* Sets current string and offset |
303
|
|
|
* @return null|integer |
304
|
|
|
*/ |
305
|
|
|
public function next() |
306
|
|
|
{ |
307
|
|
|
if (isset($this->handle)) { |
308
|
|
|
$this->line = feof($this->handle) ? null : fgets($this->handle); |
309
|
|
|
$this->position = ftell($this->handle); |
310
|
|
|
return $this->line; |
311
|
|
|
} |
312
|
|
|
|
313
|
|
|
return null; |
314
|
|
|
} |
315
|
|
|
|
316
|
|
|
/** |
317
|
|
|
* Determines if CSV line is valid |
318
|
|
|
* @return boolean |
319
|
|
|
*/ |
320
|
|
|
public function valid() |
321
|
|
|
{ |
322
|
|
|
return isset($this->line); |
323
|
|
|
} |
324
|
|
|
|
325
|
|
|
/** |
326
|
|
|
* Gets current CSV row |
327
|
|
|
* @return string |
328
|
|
|
*/ |
329
|
|
|
public function current() |
330
|
|
|
{ |
331
|
|
|
return $this->line; |
332
|
|
|
} |
333
|
|
|
|
334
|
|
|
/** |
335
|
|
|
* Row key |
336
|
|
|
* @return int |
337
|
|
|
*/ |
338
|
|
|
public function key() |
339
|
|
|
{ |
340
|
|
|
return $this->key; |
341
|
|
|
} |
342
|
|
|
|
343
|
|
|
/** |
344
|
|
|
* Gets current offset in bytes |
345
|
|
|
* @return integer |
346
|
|
|
*/ |
347
|
|
|
public function getPosition() |
348
|
|
|
{ |
349
|
|
|
return $this->position; |
350
|
|
|
} |
351
|
|
|
|
352
|
|
|
/** |
353
|
|
|
* Get latest file pointer offset in bytes |
354
|
|
|
* @return integer |
355
|
|
|
*/ |
356
|
|
|
public function getOffset() |
357
|
|
|
{ |
358
|
|
|
return $this->last; |
359
|
|
|
} |
360
|
|
|
|
361
|
|
|
/** |
362
|
|
|
* Sets initial file offset in bytes |
363
|
|
|
* @param integer $offset |
364
|
|
|
* @return $this |
365
|
|
|
*/ |
366
|
|
|
public function setOffset($offset) |
367
|
|
|
{ |
368
|
|
|
$this->offset = $offset; |
369
|
|
|
return $this; |
370
|
|
|
} |
371
|
|
|
|
372
|
|
|
/** |
373
|
|
|
* Force to skip first row (header) |
374
|
|
|
* @return $this |
375
|
|
|
*/ |
376
|
|
|
public function skipHeader() |
377
|
|
|
{ |
378
|
|
|
$this->skip_header = true; |
379
|
|
|
return $this; |
380
|
|
|
} |
381
|
|
|
|
382
|
|
|
/** |
383
|
|
|
* Parses CSV into multidimensional array |
384
|
|
|
* @return array |
385
|
|
|
*/ |
386
|
|
|
public function parse() |
387
|
|
|
{ |
388
|
|
|
if (!empty($this->total)) { |
389
|
|
|
return $this->read(); |
390
|
|
|
} |
391
|
|
|
|
392
|
|
|
return array(); |
393
|
|
|
} |
394
|
|
|
|
395
|
|
|
} |
396
|
|
|
|