1
|
|
|
<?php |
2
|
|
|
/** |
3
|
|
|
* This file is part of graze/data-file |
4
|
|
|
* |
5
|
|
|
* Copyright (c) 2016 Nature Delivered Ltd. <https://www.graze.com> |
6
|
|
|
* |
7
|
|
|
* For the full copyright and license information, please view the LICENSE |
8
|
|
|
* file that was distributed with this source code. |
9
|
|
|
* |
10
|
|
|
* @license https://github.com/graze/data-file/blob/master/LICENSE.md |
11
|
|
|
* @link https://github.com/graze/data-file |
12
|
|
|
*/ |
13
|
|
|
|
14
|
|
|
namespace Graze\DataFile\Format; |
15
|
|
|
|
16
|
|
|
use Graze\CsvToken\Csv\Bom; |
17
|
|
|
use Graze\DataFile\Helper\GetOptionTrait; |
18
|
|
|
|
19
|
|
|
class CsvFormat implements CsvFormatInterface |
20
|
|
|
{ |
21
|
|
|
use GetOptionTrait; |
22
|
|
|
|
23
|
|
|
const DEFAULT_DELIMITER = ','; |
24
|
|
|
const DEFAULT_NULL_OUTPUT = '\\N'; |
25
|
|
|
const DEFAULT_HEADER_ROW = -1; |
26
|
|
|
const DEFAULT_DATA_START = 1; |
27
|
|
|
const DEFAULT_LINE_TERMINATOR = "\n"; |
28
|
|
|
const DEFAULT_QUOTE_CHARACTER = '"'; |
29
|
|
|
const DEFAULT_ESCAPE = '\\'; |
30
|
|
|
const DEFAULT_LIMIT = -1; |
31
|
|
|
const DEFAULT_DOUBLE_QUOTE = false; |
32
|
|
|
const DEFAULT_BOM = null; |
33
|
|
|
const DEFAULT_ENCODING = 'UTF-8'; |
34
|
|
|
|
35
|
|
|
const OPTION_DELIMITER = 'delimiter'; |
36
|
|
|
const OPTION_NULL_OUTPUT = 'nullOutput'; |
37
|
|
|
const OPTION_HEADER_ROW = 'headerRow'; |
38
|
|
|
const OPTION_DATA_START = 'dataStart'; |
39
|
|
|
const OPTION_LINE_TERMINATOR = 'lineTerminator'; |
40
|
|
|
const OPTION_QUOTE_CHARACTER = 'quoteCharacter'; |
41
|
|
|
const OPTION_ESCAPE = 'escape'; |
42
|
|
|
const OPTION_LIMIT = 'limit'; |
43
|
|
|
const OPTION_DOUBLE_QUOTE = 'doubleQuote'; |
44
|
|
|
const OPTION_BOM = 'bom'; |
45
|
|
|
const OPTION_ENCODING = 'encoding'; |
46
|
|
|
|
47
|
|
|
/** @var string */ |
48
|
|
|
protected $delimiter; |
49
|
|
|
/** @var string */ |
50
|
|
|
protected $quoteCharacter; |
51
|
|
|
/** @var string */ |
52
|
|
|
protected $nullOutput; |
53
|
|
|
/** @var int */ |
54
|
|
|
protected $headerRow; |
55
|
|
|
/** @var string */ |
56
|
|
|
protected $lineTerminator; |
57
|
|
|
/** @var bool */ |
58
|
|
|
protected $nullQuotes; |
59
|
|
|
/** @var string */ |
60
|
|
|
protected $escape; |
61
|
|
|
/** @var int */ |
62
|
|
|
protected $limit; |
63
|
|
|
/** @var bool */ |
64
|
|
|
protected $doubleQuote; |
65
|
|
|
/** @var int */ |
66
|
|
|
protected $dataStart; |
67
|
|
|
/** @var string|null */ |
68
|
|
|
protected $bom; |
69
|
|
|
/** @var string */ |
70
|
|
|
protected $encoding; |
71
|
|
|
|
72
|
|
|
/** |
73
|
|
|
* @param array $options -delimiter <string> (Default: ,) Character to use between fields |
74
|
|
|
* -quoteCharacter <string> (Default: ") |
75
|
|
|
* -nullOutput <string> (Default: \N) |
76
|
|
|
* -headerRow <int> (Default: -1) -1 for no header row. (1 is the first line of the file) |
77
|
|
|
* -dataStart <int> (Default: 1) The line where the data starts (1 is the first list of the |
78
|
|
|
* file) |
79
|
|
|
* -lineTerminator <string> (Default: \n) |
80
|
|
|
* -escape <string> (Default: \\) Character to use for escaping |
81
|
|
|
* -limit <int> Total number of data rows to return |
82
|
|
|
* -doubleQuote <bool> instances of quote in fields are indicated by a double quote |
83
|
|
|
* -bom <string> (Default: null) Specify a ByteOrderMark for this file |
84
|
|
|
* -encoding <string> (Default: UTF-8) Specify the encoding of the csv file |
85
|
|
|
*/ |
86
|
16 |
|
public function __construct(array $options = []) |
87
|
|
|
{ |
88
|
16 |
|
$this->options = $options; |
89
|
16 |
|
$this->delimiter = $this->getOption(static::OPTION_DELIMITER, static::DEFAULT_DELIMITER); |
90
|
16 |
|
$this->quoteCharacter = $this->getOption(static::OPTION_QUOTE_CHARACTER, static::DEFAULT_QUOTE_CHARACTER); |
91
|
16 |
|
$this->nullOutput = $this->getOption(static::OPTION_NULL_OUTPUT, static::DEFAULT_NULL_OUTPUT); |
92
|
16 |
|
$this->headerRow = $this->getOption(static::OPTION_HEADER_ROW, static::DEFAULT_HEADER_ROW); |
93
|
16 |
|
$this->dataStart = $this->getOption(static::OPTION_DATA_START, static::DEFAULT_DATA_START); |
94
|
16 |
|
$this->lineTerminator = $this->getOption(static::OPTION_LINE_TERMINATOR, static::DEFAULT_LINE_TERMINATOR); |
95
|
16 |
|
$this->escape = $this->getOption(static::OPTION_ESCAPE, static::DEFAULT_ESCAPE); |
96
|
16 |
|
$this->limit = $this->getOption(static::OPTION_LIMIT, static::DEFAULT_LIMIT); |
97
|
16 |
|
$this->doubleQuote = $this->getOption(static::OPTION_DOUBLE_QUOTE, static::DEFAULT_DOUBLE_QUOTE); |
98
|
16 |
|
$this->setBom($this->getOption(static::OPTION_BOM, static::DEFAULT_BOM)); |
99
|
16 |
|
$this->encoding = $this->getOption(static::OPTION_ENCODING, static::DEFAULT_ENCODING); |
100
|
16 |
|
} |
101
|
|
|
|
102
|
|
|
/** |
103
|
|
|
* @return string |
104
|
|
|
*/ |
105
|
27 |
|
public function getDelimiter() |
106
|
|
|
{ |
107
|
27 |
|
return $this->delimiter; |
108
|
|
|
} |
109
|
|
|
|
110
|
|
|
/** |
111
|
|
|
* @param string $delimiter |
112
|
|
|
* |
113
|
|
|
* @return static |
114
|
|
|
*/ |
115
|
2 |
|
public function setDelimiter($delimiter) |
116
|
|
|
{ |
117
|
2 |
|
$this->delimiter = $delimiter; |
118
|
2 |
|
return $this; |
119
|
|
|
} |
120
|
|
|
|
121
|
|
|
/** |
122
|
|
|
* @return bool |
123
|
|
|
*/ |
124
|
18 |
|
public function hasQuotes() |
125
|
|
|
{ |
126
|
18 |
|
return $this->quoteCharacter <> ''; |
127
|
|
|
} |
128
|
|
|
|
129
|
|
|
/** |
130
|
|
|
* @return string |
131
|
|
|
*/ |
132
|
16 |
|
public function getNullOutput() |
133
|
|
|
{ |
134
|
16 |
|
return $this->nullOutput; |
135
|
|
|
} |
136
|
|
|
|
137
|
|
|
/** |
138
|
|
|
* @param string $nullOutput |
139
|
|
|
* |
140
|
|
|
* @return static |
141
|
|
|
*/ |
142
|
1 |
|
public function setNullOutput($nullOutput) |
143
|
|
|
{ |
144
|
1 |
|
$this->nullOutput = $nullOutput; |
145
|
1 |
|
return $this; |
146
|
|
|
} |
147
|
|
|
|
148
|
|
|
/** |
149
|
|
|
* @return bool |
150
|
|
|
*/ |
151
|
13 |
|
public function hasHeaderRow() |
152
|
|
|
{ |
153
|
13 |
|
return $this->headerRow > 0; |
154
|
|
|
} |
155
|
|
|
|
156
|
|
|
/** |
157
|
|
|
* @param int $headerRow |
158
|
|
|
* |
159
|
|
|
* @return static |
160
|
|
|
*/ |
161
|
2 |
|
public function setHeaderRow($headerRow) |
162
|
|
|
{ |
163
|
2 |
|
$this->headerRow = $headerRow; |
164
|
2 |
|
return $this; |
165
|
|
|
} |
166
|
|
|
|
167
|
|
|
/** |
168
|
|
|
* @return int |
169
|
|
|
*/ |
170
|
7 |
|
public function getHeaderRow() |
171
|
|
|
{ |
172
|
7 |
|
return $this->headerRow; |
173
|
|
|
} |
174
|
|
|
|
175
|
|
|
/** |
176
|
|
|
* @param int $row |
177
|
|
|
* |
178
|
|
|
* @return static |
179
|
|
|
*/ |
180
|
2 |
|
public function setDataStart($row) |
181
|
|
|
{ |
182
|
2 |
|
$this->dataStart = $row; |
183
|
2 |
|
return $this; |
184
|
|
|
} |
185
|
|
|
|
186
|
|
|
/** |
187
|
|
|
* @return int |
188
|
|
|
*/ |
189
|
13 |
|
public function getDataStart() |
190
|
|
|
{ |
191
|
13 |
|
if ($this->hasHeaderRow() && $this->getHeaderRow() >= $this->dataStart) { |
192
|
4 |
|
return max(1, $this->getHeaderRow() + 1); |
193
|
|
|
} |
194
|
10 |
|
return max(1, $this->dataStart); |
195
|
|
|
} |
196
|
|
|
|
197
|
|
|
/** |
198
|
|
|
* @return string |
199
|
|
|
*/ |
200
|
16 |
|
public function getLineTerminator() |
201
|
|
|
{ |
202
|
16 |
|
return $this->lineTerminator; |
203
|
|
|
} |
204
|
|
|
|
205
|
|
|
/** |
206
|
|
|
* @param string $lineTerminator |
207
|
|
|
* |
208
|
|
|
* @return static |
209
|
|
|
*/ |
210
|
1 |
|
public function setLineTerminator($lineTerminator) |
211
|
|
|
{ |
212
|
1 |
|
$this->lineTerminator = $lineTerminator; |
213
|
1 |
|
return $this; |
214
|
|
|
} |
215
|
|
|
|
216
|
|
|
/** |
217
|
|
|
* @note Csv Rfc spec defines escaping of quotes to be done using double quotes `""` |
218
|
|
|
* |
219
|
|
|
* @return string |
220
|
|
|
*/ |
221
|
26 |
|
public function getQuoteCharacter() |
222
|
|
|
{ |
223
|
26 |
|
return $this->quoteCharacter; |
224
|
|
|
} |
225
|
|
|
|
226
|
|
|
/** |
227
|
|
|
* @param string $quoteCharacter |
228
|
|
|
* |
229
|
|
|
* @return static |
230
|
|
|
*/ |
231
|
1 |
|
public function setQuoteCharacter($quoteCharacter) |
232
|
|
|
{ |
233
|
1 |
|
$this->quoteCharacter = $quoteCharacter; |
234
|
1 |
|
return $this; |
235
|
|
|
} |
236
|
|
|
|
237
|
|
|
/** |
238
|
|
|
* Type type of file format (defined in FileFormatType::) |
239
|
|
|
* |
240
|
|
|
* @return string |
241
|
|
|
*/ |
242
|
3 |
|
public function getType() |
243
|
|
|
{ |
244
|
3 |
|
return 'csv'; |
245
|
|
|
} |
246
|
|
|
|
247
|
|
|
/** |
248
|
|
|
* @return string |
249
|
|
|
*/ |
250
|
27 |
|
public function getEscapeCharacter() |
251
|
|
|
{ |
252
|
27 |
|
return $this->escape; |
253
|
|
|
} |
254
|
|
|
|
255
|
|
|
/** |
256
|
|
|
* @param string $escape |
257
|
|
|
* |
258
|
|
|
* @return static |
259
|
|
|
*/ |
260
|
1 |
|
public function setEscapeCharacter($escape) |
261
|
|
|
{ |
262
|
1 |
|
$this->escape = $escape; |
263
|
1 |
|
return $this; |
264
|
|
|
} |
265
|
|
|
|
266
|
|
|
/** |
267
|
|
|
* @return bool |
268
|
|
|
*/ |
269
|
3 |
|
public function hasEscapeCharacter() |
270
|
|
|
{ |
271
|
3 |
|
return $this->escape !== ''; |
272
|
|
|
} |
273
|
|
|
|
274
|
|
|
/** |
275
|
|
|
* Get the limit that should be returned (-1 for no limit) |
276
|
|
|
* |
277
|
|
|
* @return int |
278
|
|
|
*/ |
279
|
12 |
|
public function getLimit() |
280
|
|
|
{ |
281
|
12 |
|
return $this->limit; |
282
|
|
|
} |
283
|
|
|
|
284
|
|
|
/** |
285
|
|
|
* Set the limit of the number of items to be returned (-1 for not limit) |
286
|
|
|
* |
287
|
|
|
* @param int $limit |
288
|
|
|
* |
289
|
|
|
* @return static |
290
|
|
|
*/ |
291
|
1 |
|
public function setLimit($limit) |
292
|
|
|
{ |
293
|
1 |
|
$this->limit = $limit; |
294
|
1 |
|
return $this; |
295
|
|
|
} |
296
|
|
|
|
297
|
|
|
/** |
298
|
|
|
* @return bool |
299
|
|
|
*/ |
300
|
24 |
|
public function isDoubleQuote() |
301
|
|
|
{ |
302
|
24 |
|
return $this->doubleQuote; |
303
|
|
|
} |
304
|
|
|
|
305
|
|
|
/** |
306
|
|
|
* @param bool $doubleQuote |
307
|
|
|
* |
308
|
|
|
* @return static |
309
|
|
|
*/ |
310
|
1 |
|
public function setDoubleQuote($doubleQuote) |
311
|
|
|
{ |
312
|
1 |
|
$this->doubleQuote = $doubleQuote; |
313
|
1 |
|
return $this; |
314
|
|
|
} |
315
|
|
|
|
316
|
|
|
/** |
317
|
|
|
* @return null|string |
318
|
|
|
*/ |
319
|
27 |
|
public function getBom() |
320
|
|
|
{ |
321
|
27 |
|
return $this->bom; |
322
|
|
|
} |
323
|
|
|
|
324
|
|
|
/** |
325
|
|
|
* @param null|string $bom |
326
|
|
|
* |
327
|
|
|
* @return static |
328
|
|
|
*/ |
329
|
16 |
|
public function setBom($bom) |
330
|
|
|
{ |
331
|
16 |
|
$this->bom = $bom; |
332
|
16 |
|
if (!is_null($bom)) { |
333
|
4 |
|
Bom::getEncoding($bom); |
334
|
|
|
} |
335
|
16 |
|
return $this; |
336
|
|
|
} |
337
|
|
|
|
338
|
|
|
/** |
339
|
|
|
* @return string |
340
|
|
|
*/ |
341
|
24 |
|
public function getEncoding() |
342
|
|
|
{ |
343
|
24 |
|
if (!is_null($this->bom)) { |
344
|
3 |
|
return Bom::getEncoding($this->bom); |
345
|
|
|
} |
346
|
22 |
|
return $this->encoding; |
347
|
|
|
} |
348
|
|
|
|
349
|
|
|
/** |
350
|
|
|
* @param string $encoding |
351
|
|
|
* |
352
|
|
|
* @return static |
353
|
|
|
*/ |
354
|
1 |
|
public function setEncoding($encoding) |
355
|
|
|
{ |
356
|
1 |
|
$this->encoding = $encoding; |
357
|
1 |
|
return $this; |
358
|
|
|
} |
359
|
|
|
} |
360
|
|
|
|