|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* CSVelte: Slender, elegant CSV for PHP |
|
4
|
|
|
* |
|
5
|
|
|
* Inspired by Python's CSV module and Frictionless Data and the W3C's CSV |
|
6
|
|
|
* standardization efforts, CSVelte was written in an effort to take all the |
|
7
|
|
|
* suck out of working with CSV. |
|
8
|
|
|
* |
|
9
|
|
|
* @copyright Copyright (c) 2018 Luke Visinoni |
|
10
|
|
|
* @author Luke Visinoni <[email protected]> |
|
11
|
|
|
* @license See LICENSE file (MIT license) |
|
12
|
|
|
*/ |
|
13
|
|
|
namespace CSVelte; |
|
14
|
|
|
|
|
15
|
|
|
use Traversable; |
|
16
|
|
|
use function Noz\collect, |
|
17
|
|
|
Noz\to_array; |
|
18
|
|
|
|
|
19
|
|
|
/** |
|
20
|
|
|
* CSV Dialect - Default dialect |
|
21
|
|
|
* |
|
22
|
|
|
* Due to CSV being without any definitive format definition for so long, many dialects of it exist. This class allows |
|
23
|
|
|
* the creation of reusable "dialects" for commonly used flavors of CSV. You can make one for Excel CSV, another for |
|
24
|
|
|
* tab delimited CSV, another for pipe-delimited, etc. |
|
25
|
|
|
*/ |
|
26
|
|
|
class Dialect |
|
27
|
|
|
{ |
|
28
|
|
|
/** Quote none - never quote any columns */ |
|
29
|
|
|
const QUOTE_NONE = 0; |
|
30
|
|
|
/** Quote all - always quote all columns */ |
|
31
|
|
|
const QUOTE_ALL = 1; |
|
32
|
|
|
/** Quote minimal - quote only those columns that contain the delimiter or quote character */ |
|
33
|
|
|
const QUOTE_MINIMAL = 2; |
|
34
|
|
|
/** Quote non-numeric - quote only those columns that contain non-numeric values */ |
|
35
|
|
|
const QUOTE_NONNUMERIC = 3; |
|
36
|
|
|
|
|
37
|
|
|
/** Trim all - trim empty space from start and end */ |
|
38
|
|
|
const TRIM_ALL = true; |
|
39
|
|
|
/** Trim none - do not trim at all */ |
|
40
|
|
|
const TRIM_NONE = false; |
|
41
|
|
|
/** Trim start - trim empty space from the start (left) */ |
|
42
|
|
|
const TRIM_START = 'start'; |
|
43
|
|
|
/** Trim end - trim empty space from the end (right) */ |
|
44
|
|
|
const TRIM_END = 'end'; |
|
45
|
|
|
|
|
46
|
|
|
/** Standard attributes (from W3 CSVW working group) */ |
|
47
|
|
|
|
|
48
|
|
|
/** @var string The character to use to begin a comment line */ |
|
49
|
|
|
protected $commentPrefix = "#"; |
|
50
|
|
|
|
|
51
|
|
|
/** @var string The character to delimit columns with */ |
|
52
|
|
|
protected $delimiter = ","; |
|
53
|
|
|
|
|
54
|
|
|
/** @var bool Whether to escape quotes within a column by preceding them with another quote */ |
|
55
|
|
|
protected $doubleQuote = true; |
|
56
|
|
|
|
|
57
|
|
|
/** @var string The character encoding for this dialect */ |
|
58
|
|
|
protected $encoding = "utf-8"; |
|
59
|
|
|
|
|
60
|
|
|
/** @var bool Whether the dialect expects a header row (or rows) within the data */ |
|
61
|
|
|
protected $header = true; |
|
62
|
|
|
|
|
63
|
|
|
/** @var int How many header rows are expected within the data */ |
|
64
|
|
|
protected $headerRowCount = 1; |
|
65
|
|
|
|
|
66
|
|
|
/** @var string The line ending character or character sequence */ |
|
67
|
|
|
protected $lineTerminator = "\n"; |
|
68
|
|
|
|
|
69
|
|
|
/** @var string The quoting character (used to quote columns depending on quoteStyle) */ |
|
70
|
|
|
protected $quoteChar = '"'; |
|
71
|
|
|
|
|
72
|
|
|
/** @var bool Whether blank rows within the data should be skipped/ignored */ |
|
73
|
|
|
protected $skipBlankRows = false; |
|
74
|
|
|
|
|
75
|
|
|
/** @var int How many columns to skip/ignore */ |
|
76
|
|
|
protected $skipColumns = 0; |
|
77
|
|
|
|
|
78
|
|
|
/** @var bool Whether to skip/ignore initial space within a column */ |
|
79
|
|
|
protected $skipInitialSpace = false; |
|
80
|
|
|
|
|
81
|
|
|
/** @var int How many rows to skip/ignore */ |
|
82
|
|
|
protected $skipRows = 0; |
|
83
|
|
|
|
|
84
|
|
|
/** @var bool|string Whether to trim empty space and where (see TRIM_* constants above) */ |
|
85
|
|
|
protected $trim = self::TRIM_ALL; |
|
86
|
|
|
|
|
87
|
|
|
/** Non-standard attributes (my own additions) */ |
|
88
|
|
|
|
|
89
|
|
|
/** @var int The quoting style (see QUOTE_* constants above) */ |
|
90
|
|
|
protected $quoteStyle = self::QUOTE_MINIMAL; |
|
91
|
|
|
|
|
92
|
|
|
/** |
|
93
|
|
|
* Dialect constructor. |
|
94
|
|
|
* |
|
95
|
|
|
* Any of the above properties may be set within the $attribs array to initialize the dialect with different |
|
96
|
|
|
* attributes than are defined above. |
|
97
|
|
|
* |
|
98
|
|
|
* @param array|Traversable $attribs An array of dialect attributes |
|
99
|
|
|
*/ |
|
100
|
16 |
|
public function __construct($attribs = null) |
|
101
|
|
|
{ |
|
102
|
16 |
|
$attribs = to_array($attribs, true); |
|
103
|
16 |
|
foreach ($attribs as $attr => $val) { |
|
104
|
8 |
|
if (property_exists($this, $attr)) { |
|
105
|
|
|
// find the appropriate setter... |
|
106
|
8 |
|
foreach (['set', 'setIs', 'setHas'] as $prefix) { |
|
107
|
8 |
|
$setter = $prefix . ucfirst(strtolower($attr)); |
|
108
|
8 |
|
if (method_exists($this, $setter)) { |
|
109
|
8 |
|
$this->{$setter}($val); |
|
110
|
8 |
|
} |
|
111
|
8 |
|
} |
|
112
|
8 |
|
} |
|
113
|
16 |
|
} |
|
114
|
16 |
|
} |
|
115
|
|
|
|
|
116
|
|
|
/** |
|
117
|
|
|
* Set comment prefix character(s) |
|
118
|
|
|
* |
|
119
|
|
|
* @param string $commentPrefix The character(s) used to begin a comment |
|
120
|
|
|
* |
|
121
|
|
|
* @return self |
|
122
|
|
|
*/ |
|
123
|
2 |
|
public function setCommentPrefix($commentPrefix) |
|
124
|
|
|
{ |
|
125
|
2 |
|
$this->commentPrefix = (string) $commentPrefix; |
|
126
|
2 |
|
return $this; |
|
127
|
|
|
} |
|
128
|
|
|
|
|
129
|
|
|
/** |
|
130
|
|
|
* Get comment prefix character(s) |
|
131
|
|
|
* |
|
132
|
|
|
* @return string |
|
133
|
|
|
*/ |
|
134
|
3 |
|
public function getCommentPrefix() |
|
135
|
|
|
{ |
|
136
|
3 |
|
return $this->commentPrefix; |
|
137
|
|
|
} |
|
138
|
|
|
|
|
139
|
|
|
/** |
|
140
|
|
|
* Set delimiter character(s) |
|
141
|
|
|
* |
|
142
|
|
|
* @param string $delimiter The character(s) used to delimit data |
|
143
|
|
|
* |
|
144
|
|
|
* @return self |
|
145
|
|
|
*/ |
|
146
|
2 |
|
public function setDelimiter($delimiter) |
|
147
|
|
|
{ |
|
148
|
2 |
|
$this->delimiter = (string) $delimiter; |
|
149
|
2 |
|
return $this; |
|
150
|
|
|
} |
|
151
|
|
|
|
|
152
|
|
|
/** |
|
153
|
|
|
* Set delimiter character(s) |
|
154
|
|
|
* |
|
155
|
|
|
* @return string |
|
156
|
|
|
*/ |
|
157
|
16 |
|
public function getDelimiter() |
|
158
|
|
|
{ |
|
159
|
16 |
|
return $this->delimiter; |
|
160
|
|
|
} |
|
161
|
|
|
|
|
162
|
|
|
/** |
|
163
|
|
|
* Set double quote |
|
164
|
|
|
* |
|
165
|
|
|
* @param bool $doubleQuote Whether to escape quote character with a preceding quote character |
|
166
|
|
|
* |
|
167
|
|
|
* @return self |
|
168
|
|
|
*/ |
|
169
|
2 |
|
public function setIsDoubleQuote($doubleQuote) |
|
170
|
|
|
{ |
|
171
|
2 |
|
$this->doubleQuote = (bool) $doubleQuote; |
|
172
|
2 |
|
return $this; |
|
173
|
|
|
} |
|
174
|
|
|
|
|
175
|
|
|
/** |
|
176
|
|
|
* Get double quote |
|
177
|
|
|
* |
|
178
|
|
|
* @return bool |
|
179
|
|
|
*/ |
|
180
|
16 |
|
public function isDoubleQuote() |
|
181
|
|
|
{ |
|
182
|
16 |
|
return $this->doubleQuote; |
|
183
|
|
|
} |
|
184
|
|
|
|
|
185
|
|
|
/** |
|
186
|
|
|
* Set character encoding |
|
187
|
|
|
* |
|
188
|
|
|
* @param string $encoding The character encoding |
|
189
|
|
|
* |
|
190
|
|
|
* @return self |
|
191
|
|
|
*/ |
|
192
|
2 |
|
public function setEncoding($encoding) |
|
193
|
|
|
{ |
|
194
|
2 |
|
$this->encoding = (string) $encoding; |
|
195
|
2 |
|
return $this; |
|
196
|
|
|
} |
|
197
|
|
|
|
|
198
|
|
|
/** |
|
199
|
|
|
* Get character encoding |
|
200
|
|
|
* |
|
201
|
|
|
* @return string |
|
202
|
|
|
*/ |
|
203
|
3 |
|
public function getEncoding() |
|
204
|
|
|
{ |
|
205
|
3 |
|
return $this->encoding; |
|
206
|
|
|
} |
|
207
|
|
|
|
|
208
|
|
|
/** |
|
209
|
|
|
* Set header row flag |
|
210
|
|
|
* |
|
211
|
|
|
* @param bool $header Whether the data has header row(s) |
|
212
|
|
|
* |
|
213
|
|
|
* @return self |
|
214
|
|
|
*/ |
|
215
|
9 |
|
public function setHasHeader($header) |
|
216
|
|
|
{ |
|
217
|
9 |
|
$this->header = (bool) $header; |
|
218
|
9 |
|
return $this; |
|
219
|
|
|
} |
|
220
|
|
|
|
|
221
|
|
|
/** |
|
222
|
|
|
* Get whether dialect expects header row(s) |
|
223
|
|
|
* |
|
224
|
|
|
* @return bool |
|
225
|
|
|
*/ |
|
226
|
16 |
|
public function hasHeader() |
|
227
|
|
|
{ |
|
228
|
16 |
|
return $this->header; |
|
229
|
|
|
} |
|
230
|
|
|
|
|
231
|
|
|
/** |
|
232
|
|
|
* Set header row count |
|
233
|
|
|
* |
|
234
|
|
|
* @param int $headerRowCount The amount of expected header rows |
|
235
|
|
|
* |
|
236
|
|
|
* @return self |
|
237
|
|
|
*/ |
|
238
|
2 |
|
public function setHeaderRowCount($headerRowCount) |
|
239
|
|
|
{ |
|
240
|
2 |
|
$this->headerRowCount = (int) $headerRowCount; |
|
241
|
2 |
|
return $this; |
|
242
|
|
|
} |
|
243
|
|
|
|
|
244
|
|
|
/** |
|
245
|
|
|
* Get header row count |
|
246
|
|
|
* |
|
247
|
|
|
* @return int |
|
248
|
|
|
*/ |
|
249
|
2 |
|
public function getHeaderRowCount() |
|
250
|
|
|
{ |
|
251
|
2 |
|
return $this->headerRowCount; |
|
252
|
|
|
} |
|
253
|
|
|
|
|
254
|
|
|
/** |
|
255
|
|
|
* Set line terminator character or character sequence |
|
256
|
|
|
* |
|
257
|
|
|
* @param string $lineTerminator The line ending character(s) |
|
258
|
|
|
* |
|
259
|
|
|
* @return self |
|
260
|
|
|
*/ |
|
261
|
2 |
|
public function setLineTerminator($lineTerminator) |
|
262
|
|
|
{ |
|
263
|
2 |
|
$this->lineTerminator = (string) $lineTerminator; |
|
264
|
2 |
|
return $this; |
|
265
|
|
|
} |
|
266
|
|
|
|
|
267
|
|
|
/** |
|
268
|
|
|
* Get line terminator |
|
269
|
|
|
* |
|
270
|
|
|
* @return string |
|
271
|
|
|
*/ |
|
272
|
16 |
|
public function getLineTerminator() |
|
273
|
|
|
{ |
|
274
|
16 |
|
return $this->lineTerminator; |
|
275
|
|
|
} |
|
276
|
|
|
|
|
277
|
|
|
/** |
|
278
|
|
|
* Set quote character |
|
279
|
|
|
* |
|
280
|
|
|
* @param string $quoteChar The quote character |
|
281
|
|
|
* |
|
282
|
|
|
* @return self |
|
283
|
|
|
*/ |
|
284
|
2 |
|
public function setQuoteChar($quoteChar) |
|
285
|
|
|
{ |
|
286
|
2 |
|
$this->quoteChar = (string) $quoteChar; |
|
287
|
2 |
|
return $this; |
|
288
|
|
|
} |
|
289
|
|
|
|
|
290
|
|
|
/** |
|
291
|
|
|
* Get quote character |
|
292
|
|
|
* |
|
293
|
|
|
* @return string |
|
294
|
|
|
*/ |
|
295
|
16 |
|
public function getQuoteChar() |
|
296
|
|
|
{ |
|
297
|
16 |
|
return $this->quoteChar; |
|
298
|
|
|
} |
|
299
|
|
|
|
|
300
|
|
|
/** |
|
301
|
|
|
* Set whether to skip blank rows |
|
302
|
|
|
* |
|
303
|
|
|
* @param bool $skipBlankRows Whether to skip blank rows |
|
304
|
|
|
* |
|
305
|
|
|
* @return self |
|
306
|
|
|
*/ |
|
307
|
2 |
|
public function setIsSkipBlankRows($skipBlankRows) |
|
308
|
|
|
{ |
|
309
|
2 |
|
$this->skipBlankRows = (bool) $skipBlankRows; |
|
310
|
2 |
|
return $this; |
|
311
|
|
|
} |
|
312
|
|
|
|
|
313
|
|
|
/** |
|
314
|
|
|
* Get skip blank rows flag |
|
315
|
|
|
* |
|
316
|
|
|
* @return bool |
|
317
|
|
|
*/ |
|
318
|
3 |
|
public function isSkipBlankRows() |
|
319
|
|
|
{ |
|
320
|
3 |
|
return $this->skipBlankRows; |
|
321
|
|
|
} |
|
322
|
|
|
|
|
323
|
|
|
/** |
|
324
|
|
|
* Set number of columns to skip/ignore |
|
325
|
|
|
* |
|
326
|
|
|
* @param int $skipColumns The number of columns to skip/ignore |
|
327
|
|
|
* |
|
328
|
|
|
* @return self |
|
329
|
|
|
*/ |
|
330
|
2 |
|
public function setSkipColumns($skipColumns) |
|
331
|
|
|
{ |
|
332
|
2 |
|
$this->skipColumns = (int) $skipColumns; |
|
333
|
2 |
|
return $this; |
|
334
|
|
|
} |
|
335
|
|
|
|
|
336
|
|
|
/** |
|
337
|
|
|
* Get number of columns to skip/ignore |
|
338
|
|
|
* |
|
339
|
|
|
* @return int |
|
340
|
|
|
*/ |
|
341
|
3 |
|
public function getSkipColumns() |
|
342
|
|
|
{ |
|
343
|
3 |
|
return $this->skipColumns; |
|
344
|
|
|
} |
|
345
|
|
|
|
|
346
|
|
|
/** |
|
347
|
|
|
* Set skip initial space flag |
|
348
|
|
|
* |
|
349
|
|
|
* @param bool $skipInitialSpace Skip initial space flag |
|
350
|
|
|
* |
|
351
|
|
|
* @return self |
|
352
|
|
|
*/ |
|
353
|
2 |
|
public function setIsSkipInitialSpace($skipInitialSpace) |
|
354
|
|
|
{ |
|
355
|
2 |
|
$this->skipInitialSpace = (bool) $skipInitialSpace; |
|
356
|
2 |
|
return $this; |
|
357
|
|
|
} |
|
358
|
|
|
|
|
359
|
|
|
/** |
|
360
|
|
|
* Get skip initial space flag |
|
361
|
|
|
* |
|
362
|
|
|
* @return bool |
|
363
|
|
|
*/ |
|
364
|
3 |
|
public function isSkipInitialSpace() |
|
365
|
|
|
{ |
|
366
|
3 |
|
return $this->skipInitialSpace; |
|
367
|
|
|
} |
|
368
|
|
|
|
|
369
|
|
|
/** |
|
370
|
|
|
* Set number of rows to skip/ignore |
|
371
|
|
|
* |
|
372
|
|
|
* @param int $skipRows Number of rows to skip/ignore |
|
373
|
|
|
* |
|
374
|
|
|
* @return self |
|
375
|
|
|
*/ |
|
376
|
2 |
|
public function setSkipRows($skipRows) |
|
377
|
|
|
{ |
|
378
|
2 |
|
$this->skipRows = (int) $skipRows; |
|
379
|
2 |
|
return $this; |
|
380
|
|
|
} |
|
381
|
|
|
|
|
382
|
|
|
/** |
|
383
|
|
|
* Get number of rows to skip/ignore |
|
384
|
|
|
* |
|
385
|
|
|
* @return int |
|
386
|
|
|
*/ |
|
387
|
3 |
|
public function getSkipRows() |
|
388
|
|
|
{ |
|
389
|
3 |
|
return $this->skipRows; |
|
390
|
|
|
} |
|
391
|
|
|
|
|
392
|
|
|
/** |
|
393
|
|
|
* Set trim type |
|
394
|
|
|
* |
|
395
|
|
|
* Allows you to set whether you want data to be trimmed on one, both, or neither sides. |
|
396
|
|
|
* |
|
397
|
|
|
* @param bool|string $trim The type trimming you want to do (see TRIM_* constants above) |
|
398
|
|
|
* |
|
399
|
|
|
* @return self |
|
400
|
|
|
*/ |
|
401
|
2 |
|
public function setTrim($trim) |
|
402
|
|
|
{ |
|
403
|
2 |
|
$this->trim = $trim; |
|
404
|
2 |
|
return $this; |
|
405
|
|
|
} |
|
406
|
|
|
|
|
407
|
|
|
/** |
|
408
|
|
|
* Get trim type |
|
409
|
|
|
* |
|
410
|
|
|
* The type will coincide with one of the TRIM_* constants defined above. |
|
411
|
|
|
* |
|
412
|
|
|
* @return bool|string |
|
413
|
|
|
*/ |
|
414
|
3 |
|
public function getTrim() |
|
415
|
|
|
{ |
|
416
|
3 |
|
return $this->trim; |
|
417
|
|
|
} |
|
418
|
|
|
|
|
419
|
|
|
/** |
|
420
|
|
|
* Set the quoting style |
|
421
|
|
|
* |
|
422
|
|
|
* Allows you to set how data is quoted |
|
423
|
|
|
* |
|
424
|
|
|
* @param int $quoteStyle The desired quoting style (see QUOTE_* constants above) |
|
425
|
|
|
* |
|
426
|
|
|
* @return self |
|
427
|
|
|
*/ |
|
428
|
2 |
|
public function setQuoteStyle($quoteStyle) |
|
429
|
|
|
{ |
|
430
|
2 |
|
$this->quoteStyle = (int) $quoteStyle; |
|
431
|
2 |
|
return $this; |
|
432
|
|
|
} |
|
433
|
|
|
|
|
434
|
|
|
/** |
|
435
|
|
|
* Get quoting style |
|
436
|
|
|
* |
|
437
|
|
|
* The quoteStyle value will coincide with one of the QUOTE_* constants defined above. |
|
438
|
|
|
* |
|
439
|
|
|
* @return int |
|
440
|
|
|
*/ |
|
441
|
3 |
|
public function getQuoteStyle() |
|
442
|
|
|
{ |
|
443
|
3 |
|
return $this->quoteStyle; |
|
444
|
|
|
} |
|
445
|
|
|
} |