1
|
|
|
<?php |
2
|
|
|
/** |
3
|
|
|
* File containing the {@link CSVHelper} class. |
4
|
|
|
* |
5
|
|
|
* @package Application Utils |
6
|
|
|
* @subpackage CSVHelper |
7
|
|
|
* @see CSVHelper |
8
|
|
|
*/ |
9
|
|
|
|
10
|
|
|
declare(strict_types=1); |
11
|
|
|
|
12
|
|
|
namespace AppUtils; |
13
|
|
|
|
14
|
|
|
use AppUtils\ConvertHelper\JSONConverter; |
15
|
|
|
use JsonException; |
16
|
|
|
use ParseCsv\Csv; |
17
|
|
|
|
18
|
|
|
/** |
19
|
|
|
* Helper class to parse and create/modify csv files or strings. |
20
|
|
|
* |
21
|
|
|
* Usage: |
22
|
|
|
* |
23
|
|
|
* ```php |
24
|
|
|
* $csv = new CSVHelper(); |
25
|
|
|
* $csv->setHeadersTop(); // has to be set before anything else. |
26
|
|
|
* |
27
|
|
|
* // parse a csv file |
28
|
|
|
* $csv->loadFile('path/to/file'); |
29
|
|
|
* |
30
|
|
|
* // parse a csv string |
31
|
|
|
* $csv->loadString($csvString); |
32
|
|
|
* |
33
|
|
|
* // retrieve data |
34
|
|
|
* $headers = $csv->getHeaders(); |
35
|
|
|
* $row = $csv->getRow(4); |
36
|
|
|
* ``` |
37
|
|
|
* |
38
|
|
|
* @package Application Utils |
39
|
|
|
* @subpackage CSVHelper |
40
|
|
|
* @author Sebastian Mordziol <[email protected]> |
41
|
|
|
*/ |
42
|
|
|
class CSVHelper |
43
|
|
|
{ |
44
|
|
|
public const ERROR_INVALID_HEADERS_POSITION = 561002; |
45
|
|
|
public const ERROR_INVALID_FILE_ENCODING = 561003; |
46
|
|
|
public const ERROR_FILE_PARSING_FAILED = 561004; |
47
|
|
|
public const ERROR_CSV_FILE_NOT_READABLE = 561005; |
48
|
|
|
public const ERROR_STRING_PARSING_FAILED = 561006; |
49
|
|
|
|
50
|
|
|
public const DELIMITER_AUTO = 'auto'; |
51
|
|
|
|
52
|
|
|
public const HEADERS_LEFT = 'hleft'; |
53
|
|
|
public const HEADERS_TOP = 'htop'; |
54
|
|
|
public const HEADERS_NONE = 'hnone'; |
55
|
|
|
|
56
|
|
|
protected string $csv = ''; |
57
|
|
|
protected string $headersPosition = self::HEADERS_NONE; |
58
|
|
|
protected string $separator = ';'; |
59
|
|
|
protected int $columnCount = 0; |
60
|
|
|
protected int $rowCount = 0; |
61
|
|
|
|
62
|
|
|
/** |
63
|
|
|
* @var string[] |
64
|
|
|
*/ |
65
|
|
|
protected array $errors = array(); |
66
|
|
|
|
67
|
|
|
/** |
68
|
|
|
* @var array<int,array<int,mixed>> |
69
|
|
|
*/ |
70
|
|
|
protected array $data = array(); |
71
|
|
|
|
72
|
|
|
/** |
73
|
|
|
* @var string[] |
74
|
|
|
*/ |
75
|
|
|
protected array $headers = array(); |
76
|
|
|
|
77
|
|
|
public function __construct() |
78
|
|
|
{ |
79
|
|
|
|
80
|
|
|
} |
81
|
|
|
|
82
|
|
|
/** |
83
|
|
|
* Creates and returns a new instance of the CSV builder which |
84
|
|
|
* can be used to build CSV from scratch. |
85
|
|
|
* |
86
|
|
|
* @return CSVHelper_Builder |
87
|
|
|
*/ |
88
|
|
|
public static function createBuilder() : CSVHelper_Builder |
89
|
|
|
{ |
90
|
|
|
return new CSVHelper_Builder(); |
91
|
|
|
} |
92
|
|
|
|
93
|
|
|
|
94
|
|
|
|
95
|
|
|
/** |
96
|
|
|
* Loads CSV data from a string. |
97
|
|
|
* |
98
|
|
|
* Note: Use the {@link hasErrors()} method to |
99
|
|
|
* check if the string could be parsed correctly |
100
|
|
|
* afterwards. |
101
|
|
|
* |
102
|
|
|
* @param string $string |
103
|
|
|
* @return $this |
104
|
|
|
*/ |
105
|
|
|
public function loadString(string $string) : self |
106
|
|
|
{ |
107
|
|
|
// remove any UTF byte order marks that may still be present in the string |
108
|
|
|
$this->csv = ConvertHelper::stripUTFBom($string); |
109
|
|
|
|
110
|
|
|
// ensure the string is valid UTF8 |
111
|
|
|
$this->csv = ConvertHelper::string2utf8($this->csv); |
112
|
|
|
|
113
|
|
|
$this->parse(); |
114
|
|
|
|
115
|
|
|
return $this; |
116
|
|
|
} |
117
|
|
|
|
118
|
|
|
/** |
119
|
|
|
* Loads CSV data from a file. |
120
|
|
|
* |
121
|
|
|
* Note: Use the {@link hasErrors()} method to |
122
|
|
|
* check if the string could be parsed correctly |
123
|
|
|
* afterwards. |
124
|
|
|
* |
125
|
|
|
* @param string $file |
126
|
|
|
* @throws FileHelper_Exception |
127
|
|
|
* @return CSVHelper |
128
|
|
|
* |
129
|
|
|
* @see FileHelper::ERROR_FILE_DOES_NOT_EXIST |
130
|
|
|
* @see FileHelper::ERROR_CANNOT_READ_FILE_CONTENTS |
131
|
|
|
*/ |
132
|
|
|
public function loadFile(string $file) : self |
133
|
|
|
{ |
134
|
|
|
$csv = FileHelper::readContents($file); |
135
|
|
|
|
136
|
|
|
return $this->loadString($csv); |
137
|
|
|
} |
138
|
|
|
|
139
|
|
|
/** |
140
|
|
|
* Specifies that headers are positioned on top, horizontally. |
141
|
|
|
* @return $this |
142
|
|
|
* |
143
|
|
|
* @throws CSVHelper_Exception |
144
|
|
|
*/ |
145
|
|
|
public function setHeadersTop() : self |
146
|
|
|
{ |
147
|
|
|
return $this->setHeadersPosition(self::HEADERS_TOP); |
148
|
|
|
} |
149
|
|
|
|
150
|
|
|
/** |
151
|
|
|
* Specifies that headers are positioned on the left, vertically. |
152
|
|
|
* @return $this |
153
|
|
|
* |
154
|
|
|
* @throws CSVHelper_Exception |
155
|
|
|
*/ |
156
|
|
|
public function setHeadersLeft() : self |
157
|
|
|
{ |
158
|
|
|
return $this->setHeadersPosition(self::HEADERS_LEFT); |
159
|
|
|
} |
160
|
|
|
|
161
|
|
|
/** |
162
|
|
|
* Specifies that there are no headers in the file (default). |
163
|
|
|
* @return $this |
164
|
|
|
* |
165
|
|
|
* @throws CSVHelper_Exception |
166
|
|
|
*/ |
167
|
|
|
public function setHeadersNone() : self |
168
|
|
|
{ |
169
|
|
|
return $this->setHeadersPosition(self::HEADERS_NONE); |
170
|
|
|
} |
171
|
|
|
|
172
|
|
|
public function isHeadersLeft() : bool |
173
|
|
|
{ |
174
|
|
|
return $this->isHeadersPosition(self::HEADERS_LEFT); |
175
|
|
|
} |
176
|
|
|
|
177
|
|
|
public function isHeadersTop() : bool |
178
|
|
|
{ |
179
|
|
|
return $this->isHeadersPosition(self::HEADERS_TOP); |
180
|
|
|
} |
181
|
|
|
|
182
|
|
|
public function isHeadersNone() : bool |
183
|
|
|
{ |
184
|
|
|
return $this->isHeadersPosition(self::HEADERS_NONE); |
185
|
|
|
} |
186
|
|
|
|
187
|
|
|
public function isHeadersPosition(string $position) : bool |
188
|
|
|
{ |
189
|
|
|
return $this->headersPosition === $position; |
190
|
|
|
} |
191
|
|
|
|
192
|
|
|
/** |
193
|
|
|
* Specifies where the headers are positioned in the |
194
|
|
|
* CSV, or turns them off entirely. Use the class constants |
195
|
|
|
* to ensure the value is correct. |
196
|
|
|
* |
197
|
|
|
* @param string $position |
198
|
|
|
* @throws CSVHelper_Exception |
199
|
|
|
* @return $this |
200
|
|
|
* |
201
|
|
|
* @see CSVHelper::HEADERS_LEFT |
202
|
|
|
* @see CSVHelper::HEADERS_TOP |
203
|
|
|
* @see CSVHelper::HEADERS_NONE |
204
|
|
|
*/ |
205
|
|
|
public function setHeadersPosition(string $position) : self |
206
|
|
|
{ |
207
|
|
|
$validPositions = array( |
208
|
|
|
self::HEADERS_LEFT, |
209
|
|
|
self::HEADERS_NONE, |
210
|
|
|
self::HEADERS_TOP |
211
|
|
|
); |
212
|
|
|
|
213
|
|
|
if(!in_array($position, $validPositions)) { |
214
|
|
|
throw new CSVHelper_Exception( |
215
|
|
|
'Invalid headers position', |
216
|
|
|
sprintf( |
217
|
|
|
'The header position [%s] is invalid. Valid positions are [%s]. '. |
218
|
|
|
'It is recommended to use the class constants, for example [%s].', |
219
|
|
|
$position, |
220
|
|
|
implode(', ', $validPositions), |
221
|
|
|
'CSVHelper::HEADERS_LEFT' |
222
|
|
|
), |
223
|
|
|
self::ERROR_INVALID_HEADERS_POSITION |
224
|
|
|
); |
225
|
|
|
} |
226
|
|
|
|
227
|
|
|
$this->headersPosition = $position; |
228
|
|
|
return $this; |
229
|
|
|
} |
230
|
|
|
|
231
|
|
|
/** |
232
|
|
|
* Resets all internal data, allowing to start entirely anew |
233
|
|
|
* with a new file, or to start building a new CSV file from |
234
|
|
|
* scratch. |
235
|
|
|
* |
236
|
|
|
* @return $this |
237
|
|
|
*/ |
238
|
|
|
public function reset() : self |
239
|
|
|
{ |
240
|
|
|
$this->data = array(); |
241
|
|
|
$this->headers = array(); |
242
|
|
|
$this->errors = array(); |
243
|
|
|
$this->columnCount = 0; |
244
|
|
|
$this->rowCount = 0; |
245
|
|
|
|
246
|
|
|
return $this; |
247
|
|
|
} |
248
|
|
|
|
249
|
|
|
/** |
250
|
|
|
* @return array<int,array<int,mixed>> |
251
|
|
|
*/ |
252
|
|
|
public function getData() : array |
253
|
|
|
{ |
254
|
|
|
return $this->data; |
255
|
|
|
} |
256
|
|
|
|
257
|
|
|
/** |
258
|
|
|
* Retrieves the row at the specified index. |
259
|
|
|
* If there is no data at the index, this will |
260
|
|
|
* return an array populated with empty strings |
261
|
|
|
* for all available columns. |
262
|
|
|
* |
263
|
|
|
* Tip: Use the {@link rowExists()} method to check |
264
|
|
|
* whether the specified row exists. |
265
|
|
|
* |
266
|
|
|
* @param integer $index |
267
|
|
|
* @return array<int,mixed> |
268
|
|
|
* @see rowExists() |
269
|
|
|
*/ |
270
|
|
|
public function getRow(int $index) : array |
271
|
|
|
{ |
272
|
|
|
return $this->data[$index] ?? array_fill(0, $this->rowCount, ''); |
273
|
|
|
} |
274
|
|
|
|
275
|
|
|
/** |
276
|
|
|
* Checks whether the specified row exists in the data set. |
277
|
|
|
* @param integer $index |
278
|
|
|
* @return boolean |
279
|
|
|
*/ |
280
|
|
|
public function rowExists(int $index) : bool |
281
|
|
|
{ |
282
|
|
|
return isset($this->data[$index]); |
283
|
|
|
} |
284
|
|
|
|
285
|
|
|
/** |
286
|
|
|
* Counts the amount of rows in the parsed CSV, |
287
|
|
|
* excluding the headers if any, depending on |
288
|
|
|
* their position. |
289
|
|
|
* |
290
|
|
|
* @return integer |
291
|
|
|
*/ |
292
|
|
|
public function countRows() : int |
293
|
|
|
{ |
294
|
|
|
return $this->rowCount; |
295
|
|
|
} |
296
|
|
|
|
297
|
|
|
/** |
298
|
|
|
* Counts the amount of rows in the parsed CSV, |
299
|
|
|
* excluding the headers if any, depending on |
300
|
|
|
* their position. |
301
|
|
|
* |
302
|
|
|
* @return integer |
303
|
|
|
*/ |
304
|
|
|
public function countColumns() : int |
305
|
|
|
{ |
306
|
|
|
return $this->columnCount; |
307
|
|
|
} |
308
|
|
|
|
309
|
|
|
/** |
310
|
|
|
* Retrieves the headers, if any. Specify the position of the |
311
|
|
|
* headers first to ensure this works correctly. |
312
|
|
|
* |
313
|
|
|
* @return string[] Indexed array with header names. |
314
|
|
|
*/ |
315
|
|
|
public function getHeaders() : array |
316
|
|
|
{ |
317
|
|
|
return $this->headers; |
318
|
|
|
} |
319
|
|
|
|
320
|
|
|
/** |
321
|
|
|
* Retrieves the column at the specified index. If there |
322
|
|
|
* is no column at the index, this returns an array |
323
|
|
|
* populated with empty strings. |
324
|
|
|
* |
325
|
|
|
* Tip: Use the {@link columnExists()} method to check |
326
|
|
|
* whether a column exists. |
327
|
|
|
* |
328
|
|
|
* @param integer $index |
329
|
|
|
* @return string[] |
330
|
|
|
* @see columnExists() |
331
|
|
|
*/ |
332
|
|
|
public function getColumn(int $index) : array |
333
|
|
|
{ |
334
|
|
|
$data = array(); |
335
|
|
|
|
336
|
|
|
for($i=0; $i < $this->rowCount; $i++) |
337
|
|
|
{ |
338
|
|
|
$value = $this->data[$i][$index] ?? ''; |
339
|
|
|
|
340
|
|
|
$data[] = $value; |
341
|
|
|
} |
342
|
|
|
|
343
|
|
|
return $data; |
344
|
|
|
} |
345
|
|
|
|
346
|
|
|
/** |
347
|
|
|
* Checks whether the specified column exists in the data set. |
348
|
|
|
* @param integer $index |
349
|
|
|
* @return boolean |
350
|
|
|
*/ |
351
|
|
|
public function columnExists(int $index) : bool |
352
|
|
|
{ |
353
|
|
|
return $index < $this->columnCount; |
354
|
|
|
} |
355
|
|
|
|
356
|
|
|
protected function parse() : void |
357
|
|
|
{ |
358
|
|
|
$this->reset(); |
359
|
|
|
|
360
|
|
|
if(empty(trim($this->csv))) { |
361
|
|
|
$this->addError('Tried to parse an empty CSV string.'); |
362
|
|
|
return; |
363
|
|
|
} |
364
|
|
|
|
365
|
|
|
// ensure that the last line in the CSV has |
366
|
|
|
// a linebreak afterwards, otherwise the line |
367
|
|
|
// will not be parsed. |
368
|
|
|
$this->csv = rtrim($this->csv).PHP_EOL; |
369
|
|
|
|
370
|
|
|
$parser = self::createParser(); |
371
|
|
|
|
372
|
|
|
if(!$parser->parse($this->csv)) { |
373
|
|
|
$this->addError('The CSV string could not be parsed.'); |
374
|
|
|
return; |
375
|
|
|
} |
376
|
|
|
|
377
|
|
|
$result = $parser->data; |
378
|
|
|
|
379
|
|
|
switch($this->headersPosition) |
380
|
|
|
{ |
381
|
|
|
case self::HEADERS_TOP: |
382
|
|
|
$this->headers = array_shift($result); |
383
|
|
|
break; |
384
|
|
|
|
385
|
|
|
case self::HEADERS_LEFT: |
386
|
|
|
$keep = array(); |
387
|
|
|
|
388
|
|
|
foreach ($result as $value) |
389
|
|
|
{ |
390
|
|
|
$row = $value; |
391
|
|
|
$this->headers[] = array_shift($row); |
392
|
|
|
$keep[] = $row; |
393
|
|
|
} |
394
|
|
|
|
395
|
|
|
$result = $keep; |
396
|
|
|
break; |
397
|
|
|
} |
398
|
|
|
|
399
|
|
|
$this->data = $result; |
400
|
|
|
$this->rowCount = count($this->data); |
401
|
|
|
|
402
|
|
|
for($i=0; $i < $this->rowCount; $i++) { |
403
|
|
|
$amount = count($this->data[$i]); |
404
|
|
|
if($amount > $this->columnCount) { |
405
|
|
|
$this->columnCount = $amount; |
406
|
|
|
} |
407
|
|
|
} |
408
|
|
|
} |
409
|
|
|
|
410
|
|
|
/** |
411
|
|
|
* Checks whether any errors have been encountered |
412
|
|
|
* while parsing the CSV. |
413
|
|
|
* |
414
|
|
|
* @return boolean |
415
|
|
|
* @see getErrorMessages() |
416
|
|
|
*/ |
417
|
|
|
public function hasErrors() : bool |
418
|
|
|
{ |
419
|
|
|
return !empty($this->errors); |
420
|
|
|
} |
421
|
|
|
|
422
|
|
|
/** |
423
|
|
|
* Retrieves all error messages. |
424
|
|
|
* @return string[] |
425
|
|
|
*/ |
426
|
|
|
public function getErrorMessages() : array |
427
|
|
|
{ |
428
|
|
|
return $this->errors; |
429
|
|
|
} |
430
|
|
|
|
431
|
|
|
protected function addError(string $error) : self |
432
|
|
|
{ |
433
|
|
|
$this->errors[] = $error; |
434
|
|
|
return $this; |
435
|
|
|
} |
436
|
|
|
|
437
|
|
|
protected function detectSeparator() : string |
438
|
|
|
{ |
439
|
|
|
$search = array( |
440
|
|
|
"\"\t\"" => "\t", |
441
|
|
|
'";"' => ';', |
442
|
|
|
'","' => ',', |
443
|
|
|
';;' => ';', |
444
|
|
|
',,' => ',' |
445
|
|
|
); |
446
|
|
|
|
447
|
|
|
foreach($search as $char => $separator) { |
448
|
|
|
if(strpos($this->csv, $char) !== false) { |
449
|
|
|
return $separator; |
450
|
|
|
} |
451
|
|
|
} |
452
|
|
|
|
453
|
|
|
return $this->separator; |
454
|
|
|
} |
455
|
|
|
|
456
|
|
|
/** |
457
|
|
|
* Creates a new CSV parser instance. |
458
|
|
|
* |
459
|
|
|
* @param string $delimiter |
460
|
|
|
* @return Csv |
461
|
|
|
*/ |
462
|
|
|
public static function createParser(string $delimiter=self::DELIMITER_AUTO) : Csv |
463
|
|
|
{ |
464
|
|
|
$csv = new Csv(); |
465
|
|
|
|
466
|
|
|
if($delimiter !== self::DELIMITER_AUTO) { |
467
|
|
|
$csv->delimiter = $delimiter; |
468
|
|
|
} |
469
|
|
|
|
470
|
|
|
return $csv; |
471
|
|
|
} |
472
|
|
|
|
473
|
|
|
/** |
474
|
|
|
* Parses a CSV file in automatic mode (to detect the delimiter and |
475
|
|
|
* enclosure), and returns the data rows, including the header row |
476
|
|
|
* if any. |
477
|
|
|
* |
478
|
|
|
* @param string $path |
479
|
|
|
* @return array<int,array<int,mixed>> |
480
|
|
|
* |
481
|
|
|
* @throws CSVHelper_Exception |
482
|
|
|
* @throws FileHelper_Exception |
483
|
|
|
* |
484
|
|
|
* @see CSVHelper::ERROR_CSV_FILE_NOT_READABLE |
485
|
|
|
* @see CSVHelper::ERROR_FILE_PARSING_FAILED |
486
|
|
|
*/ |
487
|
|
|
public static function parseFile(string $path) : array |
488
|
|
|
{ |
489
|
|
|
$path = FileHelper::requireFileReadable($path, self::ERROR_CSV_FILE_NOT_READABLE); |
490
|
|
|
|
491
|
|
|
$parser = self::createParser(); |
492
|
|
|
$parser->heading = false; |
493
|
|
|
|
494
|
|
|
$result = $parser->auto($path); |
495
|
|
|
|
496
|
|
|
if(is_string($result)) { |
497
|
|
|
return $parser->data; |
498
|
|
|
} |
499
|
|
|
|
500
|
|
|
throw new CSVHelper_Exception( |
501
|
|
|
'Cannot parse CSV file', |
502
|
|
|
sprintf( |
503
|
|
|
'The file [%s] could not be parsed.'.PHP_EOL. |
504
|
|
|
'Additional information: '.PHP_EOL. |
505
|
|
|
'%s', |
506
|
|
|
$path, |
507
|
|
|
JSONConverter::var2jsonSilent($parser->error_info) |
508
|
|
|
), |
509
|
|
|
self::ERROR_FILE_PARSING_FAILED |
510
|
|
|
); |
511
|
|
|
} |
512
|
|
|
|
513
|
|
|
/** |
514
|
|
|
* Parses a CSV string in automatic mode (to detect the delimiter and |
515
|
|
|
* enclosure), and returns the data rows, including the header row |
516
|
|
|
* if any. |
517
|
|
|
* |
518
|
|
|
* @param string $string |
519
|
|
|
* @return array<int,array<int,mixed>> |
520
|
|
|
* @throws CSVHelper_Exception |
521
|
|
|
* |
522
|
|
|
* @see CSVHelper::ERROR_STRING_PARSING_FAILED |
523
|
|
|
*/ |
524
|
|
|
public static function parseString(string $string) : array |
525
|
|
|
{ |
526
|
|
|
$parser = self::createParser(); |
527
|
|
|
$result = $parser->parse($string); |
528
|
|
|
|
529
|
|
|
if($result === true) { |
530
|
|
|
return $parser->data; |
531
|
|
|
} |
532
|
|
|
|
533
|
|
|
throw new CSVHelper_Exception( |
534
|
|
|
'Cannot parse CSV string', |
535
|
|
|
'The string could not be parsed. No additional information is available.', |
536
|
|
|
self::ERROR_STRING_PARSING_FAILED |
537
|
|
|
); |
538
|
|
|
} |
539
|
|
|
} |
540
|
|
|
|