1
|
|
|
<?php |
2
|
|
|
/** |
3
|
|
|
* This file is part of the League.csv library |
4
|
|
|
* |
5
|
|
|
* @license http://opensource.org/licenses/MIT |
6
|
|
|
* @link https://github.com/thephpleague/csv/ |
7
|
|
|
* @version 9.1.0 |
8
|
|
|
* @package League.csv |
9
|
|
|
* |
10
|
|
|
* For the full copyright and license information, please view the LICENSE |
11
|
|
|
* file that was distributed with this source code. |
12
|
|
|
*/ |
13
|
|
|
declare(strict_types=1); |
14
|
|
|
|
15
|
|
|
namespace League\Csv; |
16
|
|
|
|
17
|
|
|
use BadMethodCallException; |
18
|
|
|
use CallbackFilterIterator; |
19
|
|
|
use Countable; |
20
|
|
|
use Iterator; |
21
|
|
|
use IteratorAggregate; |
22
|
|
|
use JsonSerializable; |
23
|
|
|
use SplFileObject; |
24
|
|
|
use TypeError; |
25
|
|
|
|
26
|
|
|
/** |
27
|
|
|
* A class to select records from a CSV document |
28
|
|
|
* |
29
|
|
|
* @package League.csv |
30
|
|
|
* @since 3.0.0 |
31
|
|
|
* |
32
|
|
|
* @method array fetchOne(int $nth_record = 0) Returns a single record from the CSV |
33
|
|
|
* @method Generator fetchColumn(string|int $column_index) Returns the next value from a single CSV record field |
34
|
|
|
* @method Generator fetchPairs(string|int $offset_index = 0, string|int $value_index = 1) Fetches the next key-value pairs from the CSV document |
35
|
|
|
*/ |
36
|
|
|
class Reader extends AbstractCsv implements Countable, IteratorAggregate, JsonSerializable |
37
|
|
|
{ |
38
|
|
|
/** |
39
|
|
|
* header offset |
40
|
|
|
* |
41
|
|
|
* @var int|null |
42
|
|
|
*/ |
43
|
|
|
protected $header_offset; |
44
|
|
|
|
45
|
|
|
/** |
46
|
|
|
* header record |
47
|
|
|
* |
48
|
|
|
* @var string[] |
49
|
|
|
*/ |
50
|
|
|
protected $header = []; |
51
|
|
|
|
52
|
|
|
/** |
53
|
|
|
* records count |
54
|
|
|
* |
55
|
|
|
* @var int |
56
|
|
|
*/ |
57
|
|
|
protected $nb_records = -1; |
58
|
|
|
|
59
|
|
|
/** |
60
|
|
|
* @inheritdoc |
61
|
|
|
*/ |
62
|
|
|
protected $stream_filter_mode = STREAM_FILTER_READ; |
63
|
|
|
|
64
|
|
|
/** |
65
|
|
|
* @inheritdoc |
66
|
|
|
*/ |
67
|
2 |
|
public static function createFromPath(string $path, string $open_mode = 'r', $context = null): AbstractCsv |
68
|
|
|
{ |
69
|
2 |
|
return new static(Stream::createFromPath($path, $open_mode, $context)); |
70
|
|
|
} |
71
|
|
|
|
72
|
|
|
/** |
73
|
|
|
* Returns the header offset |
74
|
|
|
* |
75
|
|
|
* If no CSV header offset is set this method MUST return null |
76
|
|
|
* |
77
|
|
|
* @return int|null |
78
|
|
|
*/ |
79
|
10 |
|
public function getHeaderOffset() |
80
|
|
|
{ |
81
|
10 |
|
return $this->header_offset; |
82
|
|
|
} |
83
|
|
|
|
84
|
|
|
/** |
85
|
|
|
* Returns the CSV record used as header |
86
|
|
|
* |
87
|
|
|
* The returned header is represented as an array of string values |
88
|
|
|
* |
89
|
|
|
* @return string[] |
90
|
|
|
*/ |
91
|
10 |
|
public function getHeader(): array |
92
|
|
|
{ |
93
|
10 |
|
if (null === $this->header_offset) { |
94
|
8 |
|
return $this->header; |
95
|
|
|
} |
96
|
|
|
|
97
|
4 |
|
if (!empty($this->header)) { |
98
|
2 |
|
return $this->header; |
99
|
|
|
} |
100
|
|
|
|
101
|
4 |
|
$this->header = $this->setHeader($this->header_offset); |
102
|
|
|
|
103
|
4 |
|
return $this->header; |
104
|
|
|
} |
105
|
|
|
|
106
|
|
|
/** |
107
|
|
|
* Determine the CSV record header |
108
|
|
|
* |
109
|
|
|
* @param int $offset |
110
|
|
|
* |
111
|
|
|
* @throws Exception If the header offset is set and no record is found or is the empty array |
112
|
|
|
* |
113
|
|
|
* @return string[] |
114
|
|
|
*/ |
115
|
6 |
|
protected function setHeader(int $offset): array |
116
|
|
|
{ |
117
|
6 |
|
$this->document->setFlags(SplFileObject::READ_CSV | SplFileObject::READ_AHEAD | SplFileObject::SKIP_EMPTY); |
118
|
6 |
|
$this->document->setCsvControl($this->delimiter, $this->enclosure, $this->escape); |
119
|
6 |
|
$this->document->seek($offset); |
120
|
6 |
|
if (empty($header = $this->document->current())) { |
121
|
2 |
|
throw new Exception(sprintf('The header record does not exist or is empty at offset: `%s`', $offset)); |
122
|
|
|
} |
123
|
|
|
|
124
|
4 |
|
if (0 === $offset) { |
125
|
2 |
|
return $this->removeBOM($header, mb_strlen($this->getInputBOM()), $this->enclosure); |
126
|
|
|
} |
127
|
|
|
|
128
|
2 |
|
return $header; |
129
|
|
|
} |
130
|
|
|
|
131
|
|
|
/** |
132
|
|
|
* Strip the BOM sequence from a record |
133
|
|
|
* |
134
|
|
|
* @param string[] $record |
135
|
|
|
* @param int $bom_length |
136
|
|
|
* @param string $enclosure |
137
|
|
|
* |
138
|
|
|
* @return string[] |
139
|
|
|
*/ |
140
|
8 |
|
protected function removeBOM(array $record, int $bom_length, string $enclosure): array |
141
|
|
|
{ |
142
|
8 |
|
if (0 == $bom_length) { |
143
|
2 |
|
return $record; |
144
|
|
|
} |
145
|
|
|
|
146
|
6 |
|
$record[0] = mb_substr($record[0], $bom_length); |
147
|
6 |
|
if ($enclosure.$enclosure != substr($record[0].$record[0], strlen($record[0]) - 1, 2)) { |
148
|
4 |
|
return $record; |
149
|
|
|
} |
150
|
|
|
|
151
|
2 |
|
$record[0] = substr($record[0], 1, -1); |
152
|
|
|
|
153
|
2 |
|
return $record; |
154
|
|
|
} |
155
|
|
|
|
156
|
|
|
/** |
157
|
|
|
* @inheritdoc |
158
|
|
|
*/ |
159
|
6 |
|
public function __call($method, array $arguments) |
160
|
|
|
{ |
161
|
6 |
|
static $whitelisted = ['fetchColumn' => 1, 'fetchOne' => 1, 'fetchPairs' => 1]; |
162
|
6 |
|
if (isset($whitelisted[$method])) { |
163
|
2 |
|
return (new ResultSet($this->getRecords(), $this->getHeader()))->$method(...$arguments); |
164
|
|
|
} |
165
|
|
|
|
166
|
4 |
|
throw new BadMethodCallException(sprintf('%s::%s() method does not exist', __CLASS__, $method)); |
167
|
|
|
} |
168
|
|
|
|
169
|
|
|
/** |
170
|
|
|
* @inheritdoc |
171
|
|
|
*/ |
172
|
2 |
|
public function count(): int |
173
|
|
|
{ |
174
|
2 |
|
if (-1 === $this->nb_records) { |
175
|
2 |
|
$this->nb_records = iterator_count($this->getRecords()); |
176
|
|
|
} |
177
|
|
|
|
178
|
2 |
|
return $this->nb_records; |
179
|
|
|
} |
180
|
|
|
|
181
|
|
|
/** |
182
|
|
|
* @inheritdoc |
183
|
|
|
*/ |
184
|
2 |
|
public function getIterator(): Iterator |
185
|
|
|
{ |
186
|
2 |
|
return $this->getRecords(); |
187
|
|
|
} |
188
|
|
|
|
189
|
|
|
/** |
190
|
|
|
* @inheritdoc |
191
|
|
|
*/ |
192
|
2 |
|
public function jsonSerialize(): array |
193
|
|
|
{ |
194
|
2 |
|
return iterator_to_array($this->getRecords(), false); |
195
|
|
|
} |
196
|
|
|
|
197
|
|
|
/** |
198
|
|
|
* Returns the CSV records as an iterator object. |
199
|
|
|
* |
200
|
|
|
* Each CSV record is represented as a simple array containig strings or null values. |
201
|
|
|
* |
202
|
|
|
* If the CSV document has a header record then each record is combined |
203
|
|
|
* to the header record and the header record is removed from the iterator. |
204
|
|
|
* |
205
|
|
|
* If the CSV document is inconsistent. Missing record fields are |
206
|
|
|
* filled with null values while extra record fields are strip from |
207
|
|
|
* the returned object. |
208
|
|
|
* |
209
|
|
|
* @param string[] $header an optional header to use instead of the CSV document header |
210
|
|
|
* |
211
|
|
|
* @return Iterator |
212
|
|
|
*/ |
213
|
12 |
|
public function getRecords(array $header = []): Iterator |
214
|
|
|
{ |
215
|
12 |
|
$header = $this->computeHeader($header); |
216
|
10 |
|
$normalized = function ($record): bool { |
217
|
10 |
|
return is_array($record) && $record != [null]; |
218
|
10 |
|
}; |
219
|
10 |
|
$bom = $this->getInputBOM(); |
220
|
10 |
|
$this->document->setFlags(SplFileObject::READ_CSV | SplFileObject::READ_AHEAD | SplFileObject::SKIP_EMPTY); |
221
|
10 |
|
$this->document->setCsvControl($this->delimiter, $this->enclosure, $this->escape); |
222
|
|
|
|
223
|
10 |
|
$records = $this->stripBOM(new CallbackFilterIterator($this->document, $normalized), $bom); |
224
|
10 |
|
if (null !== $this->header_offset) { |
225
|
4 |
|
$records = new CallbackFilterIterator($records, function (array $record, int $offset): bool { |
226
|
4 |
|
return $offset !== $this->header_offset; |
227
|
4 |
|
}); |
228
|
|
|
} |
229
|
|
|
|
230
|
10 |
|
return $this->combineHeader($records, $header); |
231
|
|
|
} |
232
|
|
|
|
233
|
|
|
/** |
234
|
|
|
* Returns the header to be used for iteration |
235
|
|
|
* |
236
|
|
|
* @param string[] $header |
237
|
|
|
* |
238
|
|
|
* @throws Exception If the header contains non unique column name |
239
|
|
|
* |
240
|
|
|
* @return string[] |
241
|
|
|
*/ |
242
|
16 |
|
protected function computeHeader(array $header) |
243
|
|
|
{ |
244
|
16 |
|
if (empty($header)) { |
245
|
14 |
|
$header = $this->getHeader(); |
246
|
|
|
} |
247
|
|
|
|
248
|
16 |
|
if ($header === array_unique(array_filter($header, 'is_string'))) { |
249
|
14 |
|
return $header; |
250
|
|
|
} |
251
|
|
|
|
252
|
2 |
|
throw new Exception('The header record must be empty or a flat array with unique string values'); |
253
|
|
|
} |
254
|
|
|
|
255
|
|
|
/** |
256
|
|
|
* Combine the CSV header to each record if present |
257
|
|
|
* |
258
|
|
|
* @param Iterator $iterator |
259
|
|
|
* @param string[] $header |
260
|
|
|
* |
261
|
|
|
* @return Iterator |
262
|
|
|
*/ |
263
|
20 |
|
protected function combineHeader(Iterator $iterator, array $header): Iterator |
264
|
|
|
{ |
265
|
20 |
|
if (empty($header)) { |
266
|
14 |
|
return $iterator; |
267
|
|
|
} |
268
|
|
|
|
269
|
8 |
|
$field_count = count($header); |
270
|
8 |
|
$mapper = function (array $record) use ($header, $field_count): array { |
271
|
8 |
|
if (count($record) != $field_count) { |
272
|
4 |
|
$record = array_slice(array_pad($record, $field_count, null), 0, $field_count); |
273
|
|
|
} |
274
|
|
|
|
275
|
8 |
|
return array_combine($header, $record); |
276
|
8 |
|
}; |
277
|
|
|
|
278
|
8 |
|
return new MapIterator($iterator, $mapper); |
279
|
|
|
} |
280
|
|
|
|
281
|
|
|
/** |
282
|
|
|
* Strip the BOM sequence from the returned records if necessary |
283
|
|
|
* |
284
|
|
|
* @param Iterator $iterator |
285
|
|
|
* @param string $bom |
286
|
|
|
* |
287
|
|
|
* @return Iterator |
288
|
|
|
*/ |
289
|
16 |
|
protected function stripBOM(Iterator $iterator, string $bom): Iterator |
290
|
|
|
{ |
291
|
16 |
|
if ('' === $bom) { |
292
|
10 |
|
return $iterator; |
293
|
|
|
} |
294
|
|
|
|
295
|
6 |
|
$bom_length = mb_strlen($bom); |
296
|
6 |
|
$mapper = function (array $record, int $index) use ($bom_length): array { |
297
|
6 |
|
if (0 != $index) { |
298
|
2 |
|
return $record; |
299
|
|
|
} |
300
|
|
|
|
301
|
6 |
|
return $this->removeBOM($record, $bom_length, $this->enclosure); |
302
|
6 |
|
}; |
303
|
|
|
|
304
|
6 |
|
return new MapIterator($iterator, $mapper); |
305
|
|
|
} |
306
|
|
|
|
307
|
|
|
/** |
308
|
|
|
* Selects the record to be used as the CSV header |
309
|
|
|
* |
310
|
|
|
* Because the header is represented as an array, to be valid |
311
|
|
|
* a header MUST contain only unique string value. |
312
|
|
|
* |
313
|
|
|
* @param int|null $offset the header record offset |
314
|
|
|
* |
315
|
|
|
* @throws Exception if the offset is a negative integer |
316
|
|
|
* |
317
|
|
|
* @return static |
318
|
|
|
*/ |
319
|
14 |
|
public function setHeaderOffset($offset): self |
320
|
|
|
{ |
321
|
14 |
|
if ($offset === $this->header_offset) { |
322
|
8 |
|
return $this; |
323
|
|
|
} |
324
|
|
|
|
325
|
6 |
|
if (!is_nullable_int($offset)) { |
326
|
2 |
|
throw new TypeError(sprintf(__METHOD__.'() expects 1 Argument to be null or an integer %s given', gettype($offset))); |
327
|
|
|
} |
328
|
|
|
|
329
|
4 |
|
if (null !== $offset && 0 > $offset) { |
330
|
2 |
|
throw new Exception(__METHOD__.'() expects 1 Argument to be greater or equal to 0'); |
331
|
|
|
} |
332
|
|
|
|
333
|
2 |
|
$this->header_offset = $offset; |
334
|
2 |
|
$this->resetProperties(); |
335
|
|
|
|
336
|
2 |
|
return $this; |
337
|
|
|
} |
338
|
|
|
|
339
|
|
|
/** |
340
|
|
|
* @inheritdoc |
341
|
|
|
*/ |
342
|
8 |
|
protected function resetProperties() |
343
|
|
|
{ |
344
|
8 |
|
$this->nb_records = -1; |
345
|
8 |
|
$this->header = []; |
346
|
8 |
|
} |
347
|
|
|
} |
348
|
|
|
|