|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Palmtree\Csv; |
|
4
|
|
|
|
|
5
|
|
|
use Palmtree\Csv\Normalizer\NormalizerInterface; |
|
6
|
|
|
use Palmtree\Csv\Normalizer\NullNormalizer; |
|
7
|
|
|
use Palmtree\Csv\Row\Row; |
|
8
|
|
|
use Palmtree\Csv\Util\StringUtil; |
|
9
|
|
|
|
|
10
|
|
|
/** |
|
11
|
|
|
* Reads a CSV file by loading each line into memory one at a time. |
|
12
|
|
|
*/ |
|
13
|
|
|
class Reader extends AbstractCsvDocument implements \Iterator |
|
14
|
|
|
{ |
|
15
|
|
|
/** @var string */ |
|
16
|
|
|
private $defaultNormalizer = NullNormalizer::class; |
|
17
|
|
|
/** @var NormalizerInterface|null */ |
|
18
|
|
|
private $headerNormalizer; |
|
19
|
|
|
/** @var NormalizerInterface[] */ |
|
20
|
|
|
private $normalizers = []; |
|
21
|
|
|
/** @var Row|null */ |
|
22
|
|
|
private $headers; |
|
23
|
|
|
/** @var Row */ |
|
24
|
|
|
private $row; |
|
25
|
|
|
/** @var string|null */ |
|
26
|
|
|
private $stripBom = StringUtil::BOM_UTF8; |
|
27
|
|
|
/** @var int */ |
|
28
|
|
|
private $offset = 0; |
|
29
|
|
|
/** @var int */ |
|
30
|
|
|
private $headerOffset = 0; |
|
31
|
|
|
|
|
32
|
|
|
public static function read(string $filePath, bool $hasHeaders = true): self |
|
33
|
|
|
{ |
|
34
|
|
|
return new self($filePath, $hasHeaders); |
|
35
|
|
|
} |
|
36
|
|
|
|
|
37
|
3 |
|
public function getHeaders(): ?Row |
|
38
|
|
|
{ |
|
39
|
3 |
|
if ($this->hasHeaders && $this->headers === null) { |
|
40
|
3 |
|
$this->rewind(); |
|
41
|
|
|
} |
|
42
|
|
|
|
|
43
|
3 |
|
return $this->headers; |
|
44
|
|
|
} |
|
45
|
|
|
|
|
46
|
|
|
/** |
|
47
|
|
|
* @param string|int $key |
|
48
|
|
|
* |
|
49
|
|
|
* @return string|int |
|
50
|
|
|
*/ |
|
51
|
10 |
|
public function getHeader($key) |
|
52
|
|
|
{ |
|
53
|
10 |
|
return $this->headers[$key] ?? $key; |
|
54
|
|
|
} |
|
55
|
|
|
|
|
56
|
|
|
public function setHeaderNormalizer(NormalizerInterface $headerNormalizer): self |
|
57
|
|
|
{ |
|
58
|
|
|
$this->headerNormalizer = $headerNormalizer; |
|
59
|
|
|
|
|
60
|
|
|
return $this; |
|
61
|
|
|
} |
|
62
|
|
|
|
|
63
|
8 |
|
public function getHeaderNormalizer(): NormalizerInterface |
|
64
|
|
|
{ |
|
65
|
8 |
|
if ($this->headerNormalizer === null) { |
|
66
|
8 |
|
$this->headerNormalizer = new NullNormalizer(); |
|
67
|
|
|
} |
|
68
|
|
|
|
|
69
|
8 |
|
return $this->headerNormalizer; |
|
70
|
|
|
} |
|
71
|
|
|
|
|
72
|
|
|
public function addNormalizer(string $key, NormalizerInterface $normalizer): self |
|
73
|
|
|
{ |
|
74
|
|
|
$this->normalizers[$key] = $normalizer; |
|
75
|
|
|
|
|
76
|
|
|
return $this; |
|
77
|
|
|
} |
|
78
|
|
|
|
|
79
|
|
|
public function addNormalizers(iterable $normalizers): self |
|
80
|
|
|
{ |
|
81
|
|
|
foreach ($normalizers as $key => $normalizer) { |
|
82
|
|
|
$this->addNormalizer($key, $normalizer); |
|
83
|
|
|
} |
|
84
|
|
|
|
|
85
|
|
|
return $this; |
|
86
|
|
|
} |
|
87
|
|
|
|
|
88
|
|
|
/** |
|
89
|
|
|
* @param string|int $key |
|
90
|
|
|
*/ |
|
91
|
10 |
|
public function getNormalizer($key): NormalizerInterface |
|
92
|
|
|
{ |
|
93
|
10 |
|
if ($this->hasHeaders && \is_int($key)) { |
|
94
|
8 |
|
$this->normalizers[$key] = $this->getHeaderNormalizer(); |
|
95
|
|
|
} |
|
96
|
|
|
|
|
97
|
10 |
|
if (!isset($this->normalizers[$key])) { |
|
98
|
5 |
|
$class = $this->getDefaultNormalizer(); |
|
99
|
|
|
|
|
100
|
5 |
|
$this->normalizers[$key] = new $class(); |
|
101
|
|
|
} |
|
102
|
|
|
|
|
103
|
10 |
|
return $this->normalizers[$key]; |
|
104
|
|
|
} |
|
105
|
|
|
|
|
106
|
|
|
/** |
|
107
|
|
|
* Reads the next line in the CSV file and returns a Row object from it. |
|
108
|
|
|
*/ |
|
109
|
7 |
|
private function getCurrentRow(): ?Row |
|
110
|
|
|
{ |
|
111
|
7 |
|
$cells = $this->getDocument()->current(); |
|
112
|
|
|
|
|
113
|
7 |
|
if (!\is_array($cells) || $cells == [null]) { |
|
114
|
5 |
|
return null; |
|
115
|
|
|
} |
|
116
|
|
|
|
|
117
|
7 |
|
if ($this->key() === 0 && $this->stripBom) { |
|
118
|
5 |
|
$stripped = StringUtil::stripBom($cells[0], $this->stripBom); |
|
119
|
|
|
|
|
120
|
5 |
|
if ($stripped !== $cells[0]) { |
|
121
|
1 |
|
$cells[0] = \trim($stripped, $this->enclosure); |
|
122
|
|
|
} |
|
123
|
|
|
} |
|
124
|
|
|
|
|
125
|
7 |
|
return new Row($cells, $this); |
|
126
|
|
|
} |
|
127
|
|
|
|
|
128
|
|
|
/** |
|
129
|
|
|
* @inheritDoc |
|
130
|
|
|
*/ |
|
131
|
5 |
|
public function current(): Row |
|
132
|
|
|
{ |
|
133
|
5 |
|
return $this->row; |
|
134
|
|
|
} |
|
135
|
|
|
|
|
136
|
|
|
/** |
|
137
|
|
|
* @inheritDoc |
|
138
|
|
|
*/ |
|
139
|
5 |
|
public function next(): void |
|
140
|
|
|
{ |
|
141
|
5 |
|
$this->getDocument()->next(); |
|
142
|
5 |
|
} |
|
143
|
|
|
|
|
144
|
|
|
/** |
|
145
|
|
|
* @inheritDoc |
|
146
|
|
|
*/ |
|
147
|
7 |
|
public function key(): int |
|
148
|
|
|
{ |
|
149
|
7 |
|
return $this->getDocument()->key(); |
|
150
|
|
|
} |
|
151
|
|
|
|
|
152
|
|
|
/** |
|
153
|
|
|
* @inheritDoc |
|
154
|
|
|
*/ |
|
155
|
5 |
|
public function valid(): bool |
|
156
|
|
|
{ |
|
157
|
5 |
|
$this->row = $this->getCurrentRow(); |
|
158
|
|
|
|
|
159
|
5 |
|
return $this->row instanceof Row; |
|
160
|
|
|
} |
|
161
|
|
|
|
|
162
|
|
|
/** |
|
163
|
|
|
* @inheritDoc |
|
164
|
|
|
*/ |
|
165
|
7 |
|
public function rewind(): void |
|
166
|
|
|
{ |
|
167
|
7 |
|
$this->getDocument()->rewind(); |
|
168
|
|
|
|
|
169
|
7 |
|
$dataOffset = $this->offset + $this->headerOffset; |
|
170
|
7 |
|
if ($this->hasHeaders) { |
|
171
|
5 |
|
if ($this->headerOffset) { |
|
172
|
1 |
|
$this->getDocument()->seek($this->headerOffset); |
|
173
|
|
|
} |
|
174
|
|
|
|
|
175
|
|
|
// Set headers to null first so the header row is a zero-based array and can be used |
|
176
|
|
|
// to set the array keys of all other rows. |
|
177
|
5 |
|
$this->headers = null; |
|
178
|
5 |
|
$this->headers = $this->getCurrentRow(); |
|
179
|
|
|
|
|
180
|
5 |
|
++$dataOffset; |
|
181
|
|
|
} |
|
182
|
|
|
|
|
183
|
7 |
|
if ($dataOffset > 0) { |
|
184
|
5 |
|
$this->getDocument()->seek($dataOffset); |
|
185
|
|
|
} |
|
186
|
7 |
|
} |
|
187
|
|
|
|
|
188
|
|
|
public function setDefaultNormalizer(string $defaultNormalizer): self |
|
189
|
|
|
{ |
|
190
|
|
|
$this->defaultNormalizer = $defaultNormalizer; |
|
191
|
|
|
|
|
192
|
|
|
return $this; |
|
193
|
|
|
} |
|
194
|
|
|
|
|
195
|
5 |
|
public function getDefaultNormalizer(): string |
|
196
|
|
|
{ |
|
197
|
5 |
|
return $this->defaultNormalizer; |
|
198
|
|
|
} |
|
199
|
|
|
|
|
200
|
1 |
|
public function setStripBom(?string $stripBom): self |
|
201
|
|
|
{ |
|
202
|
1 |
|
$this->stripBom = $stripBom; |
|
203
|
|
|
|
|
204
|
1 |
|
return $this; |
|
205
|
|
|
} |
|
206
|
|
|
|
|
207
|
1 |
|
public function setOffset(int $offset): self |
|
208
|
|
|
{ |
|
209
|
1 |
|
$this->offset = $offset; |
|
210
|
|
|
|
|
211
|
1 |
|
return $this; |
|
212
|
|
|
} |
|
213
|
|
|
|
|
214
|
|
|
public function getOffset(): int |
|
215
|
|
|
{ |
|
216
|
|
|
return $this->offset; |
|
217
|
|
|
} |
|
218
|
|
|
|
|
219
|
1 |
|
public function setHeaderOffset(int $headerOffset): self |
|
220
|
|
|
{ |
|
221
|
1 |
|
$this->headerOffset = $headerOffset; |
|
222
|
|
|
|
|
223
|
1 |
|
return $this; |
|
224
|
|
|
} |
|
225
|
|
|
|
|
226
|
|
|
public function getHeaderOffset(): int |
|
227
|
|
|
{ |
|
228
|
|
|
return $this->headerOffset; |
|
229
|
|
|
} |
|
230
|
|
|
|
|
231
|
3 |
|
public function toArray(): array |
|
232
|
|
|
{ |
|
233
|
3 |
|
$result = []; |
|
234
|
3 |
|
foreach ($this as $row) { |
|
235
|
3 |
|
$result[] = $row->toArray(); |
|
236
|
|
|
} |
|
237
|
|
|
|
|
238
|
3 |
|
return $result; |
|
239
|
|
|
} |
|
240
|
|
|
|
|
241
|
7 |
|
protected function getOpenMode(): string |
|
242
|
|
|
{ |
|
243
|
7 |
|
return 'r'; |
|
244
|
|
|
} |
|
245
|
|
|
} |
|
246
|
|
|
|