|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace SPSS\Sav; |
|
4
|
|
|
|
|
5
|
|
|
use SPSS\Buffer; |
|
6
|
|
|
use SPSS\Exception; |
|
7
|
|
|
use SPSS\Sav\Record\Info; |
|
8
|
|
|
use SPSS\Utils; |
|
9
|
|
|
|
|
10
|
|
|
class Writer |
|
11
|
|
|
{ |
|
12
|
|
|
/** |
|
13
|
|
|
* @var Record\Header |
|
14
|
|
|
*/ |
|
15
|
|
|
public $header; |
|
16
|
|
|
|
|
17
|
|
|
/** |
|
18
|
|
|
* @var Record\Variable[] |
|
19
|
|
|
*/ |
|
20
|
|
|
public $variables = []; |
|
21
|
|
|
|
|
22
|
|
|
/** |
|
23
|
|
|
* @var Record\ValueLabel[] |
|
24
|
|
|
*/ |
|
25
|
|
|
public $valueLabels = []; |
|
26
|
|
|
|
|
27
|
|
|
/** |
|
28
|
|
|
* @var Record\Document |
|
29
|
|
|
*/ |
|
30
|
|
|
public $document; |
|
31
|
|
|
|
|
32
|
|
|
/** |
|
33
|
|
|
* @var InfoRecordSet |
|
34
|
|
|
*/ |
|
35
|
|
|
private $info = []; |
|
36
|
|
|
|
|
37
|
|
|
/** |
|
38
|
|
|
* @var Record\Data |
|
39
|
|
|
*/ |
|
40
|
|
|
public $data; |
|
41
|
|
|
|
|
42
|
|
|
/** |
|
43
|
|
|
* @var Buffer |
|
44
|
|
|
*/ |
|
45
|
|
|
protected $buffer; |
|
46
|
|
|
|
|
47
|
|
|
/** |
|
48
|
|
|
* Writer constructor. |
|
49
|
|
|
* |
|
50
|
|
|
* @param array $data |
|
51
|
|
|
* @throws \Exception |
|
52
|
|
|
*/ |
|
53
|
|
|
public function __construct($data = []) |
|
54
|
|
|
{ |
|
55
|
|
|
$this->buffer = Buffer::factory(); |
|
56
|
|
|
$this->buffer->context = $this; |
|
57
|
|
|
|
|
58
|
|
|
$this->info = new InfoRecordSet(); |
|
59
|
|
|
if (! empty($data)) { |
|
60
|
|
|
$this->write($data); |
|
61
|
|
|
} |
|
62
|
|
|
} |
|
63
|
|
|
|
|
64
|
|
|
/** |
|
65
|
|
|
* @param array $data |
|
66
|
|
|
* @throws \Exception |
|
67
|
|
|
*/ |
|
68
|
|
|
public function write($data) |
|
69
|
|
|
{ |
|
70
|
|
|
$this->header = new Record\Header($data['header']); |
|
71
|
|
|
$this->header->nominalCaseSize = 0; |
|
72
|
|
|
$this->header->casesCount = 0; |
|
73
|
|
|
|
|
74
|
|
|
$this->info->set($this->prepareInfoRecord( |
|
75
|
|
|
Record\Info\MachineInteger::class, |
|
76
|
|
|
$data |
|
77
|
|
|
)); |
|
78
|
|
|
|
|
79
|
|
|
$this->info->set($this->prepareInfoRecord( |
|
80
|
|
|
Record\Info\MachineFloatingPoint::class, |
|
81
|
|
|
$data |
|
82
|
|
|
)); |
|
83
|
|
|
|
|
84
|
|
|
$this->info->set(new Record\Info\VariableDisplayParam()); |
|
85
|
|
|
$this->info->set(new Record\Info\LongVariableNames()); |
|
86
|
|
|
$this->info->set(new Record\Info\VeryLongString()); |
|
87
|
|
|
$this->info->set($this->prepareInfoRecord( |
|
88
|
|
|
Record\Info\ExtendedNumberOfCases::class, |
|
89
|
|
|
$data |
|
90
|
|
|
)); |
|
91
|
|
|
$this->info->set(new Record\Info\VariableAttributes()); |
|
92
|
|
|
$this->info->set(new Record\Info\LongStringValueLabels()); |
|
93
|
|
|
$this->info->set(new Record\Info\LongStringMissingValues()); |
|
94
|
|
|
$this->info->set(new Record\Info\CharacterEncoding('UTF-8')); |
|
95
|
|
|
|
|
96
|
|
|
$this->data = new Record\Data(); |
|
97
|
|
|
|
|
98
|
|
|
$nominalIdx = 0; |
|
99
|
|
|
|
|
100
|
|
|
/** |
|
101
|
|
|
* @var bool[string] The variable names used in this SPSS file |
|
102
|
|
|
*/ |
|
103
|
|
|
$variableNames = []; |
|
104
|
|
|
/** @var Variable $var */ |
|
105
|
|
|
foreach (array_values($data['variables']) as $idx => $var) { |
|
106
|
|
|
if (!$var instanceof Variable) { |
|
107
|
|
|
throw new \InvalidArgumentException('Variables must be instance of ' . Variable::class); |
|
108
|
|
|
} |
|
109
|
|
|
|
|
110
|
|
|
$variable = new Record\Variable(); |
|
111
|
|
|
|
|
112
|
|
|
/** |
|
113
|
|
|
* @see \SPSS\Sav\Record\Variable::getSegmentName() |
|
114
|
|
|
* |
|
115
|
|
|
* Variable names in the SPSS file should be unique. If they are not, SPSS will rename them. |
|
116
|
|
|
* If SPSS renames them and it happens to be a long string then the segments will no longer share |
|
117
|
|
|
* the required prefix in the name. |
|
118
|
|
|
*/ |
|
119
|
|
|
$name = strtoupper(substr($var->getName(), 0, 8)); |
|
120
|
|
|
|
|
121
|
|
|
$counter = 0; |
|
122
|
|
|
/** |
|
123
|
|
|
* Using base convert we can encode 36^3 = 46656 variables with a common 5 character prefix in an 8 |
|
124
|
|
|
* character variable name. This should suffice since the current variable limit of SPSS is 32767 |
|
125
|
|
|
* variables. |
|
126
|
|
|
*/ |
|
127
|
|
|
while (isset($variableNames[$name])) { |
|
128
|
|
|
$name = strtoupper(substr($var->getName(), 0, 5) . base_convert($counter, 10, 36)); |
|
129
|
|
|
$counter++; |
|
130
|
|
|
} |
|
131
|
|
|
|
|
132
|
|
|
$variableNames[$name] = true; |
|
133
|
|
|
$variable->name = $name; |
|
134
|
|
|
|
|
135
|
|
|
if ($var->format == Variable::FORMAT_TYPE_A) { |
|
136
|
|
|
$variable->width = $var->getWidth(); |
|
137
|
|
|
} else { |
|
138
|
|
|
$variable->width = 0; |
|
139
|
|
|
} |
|
140
|
|
|
|
|
141
|
|
|
$variable->label = $var->label; |
|
142
|
|
|
$variable->print = [ |
|
143
|
|
|
0, |
|
144
|
|
|
$var->format, |
|
145
|
|
|
min($var->getWidth(), 255), |
|
146
|
|
|
$var->decimals, |
|
147
|
|
|
]; |
|
148
|
|
|
$variable->write = [ |
|
149
|
|
|
0, |
|
150
|
|
|
$var->format, |
|
151
|
|
|
min($var->getWidth(), 255), |
|
152
|
|
|
$var->decimals, |
|
153
|
|
|
]; |
|
154
|
|
|
|
|
155
|
|
|
// TODO: refactory |
|
156
|
|
|
$shortName = $variable->name; |
|
157
|
|
|
$longName = $var->getName(); |
|
158
|
|
|
|
|
159
|
|
|
if ($var->attributes) { |
|
|
|
|
|
|
160
|
|
|
$this->info[Record\Info\VariableAttributes::SUBTYPE][$longName] = $var->attributes; |
|
161
|
|
|
} |
|
162
|
|
|
|
|
163
|
|
|
if ($var->missing) { |
|
|
|
|
|
|
164
|
|
|
if ($var->getWidth() <= 8) { |
|
165
|
|
|
if (count($var->missing) >= 3) { |
|
166
|
|
|
$variable->missingValuesFormat = 3; |
|
167
|
|
|
} elseif (count($var->missing) == 2) { |
|
168
|
|
|
$variable->missingValuesFormat = -2; |
|
169
|
|
|
} else { |
|
170
|
|
|
$variable->missingValuesFormat = 1; |
|
171
|
|
|
} |
|
172
|
|
|
$variable->missingValues = $var->missing; |
|
173
|
|
|
} else { |
|
174
|
|
|
$this->info[Record\Info\LongStringMissingValues::SUBTYPE][$shortName] = $var->missing; |
|
175
|
|
|
} |
|
176
|
|
|
} |
|
177
|
|
|
|
|
178
|
|
|
$this->variables[$idx] = $variable; |
|
179
|
|
|
|
|
180
|
|
|
if ($var->values) { |
|
|
|
|
|
|
181
|
|
|
if ($variable->width > 8) { |
|
182
|
|
|
$this->info[Record\Info\LongStringValueLabels::SUBTYPE][$longName] = [ |
|
183
|
|
|
'width' => $var->getWidth(), |
|
184
|
|
|
'values' => $var->values, |
|
185
|
|
|
]; |
|
186
|
|
|
} else { |
|
187
|
|
|
$valueLabel = new Record\ValueLabel([ |
|
188
|
|
|
'variables' => $this->variables, |
|
189
|
|
|
]); |
|
190
|
|
|
foreach ($var->values as $key => $value) { |
|
191
|
|
|
$valueLabel->labels[] = [ |
|
192
|
|
|
'value' => $key, |
|
193
|
|
|
'label' => $value, |
|
194
|
|
|
]; |
|
195
|
|
|
$valueLabel->indexes = [$nominalIdx + 1]; |
|
196
|
|
|
} |
|
197
|
|
|
$this->valueLabels[] = $valueLabel; |
|
198
|
|
|
} |
|
199
|
|
|
} |
|
200
|
|
|
|
|
201
|
|
|
$this->info[Record\Info\LongVariableNames::SUBTYPE][$shortName] = $var->getName(); |
|
202
|
|
|
|
|
203
|
|
|
if (Record\Variable::isVeryLong($var->getWidth())) { |
|
204
|
|
|
$this->info[Record\Info\VeryLongString::SUBTYPE][$shortName] = $var->getWidth(); |
|
205
|
|
|
} |
|
206
|
|
|
|
|
207
|
|
|
$segmentCount = Utils::widthToSegments($var->getWidth()); |
|
208
|
|
|
for ($i = 0; $i < $segmentCount; $i++) { |
|
209
|
|
|
$this->info[Record\Info\VariableDisplayParam::SUBTYPE][] = [ |
|
210
|
|
|
$var->getMeasure(), |
|
211
|
|
|
$var->getColumns(), |
|
212
|
|
|
$var->getAlignment(), |
|
213
|
|
|
]; |
|
214
|
|
|
} |
|
215
|
|
|
|
|
216
|
|
|
// TODO: refactory |
|
217
|
|
|
$dataCount = count($var->data); |
|
218
|
|
|
if ($dataCount > $this->header->casesCount) { |
|
219
|
|
|
$this->header->casesCount = $dataCount; |
|
220
|
|
|
} |
|
221
|
|
|
|
|
222
|
|
|
foreach ($var->data as $case => $value) { |
|
223
|
|
|
$this->data->matrix[$case][$idx] = $value; |
|
224
|
|
|
} |
|
225
|
|
|
|
|
226
|
|
|
$nominalIdx += $var->getOcts(); |
|
227
|
|
|
} |
|
228
|
|
|
|
|
229
|
|
|
$this->header->nominalCaseSize = $nominalIdx; |
|
230
|
|
|
|
|
231
|
|
|
// write header |
|
232
|
|
|
$this->header->write($this->buffer); |
|
233
|
|
|
|
|
234
|
|
|
// write variables |
|
235
|
|
|
foreach ($this->variables as $variable) { |
|
236
|
|
|
$variable->write($this->buffer); |
|
237
|
|
|
} |
|
238
|
|
|
|
|
239
|
|
|
// write valueLabels |
|
240
|
|
|
foreach ($this->valueLabels as $valueLabel) { |
|
241
|
|
|
$valueLabel->write($this->buffer); |
|
242
|
|
|
} |
|
243
|
|
|
|
|
244
|
|
|
// write documents |
|
245
|
|
|
if (! empty($data['documents'])) { |
|
246
|
|
|
$this->document = new Record\Document([ |
|
247
|
|
|
'lines' => $data['documents'], |
|
248
|
|
|
] |
|
249
|
|
|
); |
|
250
|
|
|
$this->document->write($this->buffer); |
|
251
|
|
|
} |
|
252
|
|
|
|
|
253
|
|
|
foreach ($this->info as $info) { |
|
254
|
|
|
$info->write($this->buffer); |
|
255
|
|
|
} |
|
256
|
|
|
|
|
257
|
|
|
$this->data->write($this->buffer); |
|
258
|
|
|
} |
|
259
|
|
|
|
|
260
|
|
|
/** |
|
261
|
|
|
* @param $file |
|
262
|
|
|
* @return false|int |
|
263
|
|
|
*/ |
|
264
|
|
|
public function save($file) |
|
265
|
|
|
{ |
|
266
|
|
|
return $this->buffer->saveToFile($file); |
|
267
|
|
|
} |
|
268
|
|
|
|
|
269
|
|
|
/** |
|
270
|
|
|
* @return \SPSS\Buffer |
|
271
|
|
|
*/ |
|
272
|
|
|
public function getBuffer() |
|
273
|
|
|
{ |
|
274
|
|
|
return $this->buffer; |
|
275
|
|
|
} |
|
276
|
|
|
|
|
277
|
|
|
/** |
|
278
|
|
|
* @param string $className |
|
279
|
|
|
* @param array $data |
|
280
|
|
|
* @param string $group |
|
281
|
|
|
* @return array |
|
282
|
|
|
* @throws Exception |
|
283
|
|
|
*/ |
|
284
|
|
|
private function prepareInfoRecord($className, $data, $group = 'info'): Info |
|
285
|
|
|
{ |
|
286
|
|
|
if (! class_exists($className)) { |
|
287
|
|
|
throw new Exception('Unknown class'); |
|
288
|
|
|
} |
|
289
|
|
|
$key = lcfirst(substr($className, strrpos($className, '\\') + 1)); |
|
290
|
|
|
|
|
291
|
|
|
return new $className( |
|
|
|
|
|
|
292
|
|
|
isset($data[$group]) && isset($data[$group][$key]) ? |
|
293
|
|
|
$data[$group][$key] : |
|
294
|
|
|
[] |
|
295
|
|
|
); |
|
296
|
|
|
} |
|
297
|
|
|
} |
|
298
|
|
|
|
This check marks implicit conversions of arrays to boolean values in a comparison. While in PHP an empty array is considered to be equal (but not identical) to false, this is not always apparent.
Consider making the comparison explicit by using
empty(..)or! empty(...)instead.