1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace SPSS\Sav; |
4
|
|
|
|
5
|
|
|
use SPSS\Buffer; |
6
|
|
|
use SPSS\Exception; |
7
|
|
|
use SPSS\Utils; |
8
|
|
|
|
9
|
|
|
class Writer |
10
|
|
|
{ |
11
|
|
|
/** |
12
|
|
|
* @var Record\Header |
13
|
|
|
*/ |
14
|
|
|
public $header; |
15
|
|
|
|
16
|
|
|
/** |
17
|
|
|
* @var Record\Variable[] |
18
|
|
|
*/ |
19
|
|
|
public $variables = []; |
20
|
|
|
|
21
|
|
|
/** |
22
|
|
|
* @var Record\ValueLabel[] |
23
|
|
|
*/ |
24
|
|
|
public $valueLabels = []; |
25
|
|
|
|
26
|
|
|
/** |
27
|
|
|
* @var Record\Document |
28
|
|
|
*/ |
29
|
|
|
public $document; |
30
|
|
|
|
31
|
|
|
/** |
32
|
|
|
* @var Record\Info[] |
33
|
|
|
*/ |
34
|
|
|
public $info = []; |
35
|
|
|
|
36
|
|
|
/** |
37
|
|
|
* @var Record\Data |
38
|
|
|
*/ |
39
|
|
|
public $data; |
40
|
|
|
|
41
|
|
|
/** |
42
|
|
|
* @var Buffer |
43
|
|
|
*/ |
44
|
|
|
protected $buffer; |
45
|
|
|
|
46
|
|
|
/** |
47
|
|
|
* Writer constructor. |
48
|
|
|
* |
49
|
|
|
* @param array $data |
50
|
|
|
* @throws \Exception |
51
|
|
|
*/ |
52
|
5 |
|
public function __construct($data = []) |
53
|
|
|
{ |
54
|
5 |
|
$this->buffer = Buffer::factory(); |
55
|
5 |
|
$this->buffer->context = $this; |
56
|
|
|
|
57
|
5 |
|
if (! empty($data)) { |
58
|
5 |
|
$this->write($data); |
59
|
|
|
} |
60
|
5 |
|
} |
61
|
|
|
|
62
|
|
|
/** |
63
|
|
|
* @param array $data |
64
|
|
|
* @throws \Exception |
65
|
|
|
*/ |
66
|
5 |
|
public function write($data) |
67
|
|
|
{ |
68
|
5 |
|
$this->header = new Record\Header($data['header']); |
69
|
5 |
|
$this->header->nominalCaseSize = 0; |
70
|
5 |
|
$this->header->casesCount = 0; |
71
|
|
|
|
72
|
5 |
|
$this->info[Record\Info\MachineInteger::SUBTYPE] = $this->prepareInfoRecord( |
73
|
5 |
|
Record\Info\MachineInteger::class, |
74
|
|
|
$data |
75
|
|
|
); |
76
|
|
|
|
77
|
5 |
|
$this->info[Record\Info\MachineFloatingPoint::SUBTYPE] = $this->prepareInfoRecord( |
78
|
5 |
|
Record\Info\MachineFloatingPoint::class, |
79
|
|
|
$data |
80
|
|
|
); |
81
|
|
|
|
82
|
5 |
|
$this->info[Record\Info\VariableDisplayParam::SUBTYPE] = new Record\Info\VariableDisplayParam(); |
83
|
5 |
|
$this->info[Record\Info\LongVariableNames::SUBTYPE] = new Record\Info\LongVariableNames(); |
84
|
5 |
|
$this->info[Record\Info\VeryLongString::SUBTYPE] = new Record\Info\VeryLongString(); |
85
|
5 |
|
$this->info[Record\Info\ExtendedNumberOfCases::SUBTYPE] = $this->prepareInfoRecord( |
86
|
5 |
|
Record\Info\ExtendedNumberOfCases::class, |
87
|
|
|
$data |
88
|
|
|
); |
89
|
5 |
|
$this->info[Record\Info\VariableAttributes::SUBTYPE] = new Record\Info\VariableAttributes(); |
90
|
5 |
|
$this->info[Record\Info\LongStringValueLabels::SUBTYPE] = new Record\Info\LongStringValueLabels(); |
91
|
5 |
|
$this->info[Record\Info\LongStringMissingValues::SUBTYPE] = new Record\Info\LongStringMissingValues(); |
92
|
5 |
|
$this->info[Record\Info\CharacterEncoding::SUBTYPE] = new Record\Info\CharacterEncoding('UTF-8'); |
93
|
|
|
|
94
|
5 |
|
$this->data = new Record\Data(); |
95
|
|
|
|
96
|
5 |
|
$nominalIdx = 0; |
97
|
|
|
|
98
|
|
|
/** |
99
|
|
|
* @var bool[string] The variable names used in this SPSS file |
100
|
|
|
*/ |
101
|
5 |
|
$variableNames = []; |
102
|
|
|
/** @var Variable $var */ |
103
|
5 |
|
foreach (array_values($data['variables']) as $idx => $var) { |
104
|
|
|
|
105
|
5 |
|
if (is_array($var)) { |
106
|
5 |
|
$var = new Variable($var); |
107
|
|
|
} |
108
|
|
|
|
109
|
|
|
#if (! preg_match('/^[A-Za-z0-9_]+$/', $var->name)) { |
110
|
|
|
# UTF-8 and '.' characters could pass here |
111
|
5 |
|
if (! preg_match('/^[A-Za-z0-9_\.\x{4e00}-\x{9fa5}]+$/u', $var->name)) { |
112
|
|
|
throw new \InvalidArgumentException( |
113
|
|
|
sprintf('Variable name `%s` contains an illegal character.', $var->name) |
114
|
|
|
); |
115
|
|
|
} |
116
|
|
|
|
117
|
5 |
|
if (empty($var->width)) { |
118
|
|
|
throw new \InvalidArgumentException( |
119
|
|
|
sprintf('Invalid field width. Should be an integer number greater than zero.') |
120
|
|
|
); |
121
|
|
|
} |
122
|
|
|
|
123
|
5 |
|
$variable = new Record\Variable(); |
124
|
|
|
|
125
|
|
|
/** |
126
|
|
|
* @see \SPSS\Sav\Record\Variable::getSegmentName() |
127
|
|
|
* |
128
|
|
|
* Variable names in the SPSS file should be unique. If they are not, SPSS will rename them. |
129
|
|
|
* If SPSS renames them and it happens to be a long string then the segments will no longer share |
130
|
|
|
* the required prefix in the name. |
131
|
|
|
*/ |
132
|
5 |
|
$name = strtoupper(substr($var->name, 0, 8)); |
133
|
|
|
|
134
|
5 |
|
$counter = 0; |
135
|
|
|
/** |
136
|
|
|
* Using base convert we can encode 36^3 = 46656 variables with a common 5 character prefix in an 8 |
137
|
|
|
* character variable name. This should suffice since the current variable limit of SPSS is 32767 |
138
|
|
|
* variables. |
139
|
|
|
*/ |
140
|
5 |
|
while (isset($variableNames[$name])) { |
141
|
|
|
$name = strtoupper(substr($var->name, 0, 5) . base_convert($counter, 10, 36)); |
142
|
|
|
$counter++; |
143
|
|
|
} |
144
|
|
|
|
145
|
5 |
|
$variableNames[$name] = true; |
146
|
5 |
|
$variable->name = $name; |
147
|
|
|
|
148
|
5 |
|
if ($var->format == Variable::FORMAT_TYPE_A) { |
149
|
4 |
|
$variable->width = $var->width; |
150
|
|
|
} else { |
151
|
4 |
|
$variable->width = 0; |
152
|
|
|
} |
153
|
|
|
|
154
|
5 |
|
$variable->label = $var->label; |
155
|
5 |
|
$variable->print = [ |
156
|
5 |
|
0, |
157
|
5 |
|
$var->format, |
158
|
5 |
|
$var->width ? min($var->width, 255) : 8, |
159
|
5 |
|
$var->decimals, |
160
|
|
|
]; |
161
|
5 |
|
$variable->write = [ |
162
|
5 |
|
0, |
163
|
5 |
|
$var->format, |
164
|
5 |
|
$var->width ? min($var->width, 255) : 8, |
165
|
5 |
|
$var->decimals, |
166
|
|
|
]; |
167
|
|
|
|
168
|
|
|
// TODO: refactory |
169
|
5 |
|
$shortName = $variable->name; |
170
|
5 |
|
$longName = $var->name; |
171
|
|
|
|
172
|
5 |
|
if ($var->attributes) { |
173
|
2 |
|
$this->info[Record\Info\VariableAttributes::SUBTYPE][$longName] = $var->attributes; |
174
|
|
|
} |
175
|
|
|
|
176
|
5 |
|
if ($var->missing) { |
177
|
|
|
if ($var->width <= 8) { |
178
|
|
|
if (count($var->missing) >= 3) { |
179
|
|
|
$variable->missingValuesFormat = 3; |
180
|
|
|
} elseif (count($var->missing) == 2) { |
181
|
|
|
$variable->missingValuesFormat = -2; |
182
|
|
|
} else { |
183
|
|
|
$variable->missingValuesFormat = 1; |
184
|
|
|
} |
185
|
|
|
$variable->missingValues = $var->missing; |
186
|
|
|
} else { |
187
|
|
|
$this->info[Record\Info\LongStringMissingValues::SUBTYPE][$shortName] = $var->missing; |
188
|
|
|
} |
189
|
|
|
} |
190
|
|
|
|
191
|
5 |
|
$this->variables[$idx] = $variable; |
192
|
|
|
|
193
|
5 |
|
if ($var->values) { |
194
|
2 |
|
if ($variable->width > 8) { |
195
|
|
|
$this->info[Record\Info\LongStringValueLabels::SUBTYPE][$longName] = [ |
196
|
|
|
'width' => $var->width, |
197
|
|
|
'values' => $var->values, |
198
|
|
|
]; |
199
|
|
|
} else { |
200
|
2 |
|
$valueLabel = new Record\ValueLabel([ |
201
|
2 |
|
'variables' => $this->variables, |
202
|
|
|
]); |
203
|
2 |
|
foreach ($var->values as $key => $value) { |
204
|
2 |
|
$valueLabel->labels[] = [ |
205
|
2 |
|
'value' => $key, |
206
|
2 |
|
'label' => $value, |
207
|
|
|
]; |
208
|
2 |
|
$valueLabel->indexes = [$nominalIdx + 1]; |
209
|
|
|
} |
210
|
2 |
|
$this->valueLabels[] = $valueLabel; |
211
|
|
|
} |
212
|
|
|
} |
213
|
|
|
|
214
|
5 |
|
$this->info[Record\Info\LongVariableNames::SUBTYPE][$shortName] = $var->name; |
215
|
|
|
|
216
|
5 |
|
if (Record\Variable::isVeryLong($var->width)) { |
217
|
1 |
|
$this->info[Record\Info\VeryLongString::SUBTYPE][$shortName] = $var->width; |
218
|
|
|
} |
219
|
|
|
|
220
|
5 |
|
$segmentCount = Utils::widthToSegments($var->width); |
221
|
5 |
|
for ($i = 0; $i < $segmentCount; $i++) { |
222
|
5 |
|
$this->info[Record\Info\VariableDisplayParam::SUBTYPE][] = [ |
223
|
5 |
|
$var->getMeasure(), |
224
|
5 |
|
$var->getColumns(), |
225
|
5 |
|
$var->getAlignment(), |
226
|
|
|
]; |
227
|
|
|
} |
228
|
|
|
|
229
|
|
|
// TODO: refactory |
230
|
5 |
|
$dataCount = count($var->data); |
231
|
5 |
|
if ($dataCount > $this->header->casesCount) { |
232
|
4 |
|
$this->header->casesCount = $dataCount; |
233
|
|
|
} |
234
|
|
|
|
235
|
5 |
|
foreach ($var->data as $case => $value) { |
236
|
4 |
|
$this->data->matrix[$case][$idx] = $value; |
237
|
|
|
} |
238
|
|
|
|
239
|
5 |
|
$nominalIdx += Utils::widthToOcts($var->width); |
240
|
|
|
} |
241
|
|
|
|
242
|
5 |
|
$this->header->nominalCaseSize = $nominalIdx; |
243
|
|
|
|
244
|
|
|
// write header |
245
|
5 |
|
$this->header->write($this->buffer); |
246
|
|
|
|
247
|
|
|
// write variables |
248
|
5 |
|
foreach ($this->variables as $variable) { |
249
|
5 |
|
$variable->write($this->buffer); |
250
|
|
|
} |
251
|
|
|
|
252
|
|
|
// write valueLabels |
253
|
5 |
|
foreach ($this->valueLabels as $valueLabel) { |
254
|
2 |
|
$valueLabel->write($this->buffer); |
255
|
|
|
} |
256
|
|
|
|
257
|
|
|
// write documents |
258
|
5 |
|
if (! empty($data['documents'])) { |
259
|
1 |
|
$this->document = new Record\Document([ |
260
|
1 |
|
'lines' => $data['documents'], |
261
|
|
|
] |
262
|
|
|
); |
263
|
1 |
|
$this->document->write($this->buffer); |
264
|
|
|
} |
265
|
|
|
|
266
|
5 |
|
foreach ($this->info as $info) { |
267
|
5 |
|
$info->write($this->buffer); |
268
|
|
|
} |
269
|
|
|
|
270
|
5 |
|
$this->data->write($this->buffer); |
271
|
5 |
|
} |
272
|
|
|
|
273
|
|
|
/** |
274
|
|
|
* @param $file |
275
|
|
|
* @return false|int |
276
|
|
|
*/ |
277
|
|
|
public function save($file) |
278
|
|
|
{ |
279
|
|
|
return $this->buffer->saveToFile($file); |
280
|
|
|
} |
281
|
|
|
|
282
|
|
|
/** |
283
|
|
|
* @return \SPSS\Buffer |
284
|
|
|
*/ |
285
|
5 |
|
public function getBuffer() |
286
|
|
|
{ |
287
|
5 |
|
return $this->buffer; |
288
|
|
|
} |
289
|
|
|
|
290
|
|
|
/** |
291
|
|
|
* @param string $className |
292
|
|
|
* @param array $data |
293
|
|
|
* @param string $group |
294
|
|
|
* @return array |
295
|
|
|
* @throws Exception |
296
|
|
|
*/ |
297
|
5 |
|
private function prepareInfoRecord($className, $data, $group = 'info') |
298
|
|
|
{ |
299
|
5 |
|
if (! class_exists($className)) { |
300
|
|
|
throw new Exception('Unknown class'); |
301
|
|
|
} |
302
|
5 |
|
$key = lcfirst(substr($className, strrpos($className, '\\') + 1)); |
303
|
|
|
|
304
|
5 |
|
return new $className( |
|
|
|
|
305
|
5 |
|
isset($data[$group]) && isset($data[$group][$key]) ? |
306
|
|
|
$data[$group][$key] : |
307
|
5 |
|
[] |
308
|
|
|
); |
309
|
|
|
} |
310
|
|
|
} |
311
|
|
|
|