|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace UCD\Infrastructure\Repository\CharacterRepository; |
|
4
|
|
|
|
|
5
|
|
|
use UCD\Exception; |
|
6
|
|
|
use UCD\Unicode\Character; |
|
7
|
|
|
use UCD\Unicode\Character\Collection; |
|
8
|
|
|
use UCD\Unicode\Character\Properties; |
|
9
|
|
|
use UCD\Unicode\Character\Repository; |
|
10
|
|
|
use UCD\Unicode\Character\Repository\CharacterNotFoundException; |
|
11
|
|
|
use UCD\Unicode\Character\WritableRepository; |
|
12
|
|
|
use UCD\Unicode\Codepoint; |
|
13
|
|
|
use UCD\Unicode\Codepoint\AggregatorRelay; |
|
14
|
|
|
use UCD\Unicode\CodepointAssigned; |
|
15
|
|
|
|
|
16
|
|
|
use UCD\Infrastructure\Repository\CharacterRepository\FileRepository\CharacterSlicer; |
|
17
|
|
|
use UCD\Infrastructure\Repository\CharacterRepository\FileRepository\Property; |
|
18
|
|
|
use UCD\Infrastructure\Repository\CharacterRepository\FileRepository\PropertyAggregators; |
|
19
|
|
|
use UCD\Infrastructure\Repository\CharacterRepository\FileRepository\PropertyFileDirectory; |
|
20
|
|
|
use UCD\Infrastructure\Repository\CharacterRepository\FileRepository\Range; |
|
21
|
|
|
use UCD\Infrastructure\Repository\CharacterRepository\FileRepository\RangeFile; |
|
22
|
|
|
use UCD\Infrastructure\Repository\CharacterRepository\FileRepository\RangeFileDirectory; |
|
23
|
|
|
use UCD\Infrastructure\Repository\CharacterRepository\FileRepository\Serializer; |
|
24
|
|
|
|
|
25
|
|
|
class FileRepository implements WritableRepository |
|
26
|
|
|
{ |
|
27
|
|
|
use Repository\Capability\Notify; |
|
28
|
|
|
|
|
29
|
|
|
const DEFAULT_SLICE_SIZE = 1000; |
|
30
|
|
|
|
|
31
|
|
|
/** |
|
32
|
|
|
* @var RangeFileDirectory |
|
33
|
|
|
*/ |
|
34
|
|
|
private $charactersDirectory; |
|
35
|
|
|
|
|
36
|
|
|
/** |
|
37
|
|
|
* @var PropertyFileDirectory |
|
38
|
|
|
*/ |
|
39
|
|
|
private $propertiesDirectory; |
|
40
|
|
|
|
|
41
|
|
|
/** |
|
42
|
|
|
* @var PropertyAggregators|AggregatorRelay[] |
|
43
|
|
|
*/ |
|
44
|
|
|
private $aggregators; |
|
45
|
|
|
|
|
46
|
|
|
/** |
|
47
|
|
|
* @var Serializer |
|
48
|
|
|
*/ |
|
49
|
|
|
private $serializer; |
|
50
|
|
|
|
|
51
|
|
|
/** |
|
52
|
|
|
* @var int |
|
53
|
|
|
*/ |
|
54
|
|
|
private $sliceSize; |
|
55
|
|
|
|
|
56
|
|
|
/** |
|
57
|
|
|
* @param RangeFileDirectory $charactersDirectory |
|
58
|
|
|
* @param PropertyFileDirectory $propertiesDirectory |
|
59
|
|
|
* @param PropertyAggregators $aggregators |
|
60
|
|
|
* @param Serializer $serializer |
|
61
|
|
|
* @param int $sliceSize |
|
62
|
|
|
*/ |
|
63
|
|
|
public function __construct( |
|
64
|
|
|
RangeFileDirectory $charactersDirectory, |
|
65
|
|
|
PropertyFileDirectory $propertiesDirectory, |
|
66
|
|
|
PropertyAggregators $aggregators, |
|
67
|
|
|
Serializer $serializer, |
|
68
|
|
|
$sliceSize = self::DEFAULT_SLICE_SIZE |
|
69
|
|
|
) { |
|
70
|
|
|
$this->charactersDirectory = $charactersDirectory; |
|
71
|
|
|
$this->serializer = $serializer; |
|
72
|
|
|
$this->sliceSize = $sliceSize; |
|
73
|
|
|
$this->propertiesDirectory = $propertiesDirectory; |
|
74
|
|
|
$this->aggregators = $aggregators; |
|
75
|
|
|
} |
|
76
|
|
|
|
|
77
|
|
|
/** |
|
78
|
|
|
* {@inheritDoc} |
|
79
|
|
|
*/ |
|
80
|
|
|
public function getByCodepoint(Codepoint $codepoint) |
|
81
|
|
|
{ |
|
82
|
|
|
$value = $codepoint->getValue(); |
|
83
|
|
|
$file = $this->getFileByCodepoint($codepoint); |
|
84
|
|
|
$characters = $file->read(); |
|
85
|
|
|
|
|
86
|
|
|
if (!array_key_exists($value, $characters)) { |
|
87
|
|
|
throw CharacterNotFoundException::withCodepoint($codepoint); |
|
88
|
|
|
} |
|
89
|
|
|
|
|
90
|
|
|
return $this->serializer->unserialize($characters[$value]); |
|
91
|
|
|
} |
|
92
|
|
|
|
|
93
|
|
|
/** |
|
94
|
|
|
* {@inheritDoc} |
|
95
|
|
|
*/ |
|
96
|
|
|
public function getByCodepoints(Codepoint\Collection $codepoints) |
|
97
|
|
|
{ |
|
98
|
|
|
return new Character\Collection( |
|
99
|
|
|
$this->yieldByCodepoints($codepoints) |
|
100
|
|
|
); |
|
101
|
|
|
} |
|
102
|
|
|
|
|
103
|
|
|
/** |
|
104
|
|
|
* @param Codepoint\Collection $codepoints |
|
105
|
|
|
* @return \Generator |
|
106
|
|
|
*/ |
|
107
|
|
|
private function yieldByCodepoints(Codepoint\Collection $codepoints) |
|
108
|
|
|
{ |
|
109
|
|
|
foreach ($codepoints as $codepoint) { |
|
110
|
|
|
try { |
|
111
|
|
|
yield $this->getByCodepoint($codepoint); |
|
112
|
|
|
} catch (CharacterNotFoundException $e) { |
|
113
|
|
|
// Missing characters are skipped |
|
114
|
|
|
} |
|
115
|
|
|
} |
|
116
|
|
|
} |
|
117
|
|
|
|
|
118
|
|
|
/** |
|
119
|
|
|
* @param Codepoint $codepoint |
|
120
|
|
|
* @return RangeFile |
|
121
|
|
|
* @throws CharacterNotFoundException |
|
122
|
|
|
*/ |
|
123
|
|
|
private function getFileByCodepoint(Codepoint $codepoint) |
|
124
|
|
|
{ |
|
125
|
|
|
$file = $this->charactersDirectory->getFileFromValue($codepoint->getValue()); |
|
126
|
|
|
|
|
127
|
|
|
if ($file === null) { |
|
128
|
|
|
throw CharacterNotFoundException::withCodepoint($codepoint); |
|
129
|
|
|
} |
|
130
|
|
|
|
|
131
|
|
|
return $file; |
|
132
|
|
|
} |
|
133
|
|
|
|
|
134
|
|
|
/** |
|
135
|
|
|
* {@inheritDoc} |
|
136
|
|
|
*/ |
|
137
|
|
|
public function addMany(Collection $characters) |
|
138
|
|
|
{ |
|
139
|
|
|
$slices = CharacterSlicer::slice($characters, $this->sliceSize); |
|
140
|
|
|
|
|
141
|
|
|
foreach ($slices as $range => $chunk) { |
|
142
|
|
|
/** @var Range $range */ |
|
143
|
|
|
$this->createFileWithCharacters($range, $chunk); |
|
144
|
|
|
$this->addCharactersToAggregators($chunk); |
|
145
|
|
|
$this->notify(); |
|
146
|
|
|
} |
|
147
|
|
|
|
|
148
|
|
|
$this->writeAggregations(); |
|
149
|
|
|
} |
|
150
|
|
|
|
|
151
|
|
|
/** |
|
152
|
|
|
* @param Range $range |
|
153
|
|
|
* @param CodepointAssigned[] $characters |
|
154
|
|
|
* @return CodepointAssigned[] |
|
155
|
|
|
*/ |
|
156
|
|
|
private function createFileWithCharacters(Range $range, array $characters) |
|
157
|
|
|
{ |
|
158
|
|
|
$this->charactersDirectory->writeRange( |
|
159
|
|
|
$range, |
|
160
|
|
|
$this->flattenCharacters($characters) |
|
161
|
|
|
); |
|
162
|
|
|
} |
|
163
|
|
|
|
|
164
|
|
|
/** |
|
165
|
|
|
* @param CodepointAssigned[] $characters |
|
166
|
|
|
* @return string[] |
|
167
|
|
|
*/ |
|
168
|
|
|
private function flattenCharacters(array $characters) |
|
169
|
|
|
{ |
|
170
|
|
|
$flattened = []; |
|
171
|
|
|
|
|
172
|
|
|
foreach ($characters as $character) { |
|
173
|
|
|
$codepoint = $character->getCodepoint(); |
|
174
|
|
|
$key = $codepoint->getValue(); |
|
175
|
|
|
$flattened[$key] = $this->serializer->serialize($character);; |
|
176
|
|
|
} |
|
177
|
|
|
|
|
178
|
|
|
return $flattened; |
|
179
|
|
|
} |
|
180
|
|
|
|
|
181
|
|
|
/** |
|
182
|
|
|
* @param Codepoint\Range\Collection[] $ranges |
|
183
|
|
|
* @return string[] |
|
184
|
|
|
*/ |
|
185
|
|
|
public function flattenRanges(array $ranges) |
|
186
|
|
|
{ |
|
187
|
|
|
$flattened = []; |
|
188
|
|
|
|
|
189
|
|
|
foreach ($ranges as $key => $range) { |
|
190
|
|
|
$range = $range->toArray(); |
|
191
|
|
|
$flattened[$key] = $this->serializer->serialize($range); |
|
192
|
|
|
} |
|
193
|
|
|
|
|
194
|
|
|
return $flattened; |
|
195
|
|
|
} |
|
196
|
|
|
|
|
197
|
|
|
/** |
|
198
|
|
|
* @param CodepointAssigned[] $characters |
|
199
|
|
|
*/ |
|
200
|
|
|
private function addCharactersToAggregators(array $characters) |
|
201
|
|
|
{ |
|
202
|
|
|
$this->aggregators->addCharacters($characters); |
|
203
|
|
|
} |
|
204
|
|
|
|
|
205
|
|
|
/** |
|
206
|
|
|
* @return bool |
|
207
|
|
|
*/ |
|
208
|
|
|
private function writeAggregations() |
|
209
|
|
|
{ |
|
210
|
|
|
foreach ($this->aggregators as $property => $aggregator) { |
|
211
|
|
|
/** @var Property $property */ |
|
212
|
|
|
$ranges = $this->flattenRanges($aggregator->getAllRanges()); |
|
213
|
|
|
$this->propertiesDirectory->writeProperty($property, $ranges); |
|
214
|
|
|
} |
|
215
|
|
|
} |
|
216
|
|
|
|
|
217
|
|
|
/** |
|
218
|
|
|
* {@inheritDoc} |
|
219
|
|
|
*/ |
|
220
|
|
|
public function getAll() |
|
221
|
|
|
{ |
|
222
|
|
|
return new Collection( |
|
223
|
|
|
$this->readAll() |
|
224
|
|
|
); |
|
225
|
|
|
} |
|
226
|
|
|
|
|
227
|
|
|
/** |
|
228
|
|
|
* @return \Generator |
|
229
|
|
|
*/ |
|
230
|
|
|
private function readAll() |
|
231
|
|
|
{ |
|
232
|
|
|
$files = $this->charactersDirectory->getFiles(); |
|
233
|
|
|
|
|
234
|
|
|
foreach ($files as $file) { |
|
235
|
|
|
$characters = $file->read(); |
|
236
|
|
|
|
|
237
|
|
|
foreach ($characters as $i => $character) { |
|
238
|
|
|
yield $i => $this->serializer->unserialize($character); |
|
239
|
|
|
} |
|
240
|
|
|
} |
|
241
|
|
|
} |
|
242
|
|
|
|
|
243
|
|
|
/** |
|
244
|
|
|
* {@inheritDoc} |
|
245
|
|
|
*/ |
|
246
|
|
|
public function getCodepointsByBlock(Properties\General\Block $block) |
|
247
|
|
|
{ |
|
248
|
|
|
return $this->resolveCodepointsByProperty( |
|
249
|
|
|
Property::ofType(Property::BLOCK), |
|
250
|
|
|
(string)$block, |
|
251
|
|
|
Repository\BlockNotFoundException::withBlock($block) |
|
252
|
|
|
); |
|
253
|
|
|
} |
|
254
|
|
|
|
|
255
|
|
|
/** |
|
256
|
|
|
* {@inheritDoc} |
|
257
|
|
|
*/ |
|
258
|
|
|
public function getCodepointsByCategory(Properties\General\GeneralCategory $category) |
|
259
|
|
|
{ |
|
260
|
|
|
return $this->resolveCodepointsByProperty( |
|
261
|
|
|
Property::ofType(Property::GENERAL_CATEGORY), |
|
262
|
|
|
(string)$category, |
|
263
|
|
|
Repository\GeneralCategoryNotFoundException::withCategory($category) |
|
264
|
|
|
); |
|
265
|
|
|
} |
|
266
|
|
|
|
|
267
|
|
|
/** |
|
268
|
|
|
* {@inheritDoc} |
|
269
|
|
|
*/ |
|
270
|
|
|
public function getCodepointsByScript(Properties\General\Script $script) |
|
271
|
|
|
{ |
|
272
|
|
|
return $this->resolveCodepointsByProperty( |
|
273
|
|
|
Property::ofType(Property::SCRIPT), |
|
274
|
|
|
(string)$script, |
|
275
|
|
|
Repository\ScriptNotFoundException::withScript($script) |
|
276
|
|
|
); |
|
277
|
|
|
} |
|
278
|
|
|
|
|
279
|
|
|
/** |
|
280
|
|
|
* @param Property $property |
|
281
|
|
|
* @param string $key |
|
282
|
|
|
* @param Exception $notFoundException |
|
283
|
|
|
* @return Codepoint\Range[] |
|
284
|
|
|
* @throws Exception |
|
285
|
|
|
*/ |
|
286
|
|
|
private function resolveCodepointsByProperty(Property $property, $key, Exception $notFoundException) |
|
287
|
|
|
{ |
|
288
|
|
|
$file = $this->propertiesDirectory->getFileForProperty($property); |
|
289
|
|
|
$map = $file->read(); |
|
290
|
|
|
|
|
291
|
|
|
$codepoints = array_key_exists($key, $map) |
|
292
|
|
|
? $this->serializer->unserialize($map[$key]) |
|
293
|
|
|
: []; |
|
294
|
|
|
|
|
295
|
|
|
if (count($codepoints) === 0) { |
|
296
|
|
|
throw $notFoundException; |
|
297
|
|
|
} |
|
298
|
|
|
|
|
299
|
|
|
return Codepoint\Range\Collection::fromArray($codepoints); |
|
300
|
|
|
} |
|
301
|
|
|
|
|
302
|
|
|
/** |
|
303
|
|
|
* {@inheritDoc} |
|
304
|
|
|
*/ |
|
305
|
|
|
public function count() |
|
306
|
|
|
{ |
|
307
|
|
|
$files = $this->charactersDirectory->getFiles(); |
|
308
|
|
|
$tally = 0; |
|
309
|
|
|
|
|
310
|
|
|
foreach ($files as $file) { |
|
311
|
|
|
$tally += $file->getTotal(); |
|
312
|
|
|
} |
|
313
|
|
|
|
|
314
|
|
|
return $tally; |
|
315
|
|
|
} |
|
316
|
|
|
} |