|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
/* |
|
4
|
|
|
* This file is part of the sweetrdf/InMemoryStoreSqlite package and licensed under |
|
5
|
|
|
* the terms of the GPL-3 license. |
|
6
|
|
|
* |
|
7
|
|
|
* (c) Konrad Abicht <[email protected]> |
|
8
|
|
|
* (c) Benjamin Nowack |
|
9
|
|
|
* |
|
10
|
|
|
* For the full copyright and license information, please view the LICENSE |
|
11
|
|
|
* file that was distributed with this source code. |
|
12
|
|
|
*/ |
|
13
|
|
|
|
|
14
|
|
|
namespace sweetrdf\InMemoryStoreSqlite\Serializer; |
|
15
|
|
|
|
|
16
|
|
|
use sweetrdf\InMemoryStoreSqlite\NamespaceHelper; |
|
17
|
|
|
use function sweetrdf\InMemoryStoreSqlite\splitURI; |
|
18
|
|
|
|
|
19
|
|
|
class TurtleSerializer |
|
20
|
|
|
{ |
|
21
|
|
|
private array $ns = []; |
|
22
|
|
|
private array $nsp = []; |
|
23
|
|
|
private int $ns_count = 0; |
|
24
|
|
|
|
|
25
|
23 |
|
public function __construct() |
|
26
|
|
|
{ |
|
27
|
23 |
|
$this->qualifier = ['rdf:type', 'rdfs:domain', 'rdfs:range', 'rdfs:subClassOf']; |
|
|
|
|
|
|
28
|
|
|
|
|
29
|
23 |
|
$rdf = NamespaceHelper::NAMESPACE_RDF; |
|
30
|
23 |
|
$this->nsp = [$rdf => 'rdf']; |
|
31
|
23 |
|
$this->used_ns = [$rdf]; |
|
|
|
|
|
|
32
|
23 |
|
$this->ns = ['rdf' => $rdf]; |
|
33
|
23 |
|
} |
|
34
|
|
|
|
|
35
|
|
|
public function getSerializedTriples($triples, $raw = 0) |
|
36
|
|
|
{ |
|
37
|
|
|
$index = $this->getSimpleIndex($triples, 0); |
|
38
|
|
|
|
|
39
|
|
|
return $this->getSerializedIndex($index, $raw); |
|
40
|
|
|
} |
|
41
|
|
|
|
|
42
|
|
|
public function getSimpleIndex($triples, $flatten_objects = 1, $vals = '') |
|
43
|
|
|
{ |
|
44
|
|
|
$r = []; |
|
45
|
|
|
foreach ($triples as $t) { |
|
46
|
|
|
$skip_t = 0; |
|
47
|
|
|
foreach (['s', 'p', 'o'] as $term) { |
|
48
|
|
|
$$term = $t[$term]; |
|
49
|
|
|
/* template var */ |
|
50
|
|
|
if (isset($t[$term.'_type']) && ('var' == $t[$term.'_type'])) { |
|
51
|
|
|
$val = isset($vals[$$term]) ? $vals[$$term] : ''; |
|
52
|
|
|
$skip_t = isset($vals[$$term]) ? $skip_t : 1; |
|
53
|
|
|
$type = ''; |
|
54
|
|
|
$type = !$type && isset($vals[$$term.' type']) ? $vals[$$term.' type'] : $type; |
|
55
|
|
|
$type = !$type && preg_match('/^\_\:/', $val) ? 'bnode' : $type; |
|
56
|
|
|
if ('o' == $term) { |
|
57
|
|
|
$type = !$type && (preg_match('/\s/s', $val) || !preg_match('/\:/', $val)) ? 'literal' : $type; |
|
58
|
|
|
$type = !$type && !preg_match('/[\/]/', $val) ? 'literal' : $type; |
|
59
|
|
|
} |
|
60
|
|
|
$type = !$type ? 'uri' : $type; |
|
61
|
|
|
$t[$term.'_type'] = $type; |
|
62
|
|
|
$$term = $val; |
|
63
|
|
|
} |
|
64
|
|
|
} |
|
65
|
|
|
if ($skip_t) { |
|
66
|
|
|
continue; |
|
67
|
|
|
} |
|
68
|
|
|
if (!isset($r[$s])) { |
|
|
|
|
|
|
69
|
|
|
$r[$s] = []; |
|
70
|
|
|
} |
|
71
|
|
|
if (!isset($r[$s][$p])) { |
|
|
|
|
|
|
72
|
|
|
$r[$s][$p] = []; |
|
73
|
|
|
} |
|
74
|
|
|
if ($flatten_objects) { |
|
75
|
|
|
if (!\in_array($o, $r[$s][$p])) { |
|
|
|
|
|
|
76
|
|
|
$r[$s][$p][] = $o; |
|
77
|
|
|
} |
|
78
|
|
|
} else { |
|
79
|
|
|
$o = ['value' => $o]; |
|
80
|
|
|
foreach (['lang', 'type', 'datatype'] as $suffix) { |
|
81
|
|
|
if (isset($t['o_'.$suffix]) && $t['o_'.$suffix]) { |
|
82
|
|
|
$o[$suffix] = $t['o_'.$suffix]; |
|
83
|
|
|
} elseif (isset($t['o '.$suffix]) && $t['o '.$suffix]) { |
|
84
|
|
|
$o[$suffix] = $t['o '.$suffix]; |
|
85
|
|
|
} |
|
86
|
|
|
} |
|
87
|
|
|
if (!\in_array($o, $r[$s][$p])) { |
|
88
|
|
|
$r[$s][$p][] = $o; |
|
89
|
|
|
} |
|
90
|
|
|
} |
|
91
|
|
|
} |
|
92
|
|
|
|
|
93
|
|
|
return $r; |
|
94
|
|
|
} |
|
95
|
|
|
|
|
96
|
|
|
/** |
|
97
|
|
|
* @todo port to NamespaceHelper |
|
98
|
|
|
*/ |
|
99
|
23 |
|
public function getPNameNamespace($v, $connector = ':') |
|
100
|
|
|
{ |
|
101
|
23 |
|
$re = '/^([a-z0-9\_\-]+)\:([a-z0-9\_\-\.\%]+)$/i'; |
|
102
|
23 |
|
if (':' != $connector) { |
|
103
|
|
|
$connectors = ['\:', '\-', '\_', '\.']; |
|
104
|
|
|
$chars = implode('', array_diff($connectors, [$connector])); |
|
105
|
|
|
$re = '/^([a-z0-9'.$chars.']+)\\'.$connector.'([a-z0-9\_\-\.\%]+)$/i'; |
|
106
|
|
|
} |
|
107
|
23 |
|
if (!preg_match($re, $v, $m)) { |
|
108
|
23 |
|
return 0; |
|
109
|
|
|
} |
|
110
|
|
|
if (!isset($this->ns[$m[1]])) { |
|
111
|
|
|
return 0; |
|
112
|
|
|
} |
|
113
|
|
|
|
|
114
|
|
|
return $this->ns[$m[1]]; |
|
115
|
|
|
} |
|
116
|
|
|
|
|
117
|
|
|
/** |
|
118
|
|
|
* @todo port to NamespaceHelper |
|
119
|
|
|
*/ |
|
120
|
23 |
|
public function getPName($v, $connector = ':') |
|
121
|
|
|
{ |
|
122
|
|
|
/* is already a pname */ |
|
123
|
23 |
|
$ns = $this->getPNameNamespace($v, $connector); |
|
124
|
23 |
|
if ($ns) { |
|
125
|
|
|
if (!\in_array($ns, $this->used_ns)) { |
|
126
|
|
|
$this->used_ns[] = $ns; |
|
|
|
|
|
|
127
|
|
|
} |
|
128
|
|
|
|
|
129
|
|
|
return $v; |
|
130
|
|
|
} |
|
131
|
|
|
/* new pname */ |
|
132
|
23 |
|
$parts = splitURI($v); |
|
133
|
23 |
|
if ($parts) { |
|
134
|
|
|
/* known prefix */ |
|
135
|
23 |
|
foreach ($this->ns as $prefix => $ns) { |
|
136
|
23 |
|
if ($parts[0] == $ns) { |
|
137
|
23 |
|
if (!\in_array($ns, $this->used_ns)) { |
|
138
|
|
|
$this->used_ns[] = $ns; |
|
139
|
|
|
} |
|
140
|
|
|
|
|
141
|
23 |
|
return $prefix.$connector.$parts[1]; |
|
142
|
|
|
} |
|
143
|
|
|
} |
|
144
|
|
|
/* new prefix */ |
|
145
|
23 |
|
$prefix = $this->getPrefix($parts[0]); |
|
146
|
|
|
|
|
147
|
23 |
|
return $prefix.$connector.$parts[1]; |
|
148
|
|
|
} |
|
149
|
|
|
|
|
150
|
|
|
return $v; |
|
151
|
|
|
} |
|
152
|
|
|
|
|
153
|
|
|
/** |
|
154
|
|
|
* @todo port to NamespaceHelper |
|
155
|
|
|
*/ |
|
156
|
23 |
|
public function getPrefix($ns) |
|
157
|
|
|
{ |
|
158
|
23 |
|
if (!isset($this->nsp[$ns])) { |
|
159
|
23 |
|
$this->ns['ns'.$this->ns_count] = $ns; |
|
160
|
23 |
|
$this->nsp[$ns] = 'ns'.$this->ns_count; |
|
161
|
23 |
|
++$this->ns_count; |
|
162
|
|
|
} |
|
163
|
23 |
|
if (!\in_array($ns, $this->used_ns)) { |
|
164
|
23 |
|
$this->used_ns[] = $ns; |
|
|
|
|
|
|
165
|
|
|
} |
|
166
|
|
|
|
|
167
|
23 |
|
return $this->nsp[$ns]; |
|
168
|
|
|
} |
|
169
|
|
|
|
|
170
|
23 |
|
public function getTerm($v, $term = '', $qualifier = '') |
|
171
|
|
|
{ |
|
172
|
23 |
|
if (!\is_array($v)) { |
|
173
|
23 |
|
if (preg_match('/^\_\:/', $v)) { |
|
174
|
23 |
|
return $v; |
|
175
|
|
|
} |
|
176
|
23 |
|
if (('p' === $term) && ($pn = $this->getPName($v))) { |
|
177
|
23 |
|
return $pn; |
|
178
|
|
|
} |
|
179
|
|
|
if ( |
|
180
|
23 |
|
('o' === $term) |
|
181
|
23 |
|
&& \in_array($qualifier, $this->qualifier) |
|
182
|
23 |
|
&& ($pn = $this->getPName($v)) |
|
183
|
|
|
) { |
|
184
|
23 |
|
return $pn; |
|
185
|
|
|
} |
|
186
|
23 |
|
if (preg_match('/^[a-z0-9]+\:[^\s]*$/isu', $v)) { |
|
187
|
23 |
|
return '<'.$v.'>'; |
|
188
|
|
|
} |
|
189
|
|
|
|
|
190
|
23 |
|
return $this->getTerm(['type' => 'literal', 'value' => $v], $term, $qualifier); |
|
191
|
|
|
} |
|
192
|
23 |
|
if (!isset($v['type']) || ('literal' != $v['type'])) { |
|
193
|
|
|
return $this->getTerm($v['value'], $term, $qualifier); |
|
194
|
|
|
} |
|
195
|
|
|
/* literal */ |
|
196
|
23 |
|
$quot = '"'; |
|
197
|
23 |
|
if (false !== preg_match('/\"/', $v['value'])) { |
|
198
|
23 |
|
$quot = "'"; |
|
199
|
23 |
|
if (false !== preg_match('/\'/', $v['value']) || false !== preg_match('/[\x0d\x0a]/', $v['value'])) { |
|
200
|
23 |
|
$quot = '"""'; |
|
201
|
|
|
if ( |
|
202
|
23 |
|
false !== preg_match('/\"\"\"/', $v['value']) |
|
203
|
|
|
|| false !== preg_match('/\"$/', $v['value']) |
|
204
|
23 |
|
|| false !== preg_match('/^\"/', $v['value']) |
|
205
|
|
|
) { |
|
206
|
23 |
|
$quot = "'''"; |
|
207
|
23 |
|
$v['value'] = preg_replace("/'$/", "' ", $v['value']); |
|
208
|
23 |
|
$v['value'] = preg_replace("/^'/", " '", $v['value']); |
|
209
|
23 |
|
$v['value'] = str_replace("'''", '\\\'\\\'\\\'', $v['value']); |
|
210
|
|
|
} |
|
211
|
|
|
} |
|
212
|
|
|
} |
|
213
|
23 |
|
if ((1 == \strlen($quot)) && false !== preg_match('/[\x0d\x0a]/', $v['value'])) { |
|
214
|
|
|
$quot = $quot.$quot.$quot; |
|
215
|
|
|
} |
|
216
|
23 |
|
$suffix = isset($v['lang']) && $v['lang'] ? '@'.$v['lang'] : ''; |
|
217
|
23 |
|
$suffix = isset($v['datatype']) && $v['datatype'] ? '^^'.$this->getTerm($v['datatype'], 'dt') : $suffix; |
|
218
|
|
|
|
|
219
|
23 |
|
return $quot.$v['value'].$quot.$suffix; |
|
220
|
|
|
} |
|
221
|
|
|
|
|
222
|
23 |
|
public function getHead() |
|
223
|
|
|
{ |
|
224
|
23 |
|
$r = ''; |
|
225
|
23 |
|
$nl = "\n"; |
|
226
|
23 |
|
foreach ($this->used_ns as $v) { |
|
227
|
23 |
|
$r .= $r ? $nl : ''; |
|
228
|
23 |
|
foreach ($this->ns as $prefix => $ns) { |
|
229
|
23 |
|
if ($ns != $v) { |
|
230
|
23 |
|
continue; |
|
231
|
|
|
} |
|
232
|
23 |
|
$r .= '@prefix '.$prefix.': <'.$v.'> .'; |
|
233
|
23 |
|
break; |
|
234
|
|
|
} |
|
235
|
|
|
} |
|
236
|
|
|
|
|
237
|
23 |
|
return $r; |
|
238
|
|
|
} |
|
239
|
|
|
|
|
240
|
23 |
|
public function getSerializedIndex($index, $raw = 0) |
|
241
|
|
|
{ |
|
242
|
23 |
|
$r = ''; |
|
243
|
23 |
|
$nl = "\n"; |
|
244
|
23 |
|
foreach ($index as $s => $ps) { |
|
245
|
23 |
|
$r .= $r ? ' .'.$nl.$nl : ''; |
|
246
|
23 |
|
$s = $this->getTerm($s, 's'); |
|
247
|
23 |
|
$r .= $s; |
|
248
|
23 |
|
$first_p = 1; |
|
249
|
23 |
|
foreach ($ps as $p => $os) { |
|
250
|
23 |
|
if (!$os) { |
|
251
|
|
|
continue; |
|
252
|
|
|
} |
|
253
|
23 |
|
$p = $this->getTerm($p, 'p'); |
|
254
|
23 |
|
$r .= $first_p ? ' ' : ' ;'.$nl.str_pad('', \strlen($s) + 1); |
|
255
|
23 |
|
$r .= $p; |
|
256
|
23 |
|
$first_o = 1; |
|
257
|
23 |
|
if (!\is_array($os)) {/* single literal o */ |
|
258
|
|
|
$os = [['value' => $os, 'type' => 'literal']]; |
|
259
|
|
|
} |
|
260
|
23 |
|
foreach ($os as $o) { |
|
261
|
23 |
|
$r .= $first_o ? ' ' : ' ,'.$nl.str_pad('', \strlen($s) + \strlen($p) + 2); |
|
262
|
23 |
|
$o = $this->getTerm($o, 'o', $p); |
|
263
|
23 |
|
$r .= $o; |
|
264
|
23 |
|
$first_o = 0; |
|
265
|
|
|
} |
|
266
|
23 |
|
$first_p = 0; |
|
267
|
|
|
} |
|
268
|
|
|
} |
|
269
|
23 |
|
$r .= $r ? ' .' : ''; |
|
270
|
23 |
|
if ($raw) { |
|
271
|
|
|
return $r; |
|
272
|
|
|
} |
|
273
|
|
|
|
|
274
|
23 |
|
return $r ? $this->getHead().$nl.$nl.$r : ''; |
|
275
|
|
|
} |
|
276
|
|
|
} |
|
277
|
|
|
|
sweetrdf /
in-memory-store-sqlite
Loading history...