|
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\Parser; |
|
15
|
|
|
|
|
16
|
|
|
use sweetrdf\InMemoryStoreSqlite\Log\Logger; |
|
17
|
|
|
use sweetrdf\InMemoryStoreSqlite\NamespaceHelper; |
|
18
|
|
|
use sweetrdf\InMemoryStoreSqlite\StringReader; |
|
19
|
|
|
|
|
20
|
|
|
abstract class BaseParser |
|
21
|
|
|
{ |
|
22
|
|
|
/** |
|
23
|
|
|
* @var array |
|
24
|
|
|
*/ |
|
25
|
|
|
protected $added_triples; |
|
26
|
|
|
|
|
27
|
|
|
protected string $base; |
|
28
|
|
|
|
|
29
|
|
|
protected string $bnode_id; |
|
30
|
|
|
|
|
31
|
|
|
protected array $blocks; |
|
32
|
|
|
|
|
33
|
|
|
protected Logger $logger; |
|
34
|
|
|
|
|
35
|
|
|
/** |
|
36
|
|
|
* @var array<string, string> |
|
37
|
|
|
*/ |
|
38
|
|
|
protected array $prefixes; |
|
39
|
|
|
|
|
40
|
|
|
/** |
|
41
|
|
|
* Query infos container. |
|
42
|
|
|
*/ |
|
43
|
|
|
protected array $r = []; |
|
44
|
|
|
|
|
45
|
|
|
protected array $triples = []; |
|
46
|
|
|
|
|
47
|
|
|
protected int $t_count = 0; |
|
48
|
|
|
|
|
49
|
100 |
|
public function __construct(Logger $logger) |
|
50
|
|
|
{ |
|
51
|
|
|
// TODO pass as constructor param |
|
52
|
100 |
|
$this->reader = new StringReader(); |
|
|
|
|
|
|
53
|
|
|
|
|
54
|
100 |
|
$this->logger = $logger; |
|
55
|
|
|
|
|
56
|
|
|
// TODO make it a constructor param |
|
57
|
100 |
|
$this->prefixes = (new NamespaceHelper())->getNamespaces(); |
|
58
|
|
|
|
|
59
|
|
|
// generates random prefix for blank nodes |
|
60
|
100 |
|
$this->bnode_prefix = bin2hex(random_bytes(4)).'b'; |
|
|
|
|
|
|
61
|
|
|
|
|
62
|
100 |
|
$this->bnode_id = 0; |
|
63
|
100 |
|
} |
|
64
|
|
|
|
|
65
|
100 |
|
public function getQueryInfos() |
|
66
|
|
|
{ |
|
67
|
100 |
|
return $this->r; |
|
68
|
|
|
} |
|
69
|
|
|
|
|
70
|
23 |
|
public function getTriples() |
|
71
|
|
|
{ |
|
72
|
23 |
|
return $this->triples; |
|
73
|
|
|
} |
|
74
|
|
|
|
|
75
|
23 |
|
public function getSimpleIndex($flatten_objects = 1, $vals = ''): array |
|
76
|
|
|
{ |
|
77
|
23 |
|
return $this->_getSimpleIndex($this->getTriples(), $flatten_objects, $vals); |
|
78
|
|
|
} |
|
79
|
|
|
|
|
80
|
|
|
/** |
|
81
|
|
|
* @todo port from ARC2::getSimpleIndex; refactor and merge it with $this->getSimpleIndex |
|
82
|
|
|
*/ |
|
83
|
23 |
|
private function _getSimpleIndex($triples, $flatten_objects = 1, $vals = ''): array |
|
84
|
|
|
{ |
|
85
|
23 |
|
$r = []; |
|
86
|
23 |
|
foreach ($triples as $t) { |
|
87
|
23 |
|
$skip_t = 0; |
|
88
|
23 |
|
foreach (['s', 'p', 'o'] as $term) { |
|
89
|
23 |
|
$$term = $t[$term]; |
|
90
|
|
|
/* template var */ |
|
91
|
23 |
|
if (isset($t[$term.'_type']) && ('var' == $t[$term.'_type'])) { |
|
92
|
|
|
$val = isset($vals[$$term]) ? $vals[$$term] : ''; |
|
93
|
|
|
$skip_t = isset($vals[$$term]) ? $skip_t : 1; |
|
94
|
|
|
$type = ''; |
|
95
|
|
|
$type = !$type && isset($vals[$$term.' type']) ? $vals[$$term.' type'] : $type; |
|
96
|
|
|
$type = !$type && preg_match('/^\_\:/', $val) ? 'bnode' : $type; |
|
97
|
|
|
if ('o' == $term) { |
|
98
|
|
|
$type = !$type && (preg_match('/\s/s', $val) || !preg_match('/\:/', $val)) ? 'literal' : $type; |
|
99
|
|
|
$type = !$type && !preg_match('/[\/]/', $val) ? 'literal' : $type; |
|
100
|
|
|
} |
|
101
|
|
|
$type = !$type ? 'uri' : $type; |
|
102
|
|
|
$t[$term.'_type'] = $type; |
|
103
|
|
|
$$term = $val; |
|
104
|
|
|
} |
|
105
|
|
|
} |
|
106
|
23 |
|
if ($skip_t) { |
|
107
|
|
|
continue; |
|
108
|
|
|
} |
|
109
|
23 |
|
if (!isset($r[$s])) { |
|
|
|
|
|
|
110
|
23 |
|
$r[$s] = []; |
|
111
|
|
|
} |
|
112
|
23 |
|
if (!isset($r[$s][$p])) { |
|
|
|
|
|
|
113
|
23 |
|
$r[$s][$p] = []; |
|
114
|
|
|
} |
|
115
|
23 |
|
if ($flatten_objects) { |
|
116
|
23 |
|
if (!\in_array($o, $r[$s][$p])) { |
|
|
|
|
|
|
117
|
23 |
|
$r[$s][$p][] = $o; |
|
118
|
|
|
} |
|
119
|
|
|
} else { |
|
120
|
|
|
$o = ['value' => $o]; |
|
121
|
|
|
foreach (['lang', 'type', 'datatype'] as $suffix) { |
|
122
|
|
|
if (isset($t['o_'.$suffix]) && $t['o_'.$suffix]) { |
|
123
|
|
|
$o[$suffix] = $t['o_'.$suffix]; |
|
124
|
|
|
} elseif (isset($t['o '.$suffix]) && $t['o '.$suffix]) { |
|
125
|
|
|
$o[$suffix] = $t['o '.$suffix]; |
|
126
|
|
|
} |
|
127
|
|
|
} |
|
128
|
|
|
if (!\in_array($o, $r[$s][$p])) { |
|
129
|
|
|
$r[$s][$p][] = $o; |
|
130
|
|
|
} |
|
131
|
|
|
} |
|
132
|
|
|
} |
|
133
|
|
|
|
|
134
|
23 |
|
return $r; |
|
135
|
|
|
} |
|
136
|
|
|
} |
|
137
|
|
|
|