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\NamespaceHelper; |
17
|
|
|
use sweetrdf\InMemoryStoreSqlite\StringReader; |
18
|
|
|
|
19
|
|
|
abstract class BaseParser |
20
|
|
|
{ |
21
|
|
|
/** |
22
|
|
|
* @var array |
23
|
|
|
*/ |
24
|
|
|
protected $added_triples; |
25
|
|
|
|
26
|
|
|
protected string $base; |
27
|
|
|
|
28
|
|
|
protected string $bnode_id; |
29
|
|
|
|
30
|
|
|
protected array $blocks; |
31
|
|
|
|
32
|
|
|
private array $errors = []; |
33
|
|
|
|
34
|
|
|
/** |
35
|
|
|
* @var array<string, string> |
36
|
|
|
*/ |
37
|
|
|
protected array $prefixes; |
38
|
|
|
|
39
|
|
|
/** |
40
|
|
|
* Query infos container. |
41
|
|
|
*/ |
42
|
|
|
protected array $r = []; |
43
|
|
|
|
44
|
|
|
protected array $triples = []; |
45
|
|
|
|
46
|
|
|
protected int $t_count = 0; |
47
|
|
|
|
48
|
106 |
|
public function __construct() |
49
|
|
|
{ |
50
|
|
|
// TODO pass logger as parameter |
51
|
106 |
|
$this->reader = new StringReader(); |
|
|
|
|
52
|
|
|
|
53
|
|
|
/* |
54
|
|
|
* @todo make it a constructor param |
55
|
|
|
*/ |
56
|
106 |
|
$this->prefixes = (new NamespaceHelper())->getNamespaces(); |
57
|
|
|
|
58
|
|
|
// generates random prefix for blank nodes |
59
|
106 |
|
$this->bnode_prefix = bin2hex(random_bytes(4)).'b'; |
|
|
|
|
60
|
|
|
|
61
|
106 |
|
$this->bnode_id = 0; |
62
|
106 |
|
} |
63
|
|
|
|
64
|
|
|
/** |
65
|
|
|
* @todo replace by Logger |
66
|
|
|
*/ |
67
|
19 |
|
protected function addError(string $error): void |
68
|
|
|
{ |
69
|
19 |
|
$this->errors[] = $error; |
70
|
19 |
|
} |
71
|
|
|
|
72
|
|
|
/** |
73
|
|
|
* @todo replace by Logger |
74
|
|
|
*/ |
75
|
106 |
|
public function getErrors(): array |
76
|
|
|
{ |
77
|
106 |
|
return $this->errors; |
78
|
|
|
} |
79
|
|
|
|
80
|
106 |
|
public function getQueryInfos() |
81
|
|
|
{ |
82
|
106 |
|
return $this->r; |
83
|
|
|
} |
84
|
|
|
|
85
|
23 |
|
public function getTriples() |
86
|
|
|
{ |
87
|
23 |
|
return $this->triples; |
88
|
|
|
} |
89
|
|
|
|
90
|
23 |
|
public function getSimpleIndex($flatten_objects = 1, $vals = ''): array |
91
|
|
|
{ |
92
|
23 |
|
return $this->_getSimpleIndex($this->getTriples(), $flatten_objects, $vals); |
93
|
|
|
} |
94
|
|
|
|
95
|
|
|
/** |
96
|
|
|
* @todo port from ARC2::getSimpleIndex; refactor and merge it with $this->getSimpleIndex |
97
|
|
|
*/ |
98
|
23 |
|
private function _getSimpleIndex($triples, $flatten_objects = 1, $vals = ''): array |
99
|
|
|
{ |
100
|
23 |
|
$r = []; |
101
|
23 |
|
foreach ($triples as $t) { |
102
|
23 |
|
$skip_t = 0; |
103
|
23 |
|
foreach (['s', 'p', 'o'] as $term) { |
104
|
23 |
|
$$term = $t[$term]; |
105
|
|
|
/* template var */ |
106
|
23 |
|
if (isset($t[$term.'_type']) && ('var' == $t[$term.'_type'])) { |
107
|
|
|
$val = isset($vals[$$term]) ? $vals[$$term] : ''; |
108
|
|
|
$skip_t = isset($vals[$$term]) ? $skip_t : 1; |
109
|
|
|
$type = ''; |
110
|
|
|
$type = !$type && isset($vals[$$term.' type']) ? $vals[$$term.' type'] : $type; |
111
|
|
|
$type = !$type && preg_match('/^\_\:/', $val) ? 'bnode' : $type; |
112
|
|
|
if ('o' == $term) { |
113
|
|
|
$type = !$type && (preg_match('/\s/s', $val) || !preg_match('/\:/', $val)) ? 'literal' : $type; |
114
|
|
|
$type = !$type && !preg_match('/[\/]/', $val) ? 'literal' : $type; |
115
|
|
|
} |
116
|
|
|
$type = !$type ? 'uri' : $type; |
117
|
|
|
$t[$term.'_type'] = $type; |
118
|
|
|
$$term = $val; |
119
|
|
|
} |
120
|
|
|
} |
121
|
23 |
|
if ($skip_t) { |
122
|
|
|
continue; |
123
|
|
|
} |
124
|
23 |
|
if (!isset($r[$s])) { |
|
|
|
|
125
|
23 |
|
$r[$s] = []; |
126
|
|
|
} |
127
|
23 |
|
if (!isset($r[$s][$p])) { |
|
|
|
|
128
|
23 |
|
$r[$s][$p] = []; |
129
|
|
|
} |
130
|
23 |
|
if ($flatten_objects) { |
131
|
23 |
|
if (!\in_array($o, $r[$s][$p])) { |
|
|
|
|
132
|
23 |
|
$r[$s][$p][] = $o; |
133
|
|
|
} |
134
|
|
|
} else { |
135
|
|
|
$o = ['value' => $o]; |
136
|
|
|
foreach (['lang', 'type', 'datatype'] as $suffix) { |
137
|
|
|
if (isset($t['o_'.$suffix]) && $t['o_'.$suffix]) { |
138
|
|
|
$o[$suffix] = $t['o_'.$suffix]; |
139
|
|
|
} elseif (isset($t['o '.$suffix]) && $t['o '.$suffix]) { |
140
|
|
|
$o[$suffix] = $t['o '.$suffix]; |
141
|
|
|
} |
142
|
|
|
} |
143
|
|
|
if (!\in_array($o, $r[$s][$p])) { |
144
|
|
|
$r[$s][$p][] = $o; |
145
|
|
|
} |
146
|
|
|
} |
147
|
|
|
} |
148
|
|
|
|
149
|
23 |
|
return $r; |
150
|
|
|
} |
151
|
|
|
} |
152
|
|
|
|