|
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\Store\QueryHandler; |
|
15
|
|
|
|
|
16
|
|
|
use function sweetrdf\InMemoryStoreSqlite\calcURI; |
|
17
|
|
|
use sweetrdf\InMemoryStoreSqlite\Store\TurtleLoader; |
|
18
|
|
|
|
|
19
|
|
|
class LoadQueryHandler extends QueryHandler |
|
20
|
|
|
{ |
|
21
|
|
|
private string $target_graph; |
|
22
|
|
|
|
|
23
|
|
|
/** |
|
24
|
|
|
* @todo required? |
|
25
|
|
|
*/ |
|
26
|
|
|
private int $t_count; |
|
27
|
|
|
|
|
28
|
|
|
private int $write_buffer_size = 2500; |
|
29
|
|
|
|
|
30
|
24 |
|
public function runQuery($infos, $data = '', $keep_bnode_ids = 0) |
|
31
|
|
|
{ |
|
32
|
24 |
|
$url = $infos['query']['url']; |
|
33
|
24 |
|
$graph = $infos['query']['target_graph']; |
|
34
|
24 |
|
$this->target_graph = $graph ? calcURI($graph) : calcURI($url); |
|
35
|
24 |
|
$this->keep_bnode_ids = $keep_bnode_ids; |
|
|
|
|
|
|
36
|
|
|
|
|
37
|
|
|
// remove parameters |
|
38
|
24 |
|
$loader = new TurtleLoader(); |
|
39
|
24 |
|
$loader->setCaller($this); |
|
40
|
|
|
|
|
41
|
|
|
/* logging */ |
|
42
|
24 |
|
$this->t_count = 0; |
|
43
|
24 |
|
$this->t_start = 0; |
|
|
|
|
|
|
44
|
|
|
/* load and parse */ |
|
45
|
24 |
|
$this->max_term_id = $this->getMaxTermID(); |
|
|
|
|
|
|
46
|
24 |
|
$this->max_triple_id = $this->getMaxTripleID(); |
|
|
|
|
|
|
47
|
|
|
|
|
48
|
24 |
|
$this->term_ids = []; |
|
|
|
|
|
|
49
|
24 |
|
$this->triple_ids = []; |
|
|
|
|
|
|
50
|
24 |
|
$this->sql_buffers = []; |
|
|
|
|
|
|
51
|
24 |
|
$loader->parse($url, $data); |
|
52
|
|
|
|
|
53
|
|
|
/* done */ |
|
54
|
24 |
|
$this->checkSQLBuffers(1); |
|
55
|
|
|
|
|
56
|
|
|
return [ |
|
57
|
24 |
|
't_count' => $this->t_count, |
|
58
|
24 |
|
'load_time' => 0, |
|
59
|
|
|
]; |
|
60
|
|
|
} |
|
61
|
|
|
|
|
62
|
24 |
|
public function addT($s, $p, $o, $s_type, $o_type, $o_dt = '', $o_lang = '') |
|
63
|
|
|
{ |
|
64
|
24 |
|
$type_ids = ['uri' => '0', 'bnode' => '1', 'literal' => '2']; |
|
65
|
24 |
|
$g = $this->getStoredTermID($this->target_graph, '0', 'id'); |
|
66
|
24 |
|
$s = (('bnode' == $s_type) && !$this->keep_bnode_ids) ? '_:b'.abs(crc32($g.$s)).'_'.(\strlen($s) > 12 ? substr(substr($s, 2), -10) : substr($s, 2)) : $s; |
|
67
|
24 |
|
$o = (('bnode' == $o_type) && !$this->keep_bnode_ids) ? '_:b'.abs(crc32($g.$o)).'_'.(\strlen($o) > 12 ? substr(substr($o, 2), -10) : substr($o, 2)) : $o; |
|
68
|
|
|
/* triple */ |
|
69
|
24 |
|
$t = [ |
|
70
|
24 |
|
's' => $this->getStoredTermID($s, $type_ids[$s_type], 's'), |
|
71
|
24 |
|
'p' => $this->getStoredTermID($p, '0', 'id'), |
|
72
|
24 |
|
'o' => $this->getStoredTermID($o, $type_ids[$o_type], 'o'), |
|
73
|
24 |
|
'o_lang_dt' => $this->getStoredTermID($o_dt.$o_lang, $o_dt ? '0' : '2', 'id'), |
|
74
|
24 |
|
'o_comp' => $this->getOComp($o), |
|
75
|
24 |
|
's_type' => $type_ids[$s_type], |
|
76
|
24 |
|
'o_type' => $type_ids[$o_type], |
|
77
|
|
|
]; |
|
78
|
24 |
|
$t['t'] = $this->getTripleID($t); |
|
79
|
24 |
|
if (\is_array($t['t'])) {/* t exists already */ |
|
80
|
|
|
$t['t'] = $t['t'][0]; |
|
81
|
|
|
} else { |
|
82
|
24 |
|
$this->bufferTripleSQL($t); |
|
83
|
|
|
} |
|
84
|
|
|
/* g2t */ |
|
85
|
24 |
|
$g2t = ['g' => $g, 't' => $t['t']]; |
|
86
|
24 |
|
$this->bufferGraphSQL($g2t); |
|
87
|
24 |
|
++$this->t_count; |
|
88
|
|
|
/* check buffers */ |
|
89
|
24 |
|
if (0 == ($this->t_count % $this->write_buffer_size)) { |
|
90
|
|
|
$force_write = 1; |
|
91
|
|
|
$reset_buffers = (0 == ($this->t_count % ($this->write_buffer_size * 2))); |
|
92
|
|
|
$refresh_lock = (0 == ($this->t_count % 25000)); |
|
93
|
|
|
$split_tables = (0 == ($this->t_count % ($this->write_buffer_size * 10))); |
|
94
|
|
|
$this->checkSQLBuffers($force_write, $reset_buffers, $refresh_lock, $split_tables); |
|
|
|
|
|
|
95
|
|
|
} |
|
96
|
24 |
|
} |
|
97
|
|
|
|
|
98
|
24 |
|
public function getMaxTermID(): int |
|
99
|
|
|
{ |
|
100
|
24 |
|
$sql = ''; |
|
101
|
24 |
|
foreach (['id2val', 's2val', 'o2val'] as $tbl) { |
|
102
|
24 |
|
$sql .= $sql ? ' UNION ' : ''; |
|
103
|
24 |
|
$sql .= 'SELECT MAX(id) as id FROM '.$tbl; |
|
104
|
|
|
} |
|
105
|
24 |
|
$r = 0; |
|
106
|
|
|
|
|
107
|
24 |
|
$rows = $this->store->getDBObject()->fetchList($sql); |
|
108
|
|
|
|
|
109
|
24 |
|
if (\is_array($rows)) { |
|
|
|
|
|
|
110
|
24 |
|
foreach ($rows as $row) { |
|
111
|
24 |
|
$r = ($r < $row['id']) ? $row['id'] : $r; |
|
112
|
|
|
} |
|
113
|
|
|
} |
|
114
|
|
|
|
|
115
|
24 |
|
return $r + 1; |
|
116
|
|
|
} |
|
117
|
|
|
|
|
118
|
|
|
/** |
|
119
|
|
|
* @todo change DB schema and avoid using this function because it does not protect against race conditions |
|
120
|
|
|
* |
|
121
|
|
|
* @return int |
|
122
|
|
|
*/ |
|
123
|
24 |
|
public function getMaxTripleID() |
|
124
|
|
|
{ |
|
125
|
24 |
|
$sql = 'SELECT MAX(t) AS `id` FROM triple'; |
|
126
|
|
|
|
|
127
|
24 |
|
$row = $this->store->getDBObject()->fetchRow($sql); |
|
128
|
24 |
|
if (isset($row['id'])) { |
|
129
|
5 |
|
return $row['id'] + 1; |
|
130
|
|
|
} |
|
131
|
|
|
|
|
132
|
24 |
|
return 1; |
|
133
|
|
|
} |
|
134
|
|
|
|
|
135
|
24 |
|
public function getStoredTermID($val, $type_id, $tbl) |
|
136
|
|
|
{ |
|
137
|
|
|
/* buffered */ |
|
138
|
24 |
|
if (isset($this->term_ids[$val])) { |
|
139
|
24 |
|
if (!isset($this->term_ids[$val][$tbl])) { |
|
140
|
23 |
|
foreach (['id', 's', 'o'] as $other_tbl) { |
|
141
|
23 |
|
if (isset($this->term_ids[$val][$other_tbl])) { |
|
142
|
23 |
|
$this->term_ids[$val][$tbl] = $this->term_ids[$val][$other_tbl]; |
|
|
|
|
|
|
143
|
23 |
|
$this->bufferIDSQL($tbl, $this->term_ids[$val][$tbl], $val, $type_id); |
|
144
|
23 |
|
break; |
|
145
|
|
|
} |
|
146
|
|
|
} |
|
147
|
|
|
} |
|
148
|
|
|
|
|
149
|
24 |
|
return $this->term_ids[$val][$tbl]; |
|
150
|
|
|
} |
|
151
|
|
|
/* db */ |
|
152
|
24 |
|
$sub_tbls = ('id' == $tbl) |
|
153
|
24 |
|
? ['id2val', 's2val', 'o2val'] |
|
154
|
24 |
|
: ('s' == $tbl |
|
155
|
24 |
|
? ['s2val', 'id2val', 'o2val'] |
|
156
|
24 |
|
: ['o2val', 'id2val', 's2val'] |
|
157
|
|
|
); |
|
158
|
|
|
|
|
159
|
24 |
|
foreach ($sub_tbls as $sub_tbl) { |
|
160
|
24 |
|
$id = 0; |
|
161
|
|
|
/* via hash */ |
|
162
|
24 |
|
if (preg_match('/^(s2val|o2val)$/', $sub_tbl)) { |
|
163
|
24 |
|
$sql = 'SELECT id, val |
|
164
|
24 |
|
FROM '.$sub_tbl.' |
|
165
|
24 |
|
WHERE val_hash = "'.$this->getValueHash($val).'"'; |
|
166
|
|
|
|
|
167
|
24 |
|
$rows = $this->store->getDBObject()->fetchList($sql); |
|
168
|
24 |
|
if (\is_array($rows)) { |
|
169
|
24 |
|
foreach ($rows as $row) { |
|
170
|
|
|
if ($row['val'] == $val) { |
|
171
|
|
|
$id = $row['id']; |
|
172
|
|
|
break; |
|
173
|
|
|
} |
|
174
|
|
|
} |
|
175
|
|
|
} |
|
176
|
|
|
} else { |
|
177
|
24 |
|
$binaryValue = $this->store->getDBObject()->escape($val); |
|
178
|
24 |
|
if (false !== empty($binaryValue)) { |
|
179
|
24 |
|
$sql = 'SELECT id FROM '.$sub_tbl." WHERE val = '".$binaryValue."'"; |
|
180
|
|
|
|
|
181
|
24 |
|
$row = $this->store->getDBObject()->fetchRow($sql); |
|
182
|
24 |
|
if (\is_array($row) && isset($row['id'])) { |
|
183
|
5 |
|
$id = $row['id']; |
|
184
|
|
|
} |
|
185
|
|
|
} |
|
186
|
|
|
} |
|
187
|
24 |
|
if (0 < $id) { |
|
188
|
5 |
|
$this->term_ids[$val] = [$tbl => $id]; |
|
189
|
5 |
|
if ($sub_tbl != $tbl.'2val') { |
|
190
|
|
|
$this->bufferIDSQL($tbl, $id, $val, $type_id); |
|
191
|
|
|
} |
|
192
|
5 |
|
break; |
|
193
|
|
|
} |
|
194
|
|
|
} |
|
195
|
|
|
/* new */ |
|
196
|
24 |
|
if (!isset($this->term_ids[$val])) { |
|
197
|
24 |
|
$this->term_ids[$val] = [$tbl => $this->max_term_id]; |
|
198
|
24 |
|
$this->bufferIDSQL($tbl, $this->max_term_id, $val, $type_id); |
|
199
|
24 |
|
++$this->max_term_id; |
|
200
|
|
|
} |
|
201
|
|
|
|
|
202
|
24 |
|
return $this->term_ids[$val][$tbl]; |
|
203
|
|
|
} |
|
204
|
|
|
|
|
205
|
24 |
|
public function getTripleID($t) |
|
206
|
|
|
{ |
|
207
|
24 |
|
$val = serialize($t); |
|
208
|
|
|
/* buffered */ |
|
209
|
24 |
|
if (isset($this->triple_ids[$val])) { |
|
210
|
|
|
/* hack for "don't insert this triple" */ |
|
211
|
|
|
return [$this->triple_ids[$val]]; |
|
212
|
|
|
} |
|
213
|
|
|
/* db */ |
|
214
|
24 |
|
$sql = 'SELECT t |
|
215
|
|
|
FROM triple |
|
216
|
24 |
|
WHERE s = '.$t['s'].' |
|
217
|
24 |
|
AND p = '.$t['p'].' |
|
218
|
24 |
|
AND o = '.$t['o'].' |
|
219
|
24 |
|
AND o_lang_dt = '.$t['o_lang_dt'].' |
|
220
|
24 |
|
AND s_type = '.$t['s_type'].' |
|
221
|
24 |
|
AND o_type = '.$t['o_type'].' |
|
222
|
|
|
LIMIT 1'; |
|
223
|
24 |
|
$row = $this->store->getDBObject()->fetchRow($sql); |
|
224
|
24 |
|
if (isset($row['t'])) { |
|
225
|
|
|
/* hack for "don't insert this triple" */ |
|
226
|
|
|
$this->triple_ids[$val] = $row['t']; |
|
|
|
|
|
|
227
|
|
|
|
|
228
|
|
|
return [$row['t']]; |
|
229
|
|
|
} else { |
|
230
|
|
|
/* new */ |
|
231
|
24 |
|
$this->triple_ids[$val] = $this->max_triple_id; |
|
232
|
24 |
|
++$this->max_triple_id; |
|
233
|
|
|
|
|
234
|
24 |
|
return $this->triple_ids[$val]; |
|
235
|
|
|
} |
|
236
|
|
|
} |
|
237
|
|
|
|
|
238
|
25 |
|
public function getOComp($val) |
|
239
|
|
|
{ |
|
240
|
|
|
/* try date (e.g. 21 August 2007) */ |
|
241
|
|
|
if ( |
|
242
|
25 |
|
preg_match('/^[0-9]{1,2}\s+[a-z]+\s+[0-9]{4}/i', $val) |
|
243
|
25 |
|
&& ($uts = strtotime($val)) |
|
244
|
25 |
|
&& (-1 !== $uts) |
|
245
|
|
|
) { |
|
246
|
|
|
return date("Y-m-d\TH:i:s", $uts); |
|
247
|
|
|
} |
|
248
|
|
|
|
|
249
|
|
|
/* xsd date (e.g. 2009-05-28T18:03:38+09:00 2009-05-28T18:03:38GMT) */ |
|
250
|
25 |
|
if (true === (bool) strtotime($val)) { |
|
251
|
3 |
|
return date('Y-m-d\TH:i:s\Z', strtotime($val)); |
|
252
|
|
|
} |
|
253
|
|
|
|
|
254
|
24 |
|
if (is_numeric($val)) { |
|
255
|
2 |
|
$val = sprintf('%f', $val); |
|
256
|
2 |
|
if (preg_match("/([\-\+])([0-9]*)\.([0-9]*)/", $val, $m)) { |
|
257
|
1 |
|
return $m[1].sprintf('%018s', $m[2]).'.'.sprintf('%-015s', $m[3]); |
|
258
|
|
|
} |
|
259
|
2 |
|
if (preg_match("/([0-9]*)\.([0-9]*)/", $val, $m)) { |
|
260
|
2 |
|
return '+'.sprintf('%018s', $m[1]).'.'.sprintf('%-015s', $m[2]); |
|
261
|
|
|
} |
|
262
|
|
|
|
|
263
|
|
|
return $val; |
|
264
|
|
|
} |
|
265
|
|
|
|
|
266
|
|
|
/* any other string: remove tags, linebreaks etc., but keep MB-chars */ |
|
267
|
|
|
// [\PL\s]+ ( = non-Letters) kills digits |
|
268
|
24 |
|
$re = '/[\PL\s]+/isu'; |
|
|
|
|
|
|
269
|
24 |
|
$re = '/[\s\'\"\´\`]+/is'; |
|
270
|
24 |
|
$val = trim(preg_replace($re, '-', strip_tags($val))); |
|
271
|
24 |
|
if (\strlen($val) > 35) { |
|
272
|
24 |
|
$fnc = \function_exists('mb_substr') ? 'mb_substr' : 'substr'; |
|
273
|
24 |
|
$val = $fnc($val, 0, 17).'-'.$fnc($val, -17); |
|
274
|
|
|
} |
|
275
|
|
|
|
|
276
|
24 |
|
return $val; |
|
277
|
|
|
} |
|
278
|
|
|
|
|
279
|
24 |
|
public function bufferTripleSQL($t) |
|
280
|
|
|
{ |
|
281
|
24 |
|
$tbl = 'triple'; |
|
282
|
24 |
|
$sql = ', '; |
|
283
|
|
|
|
|
284
|
24 |
|
$sqlHead = 'INSERT OR IGNORE INTO '; |
|
285
|
|
|
|
|
286
|
24 |
|
if (!isset($this->sql_buffers[$tbl])) { |
|
287
|
24 |
|
$this->sql_buffers[$tbl] = $sqlHead; |
|
|
|
|
|
|
288
|
24 |
|
$this->sql_buffers[$tbl] .= $tbl; |
|
289
|
24 |
|
$this->sql_buffers[$tbl] .= ' (t, s, p, o, o_lang_dt, o_comp, s_type, o_type) VALUES'; |
|
290
|
24 |
|
$sql = ' '; |
|
291
|
|
|
} |
|
292
|
|
|
|
|
293
|
24 |
|
$oCompEscaped = $this->store->getDBObject()->escape($t['o_comp']); |
|
294
|
|
|
|
|
295
|
24 |
|
$this->sql_buffers[$tbl] .= $sql.'('.$t['t'].', '.$t['s'].', '.$t['p'].', '; |
|
296
|
24 |
|
$this->sql_buffers[$tbl] .= $t['o'].', '.$t['o_lang_dt'].", '"; |
|
297
|
24 |
|
$this->sql_buffers[$tbl] .= $oCompEscaped."', ".$t['s_type'].', '.$t['o_type'].')'; |
|
298
|
24 |
|
} |
|
299
|
|
|
|
|
300
|
24 |
|
public function bufferGraphSQL($g2t) |
|
301
|
|
|
{ |
|
302
|
24 |
|
$tbl = 'g2t'; |
|
303
|
24 |
|
$sql = ', '; |
|
304
|
|
|
|
|
305
|
|
|
/* |
|
306
|
|
|
* Use appropriate INSERT syntax, depending on the DBS. |
|
307
|
|
|
*/ |
|
308
|
24 |
|
$sqlHead = 'INSERT OR IGNORE INTO '; |
|
309
|
|
|
|
|
310
|
24 |
|
if (!isset($this->sql_buffers[$tbl])) { |
|
311
|
24 |
|
$this->sql_buffers[$tbl] = $sqlHead.$tbl.' (g, t) VALUES'; |
|
|
|
|
|
|
312
|
24 |
|
$sql = ' '; |
|
313
|
|
|
} |
|
314
|
24 |
|
$this->sql_buffers[$tbl] .= $sql.'('.$g2t['g'].', '.$g2t['t'].')'; |
|
315
|
24 |
|
} |
|
316
|
|
|
|
|
317
|
24 |
|
public function bufferIDSQL($tbl, $id, $val, $val_type) |
|
318
|
|
|
{ |
|
319
|
24 |
|
$tbl = $tbl.'2val'; |
|
320
|
24 |
|
if ('id2val' == $tbl) { |
|
321
|
24 |
|
$cols = 'id, val, val_type'; |
|
322
|
24 |
|
$vals = '('.$id.", '".$this->store->getDBObject()->escape($val)."', ".$val_type.')'; |
|
323
|
24 |
|
} elseif (preg_match('/^(s2val|o2val)$/', $tbl)) { |
|
324
|
24 |
|
$cols = 'id, val_hash, val'; |
|
325
|
24 |
|
$vals = '('.$id.", '" |
|
326
|
24 |
|
.$this->getValueHash($val) |
|
327
|
24 |
|
."', '" |
|
328
|
24 |
|
.$this->store->getDBObject()->escape($val) |
|
329
|
24 |
|
."')"; |
|
330
|
|
|
} else { |
|
331
|
|
|
$cols = 'id, val'; |
|
332
|
|
|
$vals = '('.$id.", '".$this->store->getDBObject()->escape($val)."')"; |
|
333
|
|
|
} |
|
334
|
24 |
|
if (!isset($this->sql_buffers[$tbl])) { |
|
335
|
24 |
|
$this->sql_buffers[$tbl] = ''; |
|
|
|
|
|
|
336
|
24 |
|
$sqlHead = 'INSERT OR IGNORE INTO '; |
|
337
|
|
|
|
|
338
|
24 |
|
$sql = $sqlHead.$tbl.'('.$cols.') VALUES '; |
|
339
|
|
|
} else { |
|
340
|
24 |
|
$sql = ', '; |
|
341
|
|
|
} |
|
342
|
24 |
|
$sql .= $vals; |
|
343
|
24 |
|
$this->sql_buffers[$tbl] .= $sql; |
|
344
|
24 |
|
} |
|
345
|
|
|
|
|
346
|
24 |
|
public function checkSQLBuffers($force_write = 0, $reset_id_buffers = 0) |
|
347
|
|
|
{ |
|
348
|
24 |
|
foreach (['triple', 'g2t', 'id2val', 's2val', 'o2val'] as $tbl) { |
|
349
|
24 |
|
$buffer_size = isset($this->sql_buffers[$tbl]) ? 1 : 0; |
|
350
|
24 |
|
if ($buffer_size && $force_write) { |
|
351
|
24 |
|
$this->store->getDBObject()->simpleQuery($this->sql_buffers[$tbl]); |
|
352
|
|
|
/* table error */ |
|
353
|
24 |
|
$error = $this->store->getDBObject()->getErrorMessage(); |
|
|
|
|
|
|
354
|
24 |
|
unset($this->sql_buffers[$tbl]); |
|
355
|
|
|
|
|
356
|
|
|
/* reset term id buffers */ |
|
357
|
24 |
|
if ($reset_id_buffers) { |
|
358
|
|
|
$this->term_ids = []; |
|
|
|
|
|
|
359
|
|
|
$this->triple_ids = []; |
|
|
|
|
|
|
360
|
|
|
} |
|
361
|
|
|
} |
|
362
|
|
|
} |
|
363
|
|
|
|
|
364
|
24 |
|
return 1; |
|
365
|
|
|
} |
|
366
|
|
|
} |
|
367
|
|
|
|
sweetrdf /
in-memory-store-sqlite
Loading history...