1
|
|
|
<?php |
2
|
|
|
/** |
3
|
|
|
* User: jensk |
4
|
|
|
* Date: 1-3-2017 |
5
|
|
|
* Time: 13:21 |
6
|
|
|
*/ |
7
|
|
|
|
8
|
|
|
namespace CloudControl\Cms\search\indexer; |
9
|
|
|
|
10
|
|
|
use CloudControl\Cms\search\Indexer; |
11
|
|
|
|
12
|
|
|
|
13
|
|
|
/** |
14
|
|
|
* Class TermFieldLengthNorm |
15
|
|
|
* Formula = norm(d) = 1 / √numTerms |
16
|
|
|
* |
17
|
|
|
* @package CloudControl\Cms\search\indexer |
18
|
|
|
*/ |
19
|
|
|
class TermFieldLengthNorm |
20
|
|
|
{ |
21
|
|
|
/** |
22
|
|
|
* @var \PDO |
23
|
|
|
*/ |
24
|
|
|
protected $dbHandle; |
25
|
|
|
|
26
|
|
|
/** |
27
|
|
|
* TermFieldLengthNorm constructor. |
28
|
|
|
* |
29
|
|
|
* @param \PDO $dbHandle |
30
|
|
|
*/ |
31
|
|
|
public function __construct($dbHandle) |
32
|
|
|
{ |
33
|
|
|
$this->dbHandle = $dbHandle; |
34
|
|
|
} |
35
|
|
|
|
36
|
|
|
public function execute() |
37
|
|
|
{ |
38
|
|
|
$db = $this->dbHandle; |
39
|
|
|
$db->/** @scrutinizer ignore-call */ |
40
|
|
|
sqliteCreateFunction('sqrt', 'sqrt', 1); |
41
|
|
|
$stmt = $this->getStatement($db); |
42
|
|
|
$uniqueFieldsPerDocument = $stmt->fetchAll(\PDO::FETCH_OBJ); |
43
|
|
|
$values = array(); |
44
|
|
|
$i = 0; |
45
|
|
|
foreach ($uniqueFieldsPerDocument as $fieldRow) { |
46
|
|
|
$values[] = 'UPDATE term_frequency SET termNorm = 1/sqrt(' . (int)$fieldRow->termCount . ') WHERE documentPath = ' . $db->quote($fieldRow->documentPath) . ' AND field = ' . $db->quote($fieldRow->field) . ';'; |
47
|
|
|
$i += 1; |
48
|
|
|
if ($i >= Indexer::SQLITE_MAX_COMPOUND_SELECT) { |
49
|
|
|
$this->executeUpdateTermNorm($values, $db); |
50
|
|
|
$values = array(); |
51
|
|
|
$i = 0; |
52
|
|
|
} |
53
|
|
|
} |
54
|
|
|
if (count($values) !== 0) { |
55
|
|
|
$this->executeUpdateTermNorm($values, $db); |
56
|
|
|
} |
57
|
|
|
} |
58
|
|
|
|
59
|
|
|
/** |
60
|
|
|
* @param array $values |
61
|
|
|
* @param \PDO $db |
62
|
|
|
* @throws \Exception |
63
|
|
|
*/ |
64
|
|
|
private function executeUpdateTermNorm($values, $db) |
65
|
|
|
{ |
66
|
|
|
$sql = 'BEGIN TRANSACTION;' . PHP_EOL; |
67
|
|
|
$sql .= implode(PHP_EOL, $values) . PHP_EOL; |
68
|
|
|
$sql .= 'COMMIT;'; |
69
|
|
|
if (($db->exec($sql)) === false) { |
70
|
|
|
$errorInfo = $db->errorInfo(); |
71
|
|
|
$errorMsg = $errorInfo[2]; |
72
|
|
|
throw new \RuntimeException('SQLite Exception: ' . $errorMsg . ' in SQL: <br /><pre>' . $sql . '</pre>'); |
73
|
|
|
} |
74
|
|
|
} |
75
|
|
|
|
76
|
|
|
/** |
77
|
|
|
* @param \PDO $db |
78
|
|
|
* @return \PDOStatement |
79
|
|
|
* @throws \Exception |
80
|
|
|
*/ |
81
|
|
|
protected function getStatement($db) |
82
|
|
|
{ |
83
|
|
|
$sql = ' |
84
|
|
|
SELECT documentPath, field, COUNT(`count`) AS termCount |
85
|
|
|
FROM term_count |
86
|
|
|
GROUP BY documentPath, field |
87
|
|
|
'; |
88
|
|
|
$stmt = $db->prepare($sql); |
89
|
|
|
if ($stmt === false) { |
90
|
|
|
$errorInfo = $db->errorInfo(); |
91
|
|
|
$errorMsg = $errorInfo[2]; |
92
|
|
|
throw new \RuntimeException('SQLite Exception: ' . $errorMsg . ' in SQL: <br /><pre>' . $sql . '</pre>'); |
93
|
|
|
} |
94
|
|
|
if (($stmt->execute()) === false) { |
95
|
|
|
$errorInfo = $db->errorInfo(); |
96
|
|
|
$errorMsg = $errorInfo[2]; |
97
|
|
|
throw new \RuntimeException('SQLite Exception: ' . $errorMsg . ' in SQL: <br /><pre>' . $sql . '</pre>'); |
98
|
|
|
} |
99
|
|
|
return $stmt; |
100
|
|
|
} |
101
|
|
|
} |