|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* User: jensk |
|
4
|
|
|
* Date: 1-3-2017 |
|
5
|
|
|
* Time: 10:34 |
|
6
|
|
|
*/ |
|
7
|
|
|
|
|
8
|
|
|
namespace CloudControl\Cms\search\indexer; |
|
9
|
|
|
|
|
10
|
|
|
use CloudControl\Cms\search\Indexer; |
|
11
|
|
|
|
|
12
|
|
|
/** |
|
13
|
|
|
* Calculate, relatively how often a term is used in a document |
|
14
|
|
|
* Where relativly means compared to the total of terms, how often is term X |
|
15
|
|
|
* used. For example: |
|
16
|
|
|
* doc1 has the following terms: |
|
17
|
|
|
* - term1 (count 2) |
|
18
|
|
|
* - term2 (count 1) |
|
19
|
|
|
* The total count of terms = 3 |
|
20
|
|
|
* The frequency of term1 in doc1 is: |
|
21
|
|
|
* count of term 1 / total count of terms |
|
22
|
|
|
* = |
|
23
|
|
|
* 2 / 3 = 0.66666666667 |
|
24
|
|
|
*/ |
|
25
|
|
|
class TermFrequency |
|
26
|
|
|
{ |
|
27
|
|
|
/** |
|
28
|
|
|
* @var \PDO |
|
29
|
|
|
*/ |
|
30
|
|
|
protected $dbHandle; |
|
31
|
|
|
|
|
32
|
|
|
/** |
|
33
|
|
|
* TermFrequency constructor. |
|
34
|
|
|
* |
|
35
|
|
|
* @param \PDO $dbHandle |
|
36
|
|
|
*/ |
|
37
|
|
|
public function __construct($dbHandle) |
|
38
|
|
|
{ |
|
39
|
|
|
$this->dbHandle = $dbHandle; |
|
40
|
|
|
} |
|
41
|
|
|
|
|
42
|
|
|
public function execute() |
|
43
|
|
|
{ |
|
44
|
|
|
$db = $this->dbHandle; |
|
45
|
|
|
$totalTermCountPerDocument = $this->getTotalTermCountPerDocument($db); |
|
46
|
|
|
foreach ($totalTermCountPerDocument as $documentField) { |
|
47
|
|
|
$termsForDocumentField = $this->getTermsForDocumentField($documentField->documentPath, |
|
48
|
|
|
$documentField->field); |
|
49
|
|
|
$sql = ' |
|
50
|
|
|
INSERT INTO term_frequency (documentPath, field, term, frequency) |
|
51
|
|
|
VALUES |
|
52
|
|
|
'; |
|
53
|
|
|
$quotedDocumentPath = $db->quote($documentField->documentPath); |
|
54
|
|
|
$quotedField = $db->quote($documentField->field); |
|
55
|
|
|
$values = array(); |
|
56
|
|
|
$i = 0; |
|
57
|
|
|
foreach ($termsForDocumentField as $term) { |
|
58
|
|
|
$frequency = (int)$term->count / $documentField->totalTermCount; |
|
59
|
|
|
$values[] = $quotedDocumentPath . ',' . $quotedField . ', ' . $db->quote($term->term) . ', ' . $db->quote($frequency); |
|
60
|
|
|
$i += 1; |
|
61
|
|
|
if ($i >= Indexer::SQLITE_MAX_COMPOUND_SELECT) { |
|
62
|
|
|
$this->executeStore($sql, $values, $db); |
|
63
|
|
|
$i = 0; |
|
64
|
|
|
$values = array(); |
|
65
|
|
|
} |
|
66
|
|
|
} |
|
67
|
|
|
if (count($values) != 0) { |
|
68
|
|
|
$this->executeStore($sql, $values, $db); |
|
69
|
|
|
} |
|
70
|
|
|
} |
|
71
|
|
|
} |
|
72
|
|
|
|
|
73
|
|
|
private function getTermsForDocumentField($documentPath, $field) |
|
74
|
|
|
{ |
|
75
|
|
|
$db = $this->dbHandle; |
|
76
|
|
|
$stmt = $db->prepare(' |
|
77
|
|
|
SELECT `term`, `count` |
|
78
|
|
|
FROM `term_count` |
|
79
|
|
|
WHERE `documentPath` = :documentPath |
|
80
|
|
|
AND `field` = :field |
|
81
|
|
|
'); |
|
82
|
|
|
$stmt->bindValue(':documentPath', $documentPath); |
|
83
|
|
|
$stmt->bindValue(':field', $field); |
|
84
|
|
|
$stmt->execute(); |
|
85
|
|
|
return $stmt->fetchAll(\PDO::FETCH_CLASS); |
|
86
|
|
|
} |
|
87
|
|
|
|
|
88
|
|
|
/** |
|
89
|
|
|
* @param $db |
|
90
|
|
|
* |
|
91
|
|
|
* @return mixed |
|
92
|
|
|
*/ |
|
93
|
|
|
private function getTotalTermCountPerDocument($db) |
|
94
|
|
|
{ |
|
95
|
|
|
$stmt = $db->prepare(' |
|
96
|
|
|
SELECT documentPath, field, SUM(count) AS totalTermCount |
|
97
|
|
|
FROM term_count |
|
98
|
|
|
GROUP BY documentPath, field |
|
99
|
|
|
'); |
|
100
|
|
|
$stmt->execute(); |
|
101
|
|
|
$totalTermCountPerDocument = $stmt->fetchAll(\PDO::FETCH_CLASS); |
|
102
|
|
|
|
|
103
|
|
|
return $totalTermCountPerDocument; |
|
104
|
|
|
} |
|
105
|
|
|
|
|
106
|
|
|
private function executeStore($sql, $values, $db) |
|
107
|
|
|
{ |
|
108
|
|
|
$sql .= '(' . implode('),' . PHP_EOL . '(', $values) . ');'; |
|
109
|
|
|
if (!$db->query($sql)) { |
|
110
|
|
|
$errorInfo = $db->errorInfo(); |
|
111
|
|
|
$errorMsg = $errorInfo[2]; |
|
112
|
|
|
throw new \Exception('SQLite Exception: ' . $errorMsg . ' in SQL: <br /><pre>' . $sql . '</pre>'); |
|
113
|
|
|
} |
|
114
|
|
|
} |
|
115
|
|
|
|
|
116
|
|
|
} |
jenskooij /
cloudcontrol
