|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* User: jensk |
|
4
|
|
|
* Date: 1-3-2017 |
|
5
|
|
|
* Time: 10:22 |
|
6
|
|
|
*/ |
|
7
|
|
|
|
|
8
|
|
|
namespace CloudControl\Cms\search\indexer; |
|
9
|
|
|
|
|
10
|
|
|
|
|
11
|
|
|
use CloudControl\Cms\search\DocumentTokenizer; |
|
12
|
|
|
use CloudControl\Cms\search\Indexer; |
|
13
|
|
|
use CloudControl\Cms\storage\Storage; |
|
14
|
|
|
|
|
15
|
|
|
class TermCount |
|
16
|
|
|
{ |
|
17
|
|
|
/** |
|
18
|
|
|
* @var \PDO |
|
19
|
|
|
*/ |
|
20
|
|
|
protected $dbHandle; |
|
21
|
|
|
protected $documents; |
|
22
|
|
|
protected $filters; |
|
23
|
|
|
protected $storage; |
|
24
|
|
|
|
|
25
|
|
|
/** |
|
26
|
|
|
* TermCount constructor. |
|
27
|
|
|
* |
|
28
|
|
|
* @param \PDO $dbHandle |
|
29
|
|
|
* @param array $documents |
|
30
|
|
|
* @param array $filters |
|
31
|
|
|
* @param Storage $jsonStorage |
|
32
|
|
|
*/ |
|
33
|
|
|
public function __construct($dbHandle, $documents, $filters, $jsonStorage) |
|
34
|
|
|
{ |
|
35
|
|
|
$this->dbHandle = $dbHandle; |
|
36
|
|
|
$this->documents = $documents; |
|
37
|
|
|
$this->filters = $filters; |
|
38
|
|
|
$this->storage = $jsonStorage; |
|
39
|
|
|
} |
|
40
|
|
|
|
|
41
|
|
|
public function execute() |
|
42
|
|
|
{ |
|
43
|
|
|
$this->iterateDocumentsAndCreateTermCount($this->documents); |
|
44
|
|
|
} |
|
45
|
|
|
|
|
46
|
|
|
protected function applyFilters($tokens) |
|
47
|
|
|
{ |
|
48
|
|
|
foreach ($this->filters as $filterName) { |
|
49
|
|
|
$filterClassName = '\CloudControl\Cms\search\filters\\' . $filterName; |
|
50
|
|
|
$filter = new $filterClassName($tokens); |
|
51
|
|
|
$tokens = $filter->getFilterResults(); |
|
52
|
|
|
} |
|
53
|
|
|
return $tokens; |
|
54
|
|
|
} |
|
55
|
|
|
|
|
56
|
|
|
protected function storeDocumentTermCount($document, $documentTermCount) |
|
57
|
|
|
{ |
|
58
|
|
|
$db = $this->dbHandle; |
|
59
|
|
|
$sqlStart = ' |
|
60
|
|
|
INSERT INTO `term_count` (`documentPath`, `term`, `count`, `field`) |
|
61
|
|
|
VALUES '; |
|
62
|
|
|
$sql = $sqlStart; |
|
63
|
|
|
$values = array(); |
|
64
|
|
|
$quotedDocumentPath = $db->quote($document->path); |
|
65
|
|
|
$i = 0; |
|
66
|
|
|
foreach ($documentTermCount as $field => $countArray) { |
|
67
|
|
|
$quotedField = $db->quote($field); |
|
68
|
|
|
foreach ($countArray as $term => $count) { |
|
69
|
|
|
$values[] = $quotedDocumentPath . ', ' . $db->quote($term) . ', ' . $db->quote($count) . ', ' . $quotedField; |
|
70
|
|
|
$i += 1; |
|
71
|
|
|
if ($i >= Indexer::SQLITE_MAX_COMPOUND_SELECT) { |
|
72
|
|
|
$this->executeStoreDocumentTermCount($values, $sql, $db); |
|
73
|
|
|
$values = array(); |
|
74
|
|
|
$sql = $sqlStart; |
|
75
|
|
|
$i = 0; |
|
76
|
|
|
} |
|
77
|
|
|
} |
|
78
|
|
|
} |
|
79
|
|
|
if (count($values) != 0) { |
|
80
|
|
|
$this->executeStoreDocumentTermCount($values, $sql, $db); |
|
81
|
|
|
} |
|
82
|
|
|
} |
|
83
|
|
|
|
|
84
|
|
|
/** |
|
85
|
|
|
* @param $values |
|
86
|
|
|
* @param $sql |
|
87
|
|
|
* @param $db |
|
88
|
|
|
* |
|
89
|
|
|
* @throws \Exception |
|
90
|
|
|
*/ |
|
91
|
|
|
protected function executeStoreDocumentTermCount($values, $sql, $db) |
|
92
|
|
|
{ |
|
93
|
|
|
$sql .= '(' . implode('),' . PHP_EOL . '(', $values) . ');'; |
|
94
|
|
|
$stmt = $db->prepare($sql); |
|
95
|
|
|
if ($stmt === false || !$stmt->execute()) { |
|
96
|
|
|
$errorInfo = $db->errorInfo(); |
|
97
|
|
|
$errorMsg = $errorInfo[2]; |
|
98
|
|
|
throw new \Exception('SQLite Exception: ' . $errorMsg . ' in SQL: <br /><pre>' . $sql . '</pre>'); |
|
99
|
|
|
} |
|
100
|
|
|
} |
|
101
|
|
|
|
|
102
|
|
|
/** |
|
103
|
|
|
* @param $document |
|
104
|
|
|
*/ |
|
105
|
|
|
private function createTermCountForDocument($document) |
|
106
|
|
|
{ |
|
107
|
|
|
$tokenizer = new DocumentTokenizer($document, $this->storage); |
|
108
|
|
|
$tokens = $tokenizer->getTokens(); |
|
109
|
|
|
$documentTermCount = $this->applyFilters($tokens); |
|
110
|
|
|
$this->storeDocumentTermCount($document, $documentTermCount); |
|
111
|
|
|
} |
|
112
|
|
|
|
|
113
|
|
|
/** |
|
114
|
|
|
* @param $documents |
|
115
|
|
|
*/ |
|
116
|
|
|
private function iterateDocumentsAndCreateTermCount($documents) |
|
117
|
|
|
{ |
|
118
|
|
|
foreach ($documents as $document) { |
|
119
|
|
|
if ($document->type === 'folder') { |
|
120
|
|
|
$this->iterateDocumentsAndCreateTermCount($document->content); |
|
121
|
|
|
} else { |
|
122
|
|
|
$this->createTermCountForDocument($document); |
|
123
|
|
|
} |
|
124
|
|
|
} |
|
125
|
|
|
} |
|
126
|
|
|
} |