custom_search::prepareIndexes() - Code Metrics - SimpleMachines/SMF2.1 - Measure and Improve Code Quality continuously with Scrutinizer

custom_search::prepareIndexes() B
last analyzed 2022-05-09 18:26 UTC

↳ Parent: custom_search

Complexity

Conditions	8
Paths	10

Size

Total Lines	21
Code Lines	12

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
cc	8
eloc	12
c	0
b	0
f	0
nop	4
dl	0
loc	21
rs	8.4444
nc	10

<?php

/**
 * Simple Machines Forum (SMF)
 *
 * @package SMF
 * @author Simple Machines https://www.simplemachines.org
 * @copyright 2022 Simple Machines and individual contributors
 * @license https://www.simplemachines.org/about/smf/license.php BSD
 *
 * @version 2.1.0
 */

if (!defined('SMF'))
	die('No direct access...');

/**
 * Used for the "custom search index" option
 * Class custom_search
 */
class custom_search extends search_api
{
	/**
	 * @var array Index settings
	 */
	protected $indexSettings = array();

	/**
	 * @var array An array of banned words
	 */
	protected $bannedWords = array();

	/**
	 * @var int|null Minimum word length (null for no minimum)
	 */
	protected $min_word_length = null;

	/**
	 * @var array Which databases support this method
	 */
	protected $supported_databases = array('mysql', 'postgresql');

	/**
	 * Constructor function
	 */
	public function __construct()
	{
		global $smcFunc, $modSettings, $db_type;

		// Is this database supported?
		if (!in_array($db_type, $this->supported_databases))
		{
			$this->is_supported = false;
			return;
		}

		if (empty($modSettings['search_custom_index_config']))
			return;

		$this->indexSettings = $smcFunc['json_decode']($modSettings['search_custom_index_config'], true);

		$this->bannedWords = empty($modSettings['search_stopwords']) ? array() : explode(',', $modSettings['search_stopwords']);
		$this->min_word_length = $this->indexSettings['bytes_per_word'];
	}

	/**
	 * {@inheritDoc}
	 */
	public function supportsMethod($methodName, $query_params = null)
	{
		$return = false;
		switch ($methodName)
		{
			case 'isValid':
			case 'searchSort':
			case 'prepareIndexes':
			case 'indexedWordQuery':
			case 'postCreated':
			case 'postModified':
				$return = true;
				break;

			// All other methods, too bad dunno you.
			default:
				$return = false;
		}

		// Maybe parent got support
		if (!$return)
			$return = parent::supportsMethod($methodName, $query_params);

		return $return;
	}

	/**
	 * {@inheritDoc}
	 */
	public function isValid()
	{
		global $modSettings;

		return !empty($modSettings['search_custom_index_config']);
	}

	/**
	 * {@inheritDoc}
	 */
	public function searchSort($a, $b)
	{
		global $excludedWords;

		$x = strlen($a) - (in_array($a, $excludedWords) ? 1000 : 0);
		$y = strlen($b) - (in_array($b, $excludedWords) ? 1000 : 0);

		return $y < $x ? 1 : ($y > $x ? -1 : 0);
	}

	/**
	 * {@inheritDoc}
	 */
	public function prepareIndexes($word, array &$wordsSearch, array &$wordsExclude, $isExcluded)
	{
		global $modSettings, $smcFunc;

		$subwords = text2words($word, $this->min_word_length, true);

		if (empty($modSettings['search_force_index']))
			$wordsSearch['words'][] = $word;

		// Excluded phrases don't benefit from being split into subwords.
		if (count($subwords) > 1 && $isExcluded)
			return;
		else
		{
			foreach ($subwords as $subword)
			{
				if ($smcFunc['strlen']($subword) >= $this->min_word_length && !in_array($subword, $this->bannedWords))
				{
					$wordsSearch['indexed_words'][] = $subword;
					if ($isExcluded)
						$wordsExclude[] = $subword;
				}
			}
		}
	}

	/**
	 * {@inheritDoc}
	 */
	public function indexedWordQuery(array $words, array $search_data)
	{
		global $modSettings, $smcFunc;

		$query_select = array(
			'id_msg' => 'm.id_msg',
		);
		$query_inner_join = array();
		$query_left_join = array();
		$query_where = array();
		$query_params = $search_data['params'];

		if ($query_params['id_search'])
			$query_select['id_search'] = '{int:id_search}';

		$count = 0;
		foreach ($words['words'] as $regularWord)
		{
			$query_where[] = 'm.body' . (in_array($regularWord, $query_params['excluded_words']) ? ' NOT' : '') . (empty($modSettings['search_match_words']) || $search_data['no_regexp'] ? ' LIKE ' : ' RLIKE ') . '{string:complex_body_' . $count . '}';
			$query_params['complex_body_' . $count++] = empty($modSettings['search_match_words']) || $search_data['no_regexp'] ? '%' . strtr($regularWord, array('_' => '\\_', '%' => '\\%')) . '%' : '[[:<:]]' . addcslashes(preg_replace(array('/([\[\]$.+*?|{}()])/'), array('[$1]'), $regularWord), '\\\'') . '[[:>:]]';
		}

		if ($query_params['user_query'])
			$query_where[] = '{raw:user_query}';
		if ($query_params['board_query'])
			$query_where[] = 'm.id_board {raw:board_query}';

		if ($query_params['topic'])
			$query_where[] = 'm.id_topic = {int:topic}';
		if ($query_params['min_msg_id'])
			$query_where[] = 'm.id_msg >= {int:min_msg_id}';
		if ($query_params['max_msg_id'])
			$query_where[] = 'm.id_msg <= {int:max_msg_id}';

		$count = 0;
		if (!empty($query_params['excluded_phrases']) && empty($modSettings['search_force_index']))
			foreach ($query_params['excluded_phrases'] as $phrase)
			{
				$query_where[] = 'subject NOT ' . (empty($modSettings['search_match_words']) || $search_data['no_regexp'] ? ' LIKE ' : ' RLIKE ') . '{string:exclude_subject_phrase_' . $count . '}';
				$query_params['exclude_subject_phrase_' . $count++] = empty($modSettings['search_match_words']) || $search_data['no_regexp'] ? '%' . strtr($phrase, array('_' => '\\_', '%' => '\\%')) . '%' : '[[:<:]]' . addcslashes(preg_replace(array('/([\[\]$.+*?|{}()])/'), array('[$1]'), $phrase), '\\\'') . '[[:>:]]';
			}
		$count = 0;
		if (!empty($query_params['excluded_subject_words']) && empty($modSettings['search_force_index']))
			foreach ($query_params['excluded_subject_words'] as $excludedWord)
			{
				$query_where[] = 'subject NOT ' . (empty($modSettings['search_match_words']) || $search_data['no_regexp'] ? ' LIKE ' : ' RLIKE ') . '{string:exclude_subject_words_' . $count . '}';
				$query_params['exclude_subject_words_' . $count++] = empty($modSettings['search_match_words']) || $search_data['no_regexp'] ? '%' . strtr($excludedWord, array('_' => '\\_', '%' => '\\%')) . '%' : '[[:<:]]' . addcslashes(preg_replace(array('/([\[\]$.+*?|{}()])/'), array('[$1]'), $excludedWord), '\\\'') . '[[:>:]]';
			}

		$numTables = 0;
		$prev_join = 0;
		foreach ($words['indexed_words'] as $indexedWord)
		{
			$numTables++;
			if (in_array($indexedWord, $query_params['excluded_index_words']))
			{
				$query_left_join[] = '{db_prefix}log_search_words AS lsw' . $numTables . ' ON (lsw' . $numTables . '.id_word = ' . $indexedWord . ' AND lsw' . $numTables . '.id_msg = m.id_msg)';
				$query_where[] = '(lsw' . $numTables . '.id_word IS NULL)';
			}
			else
			{
				$query_inner_join[] = '{db_prefix}log_search_words AS lsw' . $numTables . ' ON (lsw' . $numTables . '.id_msg = ' . ($prev_join === 0 ? 'm' : 'lsw' . $prev_join) . '.id_msg)';
				$query_where[] = 'lsw' . $numTables . '.id_word = ' . $indexedWord;
				$prev_join = $numTables;
			}
		}

		$ignoreRequest = $smcFunc['db_search_query']('insert_into_log_messages_fulltext', ($smcFunc['db_support_ignore'] ? ('
			INSERT IGNORE INTO {db_prefix}' . $search_data['insert_into'] . '
				(' . implode(', ', array_keys($query_select)) . ')') : '') . '
			SELECT ' . implode(', ', $query_select) . '
			FROM {db_prefix}messages AS m' . (empty($query_inner_join) ? '' : '
				INNER JOIN ' . implode('
				INNER JOIN ', $query_inner_join)) . (empty($query_left_join) ? '' : '
				LEFT JOIN ' . implode('
				LEFT JOIN ', $query_left_join)) . '
			WHERE ' . implode('
				AND ', $query_where) . (empty($search_data['max_results']) ? '' : '
			LIMIT ' . ($search_data['max_results'] - $search_data['indexed_results'])),
			$query_params
		);

		return $ignoreRequest;
	}

	/**
	 * {@inheritDoc}
	 */
	public function postCreated(array &$msgOptions, array &$topicOptions, array &$posterOptions)
	{
		global $modSettings, $smcFunc;

		$customIndexSettings = $smcFunc['json_decode']($modSettings['search_custom_index_config'], true);

		$inserts = array();
		foreach (text2words($msgOptions['body'], $customIndexSettings['bytes_per_word'], true) as $word)
			$inserts[] = array($word, $msgOptions['id']);

		if (!empty($inserts))
			$smcFunc['db_insert']('ignore',
				'{db_prefix}log_search_words',
				array('id_word' => 'int', 'id_msg' => 'int'),
				$inserts,
				array('id_word', 'id_msg')
			);
	}

	/**
	 * {@inheritDoc}
	 */
	public function postModified(array &$msgOptions, array &$topicOptions, array &$posterOptions)
	{
		global $modSettings, $smcFunc;

		if (isset($msgOptions['body']))
		{
			$customIndexSettings = $smcFunc['json_decode']($modSettings['search_custom_index_config'], true);
			$stopwords = empty($modSettings['search_stopwords']) ? array() : explode(',', $modSettings['search_stopwords']);
			$old_body = isset($msgOptions['old_body']) ? $msgOptions['old_body'] : '';

			// create thew new and old index
			$old_index = text2words($old_body, $customIndexSettings['bytes_per_word'], true);
			$new_index = text2words($msgOptions['body'], $customIndexSettings['bytes_per_word'], true);

			// Calculate the words to be added and removed from the index.
			$removed_words = array_diff(array_diff($old_index, $new_index), $stopwords);
			$inserted_words = array_diff(array_diff($new_index, $old_index), $stopwords);

			// Delete the removed words AND the added ones to avoid key constraints.
			if (!empty($removed_words))
			{
				$removed_words = array_merge($removed_words, $inserted_words);
				$smcFunc['db_query']('', '
					DELETE FROM {db_prefix}log_search_words
					WHERE id_msg = {int:id_msg}
						AND id_word IN ({array_int:removed_words})',
					array(
						'removed_words' => $removed_words,
						'id_msg' => $msgOptions['id'],
					)
				);
			}

			// Add the new words to be indexed.
			if (!empty($inserted_words))
			{
				$inserts = array();
				foreach ($inserted_words as $word)
					$inserts[] = array($word, $msgOptions['id']);
				$smcFunc['db_insert']('insert',
					'{db_prefix}log_search_words',
					array('id_word' => 'string', 'id_msg' => 'int'),
					$inserts,
					array('id_word', 'id_msg')
				);
			}
		}
	}
}

?>

1			<?php
2
3			/**
4			* Simple Machines Forum (SMF)
5			*
6			* @package SMF
7			* @author Simple Machines https://www.simplemachines.org
8			* @copyright 2022 Simple Machines and individual contributors
9			* @license https://www.simplemachines.org/about/smf/license.php BSD
10			*
11			* @version 2.1.0
12			*/
13
14			if (!defined('SMF'))
15			die('No direct access...');
16
17			/**
18			* Used for the "custom search index" option
19			* Class custom_search
20			*/
21			class custom_search extends search_api
22			{
23			/**
24			* @var array Index settings
25			*/
26			protected $indexSettings = array();
27
28			/**
29			* @var array An array of banned words
30			*/
31			protected $bannedWords = array();
32
33			/**
34			* @var int\|null Minimum word length (null for no minimum)
35			*/
36			protected $min_word_length = null;
37
38			/**
39			* @var array Which databases support this method
40			*/
41			protected $supported_databases = array('mysql', 'postgresql');
42
43			/**
44			* Constructor function
45			*/
46			public function __construct()
47			{
48			global $smcFunc, $modSettings, $db_type;
49
50			// Is this database supported?
51			if (!in_array($db_type, $this->supported_databases))
52			{
53			$this->is_supported = false;
54			return;
55			}
56
57			if (empty($modSettings['search_custom_index_config']))
58			return;
59
60			$this->indexSettings = $smcFunc['json_decode']($modSettings['search_custom_index_config'], true);
61
62			$this->bannedWords = empty($modSettings['search_stopwords']) ? array() : explode(',', $modSettings['search_stopwords']);
63			$this->min_word_length = $this->indexSettings['bytes_per_word'];
64			}
65
66			/**
67			* {@inheritDoc}
68			*/
69			public function supportsMethod($methodName, $query_params = null)
70			{
71			$return = false;
72			switch ($methodName)
73			{
74			case 'isValid':
75			case 'searchSort':
76			case 'prepareIndexes':
77			case 'indexedWordQuery':
78			case 'postCreated':
79			case 'postModified':
80			$return = true;
81			break;
82
83			// All other methods, too bad dunno you.
84			default:
85			$return = false;
86			}
87
88			// Maybe parent got support
89			if (!$return)
90			$return = parent::supportsMethod($methodName, $query_params);
91
92			return $return;
93			}
94
95			/**
96			* {@inheritDoc}
97			*/
98			public function isValid()
99			{
100			global $modSettings;
101
102			return !empty($modSettings['search_custom_index_config']);
103			}
104
105			/**
106			* {@inheritDoc}
107			*/
108			public function searchSort($a, $b)
109			{
110			global $excludedWords;
111
112			$x = strlen($a) - (in_array($a, $excludedWords) ? 1000 : 0);
113			$y = strlen($b) - (in_array($b, $excludedWords) ? 1000 : 0);
114
115			return $y < $x ? 1 : ($y > $x ? -1 : 0);
116			}
117
118			/**
119			* {@inheritDoc}
120			*/
121			public function prepareIndexes($word, array &$wordsSearch, array &$wordsExclude, $isExcluded)
122			{
123			global $modSettings, $smcFunc;
124
125			$subwords = text2words($word, $this->min_word_length, true);
126
127			if (empty($modSettings['search_force_index']))
128			$wordsSearch['words'][] = $word;
129
130			// Excluded phrases don't benefit from being split into subwords.
131			if (count($subwords) > 1 && $isExcluded)
132			return;
133			else
134			{
135			foreach ($subwords as $subword)
136			{
137			if ($smcFunc['strlen']($subword) >= $this->min_word_length && !in_array($subword, $this->bannedWords))
138			{
139			$wordsSearch['indexed_words'][] = $subword;
140			if ($isExcluded)
141			$wordsExclude[] = $subword;
142			}
143			}
144			}
145			}
146
147			/**
148			* {@inheritDoc}
149			*/
150			public function indexedWordQuery(array $words, array $search_data)
151			{
152			global $modSettings, $smcFunc;
153
154			$query_select = array(
155			'id_msg' => 'm.id_msg',
156			);
157			$query_inner_join = array();
158			$query_left_join = array();
159			$query_where = array();
160			$query_params = $search_data['params'];
161
162			if ($query_params['id_search'])
163			$query_select['id_search'] = '{int:id_search}';
164
165			$count = 0;
166			foreach ($words['words'] as $regularWord)
167			{
168			$query_where[] = 'm.body' . (in_array($regularWord, $query_params['excluded_words']) ? ' NOT' : '') . (empty($modSettings['search_match_words']) \|\| $search_data['no_regexp'] ? ' LIKE ' : ' RLIKE ') . '{string:complex_body_' . $count . '}';
169			$query_params['complex_body_' . $count++] = empty($modSettings['search_match_words']) \|\| $search_data['no_regexp'] ? '%' . strtr($regularWord, array('_' => '\\_', '%' => '\\%')) . '%' : '[[:<:]]' . addcslashes(preg_replace(array('/([\[\]$.+*?\|{}()])/'), array('[$1]'), $regularWord), '\\\'') . '[[:>:]]';
170			}
171
172			if ($query_params['user_query'])
173			$query_where[] = '{raw:user_query}';
174			if ($query_params['board_query'])
175			$query_where[] = 'm.id_board {raw:board_query}';
176
177			if ($query_params['topic'])
178			$query_where[] = 'm.id_topic = {int:topic}';
179			if ($query_params['min_msg_id'])
180			$query_where[] = 'm.id_msg >= {int:min_msg_id}';
181			if ($query_params['max_msg_id'])
182			$query_where[] = 'm.id_msg <= {int:max_msg_id}';
183
184			$count = 0;
185			if (!empty($query_params['excluded_phrases']) && empty($modSettings['search_force_index']))
186			foreach ($query_params['excluded_phrases'] as $phrase)
187			{
188			$query_where[] = 'subject NOT ' . (empty($modSettings['search_match_words']) \|\| $search_data['no_regexp'] ? ' LIKE ' : ' RLIKE ') . '{string:exclude_subject_phrase_' . $count . '}';
189			$query_params['exclude_subject_phrase_' . $count++] = empty($modSettings['search_match_words']) \|\| $search_data['no_regexp'] ? '%' . strtr($phrase, array('_' => '\\_', '%' => '\\%')) . '%' : '[[:<:]]' . addcslashes(preg_replace(array('/([\[\]$.+*?\|{}()])/'), array('[$1]'), $phrase), '\\\'') . '[[:>:]]';
190			}
191			$count = 0;
192			if (!empty($query_params['excluded_subject_words']) && empty($modSettings['search_force_index']))
193			foreach ($query_params['excluded_subject_words'] as $excludedWord)
194			{
195			$query_where[] = 'subject NOT ' . (empty($modSettings['search_match_words']) \|\| $search_data['no_regexp'] ? ' LIKE ' : ' RLIKE ') . '{string:exclude_subject_words_' . $count . '}';
196			$query_params['exclude_subject_words_' . $count++] = empty($modSettings['search_match_words']) \|\| $search_data['no_regexp'] ? '%' . strtr($excludedWord, array('_' => '\\_', '%' => '\\%')) . '%' : '[[:<:]]' . addcslashes(preg_replace(array('/([\[\]$.+*?\|{}()])/'), array('[$1]'), $excludedWord), '\\\'') . '[[:>:]]';
197			}
198
199			$numTables = 0;
200			$prev_join = 0;
201			foreach ($words['indexed_words'] as $indexedWord)
202			{
203			$numTables++;
204			if (in_array($indexedWord, $query_params['excluded_index_words']))
205			{
206			$query_left_join[] = '{db_prefix}log_search_words AS lsw' . $numTables . ' ON (lsw' . $numTables . '.id_word = ' . $indexedWord . ' AND lsw' . $numTables . '.id_msg = m.id_msg)';
207			$query_where[] = '(lsw' . $numTables . '.id_word IS NULL)';
208			}
209			else
210			{
211			$query_inner_join[] = '{db_prefix}log_search_words AS lsw' . $numTables . ' ON (lsw' . $numTables . '.id_msg = ' . ($prev_join === 0 ? 'm' : 'lsw' . $prev_join) . '.id_msg)';
212			$query_where[] = 'lsw' . $numTables . '.id_word = ' . $indexedWord;
213			$prev_join = $numTables;
214			}
215			}
216
217			$ignoreRequest = $smcFunc['db_search_query']('insert_into_log_messages_fulltext', ($smcFunc['db_support_ignore'] ? ('
218			INSERT IGNORE INTO {db_prefix}' . $search_data['insert_into'] . '
219			(' . implode(', ', array_keys($query_select)) . ')') : '') . '
220			SELECT ' . implode(', ', $query_select) . '
221			FROM {db_prefix}messages AS m' . (empty($query_inner_join) ? '' : '
222			INNER JOIN ' . implode('
223			INNER JOIN ', $query_inner_join)) . (empty($query_left_join) ? '' : '
224			LEFT JOIN ' . implode('
225			LEFT JOIN ', $query_left_join)) . '
226			WHERE ' . implode('
227			AND ', $query_where) . (empty($search_data['max_results']) ? '' : '
228			LIMIT ' . ($search_data['max_results'] - $search_data['indexed_results'])),
229			$query_params
230			);
231
232			return $ignoreRequest;
233			}
234
235			/**
236			* {@inheritDoc}
237			*/
238			public function postCreated(array &$msgOptions, array &$topicOptions, array &$posterOptions)
239			{
240			global $modSettings, $smcFunc;
241
242			$customIndexSettings = $smcFunc['json_decode']($modSettings['search_custom_index_config'], true);
243
244			$inserts = array();
245			foreach (text2words($msgOptions['body'], $customIndexSettings['bytes_per_word'], true) as $word)
246			$inserts[] = array($word, $msgOptions['id']);
247
248			if (!empty($inserts))
249			$smcFunc['db_insert']('ignore',
250			'{db_prefix}log_search_words',
251			array('id_word' => 'int', 'id_msg' => 'int'),
252			$inserts,
253			array('id_word', 'id_msg')
254			);
255			}
256
257			/**
258			* {@inheritDoc}
259			*/
260			public function postModified(array &$msgOptions, array &$topicOptions, array &$posterOptions)
261			{
262			global $modSettings, $smcFunc;
263
264			if (isset($msgOptions['body']))
265			{
266			$customIndexSettings = $smcFunc['json_decode']($modSettings['search_custom_index_config'], true);
267			$stopwords = empty($modSettings['search_stopwords']) ? array() : explode(',', $modSettings['search_stopwords']);
268			$old_body = isset($msgOptions['old_body']) ? $msgOptions['old_body'] : '';
269
270			// create thew new and old index
271			$old_index = text2words($old_body, $customIndexSettings['bytes_per_word'], true);
272			$new_index = text2words($msgOptions['body'], $customIndexSettings['bytes_per_word'], true);
273
274			// Calculate the words to be added and removed from the index.
275			$removed_words = array_diff(array_diff($old_index, $new_index), $stopwords);
276			$inserted_words = array_diff(array_diff($new_index, $old_index), $stopwords);
277
278			// Delete the removed words AND the added ones to avoid key constraints.
279			if (!empty($removed_words))
280			{
281			$removed_words = array_merge($removed_words, $inserted_words);
282			$smcFunc['db_query']('', '
283			DELETE FROM {db_prefix}log_search_words
284			WHERE id_msg = {int:id_msg}
285			AND id_word IN ({array_int:removed_words})',
286			array(
287			'removed_words' => $removed_words,
288			'id_msg' => $msgOptions['id'],
289			)
290			);
291			}
292
293			// Add the new words to be indexed.
294			if (!empty($inserted_words))
295			{
296			$inserts = array();
297			foreach ($inserted_words as $word)
298			$inserts[] = array($word, $msgOptions['id']);
299			$smcFunc['db_insert']('insert',
300			'{db_prefix}log_search_words',
301			array('id_word' => 'string', 'id_msg' => 'int'),
302			$inserts,
303			array('id_word', 'id_msg')
304			);
305			}
306			}
307			}
308			}
309
310			?>

SimpleMachines / SMF2.1

custom_search::prepareIndexes() B last analyzed 2022-05-09 18:26 UTC

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like

custom_search::prepareIndexes() B
last analyzed 2022-05-09 18:26 UTC