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

custom_search::prepareIndexes() B
last analyzed 2025-07-20 06:27 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 2025 Simple Machines and individual contributors
 * @license https://www.simplemachines.org/about/smf/license.php BSD
 *
 * @version 2.1.5
 */

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;

		// Specify the function to search with. Regex is for word boundaries.
		$is_search_regex = !empty($modSettings['search_match_words']) && !$search_data['no_regexp'];
		$query_match_type = $is_search_regex ? 'RLIKE' : 'LIKE';
		$word_boundary_wrapper = function(string $str) use ($smcFunc): string
		{
			return sprintf($smcFunc['db_supports_pcre'] ? '\\b%s\\b' : '[[:<:]]%s[[:>:]]', $str);
		};
		$escape_sql_regex = function(string $str): string
		{
			return addcslashes(preg_replace('/[\[\]$.+*?&^|{}()]/', '[$0]', $str), '\\\'');
		};

		$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 $regularWord)
		{
			if (in_array($regularWord, $query_params['excluded_words']))
				$query_where[] = 'm.body NOT ' . $query_match_type . ' {string:complex_body_' . $count . '}';
			else
				$query_where[] = 'm.body ' . $query_match_type . ' {string:complex_body_' . $count . '}';

			if ($is_search_regex)
				$query_params['complex_body_' . $count++] = $word_boundary_wrapper($escape_sql_regex($regularWord));
			else
				$query_params['complex_body_' . $count++] = '%' . $smcFunc['db_escape_wildcard_string']($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 ' . $query_match_type . ' {string:exclude_subject_words_' . $count . '}';

				if ($is_search_regex)
					$query_params['exclude_subject_words_' . $count++] = $word_boundary_wrapper($escape_sql_regex($phrase));
				else
					$query_params['exclude_subject_words_' . $count++] = '%' . $smcFunc['db_escape_wildcard_string']($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 ' . $query_match_type . ' {string:exclude_subject_words_' . $count . '}';

				if ($is_search_regex)
					$query_params['exclude_subject_words_' . $count++] = $word_boundary_wrapper($escape_sql_regex($excludedWord));
				else
					$query_params['exclude_subject_words_' . $count++] = '%' . $smcFunc['db_escape_wildcard_string']($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 2025 Simple Machines and individual contributors
9			* @license https://www.simplemachines.org/about/smf/license.php BSD
10			*
11			* @version 2.1.5
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			// Specify the function to search with. Regex is for word boundaries.
155			$is_search_regex = !empty($modSettings['search_match_words']) && !$search_data['no_regexp'];
156			$query_match_type = $is_search_regex ? 'RLIKE' : 'LIKE';
157			$word_boundary_wrapper = function(string $str) use ($smcFunc): string
158			{
159			return sprintf($smcFunc['db_supports_pcre'] ? '\\b%s\\b' : '[[:<:]]%s[[:>:]]', $str);
160			};
161			$escape_sql_regex = function(string $str): string
162			{
163			return addcslashes(preg_replace('/[\[\]$.+*?&^\|{}()]/', '[$0]', $str), '\\\'');
164			};
165
166			$query_select = array(
167			'id_msg' => 'm.id_msg',
168			);
169			$query_inner_join = array();
170			$query_left_join = array();
171			$query_where = array();
172			$query_params = $search_data['params'];
173
174			if ($query_params['id_search'])
175			$query_select['id_search'] = '{int:id_search}';
176
177			$count = 0;
178			foreach ($words['words'] as $regularWord)
179			{
180			if (in_array($regularWord, $query_params['excluded_words']))
181			$query_where[] = 'm.body NOT ' . $query_match_type . ' {string:complex_body_' . $count . '}';
182			else
183			$query_where[] = 'm.body ' . $query_match_type . ' {string:complex_body_' . $count . '}';
184
185			if ($is_search_regex)
186			$query_params['complex_body_' . $count++] = $word_boundary_wrapper($escape_sql_regex($regularWord));
187			else
188			$query_params['complex_body_' . $count++] = '%' . $smcFunc['db_escape_wildcard_string']($regularWord) . '%';
189			}
190
191			if ($query_params['user_query'])
192			$query_where[] = '{raw:user_query}';
193			if ($query_params['board_query'])
194			$query_where[] = 'm.id_board {raw:board_query}';
195
196			if ($query_params['topic'])
197			$query_where[] = 'm.id_topic = {int:topic}';
198			if ($query_params['min_msg_id'])
199			$query_where[] = 'm.id_msg >= {int:min_msg_id}';
200			if ($query_params['max_msg_id'])
201			$query_where[] = 'm.id_msg <= {int:max_msg_id}';
202
203			$count = 0;
204			if (!empty($query_params['excluded_phrases']) && empty($modSettings['search_force_index']))
205			foreach ($query_params['excluded_phrases'] as $phrase)
206			{
207			$query_where[] = 'subject NOT ' . $query_match_type . ' {string:exclude_subject_words_' . $count . '}';
208
209			if ($is_search_regex)
210			$query_params['exclude_subject_words_' . $count++] = $word_boundary_wrapper($escape_sql_regex($phrase));
211			else
212			$query_params['exclude_subject_words_' . $count++] = '%' . $smcFunc['db_escape_wildcard_string']($phrase) . '%';
213			}
214			$count = 0;
215			if (!empty($query_params['excluded_subject_words']) && empty($modSettings['search_force_index']))
216			foreach ($query_params['excluded_subject_words'] as $excludedWord)
217			{
218			$query_where[] = 'subject NOT ' . $query_match_type . ' {string:exclude_subject_words_' . $count . '}';
219
220			if ($is_search_regex)
221			$query_params['exclude_subject_words_' . $count++] = $word_boundary_wrapper($escape_sql_regex($excludedWord));
222			else
223			$query_params['exclude_subject_words_' . $count++] = '%' . $smcFunc['db_escape_wildcard_string']($excludedWord) . '%';
224			}
225
226			$numTables = 0;
227			$prev_join = 0;
228			foreach ($words['indexed_words'] as $indexedWord)
229			{
230			$numTables++;
231			if (in_array($indexedWord, $query_params['excluded_index_words']))
232			{
233			$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)';
234			$query_where[] = '(lsw' . $numTables . '.id_word IS NULL)';
235			}
236			else
237			{
238			$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)';
239			$query_where[] = 'lsw' . $numTables . '.id_word = ' . $indexedWord;
240			$prev_join = $numTables;
241			}
242			}
243
244			$ignoreRequest = $smcFunc['db_search_query']('insert_into_log_messages_fulltext', ($smcFunc['db_support_ignore'] ? ('
245			INSERT IGNORE INTO {db_prefix}' . $search_data['insert_into'] . '
246			(' . implode(', ', array_keys($query_select)) . ')') : '') . '
247			SELECT ' . implode(', ', $query_select) . '
248			FROM {db_prefix}messages AS m' . (empty($query_inner_join) ? '' : '
249			INNER JOIN ' . implode('
250			INNER JOIN ', $query_inner_join)) . (empty($query_left_join) ? '' : '
251			LEFT JOIN ' . implode('
252			LEFT JOIN ', $query_left_join)) . '
253			WHERE ' . implode('
254			AND ', $query_where) . (empty($search_data['max_results']) ? '' : '
255			LIMIT ' . ($search_data['max_results'] - $search_data['indexed_results'])),
256			$query_params
257			);
258
259			return $ignoreRequest;
260			}
261
262			/**
263			* {@inheritDoc}
264			*/
265			public function postCreated(array &$msgOptions, array &$topicOptions, array &$posterOptions)
266			{
267			global $modSettings, $smcFunc;
268
269			$customIndexSettings = $smcFunc['json_decode']($modSettings['search_custom_index_config'], true);
270
271			$inserts = array();
272			foreach (text2words($msgOptions['body'], $customIndexSettings['bytes_per_word'], true) as $word)
273			$inserts[] = array($word, $msgOptions['id']);
274
275			if (!empty($inserts))
276			$smcFunc['db_insert']('ignore',
277			'{db_prefix}log_search_words',
278			array('id_word' => 'int', 'id_msg' => 'int'),
279			$inserts,
280			array('id_word', 'id_msg')
281			);
282			}
283
284			/**
285			* {@inheritDoc}
286			*/
287			public function postModified(array &$msgOptions, array &$topicOptions, array &$posterOptions)
288			{
289			global $modSettings, $smcFunc;
290
291			if (isset($msgOptions['body']))
292			{
293			$customIndexSettings = $smcFunc['json_decode']($modSettings['search_custom_index_config'], true);
294			$stopwords = empty($modSettings['search_stopwords']) ? array() : explode(',', $modSettings['search_stopwords']);
295			$old_body = isset($msgOptions['old_body']) ? $msgOptions['old_body'] : '';
296
297			// create thew new and old index
298			$old_index = text2words($old_body, $customIndexSettings['bytes_per_word'], true);
299			$new_index = text2words($msgOptions['body'], $customIndexSettings['bytes_per_word'], true);
300
301			// Calculate the words to be added and removed from the index.
302			$removed_words = array_diff(array_diff($old_index, $new_index), $stopwords);
303			$inserted_words = array_diff(array_diff($new_index, $old_index), $stopwords);
304
305			// Delete the removed words AND the added ones to avoid key constraints.
306			if (!empty($removed_words))
307			{
308			$removed_words = array_merge($removed_words, $inserted_words);
309			$smcFunc['db_query']('', '
310			DELETE FROM {db_prefix}log_search_words
311			WHERE id_msg = {int:id_msg}
312			AND id_word IN ({array_int:removed_words})',
313			array(
314			'removed_words' => $removed_words,
315			'id_msg' => $msgOptions['id'],
316			)
317			);
318			}
319
320			// Add the new words to be indexed.
321			if (!empty($inserted_words))
322			{
323			$inserts = array();
324			foreach ($inserted_words as $word)
325			$inserts[] = array($word, $msgOptions['id']);
326			$smcFunc['db_insert']('insert',
327			'{db_prefix}log_search_words',
328			array('id_word' => 'string', 'id_msg' => 'int'),
329			$inserts,
330			array('id_word', 'id_msg')
331			);
332			}
333			}
334			}
335			}
336
337			?>

albertlast / SMF2.1

custom_search::prepareIndexes() B last analyzed 2025-07-20 06:27 UTC

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like

custom_search::prepareIndexes() B
last analyzed 2025-07-20 06:27 UTC