xapian_get_all_terms() - Code Metrics - Inspection of "Merge pull request #7159 from christianbeeznest/GH..." - chamilo/chamilo-lms - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — master ( ab6f49...c762ff )

unknown

created 2025-12-12 19:30 UTC

xapian_get_all_terms() A

↳ Parent: Project

Complexity

Conditions	6
Paths	17

Size

Total Lines	26
Code Lines	16

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
cc	6
eloc	16
c	0
b	0
f	0
nc	17
nop	3
dl	0
loc	26
rs	9.1111

<?php

/* For licensing terms, see /license.txt */

require_once 'xapian.php';
// TODO: think another way without including specific fields here
require_once api_get_path(LIBRARY_PATH) . 'specific_fields_manager.lib.php';

use Chamilo\CoreBundle\Framework\Container;

/**
 * Legacy default index path (Chamilo 1).
 *
 * In Chamilo 2, we will prefer the Symfony SearchIndexPathResolver
 * but we keep this constant as a fallback for older installs.
 */
define('XAPIAN_DB', api_get_path(SYS_UPLOAD_PATH) . 'plugins/xapian/searchdb/');


/**
 * Returns a XapianDatabase instance using the configured index directory.
 *
 * In Chamilo 2, this will prefer the Symfony SearchIndexPathResolver
 * (var/search or whatever is configured) and fall back to the legacy
 * upload path when needed.
 *
 * @param XapianDatabase|null $db Existing database instance (optional)
 *
 * @return XapianDatabase
 *
 * @throws Exception If the database cannot be opened
 */
function xapian_get_database($db = null)
{
    if ($db instanceof XapianDatabase) {
        return $db;
    }

    // Default: legacy path (Chamilo 1 behavior)
    $path = XAPIAN_DB;

    // If Chamilo 2 container is available, try to use the new index dir
    if (class_exists(Container::class)) {
        try {
            /** @var \Chamilo\CoreBundle\Search\Xapian\SearchIndexPathResolver $resolver */
            $resolver = Container::getSearchIndexPathResolver();
            $resolver->ensureIndexDirectoryExists();
            $path = $resolver->getIndexDir();
        } catch (\Throwable $e) {
            // Fallback to legacy path if resolver or container are not available.
            // This keeps backward compatibility and avoids hard failures.
        }
    }

    return new XapianDatabase($path);
}

/**
 * Queries the database.
 * The xapian_query function queries the database using both a query string
 * and application-defined terms. Based on drupal-xapian.
 *
 * @param string              $query_string The search string. This string will
 *                                          be parsed and stemmed automatically.
 * @param XapianDatabase|null $db           Xapian database to connect
 * @param int                 $start        An integer defining the first
 *                                          document to return
 * @param int                 $length       The number of results to return
 * @param array               $extra        An array containing arrays of
 *                                          extra terms to search for
 * @param int                 $count_type   How to compute the match count:
 *                                          0 = best estimate,
 *                                          1 = lower bound,
 *                                          2 = upper bound
 *
 * @return array|null [int $count, array $results] or null on error
 */
function xapian_query($query_string, $db = null, $start = 0, $length = 10, $extra = [], $count_type = 0)
{
    try {
        $db = xapian_get_database($db);

        // Build subqueries from $extra array. Now only used by tags search filter on search widget.
        $subqueries = [];
        foreach ($extra as $subquery) {
            if (!empty($subquery)) {
                $subqueries[] = new XapianQuery($subquery);
            }
        }

        $query = null;
        $enquire = new XapianEnquire($db);

        if (!empty($query_string)) {
            $query_parser = new XapianQueryParser();
            // TODO: choose stemmer based on platform/user language if needed
            $stemmer = new XapianStem('english');
            $query_parser->set_stemmer($stemmer);
            $query_parser->set_database($db);
            $query_parser->set_stemming_strategy(XapianQueryParser::STEM_SOME);
            $query_parser->add_boolean_prefix('courseid', XAPIAN_PREFIX_COURSEID);
            $query_parser->add_boolean_prefix('toolid', XAPIAN_PREFIX_TOOLID);

            $parsedQuery = $query_parser->parse_query($query_string);
            $final_array = array_merge($subqueries, [$parsedQuery]);
            $query = new XapianQuery(XapianQuery::OP_AND, $final_array);
        } else {
            // No free-text query: OR all subqueries (e.g. tag-only search)
            $query = new XapianQuery(XapianQuery::OP_OR, $subqueries);
        }

        $enquire->set_query($query);

        $matches = $enquire->get_mset((int) $start, (int) $length);

        $specific_fields = get_specific_field_list();

        $results = [];
        $i = $matches->begin();

        $count = 0;

        while (!$i->equals($matches->end())) {
            $count++;
            $document = $i->get_document();

            if (is_object($document)) {
                // Process one item terms (course id, tool id)
                $courseid_terms = xapian_get_doc_terms($document, XAPIAN_PREFIX_COURSEID);
                $results[$count]['courseid'] = substr($courseid_terms[0]['name'], 1);

                $toolid_terms = xapian_get_doc_terms($document, XAPIAN_PREFIX_TOOLID);
                $results[$count]['toolid'] = substr($toolid_terms[0]['name'], 1);

                // Process each specific field prefix
                foreach ($specific_fields as $specific_field) {
                    $results[$count]['sf-' . $specific_field['code']] = xapian_get_doc_terms(
                        $document,
                        $specific_field['code']
                    );
                }

                // Rest of data
                $results[$count]['xapian_data'] = unserialize($document->get_data());
                $results[$count]['score'] = $i->get_percent();
            }

            $i->next();
        }

        // Compute match count according to requested type
        switch ($count_type) {
            case 1: // Lower bound
                $count = $matches->get_matches_lower_bound();
                break;

            case 2: // Upper bound
                $count = $matches->get_matches_upper_bound();
                break;

            case 0: // Best estimate
            default:
                $count = $matches->get_matches_estimated();
                break;
        }

        return [$count, $results];
    } catch (Exception $e) {
        display_xapian_error($e->getMessage());

        return null;
    }
}

/**
 * Build a boolean query.
 *
 * @param string $term The term string
 *
 * @return XapianQuery
 */
function xapian_get_boolean_query($term)
{
    return new XapianQuery($term);
}

/**
 * Retrieve a list of database terms.
 *
 * @param int                 $count  Number of terms to retrieve (0 means "no limit")
 * @param string              $prefix The prefix of the term to retrieve
 * @param XapianDatabase|null $db     Xapian database to connect
 *
 * @return array|null
 */
function xapian_get_all_terms($count = 0, $prefix, $db = null)
{
    try {
        $db = xapian_get_database($db);

        if (!empty($prefix)) {
            $termi = $db->allterms_begin($prefix);
        } else {
            $termi = $db->allterms_begin();
        }

        $terms = [];
        $i = 0;

        for (; !$termi->equals($db->allterms_end()) && (++$i <= $count || $count === 0); $termi->next()) {
            $terms[] = [
                'frequency' => $termi->get_termfreq(),
                'name' => $termi->get_term(),
            ];
        }

        return $terms;
    } catch (Exception $e) {
        display_xapian_error($e->getMessage());

        return null;
    }
}

/**
 * Retrieve all terms of a document filtered by prefix.
 *
 * @param XapianDocument|null $doc    Document to inspect
 * @param string              $prefix Prefix used to filter the terms
 *
 * @return array|null
 */
function xapian_get_doc_terms($doc = null, $prefix)
{
    try {
        if (!is_a($doc, 'XapianDocument')) {
            return null;
        }

        // TODO: make the filter by prefix on xapian if possible
        // ojwb marvil07: use Document::termlist_begin() and then skip_to(prefix) on the TermIterator
        // ojwb you'll need to check the end condition by hand though
        $terms = [];

        for ($termi = $doc->termlist_begin(); !$termi->equals($doc->termlist_end()); $termi->next()) {
            $term = [
                'frequency' => $termi->get_termfreq(),
                'name' => $termi->get_term(),
            ];

            if ($term['name'][0] === $prefix) {
                $terms[] = $term;
            }
        }

        return $terms;
    } catch (Exception $e) {
        display_xapian_error($e->getMessage());

        return null;
    }
}

/**
 * Join Xapian queries.
 *
 * @param XapianQuery|array      $query1 First query or array of queries
 * @param XapianQuery|array|null $query2 Second query or array of queries (optional)
 * @param string                 $op     Logical operator: 'or' or 'and'
 *
 * @return XapianQuery
 */
function xapian_join_queries($query1, $query2 = null, $op = 'or')
{
    // Decide how to join, avoiding including xapian.php outside
    switch ($op) {
        case 'and':
            $op = XapianQuery::OP_AND;
            break;
        case 'or':
        default:
            $op = XapianQuery::OP_OR;
            break;
    }

    // Normalize parameters to arrays
    if (!is_array($query1)) {
        $query1 = [$query1];
    }

    if ($query2 === null) {
        // Join an array of queries with $op
        return new XapianQuery($op, $query1);
    }

    if (!is_array($query2)) {
        $query2 = [$query2];
    }

    return new XapianQuery($op, array_merge($query1, $query2));
}

/**
 * Maps Xapian errors to human-readable messages.
 *
 * @author Isaac flores paz <[email protected]>
 *
 * @param string $xapian_error_message The Xapian error message
 *
 * @return void
 */
function display_xapian_error($xapian_error_message)
{
    $message = explode(':', $xapian_error_message);
    $type_error_message = $message[0];

    if ($type_error_message === 'DatabaseOpeningError') {
        $message_error = get_lang('Failed to open the search database');
    } elseif ($type_error_message === 'DatabaseVersionError') {
        $message_error = get_lang('The search database uses an unsupported format');
    } elseif ($type_error_message === 'DatabaseModifiedError') {
        $message_error = get_lang('The search database has been modified/broken');
    } elseif ($type_error_message === 'DatabaseLockError') {
        $message_error = get_lang('Failed to lock the search database');
    } elseif ($type_error_message === 'DatabaseCreateError') {
        $message_error = get_lang('Failed to create the search database');
    } elseif ($type_error_message === 'DatabaseCorruptError') {
        $message_error = get_lang('The search database has suffered corruption');
    } elseif ($type_error_message === 'NetworkTimeoutError') {
        $message_error = get_lang('Connection timed out while communicating with the remote search database');
    } else {
        $message_error = get_lang('Error in search engine');
    }

    $display_message = get_lang('Error') . ' : ' . $message_error;
    echo Display::return_message($display_message, 'error');
}


1			<?php
2
3			/* For licensing terms, see /license.txt */
4
5			require_once 'xapian.php';
6			// TODO: think another way without including specific fields here
7			require_once api_get_path(LIBRARY_PATH) . 'specific_fields_manager.lib.php';
8
9			use Chamilo\CoreBundle\Framework\Container;
10
11			/**
12			* Legacy default index path (Chamilo 1).
13			*
14			* In Chamilo 2, we will prefer the Symfony SearchIndexPathResolver
15			* but we keep this constant as a fallback for older installs.
16			*/
17			define('XAPIAN_DB', api_get_path(SYS_UPLOAD_PATH) . 'plugins/xapian/searchdb/');
			0 ignored issues – show Bug introduced 2021-02-07 16:08 UTC by Report Bug Copy Issue Report The constant `SYS_UPLOAD_PATH` was not found. Maybe you did not declare it correctly or list all dependencies? Loading history...
18
19			/**
20			* Returns a XapianDatabase instance using the configured index directory.
21			*
22			* In Chamilo 2, this will prefer the Symfony SearchIndexPathResolver
23			* (var/search or whatever is configured) and fall back to the legacy
24			* upload path when needed.
25			*
26			* @param XapianDatabase\|null $db Existing database instance (optional)
27			*
28			* @return XapianDatabase
29			*
30			* @throws Exception If the database cannot be opened
31			*/
32			function xapian_get_database($db = null)
33			{
34			if ($db instanceof XapianDatabase) {
35			return $db;
36			}
37
38			// Default: legacy path (Chamilo 1 behavior)
39			$path = XAPIAN_DB;
40
41			// If Chamilo 2 container is available, try to use the new index dir
42			if (class_exists(Container::class)) {
43			try {
44			/** @var \Chamilo\CoreBundle\Search\Xapian\SearchIndexPathResolver $resolver */
45			$resolver = Container::getSearchIndexPathResolver();
46			$resolver->ensureIndexDirectoryExists();
47			$path = $resolver->getIndexDir();
48			} catch (\Throwable $e) {
49			// Fallback to legacy path if resolver or container are not available.
50			// This keeps backward compatibility and avoids hard failures.
51			}
52			}
53
54			return new XapianDatabase($path);
55			}
56
57			/**
58			* Queries the database.
59			* The xapian_query function queries the database using both a query string
60			* and application-defined terms. Based on drupal-xapian.
61			*
62			* @param string $query_string The search string. This string will
63			* be parsed and stemmed automatically.
64			* @param XapianDatabase\|null $db Xapian database to connect
65			* @param int $start An integer defining the first
66			* document to return
67			* @param int $length The number of results to return
68			* @param array $extra An array containing arrays of
69			* extra terms to search for
70			* @param int $count_type How to compute the match count:
71			* 0 = best estimate,
72			* 1 = lower bound,
73			* 2 = upper bound
74			*
75			* @return array\|null [int $count, array $results] or null on error
76			*/
77			function xapian_query($query_string, $db = null, $start = 0, $length = 10, $extra = [], $count_type = 0)
78			{
79			try {
80			$db = xapian_get_database($db);
81
82			// Build subqueries from $extra array. Now only used by tags search filter on search widget.
83			$subqueries = [];
84			foreach ($extra as $subquery) {
85			if (!empty($subquery)) {
86			$subqueries[] = new XapianQuery($subquery);
87			}
88			}
89
90			$query = null;
91			$enquire = new XapianEnquire($db);
92
93			if (!empty($query_string)) {
94			$query_parser = new XapianQueryParser();
95			// TODO: choose stemmer based on platform/user language if needed
96			$stemmer = new XapianStem('english');
97			$query_parser->set_stemmer($stemmer);
98			$query_parser->set_database($db);
99			$query_parser->set_stemming_strategy(XapianQueryParser::STEM_SOME);
100			$query_parser->add_boolean_prefix('courseid', XAPIAN_PREFIX_COURSEID);
101			$query_parser->add_boolean_prefix('toolid', XAPIAN_PREFIX_TOOLID);
102
103			$parsedQuery = $query_parser->parse_query($query_string);
104			$final_array = array_merge($subqueries, [$parsedQuery]);
105			$query = new XapianQuery(XapianQuery::OP_AND, $final_array);
106			} else {
107			// No free-text query: OR all subqueries (e.g. tag-only search)
108			$query = new XapianQuery(XapianQuery::OP_OR, $subqueries);
109			}
110
111			$enquire->set_query($query);
112
113			$matches = $enquire->get_mset((int) $start, (int) $length);
114
115			$specific_fields = get_specific_field_list();
116
117			$results = [];
118			$i = $matches->begin();
119
120			$count = 0;
121
122			while (!$i->equals($matches->end())) {
123			$count++;
124			$document = $i->get_document();
125
126			if (is_object($document)) {
127			// Process one item terms (course id, tool id)
128			$courseid_terms = xapian_get_doc_terms($document, XAPIAN_PREFIX_COURSEID);
129			$results[$count]['courseid'] = substr($courseid_terms[0]['name'], 1);
130
131			$toolid_terms = xapian_get_doc_terms($document, XAPIAN_PREFIX_TOOLID);
132			$results[$count]['toolid'] = substr($toolid_terms[0]['name'], 1);
133
134			// Process each specific field prefix
135			foreach ($specific_fields as $specific_field) {
136			$results[$count]['sf-' . $specific_field['code']] = xapian_get_doc_terms(
137			$document,
138			$specific_field['code']
139			);
140			}
141
142			// Rest of data
143			$results[$count]['xapian_data'] = unserialize($document->get_data());
144			$results[$count]['score'] = $i->get_percent();
145			}
146
147			$i->next();
148			}
149
150			// Compute match count according to requested type
151			switch ($count_type) {
152			case 1: // Lower bound
153			$count = $matches->get_matches_lower_bound();
154			break;
155
156			case 2: // Upper bound
157			$count = $matches->get_matches_upper_bound();
158			break;
159
160			case 0: // Best estimate
161			default:
162			$count = $matches->get_matches_estimated();
163			break;
164			}
165
166			return [$count, $results];
167			} catch (Exception $e) {
168			display_xapian_error($e->getMessage());
169
170			return null;
171			}
172			}
173
174			/**
175			* Build a boolean query.
176			*
177			* @param string $term The term string
178			*
179			* @return XapianQuery
180			*/
181			function xapian_get_boolean_query($term)
182			{
183			return new XapianQuery($term);
184			}
185
186			/**
187			* Retrieve a list of database terms.
188			*
189			* @param int $count Number of terms to retrieve (0 means "no limit")
190			* @param string $prefix The prefix of the term to retrieve
191			* @param XapianDatabase\|null $db Xapian database to connect
192			*
193			* @return array\|null
194			*/
195			function xapian_get_all_terms($count = 0, $prefix, $db = null)
196			{
197			try {
198			$db = xapian_get_database($db);
199
200			if (!empty($prefix)) {
201			$termi = $db->allterms_begin($prefix);
202			} else {
203			$termi = $db->allterms_begin();
204			}
205
206			$terms = [];
207			$i = 0;
208
209			for (; !$termi->equals($db->allterms_end()) && (++$i <= $count \|\| $count === 0); $termi->next()) {
210			$terms[] = [
211			'frequency' => $termi->get_termfreq(),
212			'name' => $termi->get_term(),
213			];
214			}
215
216			return $terms;
217			} catch (Exception $e) {
218			display_xapian_error($e->getMessage());
219
220			return null;
221			}
222			}
223
224			/**
225			* Retrieve all terms of a document filtered by prefix.
226			*
227			* @param XapianDocument\|null $doc Document to inspect
228			* @param string $prefix Prefix used to filter the terms
229			*
230			* @return array\|null
231			*/
232			function xapian_get_doc_terms($doc = null, $prefix)
233			{
234			try {
235			if (!is_a($doc, 'XapianDocument')) {
236			return null;
237			}
238
239			// TODO: make the filter by prefix on xapian if possible
240			// ojwb marvil07: use Document::termlist_begin() and then skip_to(prefix) on the TermIterator
241			// ojwb you'll need to check the end condition by hand though
242			$terms = [];
243
244			for ($termi = $doc->termlist_begin(); !$termi->equals($doc->termlist_end()); $termi->next()) {
245			$term = [
246			'frequency' => $termi->get_termfreq(),
247			'name' => $termi->get_term(),
248			];
249
250			if ($term['name'][0] === $prefix) {
251			$terms[] = $term;
252			}
253			}
254
255			return $terms;
256			} catch (Exception $e) {
257			display_xapian_error($e->getMessage());
258
259			return null;
260			}
261			}
262
263			/**
264			* Join Xapian queries.
265			*
266			* @param XapianQuery\|array $query1 First query or array of queries
267			* @param XapianQuery\|array\|null $query2 Second query or array of queries (optional)
268			* @param string $op Logical operator: 'or' or 'and'
269			*
270			* @return XapianQuery
271			*/
272			function xapian_join_queries($query1, $query2 = null, $op = 'or')
273			{
274			// Decide how to join, avoiding including xapian.php outside
275			switch ($op) {
276			case 'and':
277			$op = XapianQuery::OP_AND;
278			break;
279			case 'or':
280			default:
281			$op = XapianQuery::OP_OR;
282			break;
283			}
284
285			// Normalize parameters to arrays
286			if (!is_array($query1)) {
287			$query1 = [$query1];
288			}
289
290			if ($query2 === null) {
291			// Join an array of queries with $op
292			return new XapianQuery($op, $query1);
293			}
294
295			if (!is_array($query2)) {
296			$query2 = [$query2];
297			}
298
299			return new XapianQuery($op, array_merge($query1, $query2));
300			}
301
302			/**
303			* Maps Xapian errors to human-readable messages.
304			*
305			* @author Isaac flores paz <[email protected]>
306			*
307			* @param string $xapian_error_message The Xapian error message
308			*
309			* @return void
310			*/
311			function display_xapian_error($xapian_error_message)
312			{
313			$message = explode(':', $xapian_error_message);
314			$type_error_message = $message[0];
315
316			if ($type_error_message === 'DatabaseOpeningError') {
317			$message_error = get_lang('Failed to open the search database');
318			} elseif ($type_error_message === 'DatabaseVersionError') {
319			$message_error = get_lang('The search database uses an unsupported format');
320			} elseif ($type_error_message === 'DatabaseModifiedError') {
321			$message_error = get_lang('The search database has been modified/broken');
322			} elseif ($type_error_message === 'DatabaseLockError') {
323			$message_error = get_lang('Failed to lock the search database');
324			} elseif ($type_error_message === 'DatabaseCreateError') {
325			$message_error = get_lang('Failed to create the search database');
326			} elseif ($type_error_message === 'DatabaseCorruptError') {
327			$message_error = get_lang('The search database has suffered corruption');
328			} elseif ($type_error_message === 'NetworkTimeoutError') {
329			$message_error = get_lang('Connection timed out while communicating with the remote search database');
330			} else {
331			$message_error = get_lang('Error in search engine');
332			}
333
334			$display_message = get_lang('Error') . ' : ' . $message_error;
335			echo Display::return_message($display_message, 'error');
336			}
337

chamilo / chamilo-lms

Push — master ( ab6f49...c762ff )

xapian_get_all_terms() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like