XapianIndexer::connectDb() - Code Metrics - Inspection of "Search: enable Xapian indexing for course resource..." - chamilo/chamilo-lms - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Pull Request — master (#7159)

unknown

created 2025-12-12 19:29 UTC

XapianIndexer::connectDb() B

↳ Parent: XapianIndexer

Complexity

Conditions	8
Paths	109

Size

Total Lines	44
Code Lines	23

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
cc	8
eloc	23
c	0
b	0
f	0
nc	109
nop	3
dl	0
loc	44
rs	8.3694

<?php
/* For licensing terms, see /license.txt */

// @todo add setting to add xapian.php
// require_once 'xapian.php';

use Chamilo\CoreBundle\Framework\Container;
use Chamilo\CoreBundle\Search\Xapian\SearchIndexPathResolver;

/**
 * Abstract helper class.
 */
abstract class XapianIndexer
{
    /** @var XapianTermGenerator */
    public $indexer;

    /** @var XapianStem */
    public $stemmer;

    /** @var XapianWritableDatabase */
    protected $db;

    /** @var IndexableChunk[] */
    protected $chunks = [];

    /**
     * Class constructor.
     */
    public function __construct()
    {
        $this->db = null;
        $this->stemmer = null;
    }

    /**
     * Class destructor.
     */
    public function __destruct()
    {
        unset($this->db);
        unset($this->stemmer);
    }

    /**
     * Generates a list of languages Xapian manages.
     *
     * This method enables the definition of more matches between
     * Chamilo languages and Xapian languages (through hardcoding)
     *
     * @return array Array of languages codes -> Xapian languages
     */
    final public function xapian_languages()
    {
        /* http://xapian.org/docs/apidoc/html/classXapian_1_1Stem.html */
        return [
            'none' => 'none', // do not stem terms
            'da' => 'danish',
            'nl' => 'dutch',
            /* Martin Porter's 2002 revision of his stemmer */
            'en' => 'english',
            /* Lovin's stemmer */
            'lovins' => 'english_lovins',
            /* Porter's stemmer as described in his 1980 paper */
            'porter' => 'english_porter',
            'fi' => 'finnish',
            'fr' => 'french',
            'de' => 'german',
            'it' => 'italian',
            'no' => 'norwegian',
            'pt' => 'portuguese',
            'ru' => 'russian',
            'es' => 'spanish',
            'sv' => 'swedish',
        ];
    }

    /**
     * Connect to the database, and create it if it does not exist.
     *
     * In Chamilo 2, this will prefer the Symfony SearchIndexPathResolver
     * (var/search) and fall back to the legacy upload path when needed.
     */
    public function connectDb($path = null, $dbMode = null, $lang = 'english')
    {
        if ($this->db !== null) {
            return $this->db;
        }

        if ($dbMode === null) {
            $dbMode = Xapian::DB_CREATE_OR_OPEN;
        }

        if ($path === null) {
            // Legacy default path (Chamilo 1)
            $path = api_get_path(SYS_UPLOAD_PATH) . 'plugins/xapian/searchdb/';


            // If running under Chamilo 2, prefer the Symfony resolver and var/search
            if (class_exists(Container::class)) {
                try {
                    /** @var 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.
                }
            }
        }

        try {
            $this->db = new XapianWritableDatabase($path, $dbMode);
            $this->indexer = new XapianTermGenerator();

            if (!in_array($lang, $this->xapian_languages(), true)) {
                $lang = 'english';
            }

            $this->stemmer = new XapianStem($lang);
            $this->indexer->set_stemmer($this->stemmer);

            return $this->db;
        } catch (Exception $e) {
            echo Display::return_message($e->getMessage(), 'error');

            return 1;
        }
    }

    /**
     * Simple getter for the db attribute.
     *
     * @return object|null The db attribute
     */
    public function getDb()
    {
        return $this->db;
    }

    /**
     * Add this chunk to the chunk array attribute.
     *
     * @param mixed $chunk Chunk of text (IndexableChunk instance)
     */
    public function addChunk($chunk): void
    {
        $this->chunks[] = $chunk;
    }

    /**
     * Actually index the current data.
     *
     * @return int|null New Xapian document ID or null upon failure
     */
    public function index()
    {
        try {
            if (!empty($this->chunks)) {
                foreach ($this->chunks as $chunk) {
                    $doc = new XapianDocument();
                    $this->indexer->set_document($doc);

                    if (!empty($chunk->terms)) {
                        foreach ($chunk->terms as $term) {
                            // @todo consider using a proper weight value instead of 1
                            $doc->add_term($term['flag'] . $term['name'], 1);
                        }
                    }

                    // Free-form index all data array (title, content, etc.)
                    if (!empty($chunk->data)) {
                        foreach ($chunk->data as $key => $value) {
                            $this->indexer->index_text($value, 1);
                        }
                    }

                    $doc->set_data($chunk->xapian_data, 1);
                    $did = $this->db->add_document($doc);

                    // Make sure changes are flushed to disk
                    $this->db->flush();

                    return $did;
                }
            }
        } catch (Exception $e) {
            echo Display::return_message($e->getMessage(), 'error');
            exit(1);

        }

        return null;
    }

    /**
     * Get a specific document from Xapian db.
     *
     * @param int $did Xapian::docid
     *
     * @return XapianDocument|false XapianDocument, or false on error
     */
    public function get_document($did)
    {
        if ($this->db === null) {
            $this->connectDb();
        }

        try {
            $docid = $this->db->get_document($did);
        } catch (Exception $e) {
            // Intentionally silent here: caller will handle false result.
            return false;
        }

        return $docid;
    }

    /**
     * Get document data on a Xapian document.
     *
     * @param XapianDocument $doc Xapian document to read from
     *
     * @return mixed Xapian document data or false if error
     */
    public function get_document_data($doc)
    {
        if ($this->db === null) {
            $this->connectDb();
        }

        try {
            if (!is_a($doc, 'XapianDocument')) {
                return false;
            }

            $doc_data = $doc->get_data();

            return $doc_data;
        } catch (Exception $e) {
            // Intentionally silent here: caller will handle false result.
            return false;
        }
    }

    /**
     * Replace all terms of a document in Xapian db.
     *
     * @param int    $did    Xapian::docid
     * @param array  $terms  New terms of the document
     * @param string $prefix Prefix used to categorize the doc
     *                       (usually 'T' for title, 'A' for author)
     *
     * @return bool false on error
     */
    public function update_terms($did, $terms, $prefix): bool
    {
        $doc = $this->get_document($did);
        if ($doc === false) {
            return false;
        }

        $doc->clear_terms();

        foreach ($terms as $term) {
            // Add directly with given prefix
            $doc->add_term($prefix . $term, 1);
        }

        $this->db->replace_document($did, $doc);
        $this->db->flush();

        return true;
    }

    /**
     * Remove a document from Xapian db.
     *
     * @param int $did Xapian::docid
     */
    public function remove_document($did): void
    {
        if ($this->db === null) {
            $this->connectDb();
        }

        $did = (int) $did;

        if ($did > 0) {
            $doc = $this->get_document($did);
            if ($doc !== false) {
                $this->db->delete_document($did);
                $this->db->flush();
            }
        }
    }

    /**
     * Adds a term to the document specified.
     *
     * @param string         $term The term to add
     * @param XapianDocument $doc  The Xapian document where to add the term
     *
     * @return XapianDocument|false XapianDocument, or false on error
     */
    public function add_term_to_doc($term, $doc)
    {
        if (!is_a($doc, 'XapianDocument')) {
            return false;
        }

        try {
            $doc->add_term($term);
        } catch (Exception $e) {
            echo Display::return_message($e->getMessage(), 'error');

            return 1;
        }

        return $doc;
    }

    /**
     * Remove a term from the document specified.
     *
     * @param string         $term The term to remove
     * @param XapianDocument $doc  The Xapian document where to remove the term
     *
     * @return XapianDocument|false XapianDocument, or false on error
     */
    public function remove_term_from_doc($term, $doc)
    {
        if (!is_a($doc, 'XapianDocument')) {
            return false;
        }

        try {
            $doc->remove_term($term);
        } catch (Exception $e) {
            echo Display::return_message($e->getMessage(), 'error');

            return 1;
        }

        return $doc;
    }

    /**
     * Replace a document in the actual db.
     *
     * @param XapianDocument $doc Xapian document to push into the db
     * @param int            $did Xapian document id of the document to replace
     *
     * @return mixed
     */
    public function replace_document($doc, $did)
    {
        if (!is_a($doc, 'XapianDocument')) {
            return false;
        }

        if ($this->db === null) {
            $this->connectDb();
        }

        try {
            $this->getDb()->replace_document((int) $did, $doc);
            $this->getDb()->flush();
        } catch (Exception $e) {
            echo Display::return_message($e->getMessage(), 'error');

            return 1;
        }

        return $doc;
    }
}


1			<?php
2			/* For licensing terms, see /license.txt */
3
4			// @todo add setting to add xapian.php
5			// require_once 'xapian.php';
6
7			use Chamilo\CoreBundle\Framework\Container;
8			use Chamilo\CoreBundle\Search\Xapian\SearchIndexPathResolver;
9
10			/**
11			* Abstract helper class.
12			*/
13			abstract class XapianIndexer
14			{
15			/** @var XapianTermGenerator */
16			public $indexer;
17
18			/** @var XapianStem */
19			public $stemmer;
20
21			/** @var XapianWritableDatabase */
22			protected $db;
23
24			/** @var IndexableChunk[] */
25			protected $chunks = [];
26
27			/**
28			* Class constructor.
29			*/
30			public function __construct()
31			{
32			$this->db = null;
33			$this->stemmer = null;
34			}
35
36			/**
37			* Class destructor.
38			*/
39			public function __destruct()
40			{
41			unset($this->db);
42			unset($this->stemmer);
43			}
44
45			/**
46			* Generates a list of languages Xapian manages.
47			*
48			* This method enables the definition of more matches between
49			* Chamilo languages and Xapian languages (through hardcoding)
50			*
51			* @return array Array of languages codes -> Xapian languages
52			*/
53			final public function xapian_languages()
54			{
55			/* http://xapian.org/docs/apidoc/html/classXapian_1_1Stem.html */
56			return [
57			'none' => 'none', // do not stem terms
58			'da' => 'danish',
59			'nl' => 'dutch',
60			/* Martin Porter's 2002 revision of his stemmer */
61			'en' => 'english',
62			/* Lovin's stemmer */
63			'lovins' => 'english_lovins',
64			/* Porter's stemmer as described in his 1980 paper */
65			'porter' => 'english_porter',
66			'fi' => 'finnish',
67			'fr' => 'french',
68			'de' => 'german',
69			'it' => 'italian',
70			'no' => 'norwegian',
71			'pt' => 'portuguese',
72			'ru' => 'russian',
73			'es' => 'spanish',
74			'sv' => 'swedish',
75			];
76			}
77
78			/**
79			* Connect to the database, and create it if it does not exist.
80			*
81			* In Chamilo 2, this will prefer the Symfony SearchIndexPathResolver
82			* (var/search) and fall back to the legacy upload path when needed.
83			*/
84			public function connectDb($path = null, $dbMode = null, $lang = 'english')
85			{
86			if ($this->db !== null) {
87			return $this->db;
88			}
89
90			if ($dbMode === null) {
91			$dbMode = Xapian::DB_CREATE_OR_OPEN;
92			}
93
94			if ($path === null) {
95			// Legacy default path (Chamilo 1)
96			$path = 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...
97
98			// If running under Chamilo 2, prefer the Symfony resolver and var/search
99			if (class_exists(Container::class)) {
100			try {
101			/** @var SearchIndexPathResolver $resolver */
102			$resolver = Container::getSearchIndexPathResolver();
103			$resolver->ensureIndexDirectoryExists();
104			$path = $resolver->getIndexDir();
105			} catch (\Throwable $e) {
106			// Fallback to legacy path if resolver or container are not available.
107			// This keeps backward compatibility and avoids hard failures.
108			}
109			}
110			}
111
112			try {
113			$this->db = new XapianWritableDatabase($path, $dbMode);
114			$this->indexer = new XapianTermGenerator();
115
116			if (!in_array($lang, $this->xapian_languages(), true)) {
117			$lang = 'english';
118			}
119
120			$this->stemmer = new XapianStem($lang);
121			$this->indexer->set_stemmer($this->stemmer);
122
123			return $this->db;
124			} catch (Exception $e) {
125			echo Display::return_message($e->getMessage(), 'error');
126
127			return 1;
128			}
129			}
130
131			/**
132			* Simple getter for the db attribute.
133			*
134			* @return object\|null The db attribute
135			*/
136			public function getDb()
137			{
138			return $this->db;
139			}
140
141			/**
142			* Add this chunk to the chunk array attribute.
143			*
144			* @param mixed $chunk Chunk of text (IndexableChunk instance)
145			*/
146			public function addChunk($chunk): void
147			{
148			$this->chunks[] = $chunk;
149			}
150
151			/**
152			* Actually index the current data.
153			*
154			* @return int\|null New Xapian document ID or null upon failure
155			*/
156			public function index()
157			{
158			try {
159			if (!empty($this->chunks)) {
160			foreach ($this->chunks as $chunk) {
161			$doc = new XapianDocument();
162			$this->indexer->set_document($doc);
163
164			if (!empty($chunk->terms)) {
165			foreach ($chunk->terms as $term) {
166			// @todo consider using a proper weight value instead of 1
167			$doc->add_term($term['flag'] . $term['name'], 1);
168			}
169			}
170
171			// Free-form index all data array (title, content, etc.)
172			if (!empty($chunk->data)) {
173			foreach ($chunk->data as $key => $value) {
174			$this->indexer->index_text($value, 1);
175			}
176			}
177
178			$doc->set_data($chunk->xapian_data, 1);
179			$did = $this->db->add_document($doc);
180
181			// Make sure changes are flushed to disk
182			$this->db->flush();
183
184			return $did;
185			}
186			}
187			} catch (Exception $e) {
188			echo Display::return_message($e->getMessage(), 'error');
189			exit(1);
			0 ignored issues – show Best Practice introduced 2018-01-05 13:33 UTC by Report Bug Copy Issue Report Using `exit` here is not recommended. In general, usage of exit should be done with care and only when running in a scripting context like a CLI script. Loading history...
190			}
191
192			return null;
193			}
194
195			/**
196			* Get a specific document from Xapian db.
197			*
198			* @param int $did Xapian::docid
199			*
200			* @return XapianDocument\|false XapianDocument, or false on error
201			*/
202			public function get_document($did)
203			{
204			if ($this->db === null) {
205			$this->connectDb();
206			}
207
208			try {
209			$docid = $this->db->get_document($did);
210			} catch (Exception $e) {
211			// Intentionally silent here: caller will handle false result.
212			return false;
213			}
214
215			return $docid;
216			}
217
218			/**
219			* Get document data on a Xapian document.
220			*
221			* @param XapianDocument $doc Xapian document to read from
222			*
223			* @return mixed Xapian document data or false if error
224			*/
225			public function get_document_data($doc)
226			{
227			if ($this->db === null) {
228			$this->connectDb();
229			}
230
231			try {
232			if (!is_a($doc, 'XapianDocument')) {
233			return false;
234			}
235
236			$doc_data = $doc->get_data();
237
238			return $doc_data;
239			} catch (Exception $e) {
240			// Intentionally silent here: caller will handle false result.
241			return false;
242			}
243			}
244
245			/**
246			* Replace all terms of a document in Xapian db.
247			*
248			* @param int $did Xapian::docid
249			* @param array $terms New terms of the document
250			* @param string $prefix Prefix used to categorize the doc
251			* (usually 'T' for title, 'A' for author)
252			*
253			* @return bool false on error
254			*/
255			public function update_terms($did, $terms, $prefix): bool
256			{
257			$doc = $this->get_document($did);
258			if ($doc === false) {
259			return false;
260			}
261
262			$doc->clear_terms();
263
264			foreach ($terms as $term) {
265			// Add directly with given prefix
266			$doc->add_term($prefix . $term, 1);
267			}
268
269			$this->db->replace_document($did, $doc);
270			$this->db->flush();
271
272			return true;
273			}
274
275			/**
276			* Remove a document from Xapian db.
277			*
278			* @param int $did Xapian::docid
279			*/
280			public function remove_document($did): void
281			{
282			if ($this->db === null) {
283			$this->connectDb();
284			}
285
286			$did = (int) $did;
287
288			if ($did > 0) {
289			$doc = $this->get_document($did);
290			if ($doc !== false) {
291			$this->db->delete_document($did);
292			$this->db->flush();
293			}
294			}
295			}
296
297			/**
298			* Adds a term to the document specified.
299			*
300			* @param string $term The term to add
301			* @param XapianDocument $doc The Xapian document where to add the term
302			*
303			* @return XapianDocument\|false XapianDocument, or false on error
304			*/
305			public function add_term_to_doc($term, $doc)
306			{
307			if (!is_a($doc, 'XapianDocument')) {
308			return false;
309			}
310
311			try {
312			$doc->add_term($term);
313			} catch (Exception $e) {
314			echo Display::return_message($e->getMessage(), 'error');
315
316			return 1;
317			}
318
319			return $doc;
320			}
321
322			/**
323			* Remove a term from the document specified.
324			*
325			* @param string $term The term to remove
326			* @param XapianDocument $doc The Xapian document where to remove the term
327			*
328			* @return XapianDocument\|false XapianDocument, or false on error
329			*/
330			public function remove_term_from_doc($term, $doc)
331			{
332			if (!is_a($doc, 'XapianDocument')) {
333			return false;
334			}
335
336			try {
337			$doc->remove_term($term);
338			} catch (Exception $e) {
339			echo Display::return_message($e->getMessage(), 'error');
340
341			return 1;
342			}
343
344			return $doc;
345			}
346
347			/**
348			* Replace a document in the actual db.
349			*
350			* @param XapianDocument $doc Xapian document to push into the db
351			* @param int $did Xapian document id of the document to replace
352			*
353			* @return mixed
354			*/
355			public function replace_document($doc, $did)
356			{
357			if (!is_a($doc, 'XapianDocument')) {
358			return false;
359			}
360
361			if ($this->db === null) {
362			$this->connectDb();
363			}
364
365			try {
366			$this->getDb()->replace_document((int) $did, $doc);
367			$this->getDb()->flush();
368			} catch (Exception $e) {
369			echo Display::return_message($e->getMessage(), 'error');
370
371			return 1;
372			}
373
374			return $doc;
375			}
376			}
377

chamilo / chamilo-lms

Pull Request — master (#7159)

XapianIndexer::connectDb() B

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like