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 (#7144)

unknown

created 2025-12-12 19:05 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(): array
    {
        // See http://xapian.org/docs/apidoc/html/classXapian_1_1Stem.html for available languages.
        // The match with Chamilo's language table is made on the english_name field, not the ISO code.
        return [
            'none' => 'none', // do not stem terms
            'ar' => 'arabic',
            'ca' => 'catalan',
            'da' => 'danish',
            'de' => 'german',
            'en' => 'english',
            'es' => 'spanish',
            'eu' => 'basque',
            'fi' => 'finnish',
            'fr' => 'french',
            'ga' => 'irish',
            'hu' => 'hungarian',
            'hy' => 'armenian',
            'id' => 'indonesian',
            'it' => 'italian',
            'lt' => 'lithuanian',
            'ne' => 'nepali',
            'nl' => 'dutch',
            'no' => 'norwegian',
            'pt' => 'portuguese',
            'ro' => 'romanian',
            'ru' => 'russian',
            'sv' => 'swedish',
            'ta' => 'tamil',
            'tr' => 'turkish',
        ];
    }

    /**
     * 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($this->stemmer);
43			}
44
45			/**
46			* Generates a list of languages Xapian manages.
47			*
48			* This method enables the definition of more matches between
49			* Chamilo languages and Xapian languages (through hardcoding)
50			*
51			* @return array Array of languages codes -> Xapian languages
52			*/
53			final public function xapian_languages(): array
54			{
55			// See http://xapian.org/docs/apidoc/html/classXapian_1_1Stem.html for available languages.
56			// The match with Chamilo's language table is made on the english_name field, not the ISO code.
57			return [
58			'none' => 'none', // do not stem terms
59			'ar' => 'arabic',
60			'ca' => 'catalan',
61			'da' => 'danish',
62			'de' => 'german',
63			'en' => 'english',
64			'es' => 'spanish',
65			'eu' => 'basque',
66			'fi' => 'finnish',
67			'fr' => 'french',
68			'ga' => 'irish',
69			'hu' => 'hungarian',
70			'hy' => 'armenian',
71			'id' => 'indonesian',
72			'it' => 'italian',
73			'lt' => 'lithuanian',
74			'ne' => 'nepali',
75			'nl' => 'dutch',
76			'no' => 'norwegian',
77			'pt' => 'portuguese',
78			'ro' => 'romanian',
79			'ru' => 'russian',
80			'sv' => 'swedish',
81			'ta' => 'tamil',
82			'tr' => 'turkish',
83			];
84			}
85
86			/**
87			* Connect to the database, and create it if it does not exist.
88			*
89			* In Chamilo 2, this will prefer the Symfony SearchIndexPathResolver
90			* (var/search) and fall back to the legacy upload path when needed.
91			*/
92			public function connectDb($path = null, $dbMode = null, $lang = 'english')
93			{
94			if ($this->db !== null) {
95			return $this->db;
96			}
97
98			if ($dbMode === null) {
99			$dbMode = Xapian::DB_CREATE_OR_OPEN;
100			}
101
102			if ($path === null) {
103			// Legacy default path (Chamilo 1)
104			$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...
105
106			// If running under Chamilo 2, prefer the Symfony resolver and var/search
107			if (class_exists(Container::class)) {
108			try {
109			/** @var SearchIndexPathResolver $resolver */
110			$resolver = Container::getSearchIndexPathResolver();
111			$resolver->ensureIndexDirectoryExists();
112			$path = $resolver->getIndexDir();
113			} catch (\Throwable $e) {
114			// Fallback to legacy path if resolver or container are not available.
115			// This keeps backward compatibility and avoids hard failures.
116			}
117			}
118			}
119
120			try {
121			$this->db = new XapianWritableDatabase($path, $dbMode);
122			$this->indexer = new XapianTermGenerator();
123
124			if (!in_array($lang, $this->xapian_languages(), true)) {
125			$lang = 'english';
126			}
127
128			$this->stemmer = new XapianStem($lang);
129			$this->indexer->set_stemmer($this->stemmer);
130
131			return $this->db;
132			} catch (Exception $e) {
133			echo Display::return_message($e->getMessage(), 'error');
134
135			return 1;
136			}
137			}
138
139			/**
140			* Simple getter for the db attribute.
141			*
142			* @return object\|null The db attribute
143			*/
144			public function getDb()
145			{
146			return $this->db;
147			}
148
149			/**
150			* Add this chunk to the chunk array attribute.
151			*
152			* @param mixed $chunk Chunk of text (IndexableChunk instance)
153			*/
154			public function addChunk($chunk): void
155			{
156			$this->chunks[] = $chunk;
157			}
158
159			/**
160			* Actually index the current data.
161			*
162			* @return int\|null New Xapian document ID or null upon failure
163			*/
164			public function index()
165			{
166			try {
167			if (!empty($this->chunks)) {
168			foreach ($this->chunks as $chunk) {
169			$doc = new XapianDocument();
170			$this->indexer->set_document($doc);
171
172			if (!empty($chunk->terms)) {
173			foreach ($chunk->terms as $term) {
174			// @todo consider using a proper weight value instead of 1
175			$doc->add_term($term['flag'] . $term['name'], 1);
176			}
177			}
178
179			// Free-form index all data array (title, content, etc.)
180			if (!empty($chunk->data)) {
181			foreach ($chunk->data as $key => $value) {
182			$this->indexer->index_text($value, 1);
183			}
184			}
185
186			$doc->set_data($chunk->xapian_data, 1);
187			$did = $this->db->add_document($doc);
188
189			// Make sure changes are flushed to disk
190			$this->db->flush();
191
192			return $did;
193			}
194			}
195			} catch (Exception $e) {
196			echo Display::return_message($e->getMessage(), 'error');
197			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...
198			}
199
200			return null;
201			}
202
203			/**
204			* Get a specific document from Xapian db.
205			*
206			* @param int $did Xapian::docid
207			*
208			* @return XapianDocument\|false XapianDocument, or false on error
209			*/
210			public function get_document($did)
211			{
212			if ($this->db === null) {
213			$this->connectDb();
214			}
215
216			try {
217			$docid = $this->db->get_document($did);
218			} catch (Exception $e) {
219			// Intentionally silent here: caller will handle false result.
220			return false;
221			}
222
223			return $docid;
224			}
225
226			/**
227			* Get document data on a Xapian document.
228			*
229			* @param XapianDocument $doc Xapian document to read from
230			*
231			* @return mixed Xapian document data or false if error
232			*/
233			public function get_document_data($doc)
234			{
235			if ($this->db === null) {
236			$this->connectDb();
237			}
238
239			try {
240			if (!is_a($doc, 'XapianDocument')) {
241			return false;
242			}
243
244			$doc_data = $doc->get_data();
245
246			return $doc_data;
247			} catch (Exception $e) {
248			// Intentionally silent here: caller will handle false result.
249			return false;
250			}
251			}
252
253			/**
254			* Replace all terms of a document in Xapian db.
255			*
256			* @param int $did Xapian::docid
257			* @param array $terms New terms of the document
258			* @param string $prefix Prefix used to categorize the doc
259			* (usually 'T' for title, 'A' for author)
260			*
261			* @return bool false on error
262			*/
263			public function update_terms($did, $terms, $prefix): bool
264			{
265			$doc = $this->get_document($did);
266			if ($doc === false) {
267			return false;
268			}
269
270			$doc->clear_terms();
271
272			foreach ($terms as $term) {
273			// Add directly with given prefix
274			$doc->add_term($prefix . $term, 1);
275			}
276
277			$this->db->replace_document($did, $doc);
278			$this->db->flush();
279
280			return true;
281			}
282
283			/**
284			* Remove a document from Xapian db.
285			*
286			* @param int $did Xapian::docid
287			*/
288			public function remove_document($did): void
289			{
290			if ($this->db === null) {
291			$this->connectDb();
292			}
293
294			$did = (int) $did;
295
296			if ($did > 0) {
297			$doc = $this->get_document($did);
298			if ($doc !== false) {
299			$this->db->delete_document($did);
300			$this->db->flush();
301			}
302			}
303			}
304
305			/**
306			* Adds a term to the document specified.
307			*
308			* @param string $term The term to add
309			* @param XapianDocument $doc The Xapian document where to add the term
310			*
311			* @return XapianDocument\|false XapianDocument, or false on error
312			*/
313			public function add_term_to_doc($term, $doc)
314			{
315			if (!is_a($doc, 'XapianDocument')) {
316			return false;
317			}
318
319			try {
320			$doc->add_term($term);
321			} catch (Exception $e) {
322			echo Display::return_message($e->getMessage(), 'error');
323
324			return 1;
325			}
326
327			return $doc;
328			}
329
330			/**
331			* Remove a term from the document specified.
332			*
333			* @param string $term The term to remove
334			* @param XapianDocument $doc The Xapian document where to remove the term
335			*
336			* @return XapianDocument\|false XapianDocument, or false on error
337			*/
338			public function remove_term_from_doc($term, $doc)
339			{
340			if (!is_a($doc, 'XapianDocument')) {
341			return false;
342			}
343
344			try {
345			$doc->remove_term($term);
346			} catch (Exception $e) {
347			echo Display::return_message($e->getMessage(), 'error');
348
349			return 1;
350			}
351
352			return $doc;
353			}
354
355			/**
356			* Replace a document in the actual db.
357			*
358			* @param XapianDocument $doc Xapian document to push into the db
359			* @param int $did Xapian document id of the document to replace
360			*
361			* @return mixed
362			*/
363			public function replace_document($doc, $did)
364			{
365			if (!is_a($doc, 'XapianDocument')) {
366			return false;
367			}
368
369			if ($this->db === null) {
370			$this->connectDb();
371			}
372
373			try {
374			$this->getDb()->replace_document((int) $did, $doc);
375			$this->getDb()->flush();
376			} catch (Exception $e) {
377			echo Display::return_message($e->getMessage(), 'error');
378
379			return 1;
380			}
381
382			return $doc;
383			}
384			}
385

chamilo / chamilo-lms

Pull Request — master (#7144)

XapianIndexer::connectDb() B

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like