XapianIndexer::index() - Code Metrics - Inspection of "Minor: Format code" - chamilo/chamilo-lms - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — master ( 20a0f4...09a1a7 )

by Angel Fernando Quiroz

created 2025-12-12 22:59 UTC

XapianIndexer::index() B

↳ Parent: XapianIndexer

Complexity

Conditions	8
Paths	44

Size

Total Lines	37
Code Lines	20

Duplication

Lines	0
Ratio	0 %

Importance

Changes	1
Bugs	0	Features	0

Metric	Value
cc	8
eloc	20
c	1
b	0
f	0
nc	44
nop	0
dl	0
loc	37
rs	8.4444

<?php

declare(strict_types=1);

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

namespace Chamilo\CoreBundle\Search\Xapian;

use Exception;
use RuntimeException;
use Xapian;
use XapianDocument;
use XapianStem;
use XapianTermGenerator;
use XapianWritableDatabase;

/**
 * Base Xapian indexer for Chamilo 2.
 *
 * This is a modernized version of the legacy XapianIndexer from Chamilo 1,
 * adapted to namespaces and DI.
 */
abstract class XapianIndexer
{
    protected ?XapianTermGenerator $indexer = null;

    protected ?XapianStem $stemmer = null;

    protected ?XapianWritableDatabase $db = null;

    /**
     * @var array<int,object>
     */
    protected array $chunks = [];

    public function __construct(
        private readonly SearchIndexPathResolver $indexPathResolver,
    ) {
        // Defer DB opening until first use.
    }

    public function __destruct()
    {
        unset($this->db, $this->stemmer, $this->indexer);
    }

    /**
     * Returns the list of languages supported by Xapian.
     *
     * @return array<string,string> Language codes -> Xapian languages
     */
    final public function getSupportedLanguages(): array
    {
        return [
            'none' => 'none',
            'da' => 'danish',
            'nl' => 'dutch',
            'en' => 'english',
            'lovins' => 'english_lovins',
            'porter' => 'english_porter',
            'fi' => 'finnish',
            'fr' => 'french',
            'de' => 'german',
            'it' => 'italian',
            'no' => 'norwegian',
            'pt' => 'portuguese',
            'ru' => 'russian',
            'es' => 'spanish',
            'sv' => 'swedish',
        ];
    }

    /**
     * Connect to the Xapian writable database, creating it when needed.
     *
     * @throws RuntimeException when the DB cannot be created or opened
     */
    public function connectDb(?string $path = null, ?int $dbMode = null, string $lang = 'english'): XapianWritableDatabase
    {
        require_once 'xapian.php';

        if ($this->db instanceof XapianWritableDatabase) {
            return $this->db;
        }

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

        if (null === $path) {
            $path = $this->indexPathResolver->getIndexDir();
        }

        $this->indexPathResolver->ensureIndexDirectoryExists();

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

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

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

            return $this->db;

        } catch (Exception $e) {
            throw new RuntimeException(\sprintf('Unable to create or open Xapian index at "%s": %s', $path, $e->getMessage()), 0, $e);
        }
    }

    /**
     * Simple getter for the writable database.
     */
    public function getDb(): ?XapianWritableDatabase
    {
        return $this->db;
    }

    /**
     * Add a chunk of indexable data to the batch.
     *
     * @param object $chunk Generic DTO with: terms[], data[], xapian_data
     */
    public function addChunk(object $chunk): void
    {
        $this->chunks[] = $chunk;
    }

    /**
     * Index the current batch of chunks.
     *
     * @return int|null New Xapian document ID, or null when nothing was indexed
     */
    public function index(): ?int
    {
        if (empty($this->chunks)) {
            return null;
        }

        $this->connectDb();

        try {
            foreach ($this->chunks as $chunk) {
                $doc = new XapianDocument();
                $this->indexer?->set_document($doc);

                if (!empty($chunk->terms)) {
                    foreach ($chunk->terms as $term) {
                        $doc->add_term($term['flag'].$term['name'], 1);
                    }
                }

                if (!empty($chunk->data)) {
                    foreach ($chunk->data as $value) {
                        $this->indexer?->index_text((string) $value, 1);
                    }
                }

                $doc->set_data($chunk->xapian_data, 1);

                $did = $this->db?->add_document($doc);
                $this->db?->flush();

                return $did ?? null;
            }
        } catch (Exception $e) {
            throw new RuntimeException(\sprintf('Failed to index chunk in Xapian: %s', $e->getMessage()), 0, $e);
        }

        return null;
    }

    /**
     * Fetch a document by its Xapian docid.
     */
    public function getDocument(int $did): ?XapianDocument
    {
        if (null === $this->db) {
            $this->connectDb();
        }

        try {
            return $this->db?->get_document($did) ?: null;
        } catch (Exception) {
            return null;
        }
    }

    /**
     * Update all terms of a document in the index.
     *
     * @param int    $did    Xapian docid
     * @param string $prefix Prefix used to categorize the terms
     * @param array  $terms  New terms (strings)
     */
    public function updateTerms(int $did, array $terms, string $prefix): bool
    {
        $doc = $this->getDocument($did);
        if (null === $doc) {
            return false;
        }

        $doc->clear_terms();

        foreach ($terms as $term) {
            $doc->add_term($prefix.$term, 1);
        }

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

        return true;
    }

    /**
     * Remove a document from the index.
     */
    public function removeDocument(int $did): void
    {
        if (null === $this->db) {
            $this->connectDb();
        }

        if ($did <= 0) {
            return;
        }

        $doc = $this->getDocument($did);
        if (null === $doc) {
            return;
        }

        $this->db?->delete_document($did);
        $this->db?->flush();
    }

    /**
     * Replace a document in the index.
     */
    public function replaceDocument(XapianDocument $doc, int $did): void
    {
        if (null === $this->db) {
            $this->connectDb();
        }

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


1			<?php
2
3			declare(strict_types=1);
4
5			/* For licensing terms, see /license.txt */
6
7			namespace Chamilo\CoreBundle\Search\Xapian;
8
9			use Exception;
10			use RuntimeException;
11			use Xapian;
12			use XapianDocument;
13			use XapianStem;
14			use XapianTermGenerator;
15			use XapianWritableDatabase;
16
17			/**
18			* Base Xapian indexer for Chamilo 2.
19			*
20			* This is a modernized version of the legacy XapianIndexer from Chamilo 1,
21			* adapted to namespaces and DI.
22			*/
23			abstract class XapianIndexer
24			{
25			protected ?XapianTermGenerator $indexer = null;
26
27			protected ?XapianStem $stemmer = null;
28
29			protected ?XapianWritableDatabase $db = null;
30
31			/**
32			* @var array<int,object>
33			*/
34			protected array $chunks = [];
35
36			public function __construct(
37			private readonly SearchIndexPathResolver $indexPathResolver,
38			) {
39			// Defer DB opening until first use.
40			}
41
42			public function __destruct()
43			{
44			unset($this->db, $this->stemmer, $this->indexer);
45			}
46
47			/**
48			* Returns the list of languages supported by Xapian.
49			*
50			* @return array<string,string> Language codes -> Xapian languages
51			*/
52			final public function getSupportedLanguages(): array
53			{
54			return [
55			'none' => 'none',
56			'da' => 'danish',
57			'nl' => 'dutch',
58			'en' => 'english',
59			'lovins' => 'english_lovins',
60			'porter' => 'english_porter',
61			'fi' => 'finnish',
62			'fr' => 'french',
63			'de' => 'german',
64			'it' => 'italian',
65			'no' => 'norwegian',
66			'pt' => 'portuguese',
67			'ru' => 'russian',
68			'es' => 'spanish',
69			'sv' => 'swedish',
70			];
71			}
72
73			/**
74			* Connect to the Xapian writable database, creating it when needed.
75			*
76			* @throws RuntimeException when the DB cannot be created or opened
77			*/
78			public function connectDb(?string $path = null, ?int $dbMode = null, string $lang = 'english'): XapianWritableDatabase
79			{
80			require_once 'xapian.php';
81
82			if ($this->db instanceof XapianWritableDatabase) {
83			return $this->db;
84			}
85
86			if (null === $dbMode) {
87			$dbMode = Xapian::DB_CREATE_OR_OPEN;
88			}
89
90			if (null === $path) {
91			$path = $this->indexPathResolver->getIndexDir();
92			}
93
94			$this->indexPathResolver->ensureIndexDirectoryExists();
95
96			try {
97			$this->db = new XapianWritableDatabase($path, $dbMode);
98			$this->indexer = new XapianTermGenerator();
99
100			$supported = $this->getSupportedLanguages();
101			if (!\in_array($lang, $supported, true)) {
102			$lang = 'english';
103			}
104
105			$this->stemmer = new XapianStem($lang);
106			$this->indexer->set_stemmer($this->stemmer);
107
108			return $this->db;
			0 ignored issues – show Bug Best Practice introduced 2025-12-08 05:34 UTC by Report Bug Copy Issue Report The expression `return $this->db` returns the type `null` which is incompatible with the type-hinted return `XapianWritableDatabase`. Loading history...
109			} catch (Exception $e) {
110			throw new RuntimeException(\sprintf('Unable to create or open Xapian index at "%s": %s', $path, $e->getMessage()), 0, $e);
111			}
112			}
113
114			/**
115			* Simple getter for the writable database.
116			*/
117			public function getDb(): ?XapianWritableDatabase
118			{
119			return $this->db;
120			}
121
122			/**
123			* Add a chunk of indexable data to the batch.
124			*
125			* @param object $chunk Generic DTO with: terms[], data[], xapian_data
126			*/
127			public function addChunk(object $chunk): void
128			{
129			$this->chunks[] = $chunk;
130			}
131
132			/**
133			* Index the current batch of chunks.
134			*
135			* @return int\|null New Xapian document ID, or null when nothing was indexed
136			*/
137			public function index(): ?int
138			{
139			if (empty($this->chunks)) {
140			return null;
141			}
142
143			$this->connectDb();
144
145			try {
146			foreach ($this->chunks as $chunk) {
147			$doc = new XapianDocument();
148			$this->indexer?->set_document($doc);
149
150			if (!empty($chunk->terms)) {
151			foreach ($chunk->terms as $term) {
152			$doc->add_term($term['flag'].$term['name'], 1);
153			}
154			}
155
156			if (!empty($chunk->data)) {
157			foreach ($chunk->data as $value) {
158			$this->indexer?->index_text((string) $value, 1);
159			}
160			}
161
162			$doc->set_data($chunk->xapian_data, 1);
163
164			$did = $this->db?->add_document($doc);
165			$this->db?->flush();
166
167			return $did ?? null;
168			}
169			} catch (Exception $e) {
170			throw new RuntimeException(\sprintf('Failed to index chunk in Xapian: %s', $e->getMessage()), 0, $e);
171			}
172
173			return null;
174			}
175
176			/**
177			* Fetch a document by its Xapian docid.
178			*/
179			public function getDocument(int $did): ?XapianDocument
180			{
181			if (null === $this->db) {
182			$this->connectDb();
183			}
184
185			try {
186			return $this->db?->get_document($did) ?: null;
187			} catch (Exception) {
188			return null;
189			}
190			}
191
192			/**
193			* Update all terms of a document in the index.
194			*
195			* @param int $did Xapian docid
196			* @param string $prefix Prefix used to categorize the terms
197			* @param array $terms New terms (strings)
198			*/
199			public function updateTerms(int $did, array $terms, string $prefix): bool
200			{
201			$doc = $this->getDocument($did);
202			if (null === $doc) {
203			return false;
204			}
205
206			$doc->clear_terms();
207
208			foreach ($terms as $term) {
209			$doc->add_term($prefix.$term, 1);
210			}
211
212			$this->db?->replace_document($did, $doc);
213			$this->db?->flush();
214
215			return true;
216			}
217
218			/**
219			* Remove a document from the index.
220			*/
221			public function removeDocument(int $did): void
222			{
223			if (null === $this->db) {
224			$this->connectDb();
225			}
226
227			if ($did <= 0) {
228			return;
229			}
230
231			$doc = $this->getDocument($did);
232			if (null === $doc) {
233			return;
234			}
235
236			$this->db?->delete_document($did);
237			$this->db?->flush();
238			}
239
240			/**
241			* Replace a document in the index.
242			*/
243			public function replaceDocument(XapianDocument $doc, int $did): void
244			{
245			if (null === $this->db) {
246			$this->connectDb();
247			}
248
249			$this->db?->replace_document($did, $doc);
250			$this->db?->flush();
251			}
252			}
253

chamilo / chamilo-lms

Push — master ( 20a0f4...09a1a7 )

XapianIndexer::index() B

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like