XapianIndexService::mapLanguageToXapianStemmer() - Code Metrics - Inspection of "Merge pull request #7223 from christianbeeznest/GH..." - chamilo/chamilo-lms - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — master ( 904963...c192c4 )

unknown

created 2025-12-24 07:01 UTC

XapianIndexService::mapLanguageToXapianStemmer() B

↳ Parent: XapianIndexService

Complexity

Conditions	6
Paths	7

Size

Total Lines	54
Code Lines	37

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
cc	6
eloc	37
nc	7
nop	1
dl	0
loc	54
rs	8.7057
c	0
b	0
f	0

How to fix Long Method

<?php

declare(strict_types=1);

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

namespace Chamilo\CoreBundle\Search\Xapian;

use DateTimeImmutable;
use RuntimeException;
use Throwable;
use Xapian;
use XapianDocument;
use XapianStem;
use XapianTermGenerator;
use XapianWritableDatabase;

use const DATE_ATOM;

/**
 * Service responsible for indexing documents into the Xapian database.
 */
final class XapianIndexService
{
    private const DEFAULT_LANGUAGE = 'english';

    public function __construct(
        private readonly SearchIndexPathResolver $indexPathResolver,
    ) {}

    /**
     * Indexes a simple demo document so we can verify that search works end-to-end.
     *
     * @return int The Xapian internal document id
     *
     * @throws RuntimeException When indexing fails
     */
    public function indexDemoDocument(): int
    {
        $now = new DateTimeImmutable('now');

        $fields = [
            'title' => 'Demo test document',
            'content' => 'This is a test document indexed from XapianIndexService in Chamilo 2.',
            'created_at' => $now->format(DATE_ATOM),
        ];

        $terms = [
            'XTdemo',
            'XTchamilo',
        ];

        return $this->indexDocument($fields, $terms);
    }

    /**
     * Indexes a generic document.
     *
     * @param array<string,mixed> $fields            Arbitrary data to store and index as free-text
     * @param string[]            $terms             Optional list of additional terms to add to the document
     * @param string|null         $language          Language used for stemming (defaults to english)
     * @param array<string,mixed> $fieldValuesByCode Fielded values, e.g. ['k' => '...', 't' => '...']
     *
     * @return int The Xapian internal document id
     *
     * @throws RuntimeException When Xapian fails during indexing
     */
    public function indexDocument(
        array $fields,
        array $terms = [],
        ?string $language = null,
        array $fieldValuesByCode = [] // e.g. ['t'=>'...', 'd'=>'...', 'k'=>'...']
    ): int {
        if (!class_exists(XapianWritableDatabase::class)) {
            throw new RuntimeException('Xapian PHP extension is not loaded.');
        }

        $db = $this->openWritableDatabase();

        $doc = new XapianDocument();
        $termGen = new XapianTermGenerator();

        // normalize ISO/code into a Xapian stemmer language string
        $xapianLanguage = $this->mapLanguageToXapianStemmer($language);

        try {
            $stemmer = new XapianStem($xapianLanguage);
        } catch (Throwable $e) {
            error_log(
                '[Xapian] indexDocument: failed to init stemmer for lang='
                .var_export($xapianLanguage, true)
                .', fallback=english, error='.$e->getMessage()
            );

            $stemmer = new XapianStem(self::DEFAULT_LANGUAGE);
        }

        $termGen->set_stemmer($stemmer);
        $termGen->set_document($doc);

        // Unprefixed free-text (general search)
        foreach ($fields as $value) {
            if (null === $value) {
                continue;
            }
            $value = is_string($value) ? $value : (string) $value;
            $value = trim($value);
            if ('' === $value) {
                continue;
            }
            $termGen->index_text($value, 1);
        }

        // Prefixed dynamic fields: t:, d:, k:, etc.
        if (!empty($fieldValuesByCode)) {
            error_log('[Xapian] indexDocument: fieldValuesByCode='.json_encode(array_keys($fieldValuesByCode)));

            foreach ($fieldValuesByCode as $code => $val) {
                $code = strtolower(trim((string) $code));
                if ('' === $code) {
                    continue;
                }

                $val = is_string($val) ? $val : (string) $val;
                $val = trim($val);
                if ('' === $val) {
                    continue;
                }

                // Must match query parser convention: F + strtoupper(code)
                $prefix = 'F'.strtoupper($code);

                // This is what makes t: / d: / k: work
                $termGen->index_text($val, 1, $prefix);
            }

            // Optional: keep it in stored data for debugging
            $fields['searchFieldValues'] = $fieldValuesByCode;
        }

        // Extra terms (Tdocument, Cxx, Sxx...)
        foreach ($terms as $term) {
            $term = (string) $term;
            if ('' === $term) {
                continue;
            }
            $doc->add_term($term, 1);
        }

        $doc->set_data(serialize($fields));

        try {
            $docId = $db->add_document($doc);
            $db->flush();

            error_log('[Xapian] indexDocument: added docId='.$docId);

            return $docId;
        } catch (Throwable $e) {
            throw new RuntimeException(sprintf('Failed to index document in Xapian: %s', $e->getMessage()), 0, $e);
        }
    }

    /**
     * Deletes a document from the Xapian index using its internal document id.
     *
     * @throws RuntimeException When Xapian fails during deletion
     */
    public function deleteDocument(int $docId): void
    {
        if (!class_exists(XapianWritableDatabase::class)) {
            throw new RuntimeException('Xapian PHP extension is not loaded.');
        }

        $db = $this->openWritableDatabase();

        try {
            error_log(
                '[Xapian] XapianIndexService::deleteDocument: deleting docId='
                .var_export($docId, true)
            );

            $db->delete_document($docId);
            $db->flush();
        } catch (Throwable $e) {
            throw new RuntimeException(\sprintf('Failed to delete document in Xapian: %s', $e->getMessage()), 0, $e);
        }
    }

    /**
     * Opens the writable Xapian database using DB_CREATE_OR_OPEN.
     */
    private function openWritableDatabase(): XapianWritableDatabase
    {
        $indexDir = $this->indexPathResolver->getIndexDir();

        return new XapianWritableDatabase($indexDir, Xapian::DB_CREATE_OR_OPEN);
    }

    private function mapLanguageToXapianStemmer(?string $language): string
    {
        if (null === $language) {
            return self::DEFAULT_LANGUAGE;
        }

        $raw = strtolower(trim($language));
        if ('' === $raw) {
            return self::DEFAULT_LANGUAGE;
        }

        // If caller already provides a Xapian language name, accept it
        $known = [
            'english', 'spanish', 'french', 'portuguese', 'italian', 'german', 'dutch',
            'swedish', 'norwegian', 'danish', 'finnish', 'russian', 'arabic', 'greek',
            'turkish', 'romanian', 'hungarian', 'indonesian',
        ];

        if (in_array($raw, $known, true)) {
            return $raw;
        }

        // Normalize ISO variants: es_ES, pt-BR, en_US -> es, pt, en
        $iso = $raw;
        if (str_contains($iso, '_')) {
            $iso = explode('_', $iso, 2)[0];
        }
        if (str_contains($iso, '-')) {
            $iso = explode('-', $iso, 2)[0];
        }
        $iso = strtolower(trim($iso));

        $map = [
            'en' => 'english',
            'es' => 'spanish',
            'fr' => 'french',
            'pt' => 'portuguese',
            'it' => 'italian',
            'de' => 'german',
            'nl' => 'dutch',
            'sv' => 'swedish',
            'no' => 'norwegian',
            'da' => 'danish',
            'fi' => 'finnish',
            'ru' => 'russian',
            'ar' => 'arabic',
            'el' => 'greek',
            'tr' => 'turkish',
            'ro' => 'romanian',
            'hu' => 'hungarian',
            'id' => 'indonesian',
        ];

        return $map[$iso] ?? self::DEFAULT_LANGUAGE;
    }
}


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 DateTimeImmutable;
10			use RuntimeException;
11			use Throwable;
12			use Xapian;
13			use XapianDocument;
14			use XapianStem;
15			use XapianTermGenerator;
16			use XapianWritableDatabase;
17
18			use const DATE_ATOM;
19
20			/**
21			* Service responsible for indexing documents into the Xapian database.
22			*/
23			final class XapianIndexService
24			{
25			private const DEFAULT_LANGUAGE = 'english';
26
27			public function __construct(
28			private readonly SearchIndexPathResolver $indexPathResolver,
29			) {}
30
31			/**
32			* Indexes a simple demo document so we can verify that search works end-to-end.
33			*
34			* @return int The Xapian internal document id
35			*
36			* @throws RuntimeException When indexing fails
37			*/
38			public function indexDemoDocument(): int
39			{
40			$now = new DateTimeImmutable('now');
41
42			$fields = [
43			'title' => 'Demo test document',
44			'content' => 'This is a test document indexed from XapianIndexService in Chamilo 2.',
45			'created_at' => $now->format(DATE_ATOM),
46			];
47
48			$terms = [
49			'XTdemo',
50			'XTchamilo',
51			];
52
53			return $this->indexDocument($fields, $terms);
54			}
55
56			/**
57			* Indexes a generic document.
58			*
59			* @param array<string,mixed> $fields Arbitrary data to store and index as free-text
60			* @param string[] $terms Optional list of additional terms to add to the document
61			* @param string\|null $language Language used for stemming (defaults to english)
62			* @param array<string,mixed> $fieldValuesByCode Fielded values, e.g. ['k' => '...', 't' => '...']
63			*
64			* @return int The Xapian internal document id
65			*
66			* @throws RuntimeException When Xapian fails during indexing
67			*/
68			public function indexDocument(
69			array $fields,
70			array $terms = [],
71			?string $language = null,
72			array $fieldValuesByCode = [] // e.g. ['t'=>'...', 'd'=>'...', 'k'=>'...']
73			): int {
74			if (!class_exists(XapianWritableDatabase::class)) {
75			throw new RuntimeException('Xapian PHP extension is not loaded.');
76			}
77
78			$db = $this->openWritableDatabase();
79
80			$doc = new XapianDocument();
81			$termGen = new XapianTermGenerator();
82
83			// normalize ISO/code into a Xapian stemmer language string
84			$xapianLanguage = $this->mapLanguageToXapianStemmer($language);
85
86			try {
87			$stemmer = new XapianStem($xapianLanguage);
88			} catch (Throwable $e) {
89			error_log(
90			'[Xapian] indexDocument: failed to init stemmer for lang='
91			.var_export($xapianLanguage, true)
92			.', fallback=english, error='.$e->getMessage()
93			);
94
95			$stemmer = new XapianStem(self::DEFAULT_LANGUAGE);
96			}
97
98			$termGen->set_stemmer($stemmer);
99			$termGen->set_document($doc);
100
101			// Unprefixed free-text (general search)
102			foreach ($fields as $value) {
103			if (null === $value) {
104			continue;
105			}
106			$value = is_string($value) ? $value : (string) $value;
107			$value = trim($value);
108			if ('' === $value) {
109			continue;
110			}
111			$termGen->index_text($value, 1);
112			}
113
114			// Prefixed dynamic fields: t:, d:, k:, etc.
115			if (!empty($fieldValuesByCode)) {
116			error_log('[Xapian] indexDocument: fieldValuesByCode='.json_encode(array_keys($fieldValuesByCode)));
117
118			foreach ($fieldValuesByCode as $code => $val) {
119			$code = strtolower(trim((string) $code));
120			if ('' === $code) {
121			continue;
122			}
123
124			$val = is_string($val) ? $val : (string) $val;
125			$val = trim($val);
126			if ('' === $val) {
127			continue;
128			}
129
130			// Must match query parser convention: F + strtoupper(code)
131			$prefix = 'F'.strtoupper($code);
132
133			// This is what makes t: / d: / k: work
134			$termGen->index_text($val, 1, $prefix);
135			}
136
137			// Optional: keep it in stored data for debugging
138			$fields['searchFieldValues'] = $fieldValuesByCode;
139			}
140
141			// Extra terms (Tdocument, Cxx, Sxx...)
142			foreach ($terms as $term) {
143			$term = (string) $term;
144			if ('' === $term) {
145			continue;
146			}
147			$doc->add_term($term, 1);
148			}
149
150			$doc->set_data(serialize($fields));
151
152			try {
153			$docId = $db->add_document($doc);
154			$db->flush();
155
156			error_log('[Xapian] indexDocument: added docId='.$docId);
157
158			return $docId;
159			} catch (Throwable $e) {
160			throw new RuntimeException(sprintf('Failed to index document in Xapian: %s', $e->getMessage()), 0, $e);
161			}
162			}
163
164			/**
165			* Deletes a document from the Xapian index using its internal document id.
166			*
167			* @throws RuntimeException When Xapian fails during deletion
168			*/
169			public function deleteDocument(int $docId): void
170			{
171			if (!class_exists(XapianWritableDatabase::class)) {
172			throw new RuntimeException('Xapian PHP extension is not loaded.');
173			}
174
175			$db = $this->openWritableDatabase();
176
177			try {
178			error_log(
179			'[Xapian] XapianIndexService::deleteDocument: deleting docId='
180			.var_export($docId, true)
181			);
182
183			$db->delete_document($docId);
184			$db->flush();
185			} catch (Throwable $e) {
186			throw new RuntimeException(\sprintf('Failed to delete document in Xapian: %s', $e->getMessage()), 0, $e);
187			}
188			}
189
190			/**
191			* Opens the writable Xapian database using DB_CREATE_OR_OPEN.
192			*/
193			private function openWritableDatabase(): XapianWritableDatabase
194			{
195			$indexDir = $this->indexPathResolver->getIndexDir();
196
197			return new XapianWritableDatabase($indexDir, Xapian::DB_CREATE_OR_OPEN);
198			}
199
200			private function mapLanguageToXapianStemmer(?string $language): string
201			{
202			if (null === $language) {
203			return self::DEFAULT_LANGUAGE;
204			}
205
206			$raw = strtolower(trim($language));
207			if ('' === $raw) {
208			return self::DEFAULT_LANGUAGE;
209			}
210
211			// If caller already provides a Xapian language name, accept it
212			$known = [
213			'english', 'spanish', 'french', 'portuguese', 'italian', 'german', 'dutch',
214			'swedish', 'norwegian', 'danish', 'finnish', 'russian', 'arabic', 'greek',
215			'turkish', 'romanian', 'hungarian', 'indonesian',
216			];
217
218			if (in_array($raw, $known, true)) {
219			return $raw;
220			}
221
222			// Normalize ISO variants: es_ES, pt-BR, en_US -> es, pt, en
223			$iso = $raw;
224			if (str_contains($iso, '_')) {
225			$iso = explode('_', $iso, 2)[0];
226			}
227			if (str_contains($iso, '-')) {
228			$iso = explode('-', $iso, 2)[0];
229			}
230			$iso = strtolower(trim($iso));
231
232			$map = [
233			'en' => 'english',
234			'es' => 'spanish',
235			'fr' => 'french',
236			'pt' => 'portuguese',
237			'it' => 'italian',
238			'de' => 'german',
239			'nl' => 'dutch',
240			'sv' => 'swedish',
241			'no' => 'norwegian',
242			'da' => 'danish',
243			'fi' => 'finnish',
244			'ru' => 'russian',
245			'ar' => 'arabic',
246			'el' => 'greek',
247			'tr' => 'turkish',
248			'ro' => 'romanian',
249			'hu' => 'hungarian',
250			'id' => 'indonesian',
251			];
252
253			return $map[$iso] ?? self::DEFAULT_LANGUAGE;
254			}
255			}
256

chamilo / chamilo-lms

Push — master ( 904963...c192c4 )

XapianIndexService::mapLanguageToXapianStemmer() B

Complexity

Size

Duplication

Importance

How to fix Long Method

Long Method

Duplication Side-by-Side

Filter issues like