DocumentXapianIndexer::indexDocument() - 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

DocumentXapianIndexer::indexDocument() F

↳ Parent: DocumentXapianIndexer

Complexity

Conditions	14
Paths	387

Size

Total Lines	136
Code Lines	77

Duplication

Lines	0
Ratio	0 %

Importance

Changes	1
Bugs	0	Features	0

Metric	Value
cc	14
eloc	77
c	1
b	0
f	0
nc	387
nop	1
dl	0
loc	136
rs	2.9784

How to fix Long Method Complexity

<?php

declare(strict_types=1);

namespace Chamilo\CoreBundle\Search\Xapian;

use Chamilo\CoreBundle\Entity\ResourceLink;
use Chamilo\CoreBundle\Entity\ResourceNode;
use Chamilo\CoreBundle\Entity\SearchEngineRef;
use Chamilo\CoreBundle\Settings\SettingsManager;
use Chamilo\CourseBundle\Entity\CDocument;
use Doctrine\ORM\EntityManagerInterface;

/**
 * Handles Xapian indexing for CDocument entities.
 */
final class DocumentXapianIndexer
{
    public function __construct(
        private readonly XapianIndexService $xapianIndexService,
        private readonly EntityManagerInterface $em,
        private readonly SettingsManager $settingsManager,
    ) {
    }

    /**
     * Index a CDocument into Xapian.
     *
     * @return int|null Xapian document id or null when indexing is skipped
     */
    public function indexDocument(CDocument $document): ?int
    {
        $resourceNode = $document->getResourceNode();

        error_log('[Xapian] indexDocument: start for iid='.(string) $document->getIid()
            .', resource_node_id='.($resourceNode ? $resourceNode->getId() : 'null')
            .', filetype='.$document->getFiletype()
        );

        // 1) Check if search is globally enabled
        $enabled = (string) $this->settingsManager->getSetting('search.search_enabled', true);
        error_log('[Xapian] indexDocument: search.search_enabled='.var_export($enabled, true));

        if ($enabled !== 'true') {
            error_log('[Xapian] indexDocument: search is disabled, skipping indexing');

            return null;
        }

        if (!$resourceNode instanceof ResourceNode) {
            error_log('[Xapian] indexDocument: missing ResourceNode, skipping');

            return null;
        }

        // Do not index folders
        if ($document->getFiletype() === 'folder') {
            error_log('[Xapian] indexDocument: skipping folder document, resource_node_id='
                .$resourceNode->getId()
            );

            return null;
        }

        // 2) Resolve course, session and course root node ids
        [$courseId, $sessionId, $courseRootNodeId] = $this->resolveCourseSessionAndRootNode($resourceNode);

        error_log('[Xapian] indexDocument: courseId='.var_export($courseId, true)
            .', sessionId='.var_export($sessionId, true)
            .', courseRootNodeId='.var_export($courseRootNodeId, true)
        );

        // 3) Get textual content if any
        $content = (string) ($resourceNode->getContent() ?? '');
        error_log('[Xapian] indexDocument: content_length='.strlen($content));

        // 4) Build fields payload
        $fields = [
            'title'              => (string) $document->getTitle(),
            'description'        => (string) ($document->getComment() ?? ''),
            'content'            => $content,
            'filetype'           => (string) $document->getFiletype(),
            'resource_node_id'   => (string) $resourceNode->getId(),
            'course_id'          => $courseId !== null ? (string) $courseId : '',
            'session_id'         => $sessionId !== null ? (string) $sessionId : '',
            'course_root_node_id'=> $courseRootNodeId !== null ? (string) $courseRootNodeId : '',
            'full_path'          => $document->getFullPath(),
        ];

        // 5) Base terms
        $terms = ['Tdocument'];

        if ($courseId !== null) {
            $terms[] = 'C'.$courseId;
        }
        if ($sessionId !== null) {
            $terms[] = 'S'.$sessionId;
        }

        // 6) Extra prefilter terms from config
        $this->applyPrefilterConfigToTerms($terms, $courseId, $sessionId, $document);

        error_log('[Xapian] indexDocument: terms='.json_encode($terms));

        // 7) Existing mapping?
        /** @var SearchEngineRef|null $existingRef */
        $existingRef = $this->em
            ->getRepository(SearchEngineRef::class)
            ->findOneBy(['resourceNodeId' => $resourceNode->getId()]);

        $existingDocId = $existingRef?->getSearchDid();
        error_log('[Xapian] indexDocument: existing SearchEngineRef id='
            .($existingRef?->getId() ?? 'null')
            .', existing_did='.var_export($existingDocId, true)
        );

        // 7.1) If we already had a doc in Xapian, try to delete it first
        if ($existingDocId !== null) {
            try {
                $this->xapianIndexService->deleteDocument($existingDocId);
                error_log('[Xapian] indexDocument: previous docId deleted='
                    .var_export($existingDocId, true)
                );
            } catch (\Throwable $e) {
                error_log('[Xapian] indexDocument: failed to delete previous docId='
                    .var_export($existingDocId, true)
                    .' error='.$e->getMessage()
                );
            }
        }

        // 8) Call Xapian (create new document)
        try {
            $docId = $this->xapianIndexService->indexDocument(
                $fields,
                $terms
            );
        } catch (\Throwable $e) {
            error_log('[Xapian] indexDocument: indexDocument() failed: '.$e->getMessage());

            return null;
        }

        error_log('[Xapian] indexDocument: XapianIndexService->indexDocument returned docId='
            .var_export($docId, true)
        );

        // 9) Persist mapping resource_node_id <-> search_did
        if ($existingRef instanceof SearchEngineRef) {
            $existingRef->setSearchDid($docId);
            error_log('[Xapian] indexDocument: updating existing SearchEngineRef id='.$existingRef->getId());
        } else {
            $existingRef = new SearchEngineRef();
            $existingRef->setResourceNodeId((int) $resourceNode->getId());
            $existingRef->setSearchDid($docId);
            $this->em->persist($existingRef);
            error_log('[Xapian] indexDocument: creating new SearchEngineRef for resource_node_id='
                .$resourceNode->getId()
            );
        }

        $this->em->flush();

        error_log('[Xapian] indexDocument: SearchEngineRef saved with id='.$existingRef->getId());

        return $docId;
    }

    /**
     * Remove a document from Xapian using the resource node id.
     */
    public function deleteForResourceNodeId(int $resourceNodeId): void
    {
        error_log('[Xapian] deleteForResourceNodeId: start, resource_node_id='.$resourceNodeId);

        /** @var SearchEngineRef|null $ref */
        $ref = $this->em
            ->getRepository(SearchEngineRef::class)
            ->findOneBy(['resourceNodeId' => $resourceNodeId]);

        if (!$ref instanceof SearchEngineRef) {
            error_log('[Xapian] deleteForResourceNodeId: no SearchEngineRef found, nothing to delete');

            return;
        }

        $docId = $ref->getSearchDid();
        error_log('[Xapian] deleteForResourceNodeId: found SearchEngineRef id='.$ref->getId()
            .', search_did='.var_export($docId, true)
        );

        if ($docId !== null) {
            try {
                $this->xapianIndexService->deleteDocument($docId);
                error_log('[Xapian] deleteForResourceNodeId: deleteDocument called for did='
                    .var_export($docId, true)
                );
            } catch (\Throwable $e) {
                error_log('[Xapian] deleteForResourceNodeId: deleteDocument failed for did='
                    .var_export($docId, true)
                    .' error='.$e->getMessage()
                );
            }
        }

        $this->em->remove($ref);
        $this->em->flush();

        error_log('[Xapian] deleteForResourceNodeId: SearchEngineRef removed for resource_node_id='
            .$resourceNodeId
        );
    }

    /**
     * Resolve course id, session id and course root node id from resource links.
     *
     * @return array{0: int|null, 1: int|null, 2: int|null}
     */
    private function resolveCourseSessionAndRootNode(ResourceNode $resourceNode): array
    {
        $courseId = null;
        $sessionId = null;
        $courseRootNodeId = null;

        foreach ($resourceNode->getResourceLinks() as $link) {
            if (!$link instanceof ResourceLink) {
                continue;
            }

            if ($courseId === null && $link->getCourse()) {
                $course = $link->getCourse();
                $courseId = $course->getId();

                $courseRootNode = $course->getResourceNode();
                if ($courseRootNode instanceof ResourceNode) {
                    $courseRootNodeId = $courseRootNode->getId();
                }
            }

            if ($sessionId === null && $link->getSession()) {
                $sessionId = $link->getSession()->getId();
            }

            if ($courseId !== null && $sessionId !== null && $courseRootNodeId !== null) {
                break;
            }
        }

        return [$courseId, $sessionId, $courseRootNodeId];
    }

    /**
     * Apply configured prefilter prefixes to Xapian terms.
     *
     * Expected JSON structure in search.search_prefilter_prefix, for example:
     *
     * {
     *   "course":  { "prefix": "C", "title": "Course" },
     *   "session": { "prefix": "S", "title": "Session" },
     *   "filetype": { "prefix": "F", "title": "File type" }
     * }
     *
     * "title" is meant for UI labels, "prefix" is used here for terms.
     */
    private function applyPrefilterConfigToTerms(
        array &$terms,
        ?int $courseId,
        ?int $sessionId,
        CDocument $document
    ): void {
        $raw = (string) $this->settingsManager->getSetting('search.search_prefilter_prefix', true);
        if ($raw === '') {
            return;
        }

        $config = json_decode($raw, true);
        if (!\is_array($config)) {
            return;
        }

        foreach ($config as $key => $item) {
            if (!\is_array($item)) {
                continue;
            }

            $prefix = (string) ($item['prefix'] ?? '');
            if ($prefix === '') {
                $prefix = strtoupper((string) $key);
            }

            switch ($key) {
                case 'course':
                    if ($courseId !== null) {
                        $terms[] = $prefix.(string) $courseId;
                    }

                    break;

                case 'session':
                    if ($sessionId !== null) {
                        $terms[] = $prefix.(string) $sessionId;
                    }

                    break;

                case 'filetype':
                    $terms[] = $prefix.$document->getFiletype();

                    break;

                default:
                    // Unknown key: ignore for now
                    break;
            }
        }
    }
}


1			<?php
2
3			declare(strict_types=1);
4
5			namespace Chamilo\CoreBundle\Search\Xapian;
6
7			use Chamilo\CoreBundle\Entity\ResourceLink;
8			use Chamilo\CoreBundle\Entity\ResourceNode;
9			use Chamilo\CoreBundle\Entity\SearchEngineRef;
10			use Chamilo\CoreBundle\Settings\SettingsManager;
11			use Chamilo\CourseBundle\Entity\CDocument;
12			use Doctrine\ORM\EntityManagerInterface;
13
14			/**
15			* Handles Xapian indexing for CDocument entities.
16			*/
17			final class DocumentXapianIndexer
18			{
19			public function __construct(
20			private readonly XapianIndexService $xapianIndexService,
21			private readonly EntityManagerInterface $em,
22			private readonly SettingsManager $settingsManager,
23			) {
24			}
25
26			/**
27			* Index a CDocument into Xapian.
28			*
29			* @return int\|null Xapian document id or null when indexing is skipped
30			*/
31			public function indexDocument(CDocument $document): ?int
32			{
33			$resourceNode = $document->getResourceNode();
34
35			error_log('[Xapian] indexDocument: start for iid='.(string) $document->getIid()
36			.', resource_node_id='.($resourceNode ? $resourceNode->getId() : 'null')
37			.', filetype='.$document->getFiletype()
38			);
39
40			// 1) Check if search is globally enabled
41			$enabled = (string) $this->settingsManager->getSetting('search.search_enabled', true);
42			error_log('[Xapian] indexDocument: search.search_enabled='.var_export($enabled, true));
43
44			if ($enabled !== 'true') {
45			error_log('[Xapian] indexDocument: search is disabled, skipping indexing');
46
47			return null;
48			}
49
50			if (!$resourceNode instanceof ResourceNode) {
51			error_log('[Xapian] indexDocument: missing ResourceNode, skipping');
52
53			return null;
54			}
55
56			// Do not index folders
57			if ($document->getFiletype() === 'folder') {
58			error_log('[Xapian] indexDocument: skipping folder document, resource_node_id='
59			.$resourceNode->getId()
60			);
61
62			return null;
63			}
64
65			// 2) Resolve course, session and course root node ids
66			[$courseId, $sessionId, $courseRootNodeId] = $this->resolveCourseSessionAndRootNode($resourceNode);
67
68			error_log('[Xapian] indexDocument: courseId='.var_export($courseId, true)
69			.', sessionId='.var_export($sessionId, true)
70			.', courseRootNodeId='.var_export($courseRootNodeId, true)
71			);
72
73			// 3) Get textual content if any
74			$content = (string) ($resourceNode->getContent() ?? '');
75			error_log('[Xapian] indexDocument: content_length='.strlen($content));
76
77			// 4) Build fields payload
78			$fields = [
79			'title' => (string) $document->getTitle(),
80			'description' => (string) ($document->getComment() ?? ''),
81			'content' => $content,
82			'filetype' => (string) $document->getFiletype(),
83			'resource_node_id' => (string) $resourceNode->getId(),
84			'course_id' => $courseId !== null ? (string) $courseId : '',
85			'session_id' => $sessionId !== null ? (string) $sessionId : '',
86			'course_root_node_id'=> $courseRootNodeId !== null ? (string) $courseRootNodeId : '',
87			'full_path' => $document->getFullPath(),
88			];
89
90			// 5) Base terms
91			$terms = ['Tdocument'];
92
93			if ($courseId !== null) {
94			$terms[] = 'C'.$courseId;
95			}
96			if ($sessionId !== null) {
97			$terms[] = 'S'.$sessionId;
98			}
99
100			// 6) Extra prefilter terms from config
101			$this->applyPrefilterConfigToTerms($terms, $courseId, $sessionId, $document);
102
103			error_log('[Xapian] indexDocument: terms='.json_encode($terms));
104
105			// 7) Existing mapping?
106			/** @var SearchEngineRef\|null $existingRef */
107			$existingRef = $this->em
108			->getRepository(SearchEngineRef::class)
109			->findOneBy(['resourceNodeId' => $resourceNode->getId()]);
110
111			$existingDocId = $existingRef?->getSearchDid();
112			error_log('[Xapian] indexDocument: existing SearchEngineRef id='
113			.($existingRef?->getId() ?? 'null')
114			.', existing_did='.var_export($existingDocId, true)
115			);
116
117			// 7.1) If we already had a doc in Xapian, try to delete it first
118			if ($existingDocId !== null) {
119			try {
120			$this->xapianIndexService->deleteDocument($existingDocId);
121			error_log('[Xapian] indexDocument: previous docId deleted='
122			.var_export($existingDocId, true)
123			);
124			} catch (\Throwable $e) {
125			error_log('[Xapian] indexDocument: failed to delete previous docId='
126			.var_export($existingDocId, true)
127			.' error='.$e->getMessage()
128			);
129			}
130			}
131
132			// 8) Call Xapian (create new document)
133			try {
134			$docId = $this->xapianIndexService->indexDocument(
135			$fields,
136			$terms
137			);
138			} catch (\Throwable $e) {
139			error_log('[Xapian] indexDocument: indexDocument() failed: '.$e->getMessage());
140
141			return null;
142			}
143
144			error_log('[Xapian] indexDocument: XapianIndexService->indexDocument returned docId='
145			.var_export($docId, true)
146			);
147
148			// 9) Persist mapping resource_node_id <-> search_did
149			if ($existingRef instanceof SearchEngineRef) {
150			$existingRef->setSearchDid($docId);
151			error_log('[Xapian] indexDocument: updating existing SearchEngineRef id='.$existingRef->getId());
152			} else {
153			$existingRef = new SearchEngineRef();
154			$existingRef->setResourceNodeId((int) $resourceNode->getId());
155			$existingRef->setSearchDid($docId);
156			$this->em->persist($existingRef);
157			error_log('[Xapian] indexDocument: creating new SearchEngineRef for resource_node_id='
158			.$resourceNode->getId()
159			);
160			}
161
162			$this->em->flush();
163
164			error_log('[Xapian] indexDocument: SearchEngineRef saved with id='.$existingRef->getId());
165
166			return $docId;
167			}
168
169			/**
170			* Remove a document from Xapian using the resource node id.
171			*/
172			public function deleteForResourceNodeId(int $resourceNodeId): void
173			{
174			error_log('[Xapian] deleteForResourceNodeId: start, resource_node_id='.$resourceNodeId);
175
176			/** @var SearchEngineRef\|null $ref */
177			$ref = $this->em
178			->getRepository(SearchEngineRef::class)
179			->findOneBy(['resourceNodeId' => $resourceNodeId]);
180
181			if (!$ref instanceof SearchEngineRef) {
182			error_log('[Xapian] deleteForResourceNodeId: no SearchEngineRef found, nothing to delete');
183
184			return;
185			}
186
187			$docId = $ref->getSearchDid();
188			error_log('[Xapian] deleteForResourceNodeId: found SearchEngineRef id='.$ref->getId()
189			.', search_did='.var_export($docId, true)
190			);
191
192			if ($docId !== null) {
193			try {
194			$this->xapianIndexService->deleteDocument($docId);
195			error_log('[Xapian] deleteForResourceNodeId: deleteDocument called for did='
196			.var_export($docId, true)
197			);
198			} catch (\Throwable $e) {
199			error_log('[Xapian] deleteForResourceNodeId: deleteDocument failed for did='
200			.var_export($docId, true)
201			.' error='.$e->getMessage()
202			);
203			}
204			}
205
206			$this->em->remove($ref);
207			$this->em->flush();
208
209			error_log('[Xapian] deleteForResourceNodeId: SearchEngineRef removed for resource_node_id='
210			.$resourceNodeId
211			);
212			}
213
214			/**
215			* Resolve course id, session id and course root node id from resource links.
216			*
217			* @return array{0: int\|null, 1: int\|null, 2: int\|null}
218			*/
219			private function resolveCourseSessionAndRootNode(ResourceNode $resourceNode): array
220			{
221			$courseId = null;
222			$sessionId = null;
223			$courseRootNodeId = null;
224
225			foreach ($resourceNode->getResourceLinks() as $link) {
226			if (!$link instanceof ResourceLink) {
227			continue;
228			}
229
230			if ($courseId === null && $link->getCourse()) {
231			$course = $link->getCourse();
232			$courseId = $course->getId();
233
234			$courseRootNode = $course->getResourceNode();
235			if ($courseRootNode instanceof ResourceNode) {
236			$courseRootNodeId = $courseRootNode->getId();
237			}
238			}
239
240			if ($sessionId === null && $link->getSession()) {
241			$sessionId = $link->getSession()->getId();
242			}
243
244			if ($courseId !== null && $sessionId !== null && $courseRootNodeId !== null) {
245			break;
246			}
247			}
248
249			return [$courseId, $sessionId, $courseRootNodeId];
250			}
251
252			/**
253			* Apply configured prefilter prefixes to Xapian terms.
254			*
255			* Expected JSON structure in search.search_prefilter_prefix, for example:
256			*
257			* {
258			* "course": { "prefix": "C", "title": "Course" },
259			* "session": { "prefix": "S", "title": "Session" },
260			* "filetype": { "prefix": "F", "title": "File type" }
261			* }
262			*
263			* "title" is meant for UI labels, "prefix" is used here for terms.
264			*/
265			private function applyPrefilterConfigToTerms(
266			array &$terms,
267			?int $courseId,
268			?int $sessionId,
269			CDocument $document
270			): void {
271			$raw = (string) $this->settingsManager->getSetting('search.search_prefilter_prefix', true);
272			if ($raw === '') {
273			return;
274			}
275
276			$config = json_decode($raw, true);
277			if (!\is_array($config)) {
278			return;
279			}
280
281			foreach ($config as $key => $item) {
282			if (!\is_array($item)) {
283			continue;
284			}
285
286			$prefix = (string) ($item['prefix'] ?? '');
287			if ($prefix === '') {
288			$prefix = strtoupper((string) $key);
289			}
290
291			switch ($key) {
292			case 'course':
293			if ($courseId !== null) {
294			$terms[] = $prefix.(string) $courseId;
295			}
296
297			break;
298
299			case 'session':
300			if ($sessionId !== null) {
301			$terms[] = $prefix.(string) $sessionId;
302			}
303
304			break;
305
306			case 'filetype':
307			$terms[] = $prefix.$document->getFiletype();
308
309			break;
310
311			default:
312			// Unknown key: ignore for now
313			break;
314			}
315			}
316			}
317			}
318

chamilo / chamilo-lms

Pull Request — master (#7144)

DocumentXapianIndexer::indexDocument() F

Complexity

Size

Duplication

Importance

How to fix Long Method Complexity

Long Method

Duplication Side-by-Side

Filter issues like