CollectionHandler::addDocument() - Code Metrics - Inspection of "Merge pull request #11 from brevia-ai/feat/reindex..." - brevia-ai/brevia-bedita - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Push — main ( 6b3c3f...a8a095 )

unknown

created 2024-04-10 09:55 UTC

CollectionHandler::addDocument() B

↳ Parent: CollectionHandler

Complexity

Conditions	6
Paths	5

Size

Total Lines	35
Code Lines	26

Duplication

Lines	0
Ratio	0 %

Importance

Changes	5
Bugs	1	Features	1

Metric	Value
cc	6
eloc	26
c	5
b	1
f	1
nc	5
nop	3
dl	0
loc	35
rs	8.8817

<?php
declare(strict_types=1);

/**
 * BEdita Brevia plugin
 *
 * Copyright 2023 Atlas Srl
 */
namespace Brevia\BEdita\Index;

use BEdita\Core\Filesystem\FilesystemRegistry;
use BEdita\Core\Model\Entity\AsyncJob;
use BEdita\Core\Model\Entity\ObjectEntity;
use Brevia\BEdita\Client\BreviaClient;
use Cake\Http\Client\FormData;
use Cake\Log\LogTrait;
use Cake\ORM\Locator\LocatorAwareTrait;
use Cake\Utility\Hash;
use Laminas\Diactoros\UploadedFile;

/**
 * Handle Brevia collection via API.
 */
class CollectionHandler
{
    use LocatorAwareTrait;
    use LogTrait;

    /**
     * Brevia API client
     *
     * @var \Brevia\BEdita\Client\BreviaClient
     */
    protected BreviaClient $client;

    /**
     * List of properties to exclude when saving Brevia collection metadata
     *
     * @var array
     */
    public const COLLECTION_FIELDS_EXCLUDED = [
        'uname',
        'type',
        'created',
        'modified',
        'lang',
        'locked',
        'published',
        'created_by',
        'modified_by',
        'collection_uuid',
        'collection_updated',
        '_meta',
        '_joinData',
    ];

    /**
     * List of properties to check when updating a Brevia collection index
     *
     * @var array
     */
    public const DOCUMENT_PROPERTIES = [
        'title',
        'description',
        'body',
        'url',
    ];

    /**
     * Handler constructor
     */
    public function __construct()
    {
        $this->client = new BreviaClient();
    }

    /**
     * Create Brevia collection
     *
     * @param \BEdita\Core\Model\Entity\ObjectEntity $collection Collection entity
     * @return void
     */
    public function createCollection(ObjectEntity $collection): void
    {
        $msg = sprintf('Creating collection "%s"', $collection->get('title'));
        $this->log($msg, 'info');

        $response = $this->client->post('/collections', $this->breviaCollection($collection));
        $collection->set('collection_uuid', Hash::get($response->getJson(), 'uuid'));
        $collection->set('collection_updated', date('c'));
        $collection->getTable()->saveOrFail($collection, ['_skipAfterSave' => true]);
    }

    /**
     * Update Brevia collection
     *
     * @param \BEdita\Core\Model\Entity\ObjectEntity $collection Collection entity
     * @return void
     */
    public function updateCollection(ObjectEntity $collection): void
    {
        $msg = sprintf('Updating collection "%s"', $collection->get('title'));
        $this->log($msg, 'info');
        $path = sprintf('/collections/%s', $collection->get('collection_uuid'));
        $this->client->patch($path, $this->breviaCollection($collection));
        $collection->set('collection_updated', date('c'));
        $collection->getTable()->saveOrFail($collection, ['_skipAfterSave' => true]);
    }

    /**
     * Fetch Brevia collection fields
     *
     * @param \BEdita\Core\Model\Entity\ObjectEntity $collection Collection entity
     * @return array
     */
    protected function breviaCollection(ObjectEntity $collection): array
    {
        $fields = array_diff_key(
            $collection->toArray(),
            array_flip(static::COLLECTION_FIELDS_EXCLUDED)
        ) + ['deleted' => $collection->get('deleted')];

        return [
            'name' => $collection->get('uname'),
            'cmetadata' => array_filter($fields),
        ];
    }

    /**
     * Remove Brevia collection
     *
     * @param \BEdita\Core\Model\Entity\ObjectEntity $collection Collection entity
     * @return void
     */
    public function removeCollection(ObjectEntity $collection): void
    {
        $msg = sprintf('Removing collection "%s"', $collection->get('title'));
        $this->log($msg, 'info');
        $path = sprintf('/collections/%s', $collection->get('collection_uuid'));
        $this->client->delete($path);
        $collection->set('collection_uuid', null);
    }

    /**
     * Add document to collection index
     *
     * @param \BEdita\Core\Model\Entity\ObjectEntity $collection Collection entity
     * @param \BEdita\Core\Model\Entity\ObjectEntity $entity Document entity
     * @param bool $useJob Use async job to upload files
     * @return void
     */
    public function addDocument(ObjectEntity $collection, ObjectEntity $entity, bool $useJob = true): void
    {
        if ($entity->get('status') !== 'on' || $entity->get('deleted')) {
            $msg = sprintf('Skipping doc "%s" - ', $entity->get('title')) .
                sprintf('status "%s" - deleted %b', $entity->get('status'), (bool)$entity->get('deleted'));
            $this->log($msg, 'info');

            return;
        }
        if ($entity->get('type') === 'files') {
            if ($useJob) {
                $this->uploadDocumentJob($collection, $entity);
            } else {
                $this->uploadDocument($collection, $entity);
            }

            return;
        }
        $body = [
            'collection_id' => $collection->get('collection_uuid'),
            'document_id' => (string)$entity->get('id'),
            'metadata' => $this->documentMetadata($collection, $entity),
        ];
        if ($entity->get('type') === 'links') {
            $body['link'] = (string)$entity->get('url');
            $body['options'] = Hash::get((array)$entity->get('extra'), 'brevia.options');
            $body['metadata']['url'] = $body['link'];
            $this->client->post('/index/link', array_filter($body));
        } else {
            $body['content'] = sprintf("%s\n%s", (string)$entity->get('title'), strip_tags((string)$entity->get('body')));
            $this->client->post('/index', $body);
        }
        $entity->set('index_updated', date('c'));
        $entity->set('index_status', 'done');
        $entity->getTable()->saveOrFail($entity, ['_skipAfterSave' => true]);
    }

    /**
     * Define metadata looking at current metadata and `extra.brevia.metadata`
     */
    protected function documentMetadata(ObjectEntity $collection, ObjectEntity $entity, array $addMeta = []): array
    {
        $defaultMetadata = ['type' => $entity->get('type')];
        $metadata = Hash::get((array)$entity->get('extra'), 'brevia.metadata', $defaultMetadata);

        $path = sprintf('/index/%s/%s', $collection->get('collection_uuid'), $entity->get('id'));
        $response = $this->client->get($path);

        return (array)Hash::get((array)$response->getJson(), '0.cmetadata', $metadata) + $addMeta;
    }

    /**
     * Upload file to index
     *
     * @param \BEdita\Core\Model\Entity\ObjectEntity $collection Collection entity
     * @param \BEdita\Core\Model\Entity\ObjectEntity $entity Document entity
     * @return void
     */
    public function uploadDocument(ObjectEntity $collection, ObjectEntity $entity): void
    {
        $form = new FormData();
        if (empty($entity->get('streams'))) {
            $entity->getTable()->loadInto($entity, ['Streams']);
        }
        /** @var \BEdita\Core\Model\Entity\Stream|null $stream */
        $stream = Hash::get($entity, 'streams.0');
        if (empty($stream)) {
            return;
        }
        $resource = FilesystemRegistry::getMountManager()->readStream($stream->uri);
        $file = new UploadedFile(
            $resource,
            $stream->file_size,
            UPLOAD_ERR_OK,
            $stream->file_name,
            $stream->mime_type,
        );
        $form->addFile('file', $file);
        // read options in `extra.brevia` if available
        $options = Hash::get((array)$entity->get('extra'), 'brevia.options');
        $fileMetadata = ['file' => $stream->file_name];
        $form->addMany(array_filter([
            'collection_id' => $collection->get('collection_uuid'),
            'document_id' => (string)$entity->get('id'),
            'metadata' => json_encode($this->documentMetadata($collection, $entity, $fileMetadata)),
            'options' => $options ? json_encode($options) : null,
        ]));
        $this->client->postMultipart(
            '/index/upload',
            $form
        );
        $entity->set('index_updated', date('c'));
        $entity->set('index_status', 'done');
        $entity->getTable()->saveOrFail($entity, ['_skipAfterSave' => true]);
    }

    /**
     * Create async job to upload file to index
     *
     * @param \BEdita\Core\Model\Entity\ObjectEntity $collection Collection entity
     * @param \BEdita\Core\Model\Entity\ObjectEntity $entity File entity
     * @return void
     */
    protected function uploadDocumentJob(ObjectEntity $collection, ObjectEntity $entity): void
    {
        $asyncJob = new AsyncJob([
            'service' => 'Brevia/BEdita.IndexFile',
            'max_attempts' => 3,
            'priority' => 5,
            'payload' => [
                'collection_id' => $collection->id,
                'file_id' => $entity->id,
            ],
        ]);
        $this->fetchTable('AsyncJobs')->saveOrFail($asyncJob);
        $entity->set('index_status', 'processing');
        $entity->getTable()->saveOrFail($entity, ['_skipAfterSave' => true]);
    }

    /**
     * Update collection index for a document
     *
     * @param \BEdita\Core\Model\Entity\ObjectEntity $collection Collection entity
     * @param \BEdita\Core\Model\Entity\ObjectEntity $entity Document entity
     * @param bool $forceAdd Force add document action
     * @return void
     */
    public function updateDocument(ObjectEntity $collection, ObjectEntity $entity, bool $forceAdd = false): void
    {
        if (
            $entity->isNew() ||
            ($entity->isDirty('deleted') && !$entity->get('deleted')) ||
            ($entity->isDirty('status') && $entity->get('status') === 'on') ||
            $forceAdd
        ) {
            $this->log($this->logMessage('Add', $collection, $entity), 'info');
            $this->addDocument($collection, $entity);

            return;
        }
        if (
            ($entity->isDirty('deleted') && $entity->get('deleted')) ||
            ($entity->isDirty('status') && in_array($entity->get('status'), ['draft', 'off']))
        ) {
            $this->removeDocument($collection, $entity);

            return;
        }
        // see if some object properties have changed (no effect on `files` objects)
        if ($entity->get('type') === 'files') {
            return;
        }

        foreach (static::DOCUMENT_PROPERTIES as $field) {
            if ($entity->isDirty($field)) {
                $this->log($this->logMessage('Update', $collection, $entity), 'info');
                $this->addDocument($collection, $entity);

                return;
            }
        }
    }

    /**
     * Log message on index action
     *
     * @param string $action Action to log
     * @param \BEdita\Core\Model\Entity\ObjectEntity $collection Collection entity
     * @param \BEdita\Core\Model\Entity\ObjectEntity $entity Document entity
     * @return string
     */
    protected function logMessage(string $action, ObjectEntity $collection, ObjectEntity $entity): string
    {
        return sprintf('%s document "%s"', $action, $entity->get('title')) .
            sprintf(' [collection "%s"]', $collection->get('title'));
    }

    /**
     * Remove document from collection index
     *
     * @param \BEdita\Core\Model\Entity\ObjectEntity $collection Collection entity
     * @param \BEdita\Core\Model\Entity\ObjectEntity $entity Document entity
     * @return void
     */
    public function removeDocument(ObjectEntity $collection, ObjectEntity $entity): void
    {
        $this->log($this->logMessage('Remove', $collection, $entity), 'info');
        $path = sprintf('/index/%s/%s', $collection->get('collection_uuid'), $entity->get('id'));
        $this->client->delete($path);
        $entity->set('index_status', null);
        $entity->set('index_updated', null);
        $entity->getTable()->saveOrFail($entity, ['_skipAfterSave' => true]);
    }
}


1			<?php
2			declare(strict_types=1);
3
4			/**
5			* BEdita Brevia plugin
6			*
7			* Copyright 2023 Atlas Srl
8			*/
9			namespace Brevia\BEdita\Index;
10
11			use BEdita\Core\Filesystem\FilesystemRegistry;
12			use BEdita\Core\Model\Entity\AsyncJob;
13			use BEdita\Core\Model\Entity\ObjectEntity;
14			use Brevia\BEdita\Client\BreviaClient;
15			use Cake\Http\Client\FormData;
16			use Cake\Log\LogTrait;
17			use Cake\ORM\Locator\LocatorAwareTrait;
18			use Cake\Utility\Hash;
19			use Laminas\Diactoros\UploadedFile;
20
21			/**
22			* Handle Brevia collection via API.
23			*/
24			class CollectionHandler
25			{
26			use LocatorAwareTrait;
27			use LogTrait;
28
29			/**
30			* Brevia API client
31			*
32			* @var \Brevia\BEdita\Client\BreviaClient
33			*/
34			protected BreviaClient $client;
35
36			/**
37			* List of properties to exclude when saving Brevia collection metadata
38			*
39			* @var array
40			*/
41			public const COLLECTION_FIELDS_EXCLUDED = [
42			'uname',
43			'type',
44			'created',
45			'modified',
46			'lang',
47			'locked',
48			'published',
49			'created_by',
50			'modified_by',
51			'collection_uuid',
52			'collection_updated',
53			'_meta',
54			'_joinData',
55			];
56
57			/**
58			* List of properties to check when updating a Brevia collection index
59			*
60			* @var array
61			*/
62			public const DOCUMENT_PROPERTIES = [
63			'title',
64			'description',
65			'body',
66			'url',
67			];
68
69			/**
70			* Handler constructor
71			*/
72			public function __construct()
73			{
74			$this->client = new BreviaClient();
75			}
76
77			/**
78			* Create Brevia collection
79			*
80			* @param \BEdita\Core\Model\Entity\ObjectEntity $collection Collection entity
81			* @return void
82			*/
83			public function createCollection(ObjectEntity $collection): void
84			{
85			$msg = sprintf('Creating collection "%s"', $collection->get('title'));
86			$this->log($msg, 'info');
87
88			$response = $this->client->post('/collections', $this->breviaCollection($collection));
89			$collection->set('collection_uuid', Hash::get($response->getJson(), 'uuid'));
90			$collection->set('collection_updated', date('c'));
91			$collection->getTable()->saveOrFail($collection, ['_skipAfterSave' => true]);
92			}
93
94			/**
95			* Update Brevia collection
96			*
97			* @param \BEdita\Core\Model\Entity\ObjectEntity $collection Collection entity
98			* @return void
99			*/
100			public function updateCollection(ObjectEntity $collection): void
101			{
102			$msg = sprintf('Updating collection "%s"', $collection->get('title'));
103			$this->log($msg, 'info');
104			$path = sprintf('/collections/%s', $collection->get('collection_uuid'));
105			$this->client->patch($path, $this->breviaCollection($collection));
106			$collection->set('collection_updated', date('c'));
107			$collection->getTable()->saveOrFail($collection, ['_skipAfterSave' => true]);
108			}
109
110			/**
111			* Fetch Brevia collection fields
112			*
113			* @param \BEdita\Core\Model\Entity\ObjectEntity $collection Collection entity
114			* @return array
115			*/
116			protected function breviaCollection(ObjectEntity $collection): array
117			{
118			$fields = array_diff_key(
119			$collection->toArray(),
120			array_flip(static::COLLECTION_FIELDS_EXCLUDED)
121			) + ['deleted' => $collection->get('deleted')];
122
123			return [
124			'name' => $collection->get('uname'),
125			'cmetadata' => array_filter($fields),
126			];
127			}
128
129			/**
130			* Remove Brevia collection
131			*
132			* @param \BEdita\Core\Model\Entity\ObjectEntity $collection Collection entity
133			* @return void
134			*/
135			public function removeCollection(ObjectEntity $collection): void
136			{
137			$msg = sprintf('Removing collection "%s"', $collection->get('title'));
138			$this->log($msg, 'info');
139			$path = sprintf('/collections/%s', $collection->get('collection_uuid'));
140			$this->client->delete($path);
141			$collection->set('collection_uuid', null);
142			}
143
144			/**
145			* Add document to collection index
146			*
147			* @param \BEdita\Core\Model\Entity\ObjectEntity $collection Collection entity
148			* @param \BEdita\Core\Model\Entity\ObjectEntity $entity Document entity
149			* @param bool $useJob Use async job to upload files
150			* @return void
151			*/
152			public function addDocument(ObjectEntity $collection, ObjectEntity $entity, bool $useJob = true): void
153			{
154			if ($entity->get('status') !== 'on' \|\| $entity->get('deleted')) {
155			$msg = sprintf('Skipping doc "%s" - ', $entity->get('title')) .
156			sprintf('status "%s" - deleted %b', $entity->get('status'), (bool)$entity->get('deleted'));
157			$this->log($msg, 'info');
158
159			return;
160			}
161			if ($entity->get('type') === 'files') {
162			if ($useJob) {
163			$this->uploadDocumentJob($collection, $entity);
164			} else {
165			$this->uploadDocument($collection, $entity);
166			}
167
168			return;
169			}
170			$body = [
171			'collection_id' => $collection->get('collection_uuid'),
172			'document_id' => (string)$entity->get('id'),
173			'metadata' => $this->documentMetadata($collection, $entity),
174			];
175			if ($entity->get('type') === 'links') {
176			$body['link'] = (string)$entity->get('url');
177			$body['options'] = Hash::get((array)$entity->get('extra'), 'brevia.options');
178			$body['metadata']['url'] = $body['link'];
179			$this->client->post('/index/link', array_filter($body));
180			} else {
181			$body['content'] = sprintf("%s\n%s", (string)$entity->get('title'), strip_tags((string)$entity->get('body')));
182			$this->client->post('/index', $body);
183			}
184			$entity->set('index_updated', date('c'));
185			$entity->set('index_status', 'done');
186			$entity->getTable()->saveOrFail($entity, ['_skipAfterSave' => true]);
187			}
188
189			/**
190			* Define metadata looking at current metadata and `extra.brevia.metadata`
191			*/
192			protected function documentMetadata(ObjectEntity $collection, ObjectEntity $entity, array $addMeta = []): array
193			{
194			$defaultMetadata = ['type' => $entity->get('type')];
195			$metadata = Hash::get((array)$entity->get('extra'), 'brevia.metadata', $defaultMetadata);
196
197			$path = sprintf('/index/%s/%s', $collection->get('collection_uuid'), $entity->get('id'));
198			$response = $this->client->get($path);
199
200			return (array)Hash::get((array)$response->getJson(), '0.cmetadata', $metadata) + $addMeta;
201			}
202
203			/**
204			* Upload file to index
205			*
206			* @param \BEdita\Core\Model\Entity\ObjectEntity $collection Collection entity
207			* @param \BEdita\Core\Model\Entity\ObjectEntity $entity Document entity
208			* @return void
209			*/
210			public function uploadDocument(ObjectEntity $collection, ObjectEntity $entity): void
211			{
212			$form = new FormData();
213			if (empty($entity->get('streams'))) {
214			$entity->getTable()->loadInto($entity, ['Streams']);
215			}
216			/** @var \BEdita\Core\Model\Entity\Stream\|null $stream */
217			$stream = Hash::get($entity, 'streams.0');
218			if (empty($stream)) {
219			return;
220			}
221			$resource = FilesystemRegistry::getMountManager()->readStream($stream->uri);
222			$file = new UploadedFile(
223			$resource,
224			$stream->file_size,
225			UPLOAD_ERR_OK,
226			$stream->file_name,
227			$stream->mime_type,
228			);
229			$form->addFile('file', $file);
230			// read options in `extra.brevia` if available
231			$options = Hash::get((array)$entity->get('extra'), 'brevia.options');
232			$fileMetadata = ['file' => $stream->file_name];
233			$form->addMany(array_filter([
234			'collection_id' => $collection->get('collection_uuid'),
235			'document_id' => (string)$entity->get('id'),
236			'metadata' => json_encode($this->documentMetadata($collection, $entity, $fileMetadata)),
237			'options' => $options ? json_encode($options) : null,
238			]));
239			$this->client->postMultipart(
240			'/index/upload',
241			$form
242			);
243			$entity->set('index_updated', date('c'));
244			$entity->set('index_status', 'done');
245			$entity->getTable()->saveOrFail($entity, ['_skipAfterSave' => true]);
246			}
247
248			/**
249			* Create async job to upload file to index
250			*
251			* @param \BEdita\Core\Model\Entity\ObjectEntity $collection Collection entity
252			* @param \BEdita\Core\Model\Entity\ObjectEntity $entity File entity
253			* @return void
254			*/
255			protected function uploadDocumentJob(ObjectEntity $collection, ObjectEntity $entity): void
256			{
257			$asyncJob = new AsyncJob([
258			'service' => 'Brevia/BEdita.IndexFile',
259			'max_attempts' => 3,
260			'priority' => 5,
261			'payload' => [
262			'collection_id' => $collection->id,
263			'file_id' => $entity->id,
264			],
265			]);
266			$this->fetchTable('AsyncJobs')->saveOrFail($asyncJob);
267			$entity->set('index_status', 'processing');
268			$entity->getTable()->saveOrFail($entity, ['_skipAfterSave' => true]);
269			}
270
271			/**
272			* Update collection index for a document
273			*
274			* @param \BEdita\Core\Model\Entity\ObjectEntity $collection Collection entity
275			* @param \BEdita\Core\Model\Entity\ObjectEntity $entity Document entity
276			* @param bool $forceAdd Force add document action
277			* @return void
278			*/
279			public function updateDocument(ObjectEntity $collection, ObjectEntity $entity, bool $forceAdd = false): void
280			{
281			if (
282			$entity->isNew() \|\|
283			($entity->isDirty('deleted') && !$entity->get('deleted')) \|\|
284			($entity->isDirty('status') && $entity->get('status') === 'on') \|\|
285			$forceAdd
286			) {
287			$this->log($this->logMessage('Add', $collection, $entity), 'info');
288			$this->addDocument($collection, $entity);
289
290			return;
291			}
292			if (
293			($entity->isDirty('deleted') && $entity->get('deleted')) \|\|
294			($entity->isDirty('status') && in_array($entity->get('status'), ['draft', 'off']))
295			) {
296			$this->removeDocument($collection, $entity);
297
298			return;
299			}
300			// see if some object properties have changed (no effect on `files` objects)
301			if ($entity->get('type') === 'files') {
302			return;
303			}
304
305			foreach (static::DOCUMENT_PROPERTIES as $field) {
306			if ($entity->isDirty($field)) {
307			$this->log($this->logMessage('Update', $collection, $entity), 'info');
308			$this->addDocument($collection, $entity);
309
310			return;
311			}
312			}
313			}
314
315			/**
316			* Log message on index action
317			*
318			* @param string $action Action to log
319			* @param \BEdita\Core\Model\Entity\ObjectEntity $collection Collection entity
320			* @param \BEdita\Core\Model\Entity\ObjectEntity $entity Document entity
321			* @return string
322			*/
323			protected function logMessage(string $action, ObjectEntity $collection, ObjectEntity $entity): string
324			{
325			return sprintf('%s document "%s"', $action, $entity->get('title')) .
326			sprintf(' [collection "%s"]', $collection->get('title'));
327			}
328
329			/**
330			* Remove document from collection index
331			*
332			* @param \BEdita\Core\Model\Entity\ObjectEntity $collection Collection entity
333			* @param \BEdita\Core\Model\Entity\ObjectEntity $entity Document entity
334			* @return void
335			*/
336			public function removeDocument(ObjectEntity $collection, ObjectEntity $entity): void
337			{
338			$this->log($this->logMessage('Remove', $collection, $entity), 'info');
339			$path = sprintf('/index/%s/%s', $collection->get('collection_uuid'), $entity->get('id'));
340			$this->client->delete($path);
341			$entity->set('index_status', null);
342			$entity->set('index_updated', null);
343			$entity->getTable()->saveOrFail($entity, ['_skipAfterSave' => true]);
344			}
345			}
346

brevia-ai / brevia-bedita

Push — main ( 6b3c3f...a8a095 )

CollectionHandler::addDocument() B

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like