Indexer::addMetaKeys() - Code Metrics - Inspection of "Refactor fulltext search functions and class Doku_..." - splitbrain/dokuwiki - Measure and Improve Code Quality continuously with Scrutinizer

Failed Conditions

Pull Request — master (#2943)

unknown

created 2020-09-21 13:42 UTC

Indexer::addMetaKeys() F

↳ Parent: Indexer

Complexity

Conditions	20
Paths	3471

Size

Total Lines

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
cc	20
nc	3471
nop	3
dl	0
loc	94
rs	0
c	0
b	0
f	0

1 Method

Rating	Name	Duplication	Size	Complexity
B	Indexer::deletePage()	0	35	11

How to fix Long Method Complexity

<?php

namespace dokuwiki\Search;

use dokuwiki\Extension\Event;
use dokuwiki\Search\Exception\IndexAccessException;
use dokuwiki\Search\Exception\IndexLockException;
use dokuwiki\Search\Exception\IndexWriteException;

// Version tag used to force rebuild on upgrade
const INDEXER_VERSION = 8;

/**
 * Class DokuWiki Indexer (Singleton)
 *
 * @license    GPL 2 (http://www.gnu.org/licenses/gpl.html)
 * @author     Andreas Gohr <[email protected]>
 * @author Tom N Harris <[email protected]>
 */
class Indexer extends AbstractIndex
{
    /** @var Indexer $instance */
    protected static $instance = null;

    /**
     * Get new or existing singleton instance of the Indexer
     *
     * @return Indexer
     */
    public static function getInstance()
    {
        if (is_null(static::$instance)) {
            static::$instance = new static();
        }
        return static::$instance;
    }

    /**
     * Dispatch Indexing request for the page, called by TaskRunner::runIndexer()
     *
     * @param string $page name of the page to index
     * @param bool $verbose print status messages
     * @param bool $force force reindexing even when the index is up to date
     * @return bool  If the function completed successfully
     *
     * @throws IndexLockException
     * @throws IndexWriteException
     * @author Satoshi Sahara <[email protected]>
     * @author Tom N Harris <[email protected]>
     */
    public function dispatch($page, $verbose = false, $force = false)
    {
        // check if page was deleted but is still in the index
        if (!page_exists($page)) {
            return $this->deletePage($page, $verbose, $force);
        }

        // update search index
        return $this->addPage($page, $verbose, $force);
    }

    /**
     * Version of the indexer taking into consideration the external tokenizer.
     * The indexer is only compatible with data written by the same version.
     *
     * @triggers INDEXER_VERSION_GET
     * Plugins that modify what gets indexed should hook this event and
     * add their version info to the event data like so:
     *     $data[$plugin_name] = $plugin_version;
     *
     * @author Tom N Harris <[email protected]>
     * @author Michael Hamann <[email protected]>
     *
     * @return int|string
     */
    public function getVersion()
    {
        static $indexer_version = null;
        if ($indexer_version == null) {
            $version = INDEXER_VERSION;

            // DokuWiki version is included for the convenience of plugins
            $data = array('dokuwiki' => $version);
            Event::createAndTrigger('INDEXER_VERSION_GET', $data, null, false);
            unset($data['dokuwiki']); // this needs to be first
            ksort($data);
            foreach ($data as $plugin => $vers) {
                $version .= '+'.$plugin.'='.$vers;
            }
            $indexer_version = $version;
        }
        return $indexer_version;
    }

    /**
     * Adds/updates the search index for the given page
     *
     * Locking is handled internally.
     *
     * @param string $page name of the page to index
     * @param bool $verbose print status messages
     * @param bool $force force reindexing even when the index is up to date
     * @return bool  If the function completed successfully
     *
     * @throws IndexLockException
     * @throws IndexWriteException
     * @author Satoshi Sahara <[email protected]>
     * @author Tom N Harris <[email protected]>
     */
    public function addPage($page, $verbose = false, $force = false)
    {
        // check if indexing needed for the existing page (full text and/or metadata indexing)
        $idxtag = metaFN($page,'.indexed');
        if (!$force && file_exists($idxtag)) {
            if (trim(io_readFile($idxtag)) == $this->getVersion()) {
                $last = @filemtime($idxtag);
                if ($last > @filemtime(wikiFN($page))) {
                    if ($verbose) dbglog("Indexer: index for {$page} up to date");
                    return true;
                }
            }
        }

        // register the page to the page.idx file, $pid is always numeric
        $pid = $this->getPID($page);

        // prepare metadata indexing
        $metadata = array();
        $metadata['title'] = p_get_metadata($page, 'title', METADATA_RENDER_UNLIMITED);

        $references = p_get_metadata($page, 'relation references', METADATA_RENDER_UNLIMITED);
        $metadata['relation_references'] = ($references !== null) ?
                array_keys($references) : array();

        $media = p_get_metadata($page, 'relation media', METADATA_RENDER_UNLIMITED);
        $metadata['relation_media'] = ($media !== null) ?
                array_keys($media) : array();

        // check if full text indexing allowed
        $indexenabled = p_get_metadata($page, 'internal index', METADATA_RENDER_UNLIMITED);
        if ($indexenabled !== false) $indexenabled = true;
        $metadata['internal_index'] = $indexenabled;

        $body = '';
        $data = compact('page', 'body', 'metadata', 'pid');
        $event = new Event('INDEXER_PAGE_ADD', $data);
        if ($event->advise_before()) $data['body'] = $data['body'].' '.rawWiki($page);
        $event->advise_after();
        unset($event);
        extract($data);
        $indexenabled = $metadata['internal_index'];
        unset($metadata['internal_index']);

        // Access to Metadata Index
        $MetadataIndex = MetadataIndex::getInstance();
        $result = $MetadataIndex->addMetaKeys($page, $metadata);
        if ($verbose) dbglog("Indexer: addMetaKeys({$page}) ".($result ? 'done' : 'failed'));
        if (!$result) {
            return false;
        }

        // Access to Fulltext Index
        $FulltextIndex = FulltextIndex::getInstance();
        if ($indexenabled) {
            $result = $FulltextIndex->addPagewords($page, $body);
            if ($verbose) dbglog("Indexer: addPageWords({$page}) ".($result ? 'done' : 'failed'));
            if (!$result) {
                return false;
            }
        } else {
            if ($verbose) dbglog("Indexer: full text indexing disabled for {$page}");
            // ensure the page content deleted from the Fulltext index
            $result = $FulltextIndex->deletePageWords($page);
            if ($verbose) dbglog("Indexer: deletePageWords({$page}) ".($result ? 'done' : 'failed'));
            if (!$result) {
                return false;
            }
        }

        // update index tag file
        io_saveFile($idxtag, $this->getVersion());
        if ($verbose) dbglog("Indexer: finished");

        return $result;
    }

    /**
     * Remove a page from the index
     *
     * Erases entries in all known indexes. Locking is handled internally.
     *
     * @param string $page name of the page to index
     * @param bool $verbose print status messages
     * @param bool $force force reindexing even when the index is up to date
     * @return bool  If the function completed successfully
     *
     * @throws IndexLockException
     * @throws IndexWriteException
     * @author Satoshi Sahara <[email protected]>
     * @author Tom N Harris <[email protected]>
     */
    public function deletePage($page, $verbose = false, $force = false)
    {
        $idxtag = metaFN($page,'.indexed');
        if (!$force && !file_exists($idxtag)) {
            if ($verbose) dbglog("Indexer: {$page}.indexed file does not exist, ignoring");
            return true;
        }

        // remove obsoleted content from Fulltext index
        $FulltextIndex = FulltextIndex::getInstance();
        $result = $FulltextIndex->deletePageWords($page);
        if ($verbose) dbglog("Indexer: deletePageWords({$page}) ".($result ? 'done' : 'failed'));
        if (!$result) {
            return false;
        }

        // delete all keys of the page from metadata index
        $MetadataIndex = MetadataIndex::getInstance();
        $result = $MetadataIndex->deleteMetaKeys($page);
        if ($verbose) dbglog("Indexer: deleteMetaKeys({$page}) ".($result ? 'done' : 'failed'));
        if (!$result) {
            return false;
        }

        // mark the page as deleted in the page.idx
        $pid = $this->getPID($page);
        $this->lock();
        $this->saveIndexKey('page', '', $pid, self::INDEX_MARK_DELETED.$page);
        if ($verbose) dbglog("Indexer: {$page} has marked as deleted in page.idx");
        $this->unlock();

        unset(static::$pidCache[$pid]);
        @unlink($idxtag);
        return $result;
    }

    /**
     * Rename a page in the search index without changing the indexed content.
     * This function doesn't check if the old or new name exists in the filesystem.
     * It returns an error if the old page isn't in the page list of the indexer
     * and it deletes all previously indexed content of the new page.
     *
     * @param string $oldpage The old page name
     * @param string $newpage The new page name
     * @return bool  If the page was successfully renamed
     * @throws IndexLockException
     * @throws IndexWriteException
     */
    public function renamePage($oldpage, $newpage)
    {
        $index = $this->getIndex('page', '');
        // check if oldpage found in page.idx
        $oldPid = array_search($oldpage, $index, true);
        if ($oldPid === false) return false;

        // check if newpage found in page.idx
        $newPid = array_search($newpage, $index, true);
        if ($newPid !== false) {
            $result = $this->deletePage($newpage);
            if (!$result) return false;
            // Note: $index is no longer valid after deletePage()!
            unset($index);
        }

        // update page.idx
        $this->lock();
        $this->saveIndexKey('page', '', $oldPid, $newpage);
        $this->unlock();

        // reset the pid cache
        $this->resetPIDCache();

        return true;
    }

    /**
     * Clear the Page Index
     *
     * @param bool $requireLock should be false only if the caller is resposible for index lock
     * @return bool  If the index has been cleared successfully
     * @throws Exception\IndexLockException
     */
    public function clear($requireLock = true)
    {
        global $conf;

        if ($requireLock) $this->lock();

        // clear Metadata Index
        $MetadataIndex = MetadataIndex::getInstance();
        $MetadataIndex->clear(false);

        // clear Fulltext Index
        $FulltextIndex = FulltextIndex::getInstance();
        $FulltextIndex->clear(false);

        @unlink($conf['indexdir'].'/page.idx');

        // clear the pid cache
        $this->resetPIDCache();

        if ($requireLock) $this->unlock();
        return true;
    }

}


1			<?php
2
3			namespace dokuwiki\Search;
4
5			use dokuwiki\Extension\Event;
6			use dokuwiki\Search\Exception\IndexAccessException;
7			use dokuwiki\Search\Exception\IndexLockException;
8			use dokuwiki\Search\Exception\IndexWriteException;
9
10			// Version tag used to force rebuild on upgrade
11			const INDEXER_VERSION = 8;
12
13			/**
14			* Class DokuWiki Indexer (Singleton)
15			*
16			* @license GPL 2 (http://www.gnu.org/licenses/gpl.html)
17			* @author Andreas Gohr <[email protected]>
18			* @author Tom N Harris <[email protected]>
19			*/
20			class Indexer extends AbstractIndex
21			{
22			/** @var Indexer $instance */
23			protected static $instance = null;
24
25			/**
26			* Get new or existing singleton instance of the Indexer
27			*
28			* @return Indexer
29			*/
30			public static function getInstance()
31			{
32			if (is_null(static::$instance)) {
33			static::$instance = new static();
34			}
35			return static::$instance;
36			}
37
38			/**
39			* Dispatch Indexing request for the page, called by TaskRunner::runIndexer()
40			*
41			* @param string $page name of the page to index
42			* @param bool $verbose print status messages
43			* @param bool $force force reindexing even when the index is up to date
44			* @return bool If the function completed successfully
45			*
46			* @throws IndexLockException
47			* @throws IndexWriteException
48			* @author Satoshi Sahara <[email protected]>
49			* @author Tom N Harris <[email protected]>
50			*/
51			public function dispatch($page, $verbose = false, $force = false)
52			{
53			// check if page was deleted but is still in the index
54			if (!page_exists($page)) {
55			return $this->deletePage($page, $verbose, $force);
56			}
57
58			// update search index
59			return $this->addPage($page, $verbose, $force);
60			}
61
62			/**
63			* Version of the indexer taking into consideration the external tokenizer.
64			* The indexer is only compatible with data written by the same version.
65			*
66			* @triggers INDEXER_VERSION_GET
67			* Plugins that modify what gets indexed should hook this event and
68			* add their version info to the event data like so:
69			* $data[$plugin_name] = $plugin_version;
70			*
71			* @author Tom N Harris <[email protected]>
72			* @author Michael Hamann <[email protected]>
73			*
74			* @return int\|string
75			*/
76			public function getVersion()
77			{
78			static $indexer_version = null;
79			if ($indexer_version == null) {
80			$version = INDEXER_VERSION;
81
82			// DokuWiki version is included for the convenience of plugins
83			$data = array('dokuwiki' => $version);
84			Event::createAndTrigger('INDEXER_VERSION_GET', $data, null, false);
85			unset($data['dokuwiki']); // this needs to be first
86			ksort($data);
87			foreach ($data as $plugin => $vers) {
88			$version .= '+'.$plugin.'='.$vers;
89			}
90			$indexer_version = $version;
91			}
92			return $indexer_version;
93			}
94
95			/**
96			* Adds/updates the search index for the given page
97			*
98			* Locking is handled internally.
99			*
100			* @param string $page name of the page to index
101			* @param bool $verbose print status messages
102			* @param bool $force force reindexing even when the index is up to date
103			* @return bool If the function completed successfully
104			*
105			* @throws IndexLockException
106			* @throws IndexWriteException
107			* @author Satoshi Sahara <[email protected]>
108			* @author Tom N Harris <[email protected]>
109			*/
110			public function addPage($page, $verbose = false, $force = false)
111			{
112			// check if indexing needed for the existing page (full text and/or metadata indexing)
113			$idxtag = metaFN($page,'.indexed');
114			if (!$force && file_exists($idxtag)) {
115			if (trim(io_readFile($idxtag)) == $this->getVersion()) {
116			$last = @filemtime($idxtag);
117			if ($last > @filemtime(wikiFN($page))) {
118			if ($verbose) dbglog("Indexer: index for {$page} up to date");
119			return true;
120			}
121			}
122			}
123
124			// register the page to the page.idx file, $pid is always numeric
125			$pid = $this->getPID($page);
126
127			// prepare metadata indexing
128			$metadata = array();
129			$metadata['title'] = p_get_metadata($page, 'title', METADATA_RENDER_UNLIMITED);
130
131			$references = p_get_metadata($page, 'relation references', METADATA_RENDER_UNLIMITED);
132			$metadata['relation_references'] = ($references !== null) ?
133			array_keys($references) : array();
134
135			$media = p_get_metadata($page, 'relation media', METADATA_RENDER_UNLIMITED);
136			$metadata['relation_media'] = ($media !== null) ?
137			array_keys($media) : array();
138
139			// check if full text indexing allowed
140			$indexenabled = p_get_metadata($page, 'internal index', METADATA_RENDER_UNLIMITED);
141			if ($indexenabled !== false) $indexenabled = true;
142			$metadata['internal_index'] = $indexenabled;
143
144			$body = '';
145			$data = compact('page', 'body', 'metadata', 'pid');
146			$event = new Event('INDEXER_PAGE_ADD', $data);
147			if ($event->advise_before()) $data['body'] = $data['body'].' '.rawWiki($page);
148			$event->advise_after();
149			unset($event);
150			extract($data);
151			$indexenabled = $metadata['internal_index'];
152			unset($metadata['internal_index']);
153
154			// Access to Metadata Index
155			$MetadataIndex = MetadataIndex::getInstance();
156			$result = $MetadataIndex->addMetaKeys($page, $metadata);
157			if ($verbose) dbglog("Indexer: addMetaKeys({$page}) ".($result ? 'done' : 'failed'));
158			if (!$result) {
159			return false;
160			}
161
162			// Access to Fulltext Index
163			$FulltextIndex = FulltextIndex::getInstance();
164			if ($indexenabled) {
165			$result = $FulltextIndex->addPagewords($page, $body);
166			if ($verbose) dbglog("Indexer: addPageWords({$page}) ".($result ? 'done' : 'failed'));
167			if (!$result) {
168			return false;
169			}
170			} else {
171			if ($verbose) dbglog("Indexer: full text indexing disabled for {$page}");
172			// ensure the page content deleted from the Fulltext index
173			$result = $FulltextIndex->deletePageWords($page);
174			if ($verbose) dbglog("Indexer: deletePageWords({$page}) ".($result ? 'done' : 'failed'));
175			if (!$result) {
176			return false;
177			}
178			}
179
180			// update index tag file
181			io_saveFile($idxtag, $this->getVersion());
182			if ($verbose) dbglog("Indexer: finished");
183
184			return $result;
185			}
186
187			/**
188			* Remove a page from the index
189			*
190			* Erases entries in all known indexes. Locking is handled internally.
191			*
192			* @param string $page name of the page to index
193			* @param bool $verbose print status messages
194			* @param bool $force force reindexing even when the index is up to date
195			* @return bool If the function completed successfully
196			*
197			* @throws IndexLockException
198			* @throws IndexWriteException
199			* @author Satoshi Sahara <[email protected]>
200			* @author Tom N Harris <[email protected]>
201			*/
202			public function deletePage($page, $verbose = false, $force = false)
203			{
204			$idxtag = metaFN($page,'.indexed');
205			if (!$force && !file_exists($idxtag)) {
206			if ($verbose) dbglog("Indexer: {$page}.indexed file does not exist, ignoring");
207			return true;
208			}
209
210			// remove obsoleted content from Fulltext index
211			$FulltextIndex = FulltextIndex::getInstance();
212			$result = $FulltextIndex->deletePageWords($page);
213			if ($verbose) dbglog("Indexer: deletePageWords({$page}) ".($result ? 'done' : 'failed'));
214			if (!$result) {
215			return false;
216			}
217
218			// delete all keys of the page from metadata index
219			$MetadataIndex = MetadataIndex::getInstance();
220			$result = $MetadataIndex->deleteMetaKeys($page);
221			if ($verbose) dbglog("Indexer: deleteMetaKeys({$page}) ".($result ? 'done' : 'failed'));
222			if (!$result) {
223			return false;
224			}
225
226			// mark the page as deleted in the page.idx
227			$pid = $this->getPID($page);
228			$this->lock();
229			$this->saveIndexKey('page', '', $pid, self::INDEX_MARK_DELETED.$page);
230			if ($verbose) dbglog("Indexer: {$page} has marked as deleted in page.idx");
231			$this->unlock();
232
233			unset(static::$pidCache[$pid]);
234			@unlink($idxtag);
235			return $result;
236			}
237
238			/**
239			* Rename a page in the search index without changing the indexed content.
240			* This function doesn't check if the old or new name exists in the filesystem.
241			* It returns an error if the old page isn't in the page list of the indexer
242			* and it deletes all previously indexed content of the new page.
243			*
244			* @param string $oldpage The old page name
245			* @param string $newpage The new page name
246			* @return bool If the page was successfully renamed
247			* @throws IndexLockException
248			* @throws IndexWriteException
249			*/
250			public function renamePage($oldpage, $newpage)
251			{
252			$index = $this->getIndex('page', '');
253			// check if oldpage found in page.idx
254			$oldPid = array_search($oldpage, $index, true);
255			if ($oldPid === false) return false;
256
257			// check if newpage found in page.idx
258			$newPid = array_search($newpage, $index, true);
259			if ($newPid !== false) {
260			$result = $this->deletePage($newpage);
261			if (!$result) return false;
262			// Note: $index is no longer valid after deletePage()!
263			unset($index);
264			}
265
266			// update page.idx
267			$this->lock();
268			$this->saveIndexKey('page', '', $oldPid, $newpage);
269			$this->unlock();
270
271			// reset the pid cache
272			$this->resetPIDCache();
273
274			return true;
275			}
276
277			/**
278			* Clear the Page Index
279			*
280			* @param bool $requireLock should be false only if the caller is resposible for index lock
281			* @return bool If the index has been cleared successfully
282			* @throws Exception\IndexLockException
283			*/
284			public function clear($requireLock = true)
285			{
286			global $conf;
287
288			if ($requireLock) $this->lock();
289
290			// clear Metadata Index
291			$MetadataIndex = MetadataIndex::getInstance();
292			$MetadataIndex->clear(false);
293
294			// clear Fulltext Index
295			$FulltextIndex = FulltextIndex::getInstance();
296			$FulltextIndex->clear(false);
297
298			@unlink($conf['indexdir'].'/page.idx');
299
300			// clear the pid cache
301			$this->resetPIDCache();
302
303			if ($requireLock) $this->unlock();
304			return true;
305			}
306
307			}
308

splitbrain / dokuwiki

Pull Request — master (#2943)

Indexer::addMetaKeys() F

Complexity

Size

Duplication

Importance

1 Method

How to fix Long Method Complexity

Long Method

Duplication Side-by-Side

Filter issues like