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 2020-11-10 10:54 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	41	10

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
 *
 * @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
{
    // page to be indexed
    protected $page;

    /**
     * Indexer constructor
     *
     * @param string $page name of the page to index
     */
    public function __construct($page = null)
    {
        if (isset($page)) $this->page = $page;
    }

    /**
     * Dispatch Indexing request for the page, called by TaskRunner::runIndexer()
     *
     * @param bool $verbose print status messages
     * @param bool $force force reindexing even when the index is up to date
     * @return bool  If the function completed successfully
     *
     * @throws IndexAccessException
     * @throws IndexLockException
     * @throws IndexWriteException
     * @author Satoshi Sahara <[email protected]>
     * @author Tom N Harris <[email protected]>
     */
    public function dispatch($verbose = false, $force = false)
    {
        if (!isset($this->page)) {
            throw new IndexAccessException('Indexer: unknow page name');
        }

        // check if page was deleted but is still in the index
        if (!page_exists($this->page)) {
            return $this->deletePage($verbose, $force);
        }

        // update search index
        return $this->addPage($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 bool $verbose print status messages
     * @param bool $force force reindexing even when the index is up to date
     * @return bool  If the function completed successfully
     *
     * @throws IndexAccessException
     * @throws IndexLockException
     * @throws IndexWriteException
     * @author Satoshi Sahara <[email protected]>
     * @author Tom N Harris <[email protected]>
     */
    public function addPage($verbose = false, $force = false)
    {
        if (!isset($this->page)) {
            throw new IndexAccessException('Indexer: invalid page name in addePage');
        } else {
            $page = $this->page;
        }

        // 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 integer
        $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
        $result = (new MetadataIndex($pid))->addMetaKeys($metadata);
        if ($verbose) dbglog("Indexer: addMetaKeys({$page}) ".($result ? 'done' : 'failed'));
        if (!$result) {
            return false;
        }

        // Access to Fulltext Index
        if ($indexenabled) {
            $result = (new FulltextIndex($pid))->addWords($body);
            if ($verbose) dbglog("Indexer: addWords() for {$page} done");
            if (!$result) {
                return false;
            }
        } else {
            if ($verbose) dbglog("Indexer: full text indexing disabled for {$page}");
            // ensure the page content deleted from the Fulltext index
            $result = (new FulltextIndex($page))->deleteWords();
            if ($verbose) dbglog("Indexer: deleteWords() for {$page} done");
            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 bool $verbose print status messages
     * @param bool $force force reindexing even when the index is up to date
     * @return bool  If the function completed successfully
     *
     * @throws IndexAccessException
     * @throws IndexLockException
     * @throws IndexWriteException
     * @author Satoshi Sahara <[email protected]>
     * @author Tom N Harris <[email protected]>
     */
    public function deletePage($verbose = false, $force = false)
    {
        if (!isset($this->page)) {
            throw new IndexAccessException('Indexer: invalid page name in deletePage');
        } else {
            $page = $this->page;
        }

        $idxtag = metaFN($page,'.indexed');
        if (!$force && !file_exists($idxtag)) {
            if ($verbose) dbglog("Indexer: {$page}.indexed file does not exist, ignoring");
            return true;
        }

        // retrieve pid from the page.idx file, $pid is always integer
        $pid = $this->getPID($page);

        // remove obsoleted content from Fulltext index
        $result = (new FulltextIndex($pid))->deleteWords();
        if ($verbose) dbglog("Indexer: deleteWords() for {$page} done");
        if (!$result) {
            return false;
        }

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

        // mark the page as deleted in the page.idx
        $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) {
            $result = (new Indexer($newpage))->deletePage();
            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
        (new MetadataIndex())->clear(false);

        // clear Fulltext Index
        (new 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
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			// page to be indexed
23			protected $page;
24
25			/**
26			* Indexer constructor
27			*
28			* @param string $page name of the page to index
29			*/
30			public function __construct($page = null)
31			{
32			if (isset($page)) $this->page = $page;
33			}
34
35			/**
36			* Dispatch Indexing request for the page, called by TaskRunner::runIndexer()
37			*
38			* @param bool $verbose print status messages
39			* @param bool $force force reindexing even when the index is up to date
40			* @return bool If the function completed successfully
41			*
42			* @throws IndexAccessException
43			* @throws IndexLockException
44			* @throws IndexWriteException
45			* @author Satoshi Sahara <[email protected]>
46			* @author Tom N Harris <[email protected]>
47			*/
48			public function dispatch($verbose = false, $force = false)
49			{
50			if (!isset($this->page)) {
51			throw new IndexAccessException('Indexer: unknow page name');
52			}
53
54			// check if page was deleted but is still in the index
55			if (!page_exists($this->page)) {
56			return $this->deletePage($verbose, $force);
57			}
58
59			// update search index
60			return $this->addPage($verbose, $force);
61			}
62
63			/**
64			* Version of the indexer taking into consideration the external tokenizer.
65			* The indexer is only compatible with data written by the same version.
66			*
67			* @triggers INDEXER_VERSION_GET
68			* Plugins that modify what gets indexed should hook this event and
69			* add their version info to the event data like so:
70			* $data[$plugin_name] = $plugin_version;
71			*
72			* @author Tom N Harris <[email protected]>
73			* @author Michael Hamann <[email protected]>
74			*
75			* @return int\|string
76			*/
77			public function getVersion()
78			{
79			static $indexer_version = null;
80			if ($indexer_version == null) {
81			$version = INDEXER_VERSION;
82
83			// DokuWiki version is included for the convenience of plugins
84			$data = array('dokuwiki' => $version);
85			Event::createAndTrigger('INDEXER_VERSION_GET', $data, null, false);
86			unset($data['dokuwiki']); // this needs to be first
87			ksort($data);
88			foreach ($data as $plugin => $vers) {
89			$version .= '+'.$plugin.'='.$vers;
90			}
91			$indexer_version = $version;
92			}
93			return $indexer_version;
94			}
95
96			/**
97			* Adds/updates the search index for the given page
98			*
99			* Locking is handled internally.
100			*
101			* @param bool $verbose print status messages
102			* @param bool $force force reindexing even when the index is up to date
103			* @return bool If the function completed successfully
104			*
105			* @throws IndexAccessException
106			* @throws IndexLockException
107			* @throws IndexWriteException
108			* @author Satoshi Sahara <[email protected]>
109			* @author Tom N Harris <[email protected]>
110			*/
111			public function addPage($verbose = false, $force = false)
112			{
113			if (!isset($this->page)) {
114			throw new IndexAccessException('Indexer: invalid page name in addePage');
115			} else {
116			$page = $this->page;
117			}
118
119			// check if indexing needed for the existing page (full text and/or metadata indexing)
120			$idxtag = metaFN($page,'.indexed');
121			if (!$force && file_exists($idxtag)) {
122			if (trim(io_readFile($idxtag)) == $this->getVersion()) {
123			$last = @filemtime($idxtag);
124			if ($last > @filemtime(wikiFN($page))) {
125			if ($verbose) dbglog("Indexer: index for {$page} up to date");
126			return true;
127			}
128			}
129			}
130
131			// register the page to the page.idx file, $pid is always integer
132			$pid = $this->getPID($page);
133
134			// prepare metadata indexing
135			$metadata = array();
136			$metadata['title'] = p_get_metadata($page, 'title', METADATA_RENDER_UNLIMITED);
137
138			$references = p_get_metadata($page, 'relation references', METADATA_RENDER_UNLIMITED);
139			$metadata['relation_references'] = ($references !== null) ?
140			array_keys($references) : array();
141
142			$media = p_get_metadata($page, 'relation media', METADATA_RENDER_UNLIMITED);
143			$metadata['relation_media'] = ($media !== null) ?
144			array_keys($media) : array();
145
146			// check if full text indexing allowed
147			$indexenabled = p_get_metadata($page, 'internal index', METADATA_RENDER_UNLIMITED);
148			if ($indexenabled !== false) $indexenabled = true;
149			$metadata['internal_index'] = $indexenabled;
150
151			$body = '';
152			$data = compact('page', 'body', 'metadata', 'pid');
153			$event = new Event('INDEXER_PAGE_ADD', $data);
154			if ($event->advise_before()) $data['body'] = $data['body'].' '.rawWiki($page);
155			$event->advise_after();
156			unset($event);
157			extract($data);
158			$indexenabled = $metadata['internal_index'];
159			unset($metadata['internal_index']);
160
161			// Access to Metadata Index
162			$result = (new MetadataIndex($pid))->addMetaKeys($metadata);
163			if ($verbose) dbglog("Indexer: addMetaKeys({$page}) ".($result ? 'done' : 'failed'));
164			if (!$result) {
165			return false;
166			}
167
168			// Access to Fulltext Index
169			if ($indexenabled) {
170			$result = (new FulltextIndex($pid))->addWords($body);
171			if ($verbose) dbglog("Indexer: addWords() for {$page} done");
172			if (!$result) {
173			return false;
174			}
175			} else {
176			if ($verbose) dbglog("Indexer: full text indexing disabled for {$page}");
177			// ensure the page content deleted from the Fulltext index
178			$result = (new FulltextIndex($page))->deleteWords();
179			if ($verbose) dbglog("Indexer: deleteWords() for {$page} done");
180			if (!$result) {
181			return false;
182			}
183			}
184
185			// update index tag file
186			io_saveFile($idxtag, $this->getVersion());
187			if ($verbose) dbglog("Indexer: finished");
188
189			return $result;
190			}
191
192			/**
193			* Remove a page from the index
194			*
195			* Erases entries in all known indexes. Locking is handled internally.
196			*
197			* @param bool $verbose print status messages
198			* @param bool $force force reindexing even when the index is up to date
199			* @return bool If the function completed successfully
200			*
201			* @throws IndexAccessException
202			* @throws IndexLockException
203			* @throws IndexWriteException
204			* @author Satoshi Sahara <[email protected]>
205			* @author Tom N Harris <[email protected]>
206			*/
207			public function deletePage($verbose = false, $force = false)
208			{
209			if (!isset($this->page)) {
210			throw new IndexAccessException('Indexer: invalid page name in deletePage');
211			} else {
212			$page = $this->page;
213			}
214
215			$idxtag = metaFN($page,'.indexed');
216			if (!$force && !file_exists($idxtag)) {
217			if ($verbose) dbglog("Indexer: {$page}.indexed file does not exist, ignoring");
218			return true;
219			}
220
221			// retrieve pid from the page.idx file, $pid is always integer
222			$pid = $this->getPID($page);
223
224			// remove obsoleted content from Fulltext index
225			$result = (new FulltextIndex($pid))->deleteWords();
226			if ($verbose) dbglog("Indexer: deleteWords() for {$page} done");
227			if (!$result) {
228			return false;
229			}
230
231			// delete all keys of the page from metadata index
232			$result = (new MetadataIndex($pid))->deleteMetaKeys();
233			if ($verbose) dbglog("Indexer: deleteMetaKeys() for {$page} done");
234			if (!$result) {
235			return false;
236			}
237
238			// mark the page as deleted in the page.idx
239			$this->lock();
240			$this->saveIndexKey('page', '', $pid, self::INDEX_MARK_DELETED.$page);
241			if ($verbose) dbglog("Indexer: {$page} has marked as deleted in page.idx");
242			$this->unlock();
243
244			unset(static::$pidCache[$pid]);
245			@unlink($idxtag);
246			return $result;
247			}
248
249			/**
250			* Rename a page in the search index without changing the indexed content.
251			* This function doesn't check if the old or new name exists in the filesystem.
252			* It returns an error if the old page isn't in the page list of the indexer
253			* and it deletes all previously indexed content of the new page.
254			*
255			* @param string $oldpage The old page name
256			* @param string $newpage The new page name
257			* @return bool If the page was successfully renamed
258			* @throws IndexLockException
259			* @throws IndexWriteException
260			*/
261			public function renamePage($oldpage, $newpage)
262			{
263			$index = $this->getIndex('page', '');
264			// check if oldpage found in page.idx
265			$oldPid = array_search($oldpage, $index, true);
266			if ($oldPid === false) return false;
267
268			// check if newpage found in page.idx
269			$newPid = array_search($newpage, $index, true);
270			if ($newPid !== false) {
271			$result = (new Indexer($newpage))->deletePage();
272			if (!$result) return false;
273			// Note: $index is no longer valid after deletePage()!
274			unset($index);
275			}
276
277			// update page.idx
278			$this->lock();
279			$this->saveIndexKey('page', '', $oldPid, $newpage);
280			$this->unlock();
281
282			// reset the pid cache
283			$this->resetPIDCache();
284
285			return true;
286			}
287
288			/**
289			* Clear the Page Index
290			*
291			* @param bool $requireLock should be false only if the caller is resposible for index lock
292			* @return bool If the index has been cleared successfully
293			* @throws Exception\IndexLockException
294			*/
295			public function clear($requireLock = true)
296			{
297			global $conf;
298
299			if ($requireLock) $this->lock();
300
301			// clear Metadata Index
302			(new MetadataIndex())->clear(false);
303
304			// clear Fulltext Index
305			(new FulltextIndex())->clear(false);
306
307			@unlink($conf['indexdir'].'/page.idx');
308
309			// clear the pid cache
310			$this->resetPIDCache();
311
312			if ($requireLock) $this->unlock();
313			return true;
314			}
315
316			}
317

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