WikiPageAction::isPageEditedAfterGetText() - Code Metrics - Inspection of "Feature : mapping "accès url" for externalLinks" - Dispositif/Wikibot - Measure and Improve Code Quality continuously with Scrutinizer

Test Failed

Push — master ( 319840...0bb54c )

by Dispositif

created 2020-08-26 15:55 UTC

WikiPageAction::isPageEditedAfterGetText() A

↳ Parent: WikiPageAction

Complexity

Conditions	3
Paths	2

Size

Total Lines	12
Code Lines	5

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	0
CRAP Score	12

Importance

Changes	1
Bugs	0	Features	0

Metric	Value
cc	3
eloc	5
c	1
b	0
f	0
nc	2
nop	0
dl	0
loc	12
ccs	0
cts	4
cp	0
crap	12
rs	10

<?php
/**
 * This file is part of dispositif/wikibot application
 * 2019 : Philippe M. <[email protected]>
 * For the full copyright and MIT license information, please view the LICENSE file.
 */

declare(strict_types=1);

namespace App\Application;

use App\Domain\Enums\Language;
use App\Infrastructure\TagParser;
use DomainException;
use Exception;
use Mediawiki\Api\MediawikiFactory;
use Mediawiki\DataModel\Content;
use Mediawiki\DataModel\EditInfo;
use Mediawiki\DataModel\Page;
use Mediawiki\DataModel\PageIdentifier;
use Mediawiki\DataModel\Revision;
use Mediawiki\DataModel\Title;
use Throwable;

class WikiPageAction
{
    const SKIP_LANG_INDICATOR = 'fr'; // skip {{fr}} before template

    /**
     * @var Page
     */
    public $page; // public for debug

    public $wiki; // api ?

    /**
     * @var string
     */
    private $title;
    /**
     * Wiki namespace
     *
     * @var int
     */
    private $ns;
    /**
     * @var Revision|null
     */
    private $lastTextRevision;

    /**
     * WikiPageAction constructor.
     *
     * @param MediawikiFactory $wiki
     * @param string           $title
     *
     * @throws Exception
     */
    public function __construct(MediawikiFactory $wiki, string $title)
    {
        $this->wiki = $wiki;
        $this->title = $title;

        try {
            $this->page = $wiki->newPageGetter()->getFromTitle($title);
            $this->ns = $this->page->getPageIdentifier()->getTitle()->getNs();
        } catch (Throwable $e) {
            throw new Exception('Erreur construct WikiPageAction '.$e->getMessage().$e->getFile().$e->getLine());
        }
    }

    /**
     * Get wiki text from the page.
     *
     * @return string|null
     */
    public function getText(): ?string
    {
        $latest = $this->getLastRevision();
        $this->lastTextRevision = $latest;

        if (empty($latest)) {
            return null;
        }

        return ($latest) ? $latest->getContent()->getData() : null;
    }

    public function getNs(): ?int
    {
        return $this->ns;
    }

    public function getLastRevision(): ?Revision
    {
        // page doesn't exist
        if (empty($this->page->getRevisions()->getLatest())) {
            return null;
        }

        return $this->page->getRevisions()->getLatest();
    }

    public function getLastEditor(): ?string
    {
        // page doesn't exist
        if (empty($this->page->getRevisions()->getLatest())) {
            return null;
        }

        $latest = $this->page->getRevisions()->getLatest();

        return ($latest) ? $latest->getUser() : null;
    }

    /**
     * Check if a frwiki disambiguation page.
     *
     * @return bool
     */
    public function isPageHomonymie(): bool
    {
        return false !== stristr($this->getText(), '{{homonymie');
    }

    /**
     * Is it page with a redirection link ?
     *
     * @return bool
     */
    public function isRedirect(): bool
    {
        return !empty($this->getRedirect());
    }

    /**
     * Get redirection page title or null.
     *
     * @return string|null
     */
    public function getRedirect(): ?string
    {
        if ($this->getText() && preg_match('/^#REDIRECT(?:ION)? ?\[\[([^]]+)]]/i', $this->getText(), $matches)) {
            return (string)trim($matches[1]);
        }

        return null;
    }

    /**
     * Edit the page with new text.
     * Opti : EditInfo optional param ?
     *
     * @param string    $newText
     * @param EditInfo  $editInfo
     * @param bool|null $checkConflict
     *
     * @return bool
     */
    public function editPage(string $newText, EditInfo $editInfo, ?bool $checkConflict = false): bool
    {
        if ($checkConflict && $this->isPageEditedAfterGetText()) {
            throw new DomainException('Wiki Conflict : Page has been edited after getText()');
            // return false ?
        }

        $revision = $this->page->getPageIdentifier();

        $content = new Content($newText);
        $revision = new Revision($content, $revision);

        // TODO try/catch UsageExceptions badtoken
        return $this->wiki->newRevisionSaver()->save($revision, $editInfo);
    }

    /**
     * Check if wiki has been edited by someone since bot's getText().
     *
     * @return bool
     */
    private function isPageEditedAfterGetText(): bool
    {
        $updatedPage = $this->wiki->newPageGetter()->getFromTitle($this->title);
        $updatedLastRevision = $updatedPage->getRevisions()->getLatest();

        // Non-strict object equality comparison
        /** @noinspection PhpNonStrictObjectEqualityInspection */
        if ($updatedLastRevision && $updatedLastRevision == $this->lastTextRevision) {
            return false;
        }

        return true;
    }

    /**
     * Create a new page.
     *
     * @param string        $text
     * @param EditInfo|null $editInfo
     *
     * @return bool
     * @throws Exception
     */
    public function createPage(string $text, ?EditInfo $editInfo = null): bool
    {
        if (!empty($this->page->getRevisions()->getLatest())) {
            throw new Exception('That page already exists');
        }

        $newContent = new Content($text);
        // $identifier = $this->page->getPageIdentifier()
        $title = new Title($this->title);
        $identifier = new PageIdentifier($title);
        $revision = new Revision($newContent, $identifier);

        return $this->wiki->newRevisionSaver()->save($revision, $editInfo);
    }

    /**
     * @param string   $addText
     * @param EditInfo $editInfo
     *
     * @return bool success
     * @throws Exception
     */
    public function addToBottomOrCreatePage(string $addText, EditInfo $editInfo): bool
    {
        if (empty($this->page->getRevisions()->getLatest())) {
            return $this->createPage($addText, $editInfo);
        }

        return $this->addToBottomOfThePage($addText, $editInfo);
    }

    /**
     * Add text to the bottom of the article.
     *
     * @param string   $addText
     * @param EditInfo $editInfo
     *
     * @return bool success
     * @throws Exception
     */
    public function addToBottomOfThePage(string $addText, EditInfo $editInfo): bool
    {
        if (empty($this->page->getRevisions()->getLatest())) {
            throw new Exception('That page does not exist');
        }
        $oldText = $this->getText();
        $newText = $oldText."\n".$addText;

        return $this->editPage($newText, $editInfo);
    }

    /**
     * Add text to the top of the page.
     *
     * @param string   $addText
     * @param EditInfo $editInfo
     *
     * @return bool success
     * @throws Exception
     */
    public function addToTopOfThePage(string $addText, EditInfo $editInfo): bool
    {
        if (empty($this->page->getRevisions()->getLatest())) {
            throw new Exception('That page does not exist');
        }
        $oldText = $this->getText();
        $newText = $addText.$oldText;

        return $this->editPage($newText, $editInfo);
    }

    /**
     * todo Move to WikiTextUtil ?
     * Replace serialized template and manage {{en}} prefix.
     * Don't delete {{fr}} on frwiki.
     *
     * @param string $text       wikitext of the page
     * @param string $tplOrigin  template text to replace
     * @param string $tplReplace new template text
     *
     * @return string|null
     */
    public static function replaceTemplateInText(string $text, string $tplOrigin, string $tplReplace): string
    {
        // "{{en}} {{zh}} {{ouvrage...}}"
        // todo test U
        if (preg_match_all(
            '#(?<langTemp>{{[a-z][a-z]}} ?{{[a-z][a-z]}}) ?'.preg_quote($tplOrigin, '#').'#i',
            $text,
            $matches
        )
        ) {
            // Skip double lang prefix (like in "{{fr}} {{en}} {template}")
            echo 'SKIP ! double lang prefix !';

            return $text;
        }

        // hack // todo: autres patterns {{en}} ?
        // OK : {{en}} \n {{ouvrage}}
        if (preg_match_all(
                "#(?<langTemp>{{(?<lang>[a-z][a-z])}} *\n?)?".preg_quote($tplOrigin, '#').'#i',
                $text,
                $matches
            ) > 0
        ) {
            foreach ($matches[0] as $num => $mention) {
                $lang = $matches['lang'][$num] ?? '';
                if (!empty($lang)) {
                    $lang = Language::all2wiki($lang);
                }

                // detect inconsistency between lang indicator and lang param
                // example : {{en}} {{template|lang=ru}}
                if (!empty($lang) && self::SKIP_LANG_INDICATOR !== $lang
                    && preg_match('#langue *=#', $tplReplace)
                    && !preg_match('#langue *= ?'.$lang.'#i', $tplReplace)
                    && !preg_match('#\| ?langue *= ?\n?\|#', $tplReplace)
                ) {
                    echo sprintf(
                        'prefix %s incompatible avec langue de %s',
                        $matches['langTemp'][$num],
                        $tplReplace
                    );

                    // skip all the replacements of that template
                    return $text; // return null ?
                }

                // FIX dirty juil 2020 : {{en}} mais aucun param/value sur new template
                if (!empty($lang) && $lang !== 'fr' && !preg_match('#\| ?langue *=#', $tplReplace) > 0) {
                    // skip all the replacements of that template

                    return $text;
                }

                // FIX dirty : {{en}} mais langue= avec value non définie sur new template...
                if (!empty($lang) && preg_match('#\| ?(langue *=) ?\n? ?\|#', $tplReplace, $matchLangue) > 0) {
                    $previousTpl = $tplReplace;
                    $tplReplace = str_replace($matchLangue[1], 'langue='.$lang, $tplReplace);
                    //dump('origin', $tplOrigin);
                    $text = str_replace($previousTpl, $tplReplace, $text);
                }

                // don't delete {{fr}} before {template} on frwiki
                if (self::SKIP_LANG_INDICATOR === $lang) {
                    $text = str_replace($tplOrigin, $tplReplace, $text);

                    continue;
                }

                // replace {template} and {{lang}} {template}
                $text = str_replace($mention, $tplReplace, $text);
                $text = str_replace(
                    $matches['langTemp'][$num].$tplReplace,
                    $tplReplace,
                    $text
                ); // si 1er replace global sans
                // {{en}}
            }
        }

        return $text;
    }

    /**
     * Extract <ref> data from text.
     *
     * @param $text string
     *
     * @return array
     * @throws Exception
     */
    public function extractRefFromText(string $text): ?array
    {
        $parser = new TagParser(); // todo ParserFactory
        $refs = $parser->importHtml($text)->getRefValues(); // []

        return (array)$refs;
    }

    /**
     * TODO $url parameter
     * TODO? refactor with : parse_str() + parse_url($url, PHP_URL_QUERY)
     * check if any ref contains a targeted website/URL.
     *
     * @param array $refs
     *
     * @return array
     */
    public function filterRefByURL(array $refs): array
    {
        $validRef = [];
        foreach ($refs as $ref) {
            if (preg_match(
                    '#(?<url>https?://(?:www\.)?lemonde\.fr/[^ \]]+)#i',
                    $ref,
                    $matches
                ) > 0
            ) {
                $validRef[] = ['url' => $matches['url'], 'raw' => $ref];
            }
        }

        return $validRef;
    }
}


1		<?php
2		/**
3		* This file is part of dispositif/wikibot application
4		* 2019 : Philippe M. <[email protected]>
5		* For the full copyright and MIT license information, please view the LICENSE file.
6		*/
7
8		declare(strict_types=1);
9
10		namespace App\Application;
11
12		use App\Domain\Enums\Language;
13		use App\Infrastructure\TagParser;
14		use DomainException;
15		use Exception;
16		use Mediawiki\Api\MediawikiFactory;
17		use Mediawiki\DataModel\Content;
18		use Mediawiki\DataModel\EditInfo;
19		use Mediawiki\DataModel\Page;
20		use Mediawiki\DataModel\PageIdentifier;
21		use Mediawiki\DataModel\Revision;
22		use Mediawiki\DataModel\Title;
23		use Throwable;
24
25		class WikiPageAction
26		{
27		const SKIP_LANG_INDICATOR = 'fr'; // skip {{fr}} before template
28
29		/**
30		* @var Page
31		*/
32		public $page; // public for debug
33
34		public $wiki; // api ?
35
36		/**
37		* @var string
38		*/
39		private $title;
40		/**
41		* Wiki namespace
42		*
43		* @var int
44		*/
45		private $ns;
46		/**
47		* @var Revision\|null
48		*/
49		private $lastTextRevision;
50
51		/**
52		* WikiPageAction constructor.
53		*
54		* @param MediawikiFactory $wiki
55		* @param string $title
56		*
57		* @throws Exception
58		*/
59		public function __construct(MediawikiFactory $wiki, string $title)
60		{
61		$this->wiki = $wiki;
62		$this->title = $title;
63
64		try {
65		$this->page = $wiki->newPageGetter()->getFromTitle($title);
66		$this->ns = $this->page->getPageIdentifier()->getTitle()->getNs();
67		} catch (Throwable $e) {
68		throw new Exception('Erreur construct WikiPageAction '.$e->getMessage().$e->getFile().$e->getLine());
69		}
70		}
71
72		/**
73		* Get wiki text from the page.
74		*
75		* @return string\|null
76		*/
77		public function getText(): ?string
78		{
79		$latest = $this->getLastRevision();
80		$this->lastTextRevision = $latest;
81
82		if (empty($latest)) {
83		return null;
84		}
85
86		return ($latest) ? $latest->getContent()->getData() : null;
87		}
88
89		public function getNs(): ?int
90		{
91		return $this->ns;
92		}
93
94		public function getLastRevision(): ?Revision
95		{
96		// page doesn't exist
97		if (empty($this->page->getRevisions()->getLatest())) {
98		return null;
99		}
100
101		return $this->page->getRevisions()->getLatest();
102		}
103
104		public function getLastEditor(): ?string
105		{
106		// page doesn't exist
107		if (empty($this->page->getRevisions()->getLatest())) {
108		return null;
109		}
110
111		$latest = $this->page->getRevisions()->getLatest();
112
113		return ($latest) ? $latest->getUser() : null;
114		}
115
116		/**
117		* Check if a frwiki disambiguation page.
118		*
119		* @return bool
120		*/
121		public function isPageHomonymie(): bool
122		{
123		return false !== stristr($this->getText(), '{{homonymie');
124		}
125
126		/**
127		* Is it page with a redirection link ?
128		*
129		* @return bool
130		*/
131		public function isRedirect(): bool
132		{
133		return !empty($this->getRedirect());
134		}
135
136		/**
137		* Get redirection page title or null.
138		*
139		* @return string\|null
140		*/
141		public function getRedirect(): ?string
142		{
143		if ($this->getText() && preg_match('/^#REDIRECT(?:ION)? ?\[\[([^]]+)]]/i', $this->getText(), $matches)) {
144		return (string)trim($matches[1]);
145		}
146
147		return null;
148		}
149
150		/**
151		* Edit the page with new text.
152		* Opti : EditInfo optional param ?
153		*
154		* @param string $newText
155		* @param EditInfo $editInfo
156		* @param bool\|null $checkConflict
157		*
158		* @return bool
159		*/
160		public function editPage(string $newText, EditInfo $editInfo, ?bool $checkConflict = false): bool
161		{
162		if ($checkConflict && $this->isPageEditedAfterGetText()) {
163		throw new DomainException('Wiki Conflict : Page has been edited after getText()');
164		// return false ?
165		}
166
167		$revision = $this->page->getPageIdentifier();
168
169		$content = new Content($newText);
170		$revision = new Revision($content, $revision);
171
172		// TODO try/catch UsageExceptions badtoken
173		return $this->wiki->newRevisionSaver()->save($revision, $editInfo);
174		}
175
176		/**
177		* Check if wiki has been edited by someone since bot's getText().
178		*
179		* @return bool
180		*/
181		private function isPageEditedAfterGetText(): bool
182		{
183		$updatedPage = $this->wiki->newPageGetter()->getFromTitle($this->title);
184		$updatedLastRevision = $updatedPage->getRevisions()->getLatest();
185
186		// Non-strict object equality comparison
187		/** @noinspection PhpNonStrictObjectEqualityInspection */
188		if ($updatedLastRevision && $updatedLastRevision == $this->lastTextRevision) {
189		return false;
190		}
191
192		return true;
193		}
194
195		/**
196		* Create a new page.
197		*
198		* @param string $text
199		* @param EditInfo\|null $editInfo
200		*
201		* @return bool
202		* @throws Exception
203		*/
204		public function createPage(string $text, ?EditInfo $editInfo = null): bool
205		{
206		if (!empty($this->page->getRevisions()->getLatest())) {
207		throw new Exception('That page already exists');
208		}
209
210		$newContent = new Content($text);
211		// $identifier = $this->page->getPageIdentifier()
212		$title = new Title($this->title);
213		$identifier = new PageIdentifier($title);
214		$revision = new Revision($newContent, $identifier);
215
216		return $this->wiki->newRevisionSaver()->save($revision, $editInfo);
217		}
218
219		/**
220		* @param string $addText
221		* @param EditInfo $editInfo
222		*
223		* @return bool success
224		* @throws Exception
225		*/
226		public function addToBottomOrCreatePage(string $addText, EditInfo $editInfo): bool
227		{
228		if (empty($this->page->getRevisions()->getLatest())) {
229		return $this->createPage($addText, $editInfo);
230		}
231
232		return $this->addToBottomOfThePage($addText, $editInfo);
233		}
234
235	6	/**
236		* Add text to the bottom of the article.
237		*
238	6	* @param string $addText
239	6	* @param EditInfo $editInfo
240	6	*
241	6	* @return bool success
242		* @throws Exception
243		*/
244		public function addToBottomOfThePage(string $addText, EditInfo $editInfo): bool
245		{
246		if (empty($this->page->getRevisions()->getLatest())) {
247		throw new Exception('That page does not exist');
248		}
249		$oldText = $this->getText();
250		$newText = $oldText."\n".$addText;
251
252	6	return $this->editPage($newText, $editInfo);
253	6	}
254	6
255	6	/**
256	6	* Add text to the top of the page.
257		*
258	6	* @param string $addText
259	6	* @param EditInfo $editInfo
260	6	*
261	6	* @return bool success
262		* @throws Exception
263		*/
264		public function addToTopOfThePage(string $addText, EditInfo $editInfo): bool
265		{
266		if (empty($this->page->getRevisions()->getLatest())) {
267	6	throw new Exception('That page does not exist');
268	6	}
269	6	$oldText = $this->getText();
270		$newText = $addText.$oldText;
271
272		return $this->editPage($newText, $editInfo);
273		}
274
275		/**
276		* todo Move to WikiTextUtil ?
277		* Replace serialized template and manage {{en}} prefix.
278		* Don't delete {{fr}} on frwiki.
279		*
280		* @param string $text wikitext of the page
281		* @param string $tplOrigin template text to replace
282	6	* @param string $tplReplace new template text
283	2	*
284	2	* @return string\|null
285	2	*/
286		public static function replaceTemplateInText(string $text, string $tplOrigin, string $tplReplace): string
287		{
288		// "{{en}} {{zh}} {{ouvrage...}}"
289	6	// todo test U
290	3	if (preg_match_all(
291		'#(?<langTemp>{{[a-z][a-z]}} ?{{[a-z][a-z]}}) ?'.preg_quote($tplOrigin, '#').'#i',
292	3	$text,
293		$matches
294		)
295		) {
296	6	// Skip double lang prefix (like in "{{fr}} {{en}} {template}")
297	6	echo 'SKIP ! double lang prefix !';
298	6
299	6	return $text;
300	6	}
301
302		// hack // todo: autres patterns {{en}} ?
303		// OK : {{en}} \n {{ouvrage}}
304		if (preg_match_all(
305		"#(?<langTemp>{{(?<lang>[a-z][a-z])}} *\n?)?".preg_quote($tplOrigin, '#').'#i',
306	6	$text,
307		$matches
308		) > 0
309		) {
310		foreach ($matches[0] as $num => $mention) {
311		$lang = $matches['lang'][$num] ?? '';
312		if (!empty($lang)) {
313		$lang = Language::all2wiki($lang);
314		}
315
316		// detect inconsistency between lang indicator and lang param
317		// example : {{en}} {{template\|lang=ru}}
318		if (!empty($lang) && self::SKIP_LANG_INDICATOR !== $lang
319		&& preg_match('#langue *=#', $tplReplace)
320		&& !preg_match('#langue *= ?'.$lang.'#i', $tplReplace)
321		&& !preg_match('#\\| ?langue *= ?\n?\\|#', $tplReplace)
322		) {
323		echo sprintf(
324		'prefix %s incompatible avec langue de %s',
325		$matches['langTemp'][$num],
326		$tplReplace
327		);
328
329		// skip all the replacements of that template
330		return $text; // return null ?
331		}
332
333		// FIX dirty juil 2020 : {{en}} mais aucun param/value sur new template
334		if (!empty($lang) && $lang !== 'fr' && !preg_match('#\\| ?langue *=#', $tplReplace) > 0) {
335		// skip all the replacements of that template
336
337		return $text;
338		}
339
340		// FIX dirty : {{en}} mais langue= avec value non définie sur new template...
341		if (!empty($lang) && preg_match('#\\| ?(langue *=) ?\n? ?\\|#', $tplReplace, $matchLangue) > 0) {
342		$previousTpl = $tplReplace;
343		$tplReplace = str_replace($matchLangue[1], 'langue='.$lang, $tplReplace);
344		//dump('origin', $tplOrigin);
345		$text = str_replace($previousTpl, $tplReplace, $text);
346		}
347
348		// don't delete {{fr}} before {template} on frwiki
349		if (self::SKIP_LANG_INDICATOR === $lang) {
350		$text = str_replace($tplOrigin, $tplReplace, $text);
351
352		continue;
353		}
354
355		// replace {template} and {{lang}} {template}
356		$text = str_replace($mention, $tplReplace, $text);
357		$text = str_replace(
358		$matches['langTemp'][$num].$tplReplace,
359		$tplReplace,
360		$text
361		); // si 1er replace global sans
362		// {{en}}
363		}
364		}
365
366		return $text;
367		}
368
369		/**
370		* Extract <ref> data from text.
371		*
372		* @param $text string
373		*
374		* @return array
375		* @throws Exception
376		*/
377		public function extractRefFromText(string $text): ?array
378		{
379		$parser = new TagParser(); // todo ParserFactory
380		$refs = $parser->importHtml($text)->getRefValues(); // []
381
382		return (array)$refs;
383		}
384
385		/**
386		* TODO $url parameter
387		* TODO? refactor with : parse_str() + parse_url($url, PHP_URL_QUERY)
388		* check if any ref contains a targeted website/URL.
389		*
390		* @param array $refs
391		*
392		* @return array
393		*/
394		public function filterRefByURL(array $refs): array
395		{
396		$validRef = [];
397		foreach ($refs as $ref) {
398		if (preg_match(
399		'#(?<url>https?://(?:www\.)?lemonde\.fr/[^ \]]+)#i',
400		$ref,
401		$matches
402		) > 0
403		) {
404		$validRef[] = ['url' => $matches['url'], 'raw' => $ref];
405		}
406		}
407
408		return $validRef;
409		}
410		}
411

Dispositif / Wikibot

Push — master ( 319840...0bb54c )

WikiPageAction::isPageEditedAfterGetText() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like