WikiPageAction::addToBottomOrCreatePage() - Code Metrics - Dispositif/Wikibot - Measure and Improve Code Quality continuously with Scrutinizer

WikiPageAction::addToBottomOrCreatePage() A
last analyzed 2024-01-18 16:03 UTC

↳ Parent: WikiPageAction

Complexity

Conditions	2
Paths	2

Size

Total Lines	7
Code Lines	3

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	0
CRAP Score	6

Importance

Changes	2
Bugs	0	Features	1

Metric	Value
cc	2
eloc	3
c	2
b	0
f	1
nc	2
nop	2
dl	0
loc	7
ccs	0
cts	0
cp	0
crap	6
rs	10

<?php
/*
 * This file is part of dispositif/wikibot application (@github)
 * 2019-2023 © Philippe M./Irønie  <[email protected]>
 * For the full copyright and MIT license information, view the license file.
 */

declare(strict_types=1);

namespace App\Application;

use App\Domain\Enums\Language;
use App\Infrastructure\Mediawiki\ExtendedMediawikiFactory;
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
{
    final public const SKIP_LANG_INDICATOR = 'fr'; // skip {{fr}} before template

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

    public $wiki;
    /**
     * Wiki namespace
     */
    private ?int $ns = null;
    private ?Revision $lastTextRevision = null;

    /**
     * @throws Exception
     */
    public function __construct(
        MediawikiFactory        $wiki, // api ?
        private readonly string $title
    )
    {
        $e = null;

        $this->wiki = $wiki;

        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(), $e->getCode(), $e);
        }
    }

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

    /**
     * todo Move to WikiTextUtil ?
     * Replace serialized template and manage {{en}} prefix.
     * Don't delete {{fr}} on frwiki.
     */
    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;
    }

    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.
     */
    public function isPageHomonymie(): bool
    {
        return false !== stristr($this->getText() ?? '', '{{homonymie');
    }

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

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

        return $latest->getContent()->getData();
    }

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

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

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

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

        return null;
    }

    /**
     * @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);
    }

    /**
     * Create a new page.
     * @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);
    }

    /**
     * Add text to the bottom of the article.
     * @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);
    }

    /**
     * Edit the page with new text.
     * Opti : EditInfo optional param ?
     */
    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
        // captchaId=12345&captchaWord=MediaWikiIsCool
        $revisionSaver = ExtendedMediawikiFactory::newRevisionSaverExtended();
        $result = $revisionSaver->save($revision, $editInfo);
        if (false === $result) {
            echo "Error editPage\n";
            print_r($revisionSaver->getErrors());
        }

        return $result;
    }

    /**
     * Check if wiki has been edited by someone since bot's getText().
     */
    private function isPageEditedAfterGetText(): bool
    {
        $updatedPage = $this->wiki->newPageGetter()->getFromTitle($this->title);
        $updatedLastRevision = $updatedPage->getRevisions()->getLatest();
        // Non-strict object equality comparison
        /** @noinspection PhpNonStrictObjectEqualityInspection */
        return !($updatedLastRevision && $updatedLastRevision == $this->lastTextRevision);
    }

    /**
     * Add text to the top of the page.
     * @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);
    }

    /**
     * Extract <ref> data from text.
     * @throws Exception
     */
    public function extractRefFromText(string $text): ?array
    {
        $parser = new TagParser(); // todo ParserFactory
        try {
            $refs = $parser->importHtml($text)->getRefValues();
        } catch (Exception $e) {
            echo "Error extractRefFromText: ".$e->getMessage()."\n";
            return [];
        }

        return $refs;
    }

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

        return $validRef;
    }
}


1		<?php
2		/*
3		* This file is part of dispositif/wikibot application (@github)
4		* 2019-2023 © Philippe M./Irønie <[email protected]>
5		* For the full copyright and MIT license information, 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\Mediawiki\ExtendedMediawikiFactory;
14		use App\Infrastructure\TagParser;
15		use DomainException;
16		use Exception;
17		use Mediawiki\Api\MediawikiFactory;
18		use Mediawiki\DataModel\Content;
19		use Mediawiki\DataModel\EditInfo;
20		use Mediawiki\DataModel\Page;
21		use Mediawiki\DataModel\PageIdentifier;
22		use Mediawiki\DataModel\Revision;
23		use Mediawiki\DataModel\Title;
24		use Throwable;
25
26		class WikiPageAction
27		{
28		final public const SKIP_LANG_INDICATOR = 'fr'; // skip {{fr}} before template
29
30		/**
31		* @var Page
32		*/
33		public $page; // public for debug
34
35		public $wiki;
36		/**
37		* Wiki namespace
38		*/
39		private ?int $ns = null;
40		private ?Revision $lastTextRevision = null;
41
42		/**
43		* @throws Exception
44		*/
45		public function __construct(
46		MediawikiFactory $wiki, // api ?
47		private readonly string $title
48		)
49		{
50		$e = null;
		0 ignored issues – show Unused Code introduced 2023-02-06 17:13 UTC by Report Bug Copy Issue Report The assignment to `$e` is dead and can be removed. Loading history...
51		$this->wiki = $wiki;
52
53		try {
54		$this->page = $wiki->newPageGetter()->getFromTitle($title);
55		$this->ns = $this->page->getPageIdentifier()->getTitle()->getNs();
56		} catch (Throwable $e) {
57		throw new Exception('Erreur construct WikiPageAction ' . $e->getMessage() . $e->getFile() . $e->getLine(), $e->getCode(), $e);
58		}
59		}
60
61		public function getNs(): ?int
62		{
63		return $this->ns;
64		}
65
66		/**
67		* todo Move to WikiTextUtil ?
68		* Replace serialized template and manage {{en}} prefix.
69		* Don't delete {{fr}} on frwiki.
70		*/
71		public static function replaceTemplateInText(string $text, string $tplOrigin, string $tplReplace): ?string
72		{
73		// "{{en}} {{zh}} {{ouvrage...}}"
74		// todo test U
75		if (preg_match_all(
76		'#(?<langTemp>{{[a-z][a-z]}} ?{{[a-z][a-z]}}) ?' . preg_quote($tplOrigin, '#') . '#i',
77		$text,
78		$matches
79		)
80		) {
81		// Skip double lang prefix (like in "{{fr}} {{en}} {template}")
82		echo 'SKIP ! double lang prefix !';
83
84		return $text;
85		}
86
87		// hack // todo: autres patterns {{en}} ?
88		// OK : {{en}} \n {{ouvrage}}
89		if (preg_match_all(
90		"#(?<langTemp>{{(?<lang>[a-z][a-z])}} *\n?)?" . preg_quote($tplOrigin, '#') . '#i',
91		$text,
92		$matches
93		) > 0
94		) {
95		foreach ($matches[0] as $num => $mention) {
96		$lang = $matches['lang'][$num] ?? '';
97		if (!empty($lang)) {
98		$lang = Language::all2wiki($lang);
99		}
100
101		// detect inconsistency between lang indicator and lang param
102		// example : {{en}} {{template\|lang=ru}}
103		if (!empty($lang) && self::SKIP_LANG_INDICATOR !== $lang
104		&& preg_match('#langue *=#', $tplReplace)
105		&& !preg_match('#langue *= ?' . $lang . '#i', $tplReplace)
106		&& !preg_match('#\\| ?langue *= ?\n?\\|#', $tplReplace)
107		) {
108		echo sprintf(
109		'prefix %s incompatible avec langue de %s',
110		$matches['langTemp'][$num],
111		$tplReplace
112		);
113
114		// skip all the replacements of that template
115		return $text; // return null ?
116		}
117
118		// FIX dirty juil 2020 : {{en}} mais aucun param/value sur new template
119		if (!empty($lang) && $lang !== 'fr' && !preg_match('#\\| ?langue *=#', $tplReplace) > 0) {
120		// skip all the replacements of that template
121
122		return $text;
123		}
124
125		// FIX dirty : {{en}} mais langue= avec value non définie sur new template...
126		if (!empty($lang) && preg_match('#\\| ?(langue *=) ?\n? ?\\|#', $tplReplace, $matchLangue) > 0) {
127		$previousTpl = $tplReplace;
128		$tplReplace = str_replace($matchLangue[1], 'langue=' . $lang, $tplReplace);
129		//dump('origin', $tplOrigin);
130		$text = str_replace($previousTpl, $tplReplace, $text);
131		}
132
133		// don't delete {{fr}} before {template} on frwiki
134		if (self::SKIP_LANG_INDICATOR === $lang) {
135		$text = str_replace($tplOrigin, $tplReplace, $text);
136
137		continue;
138		}
139
140		// replace {template} and {{lang}} {template}
141		$text = str_replace($mention, $tplReplace, $text);
142		$text = str_replace(
143		$matches['langTemp'][$num] . $tplReplace,
144		$tplReplace,
145		$text
146		); // si 1er replace global sans
147		// {{en}}
148		}
149		}
150
151		return $text;
152		}
153
154		public function getLastEditor(): ?string
155		{
156		// page doesn't exist
157		if (empty($this->page->getRevisions()->getLatest())) {
158		return null;
159		}
160
161		$latest = $this->page->getRevisions()->getLatest();
162
163		return ($latest) ? $latest->getUser() : null;
164		}
165
166		/**
167		* Check if a frwiki disambiguation page.
168		*/
169		public function isPageHomonymie(): bool
170		{
171		return false !== stristr($this->getText() ?? '', '{{homonymie');
172		}
173
174		/**
175		* Get wiki text from the page.
176		*/
177		public function getText(): ?string
178		{
179		$latest = $this->getLastRevision();
180		$this->lastTextRevision = $latest;
181
182		if (empty($latest)) {
183		return null;
184		}
185
186		return $latest->getContent()->getData();
187		}
188
189		public function getLastRevision(): ?Revision
190		{
191		// page doesn't exist
192		if (empty($this->page->getRevisions()->getLatest())) {
193		return null;
194		}
195
196		return $this->page->getRevisions()->getLatest();
197		}
198
199		/**
200		* Is it page with a redirection link ?
201		*/
202		public function isRedirect(): bool
203		{
204		return !empty($this->getRedirect());
205		}
206
207		/**
208		* Get redirection page title or null.
209		*/
210		public function getRedirect(): ?string
211		{
212		if ($this->getText() && preg_match('/^#REDIRECT(?:ION)? ?\[\[([^]]+)]]/i', $this->getText(), $matches)) {
213		return (string)trim($matches[1]);
214		}
215
216		return null;
217		}
218
219		/**
220		* @return bool success
221		* @throws Exception
222		*/
223		public function addToBottomOrCreatePage(string $addText, EditInfo $editInfo): bool
224		{
225		if (empty($this->page->getRevisions()->getLatest())) {
226		return $this->createPage($addText, $editInfo);
227		}
228
229		return $this->addToBottomOfThePage($addText, $editInfo);
230		}
231
232		/**
233		* Create a new page.
234		* @throws Exception
235	6	*/
236		public function createPage(string $text, ?EditInfo $editInfo = null): bool
237		{
238	6	if (!empty($this->page->getRevisions()->getLatest())) {
239	6	throw new Exception('That page already exists');
240	6	}
241	6
242		$newContent = new Content($text);
243		// $identifier = $this->page->getPageIdentifier()
244		$title = new Title($this->title);
245		$identifier = new PageIdentifier($title);
246		$revision = new Revision($newContent, $identifier);
247
248		return $this->wiki->newRevisionSaver()->save($revision, $editInfo);
249		}
250
251		/**
252	6	* Add text to the bottom of the article.
253	6	* @return bool success
254	6	* @throws Exception
255	6	*/
256	6	public function addToBottomOfThePage(string $addText, EditInfo $editInfo): bool
257		{
258	6	if (empty($this->page->getRevisions()->getLatest())) {
259	6	throw new Exception('That page does not exist');
260	6	}
261	6	$oldText = $this->getText();
262		$newText = $oldText . "\n" . $addText;
263
264		return $this->editPage($newText, $editInfo);
265		}
266
267	6	/**
268	6	* Edit the page with new text.
269	6	* Opti : EditInfo optional param ?
270		*/
271		public function editPage(string $newText, EditInfo $editInfo, ?bool $checkConflict = false): bool
272		{
273		if ($checkConflict && $this->isPageEditedAfterGetText()) {
274		throw new DomainException('Wiki Conflict : Page has been edited after getText()');
275		// return false ?
276		}
277
278		$revision = $this->page->getPageIdentifier();
279
280		$content = new Content($newText);
281		$revision = new Revision($content, $revision);
282	6
283	2	// TODO try/catch UsageExceptions badtoken
284	2	// captchaId=12345&captchaWord=MediaWikiIsCool
285	2	$revisionSaver = ExtendedMediawikiFactory::newRevisionSaverExtended();
286		$result = $revisionSaver->save($revision, $editInfo);
287		if (false === $result) {
288		echo "Error editPage\n";
289	6	print_r($revisionSaver->getErrors());
290	3	}
291
292	3	return $result;
293		}
294
295		/**
296	6	* Check if wiki has been edited by someone since bot's getText().
297	6	*/
298	6	private function isPageEditedAfterGetText(): bool
299	6	{
300	6	$updatedPage = $this->wiki->newPageGetter()->getFromTitle($this->title);
301		$updatedLastRevision = $updatedPage->getRevisions()->getLatest();
302		// Non-strict object equality comparison
303		/** @noinspection PhpNonStrictObjectEqualityInspection */
304		return !($updatedLastRevision && $updatedLastRevision == $this->lastTextRevision);
305		}
306	6
307		/**
308		* Add text to the top of the page.
309		* @return bool success
310		* @throws Exception
311		*/
312		public function addToTopOfThePage(string $addText, EditInfo $editInfo): bool
313		{
314		if (empty($this->page->getRevisions()->getLatest())) {
315		throw new Exception('That page does not exist');
316		}
317		$oldText = $this->getText();
318		$newText = $addText . $oldText;
319
320		return $this->editPage($newText, $editInfo);
321		}
322
323		/**
324		* Extract <ref> data from text.
325		* @throws Exception
326		*/
327		public function extractRefFromText(string $text): ?array
328		{
329		$parser = new TagParser(); // todo ParserFactory
330		try {
331		$refs = $parser->importHtml($text)->getRefValues();
332		} catch (Exception $e) {
333		echo "Error extractRefFromText: ".$e->getMessage()."\n";
334		return [];
335		}
336
337		return $refs;
338		}
339
340		/**
341		* TODO $url parameter
342		* TODO? refactor with : parse_str() + parse_url($url, PHP_URL_QUERY)
343		* check if any ref contains a targeted website/URL.
344		*/
345		public function filterRefByURL(array $refs): array
346		{
347		$validRef = [];
348		foreach ($refs as $ref) {
349		if (preg_match(
350		'#(?<url>https?://(?:www\.)?lemonde\.fr/[^ \]]+)#i',
351		(string)$ref,
352		$matches
353		) > 0
354		) {
355		$validRef[] = ['url' => $matches['url'], 'raw' => $ref];
356		}
357		}
358
359		return $validRef;
360		}
361		}
362

Dispositif / Wikibot

WikiPageAction::addToBottomOrCreatePage() A last analyzed 2024-01-18 16:03 UTC

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like

WikiPageAction::addToBottomOrCreatePage() A
last analyzed 2024-01-18 16:03 UTC