WikiPageAction::addToBottomOrCreatePage() - Code Metrics - Inspection of "Import lien Google : ajout '&printsec=frontcover'" - Dispositif/Wikibot - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Push — master ( 511693...0349ad )

by Dispositif

created 2019-12-07 16:20 UTC

WikiPageAction::addToBottomOrCreatePage() A

↳ Parent: WikiPageAction

Complexity

Conditions	2
Paths	2

Size

Total Lines	7
Code Lines	3

Duplication

Lines	0
Ratio	0 %

Importance

Changes	1
Bugs	0	Features	1

Metric	Value
cc	2
eloc	3
c	1
b	0
f	1
nc	2
nop	2
dl	0
loc	7
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 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;

    /**
     * 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);
        } catch (Throwable $e) {
            throw new Exception('Erreur construct WikiPageAction '.$e->getMessage());
        }
    }

    /**
     * Get wiki text from the page.
     *
     * @return string|null
     */
    public function getText(): ?string
    {
        // page doesn't exist
        if (empty($this->page->getRevisions()->getLatest())) {
            return null;
        }

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

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

    }

    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
     *
     * @return bool
     */
    public function editPage(string $newText, EditInfo $editInfo): bool
    {
        $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);
    }

    /**
     * Create a new page.
     *
     * @param string   $text
     *
     * @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
     * @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
     * @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);
    }

    /**
     * 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 $tplReplace): string
    {
        if (preg_match_all(
            '#(?<langTemp>{{[a-z][a-z]}} ?{{[a-z][a-z]}}) ?'.preg_quote($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}} ?
        if (preg_match_all('#(?<langTemp>{{(?<lang>[a-z][a-z])}} *)?'.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}}
                // BUG: prefix {{de}}  incompatible avec langue de {{Ouvrage |langue= |prénom1=Hartmut |nom1=Atsma
                if (!empty($lang) && self::SKIP_LANG_INDICATOR !== $lang
                    && !preg_match('#lang(ue)?='.$lang.'#i', $tplReplace)
                    && !preg_match('#\| ?langue= ?\|#', $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 : {{en}} mais pas de langue sur template...
                if ($lang && preg_match('#\| ?langue= ?\|#', $tplReplace) > 0) {
                    $previousTpl = $tplReplace;
                    $tplReplace = str_replace('langue=', 'langue='.$lang, $tplReplace);
                    $text = str_replace($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 Exception;
15			use Mediawiki\Api\MediawikiFactory;
16			use Mediawiki\DataModel\Content;
17			use Mediawiki\DataModel\EditInfo;
18			use Mediawiki\DataModel\Page;
19			use Mediawiki\DataModel\PageIdentifier;
20			use Mediawiki\DataModel\Revision;
21			use Mediawiki\DataModel\Title;
22			use Throwable;
23
24			class WikiPageAction
25			{
26			const SKIP_LANG_INDICATOR = 'fr'; // skip {{fr}} before template
27
28			/**
29			* @var Page
30			*/
31			public $page; // public for debug
32
33			public $wiki; // api ?
34			/**
35			* @var string
36			*/
37			private $title;
38
39			/**
40			* WikiPageAction constructor.
41			*
42			* @param MediawikiFactory $wiki
43			* @param string $title
44			*
45			* @throws Exception
46			*/
47			public function __construct(MediawikiFactory $wiki, string $title)
48			{
49			$this->wiki = $wiki;
50			$this->title = $title;
51
52			try {
53			$this->page = $wiki->newPageGetter()->getFromTitle($title);
54			} catch (Throwable $e) {
55			throw new Exception('Erreur construct WikiPageAction '.$e->getMessage());
56			}
57			}
58
59			/**
60			* Get wiki text from the page.
61			*
62			* @return string\|null
63			*/
64			public function getText(): ?string
65			{
66			// page doesn't exist
67			if (empty($this->page->getRevisions()->getLatest())) {
68			return null;
69			}
70
71			$latest = $this->page->getRevisions()->getLatest();
72
73			return ($latest) ? $latest->getContent()->getData() : null;
			1 ignored issue – show introduced 2019-12-07 16:25 UTC by Report Bug Copy Issue Report `$latest` is of type `Mediawiki\DataModel\Revision`, thus it always evaluated to `true`. Loading history...
74			}
75
76			public function getLastEditor(): ?string
77			{
78			// page doesn't exist
79			if (empty($this->page->getRevisions()->getLatest())) {
80			return null;
81			}
82
83			$latest = $this->page->getRevisions()->getLatest();
84
85			return ($latest) ? $latest->getUser() : null;
			1 ignored issue – show introduced 2019-12-07 16:25 UTC by Report Bug Copy Issue Report `$latest` is of type `Mediawiki\DataModel\Revision`, thus it always evaluated to `true`. Loading history...
86			}
87
88			/**
89			* Check if a frwiki disambiguation page.
90			*
91			* @return bool
92			*/
93			public function isPageHomonymie(): bool
94			{
95			return false !== stristr($this->getText(), '{{homonymie');
96			}
97
98			/**
99			* Is it page with a redirection link ?
100			*
101			* @return bool
102			*/
103			public function isRedirect(): bool
104			{
105			return !empty($this->getRedirect());
106			}
107
108			/**
109			* Get redirection page title or null.
110			*
111			* @return string\|null
112			*/
113			public function getRedirect(): ?string
114			{
115			if ($this->getText() && preg_match('/^#REDIRECT(?:ION)? ?\[\[([^]]+)]]/i', $this->getText(), $matches)) {
116			return (string)trim($matches[1]);
117			}
118
119			return null;
120			}
121
122			/**
123			* Edit the page with new text.
124			* Opti : EditInfo optional param ?
125			*
126			* @param string $newText
127			* @param EditInfo $editInfo
128			*
129			* @return bool
130			*/
131			public function editPage(string $newText, EditInfo $editInfo): bool
132			{
133			$revision = $this->page->getPageIdentifier();
134
135			$content = new Content($newText);
136			$revision = new Revision($content, $revision);
137
138			// TODO try/catch UsageExceptions badtoken
139			return $this->wiki->newRevisionSaver()->save($revision, $editInfo);
140			}
141
142			/**
143			* Create a new page.
144			*
145			* @param string $text
146			*
147			* @return bool
148			* @throws Exception
149			*/
150			public function createPage(string $text, ?EditInfo $editInfo=null): bool
151			{
152			if (!empty($this->page->getRevisions()->getLatest())) {
153			throw new \Exception('That page already exists');
154			}
155
156			$newContent = new Content($text);
157			// $identifier = $this->page->getPageIdentifier()
158			$title = new Title($this->title);
159			$identifier = new PageIdentifier($title);
160			$revision = new Revision($newContent, $identifier);
161			return $this->wiki->newRevisionSaver()->save($revision, $editInfo);
162			}
163
164			/**
165			* @param string $addText
166			* @param EditInfo $editInfo
167			*
168			* @return bool
169			* @throws Exception
170			*/
171			public function addToBottomOrCreatePage(string $addText, EditInfo $editInfo): bool
172			{
173			if (empty($this->page->getRevisions()->getLatest())) {
174			return $this->createPage($addText, $editInfo);
175			}
176
177			return $this->addToBottomOfThePage($addText, $editInfo);
178			}
179
180			/**
181			* Add text to the bottom of the article.
182			*
183			* @param string $addText
184			* @param EditInfo $editInfo
185			*
186			* @return bool
187			* @throws Exception
188			*/
189			public function addToBottomOfThePage(string $addText, EditInfo $editInfo): bool
190			{
191			if (empty($this->page->getRevisions()->getLatest())) {
192			throw new Exception('That page does not exist');
193			}
194			$oldText = $this->getText();
195			$newText = $oldText."\n".$addText;
196
197			return $this->editPage($newText, $editInfo);
198			}
199
200			/**
201			* todo Move to WikiTextUtil ?
202			* Replace serialized template and manage {{en}} prefix.
203			* Don't delete {{fr}} on frwiki.
204			*
205			* @param string $text wikitext of the page
206			* @param string $tplOrigin template text to replace
207			* @param string $tplReplace new template text
208			*
209			* @return string\|null
210			*/
211			public static function replaceTemplateInText(string $text, string $tplOrigin, string $tplReplace): string
212			{
213			if (preg_match_all(
214			'#(?<langTemp>{{[a-z][a-z]}} ?{{[a-z][a-z]}}) ?'.preg_quote($tplOrigin, '#').'#i',
215			$text,
216			$matches
217			)
218			) {
219			// Skip double lang prefix (like in "{{fr}} {{en}} {template}")
220			echo 'SKIP ! double lang prefix !';
221
222			return $text;
223			}
224
225			// hack // todo: autres patterns {{en}} ?
226			if (preg_match_all('#(?<langTemp>{{(?<lang>[a-z][a-z])}} *)?'.preg_quote($tplOrigin, '#').'#i', $text, $matches)
227			> 0
228			) {
229			foreach ($matches[0] as $num => $mention) {
230			$lang = $matches['lang'][$num] ?? '';
231			if(!empty($lang)){
232			$lang = Language::all2wiki($lang);
233			}
234
235			// detect inconsistency between lang indicator and lang param
236			// example : {{en}} {{template\|lang=ru}}
237			// BUG: prefix {{de}} incompatible avec langue de {{Ouvrage \|langue= \|prénom1=Hartmut \|nom1=Atsma
238			if (!empty($lang) && self::SKIP_LANG_INDICATOR !== $lang
239			&& !preg_match('#lang(ue)?='.$lang.'#i', $tplReplace)
240			&& !preg_match('#\\| ?langue= ?\\|#', $tplReplace)
241			) {
242			echo sprintf(
243			'prefix %s incompatible avec langue de %s',
244			$matches['langTemp'][$num],
245			$tplReplace
246			);
247
248			// skip all the replacements of that template
249			return $text; // return null ?
250			}
251
252			// FIX dirty : {{en}} mais pas de langue sur template...
253			if ($lang && preg_match('#\\| ?langue= ?\\|#', $tplReplace) > 0) {
254			$previousTpl = $tplReplace;
255			$tplReplace = str_replace('langue=', 'langue='.$lang, $tplReplace);
256			$text = str_replace($previousTpl, $tplReplace, $text);
257			}
258
259			// don't delete {{fr}} before {template} on frwiki
260			if (self::SKIP_LANG_INDICATOR === $lang) {
261			$text = str_replace($tplOrigin, $tplReplace, $text);
262
263			continue;
264			}
265
266			// replace {template} and {{lang}} {template}
267			$text = str_replace($mention, $tplReplace, $text);
268			$text = str_replace(
269			$matches['langTemp'][$num].$tplReplace,
270			$tplReplace,
271			$text
272			); // si 1er replace global sans
273			// {{en}}
274			}
275			}
276
277			return $text;
278			}
279
280			/**
281			* Extract <ref> data from text.
282			*
283			* @param $text string
284			*
285			* @return array
286			* @throws Exception
287			*/
288			public function extractRefFromText(string $text): ?array
289			{
290			$parser = new TagParser(); // todo ParserFactory
291			$refs = $parser->importHtml($text)->getRefValues(); // []
292
293			return (array)$refs;
294			}
295
296			/**
297			* TODO $url parameter
298			* TODO? refactor with : parse_str() + parse_url($url, PHP_URL_QUERY)
299			* check if any ref contains a targeted website/URL.
300			*
301			* @param array $refs
302			*
303			* @return array
304			*/
305			public function filterRefByURL(array $refs): array
306			{
307			$validRef = [];
308			foreach ($refs as $ref) {
309			if (preg_match(
310			'#(?<url>https?://(?:www\.)?lemonde\.fr/[^ \]]+)#i',
311			$ref,
312			$matches
313			) > 0
314			) {
315			$validRef[] = ['url' => $matches['url'], 'raw' => $ref];
316			}
317			}
318
319			return $validRef;
320			}
321			}
322

Dispositif / Wikibot

Push — master ( 511693...0349ad )

WikiPageAction::addToBottomOrCreatePage() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like