WikiPageAction::__construct() - Code Metrics - Inspection of "Style & logic (scrutiniz)" - Dispositif/Wikibot - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — master ( 5eccc7...1faa52 )

by Dispositif

created 2020-04-18 20:24 UTC

WikiPageAction::__construct() A

↳ Parent: WikiPageAction

Complexity

Conditions	2
Paths	3

Size

Total Lines	10
Code Lines	7

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	0
CRAP Score	6

Importance

Changes	5
Bugs	0	Features	1

Metric	Value
cc	2
eloc	7
nc	3
nop	2
dl	0
loc	10
ccs	0
cts	7
cp	0
crap	6
rs	10
c	5
b	0
f	1

<?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;
    /**
     * Wiki namespace
     *
     * @var int
     */
    private $ns;

    /**
     * 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
    {
        // 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 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
     *
     * @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 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);
    }

    /**
     * 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 $tplReplace): string
    {
        // "{{en}} {{zh}} {{ouvrage...}}"
        if (preg_match_all(
            '#(?<langTemp>{{[a-z][a-z]}} ?{{[a-z][a-z]}}) ?'.preg_quote($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}}
                // 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		/**
36		* @var string
37		*/
38		private $title;
39		/**
40		* Wiki namespace
41		*
42		* @var int
43		*/
44		private $ns;
45
46		/**
47		* WikiPageAction constructor.
48		*
49		* @param MediawikiFactory $wiki
50		* @param string $title
51		*
52		* @throws Exception
53		*/
54		public function __construct(MediawikiFactory $wiki, string $title)
55		{
56		$this->wiki = $wiki;
57		$this->title = $title;
58
59		try {
60		$this->page = $wiki->newPageGetter()->getFromTitle($title);
61		$this->ns = $this->page->getPageIdentifier()->getTitle()->getNs();
62		} catch (Throwable $e) {
63		throw new Exception('Erreur construct WikiPageAction '.$e->getMessage().$e->getFile().$e->getLine());
64		}
65		}
66
67		/**
68		* Get wiki text from the page.
69		*
70		* @return string\|null
71		*/
72		public function getText(): ?string
73		{
74		// page doesn't exist
75		if (empty($this->page->getRevisions()->getLatest())) {
76		return null;
77		}
78
79		$latest = $this->page->getRevisions()->getLatest();
80
81		return ($latest) ? $latest->getContent()->getData() : null;
82		}
83
84		public function getNs(): ?int
85		{
86		return $this->ns;
87		}
88
89		public function getLastRevision(): ?Revision
90		{
91		// page doesn't exist
92		if (empty($this->page->getRevisions()->getLatest())) {
93		return null;
94		}
95
96		return $this->page->getRevisions()->getLatest();
97		}
98
99		public function getLastEditor(): ?string
100		{
101		// page doesn't exist
102		if (empty($this->page->getRevisions()->getLatest())) {
103		return null;
104		}
105
106		$latest = $this->page->getRevisions()->getLatest();
107
108		return ($latest) ? $latest->getUser() : null;
109		}
110
111		/**
112		* Check if a frwiki disambiguation page.
113		*
114		* @return bool
115		*/
116		public function isPageHomonymie(): bool
117		{
118		return false !== stristr($this->getText(), '{{homonymie');
119		}
120
121		/**
122		* Is it page with a redirection link ?
123		*
124		* @return bool
125		*/
126		public function isRedirect(): bool
127		{
128		return !empty($this->getRedirect());
129		}
130
131		/**
132		* Get redirection page title or null.
133		*
134		* @return string\|null
135		*/
136		public function getRedirect(): ?string
137		{
138		if ($this->getText() && preg_match('/^#REDIRECT(?:ION)? ?\[\[([^]]+)]]/i', $this->getText(), $matches)) {
139		return (string)trim($matches[1]);
140		}
141
142		return null;
143		}
144
145		/**
146		* Edit the page with new text.
147		* Opti : EditInfo optional param ?
148		*
149		* @param string $newText
150		* @param EditInfo $editInfo
151		*
152		* @return bool
153		*/
154		public function editPage(string $newText, EditInfo $editInfo): bool
155		{
156		$revision = $this->page->getPageIdentifier();
157
158		$content = new Content($newText);
159		$revision = new Revision($content, $revision);
160
161		// TODO try/catch UsageExceptions badtoken
162		return $this->wiki->newRevisionSaver()->save($revision, $editInfo);
163		}
164
165		/**
166		* Create a new page.
167		*
168		* @param string $text
169		*
170		* @return bool
171		* @throws Exception
172		*/
173		public function createPage(string $text, ?EditInfo $editInfo = null): bool
174		{
175		if (!empty($this->page->getRevisions()->getLatest())) {
176		throw new Exception('That page already exists');
177		}
178
179		$newContent = new Content($text);
180		// $identifier = $this->page->getPageIdentifier()
181		$title = new Title($this->title);
182		$identifier = new PageIdentifier($title);
183		$revision = new Revision($newContent, $identifier);
184
185		return $this->wiki->newRevisionSaver()->save($revision, $editInfo);
186		}
187
188		/**
189		* @param string $addText
190		* @param EditInfo $editInfo
191		*
192		* @return bool success
193		* @throws Exception
194		*/
195		public function addToBottomOrCreatePage(string $addText, EditInfo $editInfo): bool
196		{
197		if (empty($this->page->getRevisions()->getLatest())) {
198		return $this->createPage($addText, $editInfo);
199		}
200
201		return $this->addToBottomOfThePage($addText, $editInfo);
202		}
203
204		/**
205		* Add text to the bottom of the article.
206		*
207		* @param string $addText
208		* @param EditInfo $editInfo
209		*
210		* @return bool success
211		* @throws Exception
212		*/
213		public function addToBottomOfThePage(string $addText, EditInfo $editInfo): bool
214		{
215		if (empty($this->page->getRevisions()->getLatest())) {
216		throw new Exception('That page does not exist');
217		}
218		$oldText = $this->getText();
219		$newText = $oldText."\n".$addText;
220
221		return $this->editPage($newText, $editInfo);
222		}
223
224		/**
225		* todo Move to WikiTextUtil ?
226		* Replace serialized template and manage {{en}} prefix.
227		* Don't delete {{fr}} on frwiki.
228		*
229		* @param string $text wikitext of the page
230		* @param string $tplOrigin template text to replace
231		* @param string $tplReplace new template text
232		*
233		* @return string\|null
234		*/
235	6	public static function replaceTemplateInText(string $text, string $tplOrigin, string $tplReplace): string
236		{
237		// "{{en}} {{zh}} {{ouvrage...}}"
238	6	if (preg_match_all(
239	6	'#(?<langTemp>{{[a-z][a-z]}} ?{{[a-z][a-z]}}) ?'.preg_quote($tplOrigin, '#').'#i',
240	6	$text,
241	6	$matches
242		)
243		) {
244		// Skip double lang prefix (like in "{{fr}} {{en}} {template}")
245		echo 'SKIP ! double lang prefix !';
246
247		return $text;
248		}
249
250		// hack // todo: autres patterns {{en}} ?
251		// OK : {{en}} \n {{ouvrage}}
252	6	if (preg_match_all(
253	6	"#(?<langTemp>{{(?<lang>[a-z][a-z])}} *\n?)?".preg_quote($tplOrigin, '#').'#i',
254	6	$text,
255	6	$matches
256	6	) > 0
257		) {
258	6	foreach ($matches[0] as $num => $mention) {
259	6	$lang = $matches['lang'][$num] ?? '';
260	6	if (!empty($lang)) {
261	6	$lang = Language::all2wiki($lang);
262		}
263
264		// detect inconsistency between lang indicator and lang param
265		// example : {{en}} {{template\|lang=ru}}
266		// BUG: prefix {{de}} incompatible avec langue de {{Ouvrage \|langue= \|prénom1=Hartmut \|nom1=Atsma
267	6	if (!empty($lang) && self::SKIP_LANG_INDICATOR !== $lang
268	6	&& !preg_match('#lang(ue)?='.$lang.'#i', $tplReplace)
269	6	&& !preg_match('#\\| ?langue= ?\\|#', $tplReplace)
270		) {
271		echo sprintf(
272		'prefix %s incompatible avec langue de %s',
273		$matches['langTemp'][$num],
274		$tplReplace
275		);
276
277		// skip all the replacements of that template
278		return $text; // return null ?
279		}
280
281		// FIX dirty : {{en}} mais pas de langue sur template...
282	6	if ($lang && preg_match('#\\| ?langue= ?\\|#', $tplReplace) > 0) {
283	2	$previousTpl = $tplReplace;
284	2	$tplReplace = str_replace('langue=', 'langue='.$lang, $tplReplace);
285	2	$text = str_replace($previousTpl, $tplReplace, $text);
286		}
287
288		// don't delete {{fr}} before {template} on frwiki
289	6	if (self::SKIP_LANG_INDICATOR === $lang) {
290	3	$text = str_replace($tplOrigin, $tplReplace, $text);
291
292	3	continue;
293		}
294
295		// replace {template} and {{lang}} {template}
296	6	$text = str_replace($mention, $tplReplace, $text);
297	6	$text = str_replace(
298	6	$matches['langTemp'][$num].$tplReplace,
299	6	$tplReplace,
300	6	$text
301		); // si 1er replace global sans
302		// {{en}}
303		}
304		}
305
306	6	return $text;
307		}
308
309		/**
310		* Extract <ref> data from text.
311		*
312		* @param $text string
313		*
314		* @return array
315		* @throws Exception
316		*/
317		public function extractRefFromText(string $text): ?array
318		{
319		$parser = new TagParser(); // todo ParserFactory
320		$refs = $parser->importHtml($text)->getRefValues(); // []
321
322		return (array)$refs;
323		}
324
325		/**
326		* TODO $url parameter
327		* TODO? refactor with : parse_str() + parse_url($url, PHP_URL_QUERY)
328		* check if any ref contains a targeted website/URL.
329		*
330		* @param array $refs
331		*
332		* @return array
333		*/
334		public function filterRefByURL(array $refs): array
335		{
336		$validRef = [];
337		foreach ($refs as $ref) {
338		if (preg_match(
339		'#(?<url>https?://(?:www\.)?lemonde\.fr/[^ \]]+)#i',
340		$ref,
341		$matches
342		) > 0
343		) {
344		$validRef[] = ['url' => $matches['url'], 'raw' => $ref];
345		}
346		}
347
348		return $validRef;
349		}
350		}
351

Dispositif / Wikibot

Push — master ( 5eccc7...1faa52 )

WikiPageAction::__construct() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like