WikiTextUtil::stripFinalPoint() - 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

WikiTextUtil::stripFinalPoint() A

↳ Parent: WikiTextUtil

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

Metric	Value
cc	2
eloc	3
nc	2
nop	1
dl	0
loc	7
ccs	0
cts	4
cp	0
crap	6
rs	10
c	0
b	0
f	0

<?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\Domain\Utils;

class WikiTextUtil extends TextUtil
{
    /**
     * todo {{ref}}
     *
     * @param string $text
     *
     * @return array [0=>['<ref>fu</ref>', 'fu'], 1=> ...]
     */
    public static function extractAllRefs(string $text): array
    {
        // s = "\n" include in "." // m = ^multiline$
        // Exclusion des imbrications
        if (!preg_match_all('#<ref[^>]*>((?:(?!</ref>).)*)</ref>#ism', $text, $refs, PREG_SET_ORDER)) {
            return [];
        }

        return $refs;
    }

    /**
     * remove wiki encoding : italic, bold, links [ ] and [[fu|bar]] => bar
     * replace non-breaking spaces
     * replace {{lang|en|fubar}} => fubar.
     *
     * @param      $text
     * @param bool $stripcomment
     *
     * @return string
     */
    public static function unWikify(string $text, ?bool $stripcomment = true): string
    {
        if (true === $stripcomment) {
            $text = self::removeHTMLcomments($text);
        }

        $text = str_replace(
            ['[', ']', "'''", "''", ' '],
            ['', '', '', '', ' '],
            preg_replace(
                [
                    "#\[\[[^|\]]*\|([^]]*)]]#",
                    '#{{ ?(?:lang|langue) ?\|[^|]+\| ?(?:texte=)?([^{}=]+)(?:\|dir=rtl)?}}#i',
                    "#&[\w\d]{2,7};#",
                ],
                ['$1', '$1', ''],
                $text
            )
        );
        // {{Lien|Jeffrey Robinson}} => Jeffrey Robinson
        $text = preg_replace('#{{ ?lien ?\| ?([^|}]+) ?}}#i', '${1}', $text);

        $text = strip_tags($text, '<sup><sub>');

        return $text;
    }

    public static function isWikify(string $text): bool
    {
        if (self::unWikify($text) !== $text) {
            return true;
        }

        return false;
    }

    /**
     * Generate wikilink from string.
     *
     * @param string      $label
     * @param string|null $page
     *
     * @return string
     */
    public static function wikilink(string $label, ?string $page = null): string
    {
        $label = trim($label);
        $page = trim($page);

        // fu_bar => [[fu_bar]] / Fu, fu => [[fu]]
        if (empty($page) || self::str2WikiTitle($label) === self::str2WikiTitle($page)) {
            return '[['.$label.']]';
        }

        // fu, bar => [[Bar|fu]]
        return sprintf(
            '[[%s|%s]]',
            self::str2WikiTitle($page),
            $label
        );
    }

    /**
     * "fu_bar_ " => "Fu bar".
     */
    private static function str2WikiTitle(string $str): string
    {
        return TextUtil::mb_ucfirst(trim(str_replace('_', ' ', $str)));
    }

    /**
     * Get page titles from wiki encoded links.
     * (but not others projects links like [[wikt:bla]].
     *
     * @param string $text
     *
     * @return array|null
     */
    public static function getWikilinkPages(string $text): ?array
    {
        if (preg_match_all('#\[\[([^:|\]]+)(?:\|[^|\]]*)?]]#', $text, $matches) > 0) {
            return $matches[1];
        }

        return null;
    }

    /**
     * Strip external links (http://) from wiki text.
     * "[http://google.fr Google]" => "Google"
     * "bla [http://google.fr]" => "bla"
     *
     * @param string $text
     *
     * @return string
     */
    public static function stripExternalLink(string $text): string
    {
        $text = preg_replace('#\[(https?://[^][<>\s"]+) *((?<= )[^\n\]]*|)\]#i', '${2}', $text);

        return trim($text);
    }

    /**
     * @param string $text
     *
     * @return bool
     */
    public static function isCommented(string $text): bool
    {
        $text = str_replace('<!-- Paramètre obligatoire -->', '', $text);

        //ou preg_match('#<\!--(?!-->).*-->#s', '', $text); // plus lourd mais précis
        return (preg_match('#<!--[^>]*-->#', $text) > 0) ? true : false;
    }

    /**
     * Remove '<!--', '-->', and everything between.
     * To avoid leaving blank lines, when a comment is both preceded
     * and followed by a newline (ignoring spaces), trim leading and
     * trailing spaces and one of the newlines.
     * (c) WikiMedia /includes/parser/Sanitizer.php.
     *
     * @param string $text
     *
     * @return string
     */
    public static function removeHTMLcomments(string $text)
    {
        while (false !== ($start = mb_strpos($text, '<!--'))) {
            $end = mb_strpos($text, '-->', $start + 4);
            if (false === $end) {
                // Unterminated comment; bail out
                break;
            }
            $end += 3;
            // Trim space and newline if the comment is both
            // preceded and followed by a newline
            $spaceStart = max($start - 1, 0);
            $spaceLen = $end - $spaceStart;
            while (' ' === substr($text, $spaceStart, 1) && $spaceStart > 0) {
                --$spaceStart;
                ++$spaceLen;
            }
            while (' ' === substr($text, $spaceStart + $spaceLen, 1)) {
                ++$spaceLen;
            }
            if ("\n" === substr($text, $spaceStart, 1)
                && "\n" === substr($text, $spaceStart + $spaceLen, 1)
            ) {
                // Remove the comment, leading and trailing
                // spaces, and leave only one newline.
                $text = substr_replace($text, "\n", $spaceStart, $spaceLen + 1);
            } else {
                // Remove just the comment.
                $text = substr_replace($text, '', $start, $end - $start);
            }
        }

        return $text;
    }

    /**
     * Strip the final point (".") as in <ref> ending.
     *
     * @param string $str
     *
     * @return string
     */
    public static function stripFinalPoint(string $str): string
    {
        if (substr($str, -1, 1) === '.') {
            return substr($str, 0, strlen($str) - 1);
        }

        return $str;
    }
}


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\Domain\Utils;
11
12		class WikiTextUtil extends TextUtil
13		{
14		/**
15		* todo {{ref}}
16		*
17		* @param string $text
18		*
19		* @return array [0=>['<ref>fu</ref>', 'fu'], 1=> ...]
20		*/
21		public static function extractAllRefs(string $text): array
22		{
23		// s = "\n" include in "." // m = ^multiline$
24		// Exclusion des imbrications
25		if (!preg_match_all('#<ref[^>]>((?:(?!</ref>).))</ref>#ism', $text, $refs, PREG_SET_ORDER)) {
26		return [];
27		}
28
29		return $refs;
30		}
31
32		/**
33		* remove wiki encoding : italic, bold, links [ ] and [[fu\|bar]] => bar
34		* replace non-breaking spaces
35		* replace {{lang\|en\|fubar}} => fubar.
36		*
37		* @param $text
38		* @param bool $stripcomment
39		*
40		* @return string
41		*/
42	60	public static function unWikify(string $text, ?bool $stripcomment = true): string
43		{
44	60	if (true === $stripcomment) {
45	60	$text = self::removeHTMLcomments($text);
46		}
47
48	60	$text = str_replace(
49	60	['[', ']', "'''", "''", ' '],
50	60	['', '', '', '', ' '],
51	60	preg_replace(
52		[
53	60	"#\[\[[^\|\]]\\|([^]])]]#",
54		'#{{ ?(?:lang\|langue) ?\\|[^\|]+\\| ?(?:texte=)?([^{}=]+)(?:\\|dir=rtl)?}}#i',
55		"#&[\w\d]{2,7};#",
56		],
57	60	['$1', '$1', ''],
58	60	$text
59		)
60		);
61		// {{Lien\|Jeffrey Robinson}} => Jeffrey Robinson
62	60	$text = preg_replace('#{{ ?lien ?\\| ?([^\|}]+) ?}}#i', '${1}', $text);
63
64	60	$text = strip_tags($text, '<sup><sub>');
65
66	60	return $text;
67		}
68
69	24	public static function isWikify(string $text): bool
70		{
71	24	if (self::unWikify($text) !== $text) {
72	3	return true;
73		}
74
75	21	return false;
76		}
77
78		/**
79		* Generate wikilink from string.
80		*
81		* @param string $label
82		* @param string\|null $page
83		*
84		* @return string
85		*/
86	4	public static function wikilink(string $label, ?string $page = null): string
87		{
88	4	$label = trim($label);
89	4	$page = trim($page);
90
91		// fu_bar => [[fu_bar]] / Fu, fu => [[fu]]
92	4	if (empty($page) \|\| self::str2WikiTitle($label) === self::str2WikiTitle($page)) {
93	3	return '[['.$label.']]';
94		}
95
96		// fu, bar => [[Bar\|fu]]
97	2	return sprintf(
98	2	'[[%s\|%s]]',
99	2	self::str2WikiTitle($page),
100	2	$label
101		);
102		}
103
104		/**
105		* "fu_bar_ " => "Fu bar".
106		*/
107	4	private static function str2WikiTitle(string $str): string
108		{
109	4	return TextUtil::mb_ucfirst(trim(str_replace('_', ' ', $str)));
110		}
111
112		/**
113		* Get page titles from wiki encoded links.
114		* (but not others projects links like [[wikt:bla]].
115		*
116		* @param string $text
117		*
118		* @return array\|null
119		*/
120	1	public static function getWikilinkPages(string $text): ?array
121		{
122	1	if (preg_match_all('#\[\[([^:\|\]]+)(?:\\|[^\|\]]*)?]]#', $text, $matches) > 0) {
123	1	return $matches[1];
124		}
125
126		return null;
127		}
128
129		/**
130		* Strip external links (http://) from wiki text.
131		* "[http://google.fr Google]" => "Google"
132		* "bla [http://google.fr]" => "bla"
133		*
134		* @param string $text
135		*
136		* @return string
137		*/
138	2	public static function stripExternalLink(string $text): string
139		{
140	2	$text = preg_replace('#\[(https?://[^][<>\s"]+) ((?<= )[^\n\]]\|)\]#i', '${2}', $text);
141
142	2	return trim($text);
143		}
144
145		/**
146		* @param string $text
147		*
148		* @return bool
149		*/
150	24	public static function isCommented(string $text): bool
151		{
152	24	$text = str_replace('<!-- Paramètre obligatoire -->', '', $text);
153
154		//ou preg_match('#<\!--(?!-->).*-->#s', '', $text); // plus lourd mais précis
155	24	return (preg_match('#<!--[^>]*-->#', $text) > 0) ? true : false;
156		}
157
158		/**
159		* Remove '<!--', '-->', and everything between.
160		* To avoid leaving blank lines, when a comment is both preceded
161		* and followed by a newline (ignoring spaces), trim leading and
162		* trailing spaces and one of the newlines.
163		* (c) WikiMedia /includes/parser/Sanitizer.php.
164		*
165		* @param string $text
166		*
167		* @return string
168		*/
169	61	public static function removeHTMLcomments(string $text)
170		{
171	61	while (false !== ($start = mb_strpos($text, '<!--'))) {
172	2	$end = mb_strpos($text, '-->', $start + 4);
173	2	if (false === $end) {
174		// Unterminated comment; bail out
175		break;
176		}
177	2	$end += 3;
178		// Trim space and newline if the comment is both
179		// preceded and followed by a newline
180	2	$spaceStart = max($start - 1, 0);
181	2	$spaceLen = $end - $spaceStart;
182	2	while (' ' === substr($text, $spaceStart, 1) && $spaceStart > 0) {
183		--$spaceStart;
184		++$spaceLen;
185		}
186	2	while (' ' === substr($text, $spaceStart + $spaceLen, 1)) {
187	1	++$spaceLen;
188		}
189	2	if ("\n" === substr($text, $spaceStart, 1)
190	2	&& "\n" === substr($text, $spaceStart + $spaceLen, 1)
191		) {
192		// Remove the comment, leading and trailing
193		// spaces, and leave only one newline.
194		$text = substr_replace($text, "\n", $spaceStart, $spaceLen + 1);
195		} else {
196		// Remove just the comment.
197	2	$text = substr_replace($text, '', $start, $end - $start);
198		}
199		}
200
201	61	return $text;
202		}
203
204		/**
205		* Strip the final point (".") as in <ref> ending.
206		*
207		* @param string $str
208		*
209		* @return string
210		*/
211		public static function stripFinalPoint(string $str): string
212		{
213		if (substr($str, -1, 1) === '.') {
214		return substr($str, 0, strlen($str) - 1);
215		}
216
217		return $str;
218		}
219		}
220

Dispositif / Wikibot

Push — master ( 5eccc7...1faa52 )

WikiTextUtil::stripFinalPoint() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like