TemplateParser::findUserStyleSeparator() - Code Metrics - Inspection of "Style + maxlag, ISBN_SKIP" - Dispositif/Wikibot - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — dev ( eba0dc...b8ced8 )

by Dispositif

created 2020-04-04 13:13 UTC

TemplateParser::findUserStyleSeparator() A

↳ Parent: TemplateParser

Complexity

Conditions	4
Paths	4

Size

Total Lines	16
Code Lines	7

Duplication

Lines	0
Ratio	0 %

Importance

Changes	1
Bugs	0	Features	0

Metric	Value
cc	4
eloc	7
c	1
b	0
f	0
nc	4
nop	1
dl	0
loc	16
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\Domain\Utils;

use App\Domain\Models\Wiki\AbstractWikiTemplate;
use App\Domain\WikiTemplateFactory;
use Exception;
use LogicException;

/**
 * todo legacy.
 *
 * Class TemplateParser.
 */
abstract class TemplateParser extends WikiTextUtil
{
    /**
     * todo : simplify array if only one occurrence ?
     * todo refac extract/logic.
     *
     * @param string $tplName
     * @param string $text
     *
     * @return array
     *
     * @throws Exception
     */
    public static function parseAllTemplateByName(string $tplName, string $text): array
    {
        // Extract wikiText from that template
        $arrayTplText = self::findAllTemplatesByName($tplName, $text);

        if (empty($arrayTplText) || empty($arrayTplText[0])) {
            return [];
        }

        $result[$tplName] = [];

        $inc = -1;
        foreach ($arrayTplText as $tplText) {
            ++$inc;
            // store the raw text of the template
            $result[$tplName][$inc] = ['raw' => $tplText];

            // create an object of the template
            /**
             * @var AbstractWikiTemplate
             */
            try{
                $tplObject = WikiTemplateFactory::create($tplName);
            }catch (\Throwable $e){
                unset($e);
                continue;
            }

            if (!is_object($tplObject) || !is_subclass_of($tplObject, AbstractWikiTemplate::class)) {
                continue;
            }

            $data = self::parseDataFromTemplate($tplName, $tplText);
            $tplObject->hydrate($data);
            $tplObject->detectUserSeparator($tplText);

            $result[$tplName][$inc] += ['model' => $tplObject];
        }

        return (array) $result;
    }

    /**
     * Find all the recurrences of a wiki's template in a text.
     * Compatible with inclusion of sub-templates.
     * Example :
     * {{Infobox |pays={{pays|France}} }}
     * retourne array {{modèle|...}}.
     *
     * @param $templateName
     * @param $text
     *
     * @return array [ 0=>{{bla|...}}, 1=>{{bla|...}} ]
     */
    public static function findAllTemplatesByName(string $templateName, string $text): array
    {
        // TODO check {{fr}}
        $res = preg_match_all(
            "#{{[ \n]*".preg_quote(trim($templateName), '#')
            ."[ \t \n\r]*\|[^{}]*(?:{{[^{}]+}}[^{}]*)*}}#i",
            $text,
            $matches
        );

        if (false === $res) {
            return [];
        }

        return $matches[0];
        //OK : preg_match_all("#\{\{".preg_quote(trim($nommodele), '#')."[ \t \n\r]*\|([^\{\}]*(\{\{[^\{\}]+\}\}[^\{\}]*)*)\}\}#i", $text, $matches);
    }

    /**
     * Parsing of any wiki template from text and templateName
     * Using the first {{template}} definition found in text
     * todo legacy.
     *
     * @param string $tplName
     * @param string $text
     *
     * @return array
     */
    public static function parseDataFromTemplate(string $tplName, string $text): array
    {
        $text = str_replace("\n", '', $text);

        // check {{template}} in text
        $tplFounded = self::findFirstTemplateInText($tplName, $text);
class A
{
    function getObject()
    {
        return null;
    }

}

$a = new A();
$object = $a->getObject();

        // $matches[0] : {{template|...}}
        if (null === $tplFounded) {

            throw new LogicException("Template $tplName not found in text");
        }
        // $matches[1] : url=blabla|titre=Popo
        if (false === $tplFounded[1]) {
            throw new LogicException("No parameters found in $tplName");
        }
        // sub-template pipe | encoding
        $tplFounded[1] = self::encodeTemplatePipes($tplFounded[1]);

        // x flag => "\ " for space
        $res = preg_match_all(
            "/
			(
	  			[^|=]*=?                          # parameter name (or nothing)
		 		(
					[^|{}\[\]<>]*               # reject <i>,<ref>
					(?:\[[^\[\]]+])?              # [url text] or [text]
					(?:<!--[^<>]+-->)?    # commentary <!-- -->
					(?:{{[^}{]+}})?          # {{template}} but KO with {{tmp|...}}
					                               # test : {{bla@PIPE@bla}}
					(?:\[\[[^]]+]])?            # [[fu|bar]]
					[^|{}\[\]]*                 # accept <i>,<ref>
		 		)*
	 		)\|?
		/x",
            $tplFounded[1],
            $wikiParams
        );

        if (false === $res || 0 === $res || empty($wikiParams[1])) {
            throw new LogicException("Parameters from template '$tplName' can't be parsed");
        }

        return self::explodeParameterValue($wikiParams[1]);
    }

    /**
     * For multiple occurrences see findAllTemplatesByName().
     *
     * @param string $templateName
     * @param string $text
     *
     * @return array|null
     */
    private static function findFirstTemplateInText(string $templateName, string $text): ?array
    {
        // BUG marche pas avec :
        //        $text = '{{Ouvrage|auteur1 = Clément|titre = Les Borgia {{nobr|Alexandre {{VI}}}}}}'; // to debug
        //        $templateName = 'ouvrage'; // to debug

        //        $text = str_replace("\n", '', $text); // ??? todo regex multiline or encode char

        // todo: replace <!-- --> by encode char and memorize in var

        // hack : replace solitary { and } by encoded string CURLYBRACKET
        $text = preg_replace('#([^{]){([^{])#', '${1}CURLYBRACKETO$2', $text);
        $text = preg_replace('#([^}])}([^}])#', '${1}CURLYBRACKETC$2', $text);

        // TODO: implement better regex :(
        if (preg_match(
                '~{{ ?'.preg_quote($templateName, '~')."[ \t \n\r]*\|([^{}]*(?:{{[^{}]+}}[^{}]*)*)}}~i",
                $text,
                $matches
            ) > 0
        ) {
            array_walk(
                $matches,
                function (&$value) {
                    $value = str_replace(['CURLYBRACKETO', 'CURLYBRACKETC'], ['{', '}'], $value);
                }
            );

            return $matches;
        }

        return null;
    }

    /**
     * replace sub-templates pipes | by @PIPE@ in text.
     *
     * @param string $text
     *
     * @return string
     */
    protected static function encodeTemplatePipes(string $text): string
    {
        if (preg_match_all('#{{(?:[^{}]+)}}#m', $text, $subTmpl) > 0) {
            foreach ($subTmpl[0] as $sub) {
                $subSanit = str_replace('|', '@PIPE@', $sub);
                $text = str_replace($sub, $subSanit, $text);
            }
        }

        return $text;
    }

    /**
     * From ['fr', 'url=blabla', 'titre=popo']
     * To [ '1'=> 'fr', url' => 'blabla', 'titre' => 'popo' ].
     *
     * @param array $wikiLines ['url=blabla', 'titre=popo']
     *
     * @return array
     */
    protected static function explodeParameterValue(array $wikiLines): array
    {
        $data = [];
        $keyNum = 1;
        foreach ($wikiLines as $line) {
            if (empty($line)) {
                continue;
            }
            $line = str_replace(
                ["\t", "\n", "\r", ' '],
                ['', '', '', ' '],
                $line
            ); // perte cosmétique : est-ce bien ? + espace insécable remplacé par espace sécable

            // $line : fu = bar (OK : fu=bar=coco)
            $pos = strpos($line, '=');
            if (is_int($pos) && $pos >= 0) {
                $param = mb_strtolower(substr($line, 0, $pos), 'UTF-8');
                $value = substr($line, $pos + 1);
            }
            // No param name => take $keyNum as param name
            if (false === $pos) {
                $param = (string) $keyNum;
                $value = $line;
                ++$keyNum;
            }

            if (!isset($param) || !isset($value)) {
                throw new LogicException('param/value variable not defined');
            }

            // TODO : accept empty value ?
            if (!isset($value) || 0 === strlen(trim($value))) {
                continue;
            }
            // reverse the sub-template pipe encoding
            $value = str_replace('@PIPE@', '|', $value);
            $data[trim($param)] = trim($value);

        }

        return $data;
    }

    /**
     * Find text style of template : only pipe, space-pipe-space, pipe-space, return-pipe, etc.
     *
     * @param string $tplText
     *
     * @return string
     */
    public static function findUserStyleSeparator(string $tplText): string
    {
        // Fixed : {{fu\n    | bar}}
        if (preg_match('#{{[^}|]+\n +\|( ?)[^}]+}}#i', $tplText, $matches) > 0) {
            return "\n |".$matches[1];
        }
        // {{fu | bar}} (duplicate : because [^}|\n]+ allows final space...)
        if (preg_match('#{{[^}|\n]+([ \n]\|[ \n]?)[^}]+}}#i', $tplText, $matches) > 0) {
            return $matches[1];
        }
        // others : {{fu|bar}} ; {{fu\n|bar}} ; {{fu |bar}} ...
        if (preg_match('#{{[^}|\n]+([ \n]?\|[ \n]?)[^}]+}}#i', $tplText, $matches) > 0) {
            return $matches[1];
        }

        return ' |';
    }

    /**
     * Detect if "param     = bla".
     *
     * @param string $tplText
     *
     * @return bool
     */
    public static function isMultispacedTemplate(string $tplText):bool
    {
        // detect 4 spaces chars
        if (preg_match('#{{[^}]+ {4}[^}]+}}#i', $tplText)) {
            return true;
        }
        return false;
    }
}


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			use App\Domain\Models\Wiki\AbstractWikiTemplate;
13			use App\Domain\WikiTemplateFactory;
14			use Exception;
15			use LogicException;
16
17			/**
18			* todo legacy.
19			*
20			* Class TemplateParser.
21			*/
22			abstract class TemplateParser extends WikiTextUtil
23			{
24			/**
25			* todo : simplify array if only one occurrence ?
26			* todo refac extract/logic.
27			*
28			* @param string $tplName
29			* @param string $text
30			*
31			* @return array
32			*
33			* @throws Exception
34			*/
35			public static function parseAllTemplateByName(string $tplName, string $text): array
36			{
37			// Extract wikiText from that template
38			$arrayTplText = self::findAllTemplatesByName($tplName, $text);
39
40			if (empty($arrayTplText) \|\| empty($arrayTplText[0])) {
41			return [];
42			}
43
44			$result[$tplName] = [];
			1 ignored issue – show Comprehensibility Best Practice introduced 2019-11-25 16:53 UTC by Report Bug Copy Issue Report `$result` was never initialized. Although not strictly required by PHP, it is generally a good practice to add `$result = array();` before regardless. Loading history...
45			$inc = -1;
46			foreach ($arrayTplText as $tplText) {
47			++$inc;
48			// store the raw text of the template
49			$result[$tplName][$inc] = ['raw' => $tplText];
50
51			// create an object of the template
52			/**
53			* @var AbstractWikiTemplate
54			*/
55			try{
56			$tplObject = WikiTemplateFactory::create($tplName);
57			}catch (\Throwable $e){
58			unset($e);
59			continue;
60			}
61
62			if (!is_object($tplObject) \|\| !is_subclass_of($tplObject, AbstractWikiTemplate::class)) {
63			continue;
64			}
65
66			$data = self::parseDataFromTemplate($tplName, $tplText);
67			$tplObject->hydrate($data);
68			$tplObject->detectUserSeparator($tplText);
69
70			$result[$tplName][$inc] += ['model' => $tplObject];
71			}
72
73			return (array) $result;
74			}
75
76			/**
77			* Find all the recurrences of a wiki's template in a text.
78			* Compatible with inclusion of sub-templates.
79			* Example :
80			* {{Infobox \|pays={{pays\|France}} }}
81			* retourne array {{modèle\|...}}.
82			*
83			* @param $templateName
84			* @param $text
85			*
86			* @return array [ 0=>{{bla\|...}}, 1=>{{bla\|...}} ]
87			*/
88			public static function findAllTemplatesByName(string $templateName, string $text): array
89			{
90			// TODO check {{fr}}
91			$res = preg_match_all(
92			"#{{[ \n]*".preg_quote(trim($templateName), '#')
93			."[ \t \n\r]\\|[^{}](?:{{[^{}]+}}[^{}])}}#i",
94			$text,
95			$matches
96			);
97
98			if (false === $res) {
99			return [];
100			}
101
102			return $matches[0];
103			//OK : preg_match_all("#\{\{".preg_quote(trim($nommodele), '#')."[ \t \n\r]\\|([^\{\}](\{\{[^\{\}]+\}\}[^\{\}]))\}\}#i", $text, $matches);
104			}
105
106			/**
107			* Parsing of any wiki template from text and templateName
108			* Using the first {{template}} definition found in text
109			* todo legacy.
110			*
111			* @param string $tplName
112			* @param string $text
113			*
114			* @return array
115			*/
116			public static function parseDataFromTemplate(string $tplName, string $text): array
117			{
118			$text = str_replace("\n", '', $text);
119
120			// check {{template}} in text
121			$tplFounded = self::findFirstTemplateInText($tplName, $text);
			1 ignored issue – show Bug introduced 2019-11-25 16:53 UTC by Report Bug Copy Issue Report Are you sure the assignment to `$tplFounded` is correct as `self::findFirstTemplateInText($tplName, $text)` targeting `App\Domain\Utils\Templat...ndFirstTemplateInText()` seems to always return null. This check looks for function or method calls that always return null and whose return value is assigned to a variable. class A { function getObject() { return null; } } $a = new A(); $object = $a->getObject(); The method `getObject()` can return nothing but null, so it makes no sense to assign that value to a variable. The reason is most likely that a function or method is imcomplete or has been reduced for debug purposes. Loading history...
122
123			// $matches[0] : {{template\|...}}
124			if (null === $tplFounded) {
			0 ignored issues – show introduced 2019-11-25 16:53 UTC by Report Bug Copy Issue Report The condition `null === $tplFounded` is always `true`. Loading history...
125			throw new LogicException("Template $tplName not found in text");
126			}
127			// $matches[1] : url=blabla\|titre=Popo
128			if (false === $tplFounded[1]) {
129			throw new LogicException("No parameters found in $tplName");
130			}
131			// sub-template pipe \| encoding
132			$tplFounded[1] = self::encodeTemplatePipes($tplFounded[1]);
133
134			// x flag => "\ " for space
135			$res = preg_match_all(
136			"/
137			(
138			[^\|=]*=? # parameter name (or nothing)
139			(
140			[^\|{}\[\]<>]* # reject <i>,<ref>
141			(?:\[[^\[\]]+])? # [url text] or [text]
142			(?:<!--[^<>]+-->)? # commentary <!-- -->
143			(?:{{[^}{]+}})? # {{template}} but KO with {{tmp\|...}}
144			# test : {{bla@PIPE@bla}}
145			(?:\[\[[^]]+]])? # [[fu\|bar]]
146			[^\|{}\[\]]* # accept <i>,<ref>
147			)*
148			)\\|?
149			/x",
150			$tplFounded[1],
151			$wikiParams
152			);
153
154			if (false === $res \|\| 0 === $res \|\| empty($wikiParams[1])) {
155			throw new LogicException("Parameters from template '$tplName' can't be parsed");
156			}
157
158			return self::explodeParameterValue($wikiParams[1]);
159			}
160
161			/**
162			* For multiple occurrences see findAllTemplatesByName().
163			*
164			* @param string $templateName
165			* @param string $text
166			*
167			* @return array\|null
168			*/
169			private static function findFirstTemplateInText(string $templateName, string $text): ?array
170			{
171			// BUG marche pas avec :
172			// $text = '{{Ouvrage\|auteur1 = Clément\|titre = Les Borgia {{nobr\|Alexandre {{VI}}}}}}'; // to debug
173			// $templateName = 'ouvrage'; // to debug
174
175			// $text = str_replace("\n", '', $text); // ??? todo regex multiline or encode char
176
177			// todo: replace <!-- --> by encode char and memorize in var
178
179			// hack : replace solitary { and } by encoded string CURLYBRACKET
180			$text = preg_replace('#([^{]){([^{])#', '${1}CURLYBRACKETO$2', $text);
181			$text = preg_replace('#([^}])}([^}])#', '${1}CURLYBRACKETC$2', $text);
182
183			// TODO: implement better regex :(
184			if (preg_match(
185			'~{{ ?'.preg_quote($templateName, '~')."[ \t \n\r]\\|([^{}](?:{{[^{}]+}}[^{}]))}}~i",
186			$text,
187			$matches
188			) > 0
189			) {
190			array_walk(
191			$matches,
192			function (&$value) {
193			$value = str_replace(['CURLYBRACKETO', 'CURLYBRACKETC'], ['{', '}'], $value);
194			}
195			);
196
197			return $matches;
198			}
199
200			return null;
201			}
202
203			/**
204			* replace sub-templates pipes \| by @PIPE@ in text.
205			*
206			* @param string $text
207			*
208			* @return string
209			*/
210			protected static function encodeTemplatePipes(string $text): string
211			{
212			if (preg_match_all('#{{(?:[^{}]+)}}#m', $text, $subTmpl) > 0) {
213			foreach ($subTmpl[0] as $sub) {
214			$subSanit = str_replace('\|', '@PIPE@', $sub);
215			$text = str_replace($sub, $subSanit, $text);
216			}
217			}
218
219			return $text;
220			}
221
222			/**
223			* From ['fr', 'url=blabla', 'titre=popo']
224			* To [ '1'=> 'fr', url' => 'blabla', 'titre' => 'popo' ].
225			*
226			* @param array $wikiLines ['url=blabla', 'titre=popo']
227			*
228			* @return array
229			*/
230			protected static function explodeParameterValue(array $wikiLines): array
231			{
232			$data = [];
233			$keyNum = 1;
234			foreach ($wikiLines as $line) {
235			if (empty($line)) {
236			continue;
237			}
238			$line = str_replace(
239			["\t", "\n", "\r", ' '],
240			['', '', '', ' '],
241			$line
242			); // perte cosmétique : est-ce bien ? + espace insécable remplacé par espace sécable
243
244			// $line : fu = bar (OK : fu=bar=coco)
245			$pos = strpos($line, '=');
246			if (is_int($pos) && $pos >= 0) {
247			$param = mb_strtolower(substr($line, 0, $pos), 'UTF-8');
248			$value = substr($line, $pos + 1);
249			}
250			// No param name => take $keyNum as param name
251			if (false === $pos) {
252			$param = (string) $keyNum;
253			$value = $line;
254			++$keyNum;
255			}
256
257			if (!isset($param) \|\| !isset($value)) {
258			throw new LogicException('param/value variable not defined');
259			}
260
261			// TODO : accept empty value ?
262			if (!isset($value) \|\| 0 === strlen(trim($value))) {
263			continue;
264			}
265			// reverse the sub-template pipe encoding
266			$value = str_replace('@PIPE@', '\|', $value);
267			$data[trim($param)] = trim($value);
			1 ignored issue – show Comprehensibility Best Practice introduced 2019-11-25 16:53 UTC by Report Bug Copy Issue Report The variable `$param` does not seem to be defined for all execution paths leading up to this point. Loading history...
268			}
269
270			return $data;
271			}
272
273			/**
274			* Find text style of template : only pipe, space-pipe-space, pipe-space, return-pipe, etc.
275			*
276			* @param string $tplText
277			*
278			* @return string
279			*/
280			public static function findUserStyleSeparator(string $tplText): string
281			{
282			// Fixed : {{fu\n \| bar}}
283			if (preg_match('#{{[^}\|]+\n +\\|( ?)[^}]+}}#i', $tplText, $matches) > 0) {
284			return "\n \|".$matches[1];
285			}
286			// {{fu \| bar}} (duplicate : because [^}\|\n]+ allows final space...)
287			if (preg_match('#{{[^}\|\n]+([ \n]\\|[ \n]?)[^}]+}}#i', $tplText, $matches) > 0) {
288			return $matches[1];
289			}
290			// others : {{fu\|bar}} ; {{fu\n\|bar}} ; {{fu \|bar}} ...
291			if (preg_match('#{{[^}\|\n]+([ \n]?\\|[ \n]?)[^}]+}}#i', $tplText, $matches) > 0) {
292			return $matches[1];
293			}
294
295			return ' \|';
296			}
297
298			/**
299			* Detect if "param = bla".
300			*
301			* @param string $tplText
302			*
303			* @return bool
304			*/
305			public static function isMultispacedTemplate(string $tplText):bool
306			{
307			// detect 4 spaces chars
308			if (preg_match('#{{[^}]+ {4}[^}]+}}#i', $tplText)) {
309			return true;
310			}
311			return false;
312			}
313			}
314

Dispositif / Wikibot

Push — dev ( eba0dc...b8ced8 )

TemplateParser::findUserStyleSeparator() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like