TemplateParser::explodeParameterValue() - 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

TemplateParser::explodeParameterValue() B

↳ Parent: TemplateParser

Complexity

Conditions	10
Paths	14

Size

Total Lines	41
Code Lines	24

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	24
CRAP Score	10.0064

Importance

Changes	1
Bugs	0	Features	0

Metric	Value
cc	10
eloc	24
nc	14
nop	1
dl	0
loc	41
ccs	24
cts	25
cp	0.96
crap	10.0064
rs	7.6666
c	1
b	0
f	0

How to fix Complexity

<?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;
use Throwable;

/**
 * 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);
    }

    /**
     * todo refactor + check if @notused
     * 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); // todo WTF ?

        // check {{template}} in text
        $tplFounded = self::findFirstTemplateInText($tplName, $text);

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

            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 (empty($param) || !isset($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		use Throwable;
17
18		/**
19		* todo legacy.
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		* @throws Exception
33		*/
34	2	public static function parseAllTemplateByName(string $tplName, string $text): array
35		{
36		// Extract wikiText from that template
37	2	$arrayTplText = self::findAllTemplatesByName($tplName, $text);
38
39	2	if (empty($arrayTplText) \|\| empty($arrayTplText[0])) {
40		return [];
41		}
42
43	2	$result[$tplName] = [];
44	2	$inc = -1;
45	2	foreach ($arrayTplText as $tplText) {
46	2	++$inc;
47		// store the raw text of the template
48	2	$result[$tplName][$inc] = ['raw' => $tplText];
49
50		// create an object of the template
51		/**
52		* @var AbstractWikiTemplate
53		*/
54		try {
55	2	$tplObject = WikiTemplateFactory::create($tplName);
56		} catch (Throwable $e) {
57		unset($e);
58		continue;
59		}
60
61	2	if (!is_object($tplObject) \|\| !is_subclass_of($tplObject, AbstractWikiTemplate::class)) {
62		continue;
63		}
64
65	2	$data = self::parseDataFromTemplate($tplName, $tplText);
66	2	$tplObject->hydrate($data);
67	2	$tplObject->detectUserSeparator($tplText);
68
69	2	$result[$tplName][$inc] += ['model' => $tplObject];
70		}
71
72	2	return (array)$result;
73		}
74
75		/**
76		* Find all the recurrences of a wiki's template in a text.
77		* Compatible with inclusion of sub-templates.
78		* Example :
79		* {{Infobox \|pays={{pays\|France}} }}
80		* retourne array {{modèle\|...}}.
81		*
82		* @param $templateName
83		* @param $text
84		*
85		* @return array [ 0=>{{bla\|...}}, 1=>{{bla\|...}} ]
86		*/
87	2	public static function findAllTemplatesByName(string $templateName, string $text): array
88		{
89		// TODO check {{fr}}
90	2	$res = preg_match_all(
91	2	"#{{[ \n]".preg_quote(trim($templateName), '#')."[ \t \n\r]\\|[^{}](?:{{[^{}]+}}[^{}])*}}#i",
92	2	$text,
93	2	$matches
94		);
95
96	2	if (false === $res) {
97		return [];
98		}
99
100	2	return $matches[0];
101		//OK : preg_match_all("#\{\{".preg_quote(trim($nommodele), '#')."[ \t \n\r]\\|([^\{\}](\{\{[^\{\}]+\}\}[^\{\}]))\}\}#i", $text, $matches);
102		}
103
104		/**
105		* todo refactor + check if @notused
106		* Parsing of any wiki template from text and templateName
107		* Using the first {{template}} definition found in text
108		* todo legacy.
109		*
110		* @param string $tplName
111		* @param string $text
112		*
113		* @return array
114		*/
115	27	public static function parseDataFromTemplate(string $tplName, string $text): array
116		{
117	27	$text = str_replace("\n", '', $text); // todo WTF ?
118
119		// check {{template}} in text
120	27	$tplFounded = self::findFirstTemplateInText($tplName, $text);
121
122		// $matches[0] : {{template\|...}}
123	27	if ($tplFounded === null) {
		1 ignored issue – show introduced 2020-04-11 02:40 UTC by Report Bug Copy Issue Report The condition `$tplFounded === null` is always `true`. Loading history...
124		throw new LogicException("Template $tplName not found in text");
125		}
126		// $matches[1] : url=blabla\|titre=Popo
127	27	if (false === $tplFounded[1]) {
128		throw new LogicException("No parameters found in $tplName");
129		}
130		// sub-template pipe \| encoding
131	27	$tplFounded[1] = self::encodeTemplatePipes($tplFounded[1]);
132
133		// x flag => "\ " for space
134	27	$res = preg_match_all(
135	27	"/
136		(
137		[^\|=]*=? # parameter name (or nothing)
138		(
139		[^\|{}\[\]<>]* # reject <i>,<ref>
140		(?:\[[^\[\]]+])? # [url text] or [text]
141		(?:<!--[^<>]+-->)? # commentary <!-- -->
142		(?:{{[^}{]+}})? # {{template}} but KO with {{tmp\|...}}
143		# test : {{bla@PIPE@bla}}
144		(?:\[\[[^]]+]])? # [[fu\|bar]]
145		[^\|{}\[\]]* # accept <i>,<ref>
146		)*
147		)\\|?
148		/x",
149	27	$tplFounded[1],
150	27	$wikiParams
151		);
152
153	27	if (false === $res \|\| 0 === $res \|\| empty($wikiParams[1])) {
154		throw new LogicException("Parameters from template '$tplName' can't be parsed");
155		}
156
157	27	return self::explodeParameterValue($wikiParams[1]);
158		}
159
160		/**
161		* For multiple occurrences see findAllTemplatesByName().
162		*
163		* @param string $templateName
164		* @param string $text
165		*
166		* @return array\|null
167		*/
168	27	private static function findFirstTemplateInText(string $templateName, string $text): ?array
169		{
170		// BUG marche pas avec :
171		// $text = '{{Ouvrage\|auteur1 = Clément\|titre = Les Borgia {{nobr\|Alexandre {{VI}}}}}}'; // to debug
172		// $templateName = 'ouvrage'; // to debug
173
174		// $text = str_replace("\n", '', $text); // ??? todo regex multiline or encode char
175
176		// todo: replace <!-- --> by encode char and memorize in var
177
178		// hack : replace solitary { and } by encoded string CURLYBRACKET
179	27	$text = preg_replace('#([^{]){([^{])#', '${1}CURLYBRACKETO$2', $text);
180	27	$text = preg_replace('#([^}])}([^}])#', '${1}CURLYBRACKETC$2', $text);
181
182		// TODO: implement better regex :(
183	27	if (preg_match(
184	27	'~{{ ?'.preg_quote($templateName, '~')."[ \t \n\r]\\|([^{}](?:{{[^{}]+}}[^{}]))}}~i",
185	27	$text,
186	27	$matches
187	27	) > 0
188		) {
189	27	array_walk(
190	27	$matches,
191		function (&$value) {
192	27	$value = str_replace(['CURLYBRACKETO', 'CURLYBRACKETC'], ['{', '}'], $value);
193	27	}
194		);
195
196	27	return $matches;
197		}
198
199		return null;
200		}
201
202		/**
203		* replace sub-templates pipes \| by @PIPE@ in text.
204		*
205		* @param string $text
206		*
207		* @return string
208		*/
209	27	protected static function encodeTemplatePipes(string $text): string
210		{
211	27	if (preg_match_all('#{{(?:[^{}]+)}}#m', $text, $subTmpl) > 0) {
212	1	foreach ($subTmpl[0] as $sub) {
213	1	$subSanit = str_replace('\|', '@PIPE@', $sub);
214	1	$text = str_replace($sub, $subSanit, $text);
215		}
216		}
217
218	27	return $text;
219		}
220
221		/**
222		* From ['fr', 'url=blabla', 'titre=popo']
223		* To [ '1'=> 'fr', url' => 'blabla', 'titre' => 'popo' ].
224		*
225		* @param array $wikiLines ['url=blabla', 'titre=popo']
226		*
227		* @return array
228		*/
229	27	protected static function explodeParameterValue(array $wikiLines): array
230		{
231	27	$data = [];
232	27	$keyNum = 1;
233	27	foreach ($wikiLines as $line) {
234	27	if (empty($line)) {
235	27	continue;
236		}
237	27	$line = str_replace(
238	27	["\t", "\n", "\r", ' '],
239	27	['', '', '', ' '],
240	27	$line
241		); // perte cosmétique : est-ce bien ? + espace insécable remplacé par espace sécable
242
243		// $line : fu = bar (OK : fu=bar=coco)
244	27	$pos = strpos($line, '=');
245	27	if (is_int($pos) && $pos >= 0) {
246	26	$param = mb_strtolower(substr($line, 0, $pos), 'UTF-8');
247	26	$value = substr($line, $pos + 1);
248		}
249		// No param name => take $keyNum as param name
250	27	if (false === $pos) {
251	2	$param = (string)$keyNum;
252	2	$value = $line;
253	2	++$keyNum;
254		}
255
256	27	if (empty($param) \|\| !isset($value)) {
257		throw new LogicException('param/value variable not defined');
258		}
259
260		// TODO : accept empty value ?
261	27	if (!isset($value) \|\| 0 === strlen(trim($value))) {
262	5	continue;
263		}
264		// reverse the sub-template pipe encoding
265	27	$value = str_replace('@PIPE@', '\|', $value);
266	27	$data[trim($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...
267		}
268
269	27	return $data;
270		}
271
272		/**
273		* Find text style of template : only pipe, space-pipe-space, pipe-space, return-pipe, etc.
274		*
275		* @param string $tplText
276		*
277		* @return string
278		*/
279	31	public static function findUserStyleSeparator(string $tplText): string
280		{
281		// Fixed : {{fu\n \| bar}}
282	31	if (preg_match('#{{[^}\|]+\n +\\|( ?)[^}]+}}#i', $tplText, $matches) > 0) {
283	2	return "\n \|".$matches[1];
284		}
285		// {{fu \| bar}} (duplicate : because [^}\|\n]+ allows final space...)
286	29	if (preg_match('#{{[^}\|\n]+([ \n]\\|[ \n]?)[^}]+}}#i', $tplText, $matches) > 0) {
287	6	return $matches[1];
288		}
289		// others : {{fu\|bar}} ; {{fu\n\|bar}} ; {{fu \|bar}} ...
290	24	if (preg_match('#{{[^}\|\n]+([ \n]?\\|[ \n]?)[^}]+}}#i', $tplText, $matches) > 0) {
291	24	return $matches[1];
292		}
293
294		return ' \|';
295		}
296
297		/**
298		* Detect if "param = bla".
299		*
300		* @param string $tplText
301		*
302		* @return bool
303		*/
304	25	public static function isMultispacedTemplate(string $tplText): bool
305		{
306		// detect 4 spaces chars
307	25	if (preg_match('#{{[^}]+ {4}[^}]+}}#i', $tplText)) {
308		return true;
309		}
310
311	25	return false;
312		}
313		}
314

Dispositif / Wikibot

Push — master ( 5eccc7...1faa52 )

TemplateParser::explodeParameterValue() B

Complexity

Size

Duplication

Code Coverage

Importance

How to fix Complexity

Long Method

Duplication Side-by-Side

Filter issues like