HtmlUniParser::setSiteBaseUrl() - Code Metrics - kosuha606/html-uni-parser - Measure and Improve Code Quality continuously with Scrutinizer

HtmlUniParser::setSiteBaseUrl() A
last analyzed 2020-03-07 19:38 UTC

↳ Parent: HtmlUniParser

Complexity

Conditions	1
Paths	1

Size

Total Lines	5
Code Lines	2

Duplication

Lines	0
Ratio	0 %

Importance

Changes	1
Bugs	0	Features	0

Metric	Value
cc	1
eloc	2
c	1
b	0
f	0
nc	1
nop	1
dl	0
loc	5
rs	10

<?php
declare(strict_types=1);

namespace kosuha606\HtmlUniParser;

use Assert\Assertion;
use kosuha606\HtmlUniParser\action\ComposeHtmlAction;
use kosuha606\HtmlUniParser\action\GetFirstMatchAction;
use kosuha606\HtmlUniParser\action\InitializeHtmlUniParserAction;
use kosuha606\HtmlUniParser\action\ValueStubAction;

/**
 * The main intrance point for work with the package
 * @package kosuha606\HtmlUniParser
 */
class HtmlUniParser extends BaseObject
{
    /**
     * Парсинг по сценарию каталога
     * @var
     */
    protected $catalogUrl;

    /**
     * Парсинг по сценарию поиска на сайте
     * @var
     */
    protected $searchUrl;

    /**
     * Парсинг по сценарию получения данных от одной страницы
     * @var
     */
    protected $pageUrl;

    /**
     * Заставить парсер получать внешний html
     * @var boolean
     */
    protected $forceOuterHtml = false;

    /**
     * Парсинг по урлам, сгенерированным генератором
     * @var
     */
    protected $urlGenerator;

    /** @var */
    protected $beforeDomCallback;

    /**
     * Кодировка сайта
     * @var string
     */
    protected $encoding = 'UTF-8';

    /**
     * @var string
     */
    protected $siteBaseUrl = '/';

    /**
     * @var bool
     */
    protected $resultLimit = false;

    /**
     * @var int
     */
    protected $sleepAfterRequest = 0;

    /**
     * @var bool
     */
    protected $goIntoCard = false;

    /**
     * @var string
     */
    protected $xpathItem;

    /**
     * @var string
     */
    protected $xpathLink;

    /**
     * @var string
     */
    protected $xpathTitle;

    /** @var string */
    protected $typeMech;

    /**
     * @var array
     */
    protected $xpathOnCard = [];

    /**
     * @var array
     */
    protected $callbacks = [];

    /**
     * Результат должен быть множественным
     * @var array
     */
    private $xpathOnCardMany = [];

    /**
     * Результат в формате HTML
     * @var array
     */
    private $xpathOnCardHtml = [];

    /** @var ZendBasedParser */
    private $zendParser;

    /**
     * HtmlUniParser constructor.
     * @param $config
     * @throws \Assert\AssertionFailedException
     */
    public function __construct($config, ZendBasedParser $parser)
    {
        parent::__construct($config);
        $this->checkPhpExtensions();
        InitializeHtmlUniParserAction::do($this, $parser);
    }

    /**
     * @throws \Assert\AssertionFailedException
     */
    private function checkPhpExtensions()
    {
        Assertion::keyNotExists(
            get_loaded_extensions(),
            'dom',
            'The dom extension in not loaded in system. HtmlUniParser cant work'
        );
        Assertion::keyNotExists(
            get_loaded_extensions(),
            'iconv',
            'The iconv extension in not loaded in system. HtmlUniParser cant work'
        );
    }

    /**
     * @return array
     */
    public function getXpathOnCard()
    {
        return $this->xpathOnCard;
    }

    /**
     * You can create instances of this class
     * by yourself or you can use this method
     * @param $config
     * @return HtmlUniParser
     * @throws \Assert\AssertionFailedException
     * @throws \ReflectionException
     */
    public static function create($config)
    {
        return Factory::createObject(
            [
                'class' => static::class,
            ],
            [
                $config,
                Factory::createObject(
                    [
                        'class' => ZendBasedParser::class,
                    ]
                ),
            ]
        );
    }

    /**
     * @param mixed $callbacs
     */
    public function addCallback($callbacs)
    {
        $this->callbacks[] = $callbacs;
    }

    /**
     * @param $node
     * @return string
     * @throws \Assert\AssertionFailedException
     */
    public function composeHtml($node): string
    {
        return ComposeHtmlAction::do($this, $node);
    }

    /**
     * @param $node
     * @return mixed
     */
    public function queryOuterHtml($node)
    {
        return $node->ownerDocument->saveXML($node);
    }

    /**
     * @param $nodes
     * @return string
     * @throws \Assert\AssertionFailedException
     */
    public function getFirstValue($nodes)
    {
        $val = GetFirstMatchAction::do($nodes);

        return $this->getValue($val);
    }

    /**
     * @param $nodes
     * @return string|string[]|null
     * @throws \Assert\AssertionFailedException
     */
    public function getFirstValueHtml($nodes)
    {
        /** @var \DOMElement $val */
        $val = GetFirstMatchAction::do($nodes);
        $html = $val->ownerDocument->saveHTML($val);
        // Удаляем картинки из спарсенного текста
        $html = preg_replace("/<img[^>]+\>/i", "", $html);

        return $this->proccessValue($html);
    }

    /**
     * @param $val
     * @return string
     * @throws \Assert\AssertionFailedException
     */
    public function getValue($val)
    {
        $result = null;
        if ($val instanceof \DOMAttr) {
            $result = ValueStubAction::do($val, 'value');
        }
        if ($val instanceof \DOMElement) {
            $result = ValueStubAction::do($val, 'nodeValue');
        }

        return $this->proccessValue(trim($result));
    }

    /**
     * @param $nodes
     * @return array
     */
    public function getAllValues($nodes)
    {
        $result = [];
        foreach ($nodes as $node) {
            $result[] = $this->getValue($node);
        }

        return $result;
    }

    /**
     * Спарсить ссылку, возможно каталог
     */
    public function parseUrl()
    {
        $this->zendParser->setSleepAfterRequest($this->sleepAfterRequest);
        $this->zendParser->setUrl($this->catalogUrl);
        $this->onBeforeDom();
        $items = $this->zendParser->dom($this->getEncoding(), $this->getTypeMech())->queryXpath($this->xpathItem);
        $pageHtml = $this->zendParser->getHtmlBuffer();
        $result = [];
        foreach ($items as $index => $item) {
            $newItem = [];
            $html = $this->composeHtml($item);
            $this->zendParser->setRawHtml($html);
            $link = $this->zendParser->dom($this->getEncoding(), $this->getTypeMech())->queryXpath($this->xpathLink);
            $link = $this->getFirstValue($link);
            if (preg_match('/^http(s)?:\/\/.*$/i', $link)) {
                $newItem['link'] = $link;
            } else {
                $newItem['link'] = $this->siteBaseUrl.$link;
            }
            if ($this->xpathTitle) {
                $this->zendParser->setRawHtml($pageHtml);
                $title = $this->zendParser->dom($this->getEncoding(), $this->getTypeMech())->queryXpath(
                    $this->xpathTitle
                );
                $title = $this->getFirstValue($title);
                $newItem['title'] = $title;
            }
            if ($this->goIntoCard && $newItem['link']) {
                $this->zendParser->setUrl($newItem['link']);
                foreach ($this->xpathOnCard as $param => $xpath) {
                    $this->onBeforeDom();
                    $temParam = $this->zendParser->dom($this->getEncoding(), $this->getTypeMech())->queryXpath($xpath);
                    if (in_array($param, $this->xpathOnCardMany)) {
                        $newItem[$param] = $this->getAllValues($temParam);
                    } elseif (in_array($param, $this->xpathOnCardHtml)) {
                        $newItem[$param] = $this->getFirstValueHtml($temParam);
                    } else {
                        $newItem[$param] = $this->getFirstValue($temParam);
                    }
                }
            }
            $this->handleCallbacks($newItem);
            $result[] = $newItem;
            // Не даем спарсить больше предела если установлен предел
            if ($this->resultLimit && $this->resultLimit <= ($index + 1)) {
                break;
            }
        }

        return $result;
    }

    /**
     * Спарсить результаты поиска
     */
    public function parseSearch($textQuery)
    {
        $this->catalogUrl = $this->searchUrl.$textQuery;

        return $this->parseUrl();
    }

    /**
     * Спарсить одну карточку
     */
    public function parseCard()
    {
        $this->zendParser->setUrl($this->pageUrl);
        foreach ($this->xpathOnCard as $param => $xpath) {
            $this->onBeforeDom();
            $temParam = $this->zendParser->dom($this->getEncoding(), $this->getTypeMech())->queryXpath($xpath);
            if (in_array($param, $this->xpathOnCardMany)) {
                $newItem[$param] = $this->getAllValues($temParam);
            } elseif (in_array($param, $this->xpathOnCardHtml)) {
                $newItem[$param] = $this->getFirstValueHtml($temParam);
            } else {
                $newItem[$param] = $this->getFirstValue($temParam);
            }
        }
        $this->handleCallbacks($newItem);

        return $newItem;
    }

    /**
     * @return array
     */
    public function parseGenerator()
    {
        $generator = $this->urlGenerator;
        $urls = $generator();
        $results = [];
        foreach ($urls as $url) {
            $this->pageUrl = $url;
            $results[] = $this->parseCard();
        }

        return $results;
    }

    /**
     * @param $lastItem
     * @return HtmlUniParser
     */
    public function handleCallbacks(&$lastItem)
    {
        foreach ($this->callbacks as &$callbac) {
            $callbac($lastItem, $this->pageUrl);
        }

        return $this;
    }

    /**
     * @return $this
     */
    public function onBeforeDom()
    {
        $callback = $this->beforeDomCallback;
        if ($callback) {
            $callback($this);
        }

        return $this;
    }

    /**
     * @return string
     */
    public function getEncoding(): string
    {
        return $this->encoding;
    }

    /**
     * @param string $encoding
     * @return HtmlUniParser
     */
    public function setEncoding(string $encoding)
    {
        $this->encoding = $encoding;

        return $this;
    }

    /**
     * @param $value
     * @return false|string
     */
    private function proccessValue($value)
    {
        if ($this->getEncoding() === 'UTF-8') {
            return $value;
        }
        $result = \iconv($this->getEncoding(), 'UTF-8', $value);

        return $result;
    }

    /**
     * @return string
     */
    public function getTypeMech(): string
    {
        return $this->typeMech ?: 'curl';
    }

    /**
     * @return ZendBasedParser
     */
    public function getZendParser(): ZendBasedParser
    {
        return $this->zendParser;
    }

    /**
     * @param string $typeMech
     * @return HtmlUniParser
     */
    public function setTypeMech(string $typeMech)
    {
        $this->typeMech = $typeMech;

        return $this;
    }

    /**
     * @param mixed $catalogUrl
     * @return HtmlUniParser
     */
    public function setCatalogUrl($catalogUrl)
    {
        $this->catalogUrl = $catalogUrl;

        return $this;
    }

    /**
     * @param mixed $searchUrl
     * @return HtmlUniParser
     */
    public function setSearchUrl($searchUrl)
    {
        $this->searchUrl = $searchUrl;

        return $this;
    }

    /**
     * @param mixed $pageUrl
     * @return HtmlUniParser
     */
    public function setPageUrl($pageUrl)
    {
        $this->pageUrl = $pageUrl;

        return $this;
    }

    /**
     * @param bool $forceOuterHtml
     * @return HtmlUniParser
     */
    public function setForceOuterHtml(bool $forceOuterHtml)
    {
        $this->forceOuterHtml = $forceOuterHtml;

        return $this;
    }

    /**
     * @return bool
     */
    public function isForceOuterHtml(): bool
    {
        return $this->forceOuterHtml;
    }

    /**
     * @param mixed $urlGenerator
     * @return HtmlUniParser
     */
    public function setUrlGenerator($urlGenerator)
    {
        $this->urlGenerator = $urlGenerator;

        return $this;
    }

    /**
     * @param string $siteBaseUrl
     * @return HtmlUniParser
     */
    public function setSiteBaseUrl(string $siteBaseUrl)
    {
        $this->siteBaseUrl = $siteBaseUrl;

        return $this;
    }

    /**
     * @param bool $resultLimit
     * @return HtmlUniParser
     */
    public function setResultLimit($resultLimit)
    {
        $this->resultLimit = $resultLimit;

        return $this;
    }

    /**
     * @param int $sleepAfterRequest
     * @return HtmlUniParser
     */
    public function setSleepAfterRequest(int $sleepAfterRequest)
    {
        $this->sleepAfterRequest = $sleepAfterRequest;

        return $this;
    }

    /**
     * @param bool $goIntoCard
     * @return HtmlUniParser
     */
    public function setGoIntoCard(bool $goIntoCard)
    {
        $this->goIntoCard = $goIntoCard;

        return $this;
    }

    /**
     * @param string $xpathItem
     * @return HtmlUniParser
     */
    public function setXpathItem(string $xpathItem)
    {
        $this->xpathItem = $xpathItem;

        return $this;
    }

    /**
     * @param string $xpathLink
     * @return HtmlUniParser
     */
    public function setXpathLink(string $xpathLink)
    {
        $this->xpathLink = $xpathLink;

        return $this;
    }

    /**
     * @param array $xpathOnCard
     * @return HtmlUniParser
     */
    public function setXpathOnCard(array $xpathOnCard)
    {
        $this->xpathOnCard = $xpathOnCard;

        return $this;
    }

    /**
     * @param array $callbacks
     * @return HtmlUniParser
     */
    public function setCallbacks(array $callbacks)
    {
        $this->callbacks = $callbacks;

        return $this;
    }

    /**
     * @param array $xpathOnCardMany
     * @return HtmlUniParser
     */
    public function setXpathOnCardMany(array $xpathOnCardMany)
    {
        $this->xpathOnCardMany = $xpathOnCardMany;

        return $this;
    }

    /**
     * @param array $xpathOnCardHtml
     * @return HtmlUniParser
     */
    public function setXpathOnCardHtml(array $xpathOnCardHtml)
    {
        $this->xpathOnCardHtml = $xpathOnCardHtml;

        return $this;
    }

    /**
     * @param ZendBasedParser $zendParser
     * @return HtmlUniParser
     */
    public function setZendParser(ZendBasedParser $zendParser)
    {
        $this->zendParser = $zendParser;

        return $this;
    }

    /**
     * @param mixed $beforeDomCallback
     * @return HtmlUniParser
     */
    public function setBeforeDomCallback($beforeDomCallback)
    {
        $this->beforeDomCallback = $beforeDomCallback;

        return $this;
    }

    /**
     * @return string
     */
    public function getXpathTitle()
    {
        return $this->xpathTitle;
    }

    /**
     * @param string $xpathTitle
     * @return HtmlUniParser
     */
    public function setXpathTitle($xpathTitle)
    {
        $this->xpathTitle = $xpathTitle;

        return $this;
    }
}


1			<?php
2			declare(strict_types=1);
3
4			namespace kosuha606\HtmlUniParser;
5
6			use Assert\Assertion;
7			use kosuha606\HtmlUniParser\action\ComposeHtmlAction;
8			use kosuha606\HtmlUniParser\action\GetFirstMatchAction;
9			use kosuha606\HtmlUniParser\action\InitializeHtmlUniParserAction;
10			use kosuha606\HtmlUniParser\action\ValueStubAction;
11
12			/**
13			* The main intrance point for work with the package
14			* @package kosuha606\HtmlUniParser
15			*/
16			class HtmlUniParser extends BaseObject
17			{
18			/**
19			* Парсинг по сценарию каталога
20			* @var
21			*/
22			protected $catalogUrl;
23
24			/**
25			* Парсинг по сценарию поиска на сайте
26			* @var
27			*/
28			protected $searchUrl;
29
30			/**
31			* Парсинг по сценарию получения данных от одной страницы
32			* @var
33			*/
34			protected $pageUrl;
35
36			/**
37			* Заставить парсер получать внешний html
38			* @var boolean
39			*/
40			protected $forceOuterHtml = false;
41
42			/**
43			* Парсинг по урлам, сгенерированным генератором
44			* @var
45			*/
46			protected $urlGenerator;
47
48			/** @var */
49			protected $beforeDomCallback;
50
51			/**
52			* Кодировка сайта
53			* @var string
54			*/
55			protected $encoding = 'UTF-8';
56
57			/**
58			* @var string
59			*/
60			protected $siteBaseUrl = '/';
61
62			/**
63			* @var bool
64			*/
65			protected $resultLimit = false;
66
67			/**
68			* @var int
69			*/
70			protected $sleepAfterRequest = 0;
71
72			/**
73			* @var bool
74			*/
75			protected $goIntoCard = false;
76
77			/**
78			* @var string
79			*/
80			protected $xpathItem;
81
82			/**
83			* @var string
84			*/
85			protected $xpathLink;
86
87			/**
88			* @var string
89			*/
90			protected $xpathTitle;
91
92			/** @var string */
93			protected $typeMech;
94
95			/**
96			* @var array
97			*/
98			protected $xpathOnCard = [];
99
100			/**
101			* @var array
102			*/
103			protected $callbacks = [];
104
105			/**
106			* Результат должен быть множественным
107			* @var array
108			*/
109			private $xpathOnCardMany = [];
110
111			/**
112			* Результат в формате HTML
113			* @var array
114			*/
115			private $xpathOnCardHtml = [];
116
117			/** @var ZendBasedParser */
118			private $zendParser;
119
120			/**
121			* HtmlUniParser constructor.
122			* @param $config
123			* @throws \Assert\AssertionFailedException
124			*/
125			public function __construct($config, ZendBasedParser $parser)
126			{
127			parent::__construct($config);
128			$this->checkPhpExtensions();
129			InitializeHtmlUniParserAction::do($this, $parser);
130			}
131
132			/**
133			* @throws \Assert\AssertionFailedException
134			*/
135			private function checkPhpExtensions()
136			{
137			Assertion::keyNotExists(
138			get_loaded_extensions(),
139			'dom',
140			'The dom extension in not loaded in system. HtmlUniParser cant work'
141			);
142			Assertion::keyNotExists(
143			get_loaded_extensions(),
144			'iconv',
145			'The iconv extension in not loaded in system. HtmlUniParser cant work'
146			);
147			}
148
149			/**
150			* @return array
151			*/
152			public function getXpathOnCard()
153			{
154			return $this->xpathOnCard;
155			}
156
157			/**
158			* You can create instances of this class
159			* by yourself or you can use this method
160			* @param $config
161			* @return HtmlUniParser
162			* @throws \Assert\AssertionFailedException
163			* @throws \ReflectionException
164			*/
165			public static function create($config)
166			{
167			return Factory::createObject(
168			[
169			'class' => static::class,
170			],
171			[
172			$config,
173			Factory::createObject(
174			[
175			'class' => ZendBasedParser::class,
176			]
177			),
178			]
179			);
180			}
181
182			/**
183			* @param mixed $callbacs
184			*/
185			public function addCallback($callbacs)
186			{
187			$this->callbacks[] = $callbacs;
188			}
189
190			/**
191			* @param $node
192			* @return string
193			* @throws \Assert\AssertionFailedException
194			*/
195			public function composeHtml($node): string
196			{
197			return ComposeHtmlAction::do($this, $node);
198			}
199
200			/**
201			* @param $node
202			* @return mixed
203			*/
204			public function queryOuterHtml($node)
205			{
206			return $node->ownerDocument->saveXML($node);
207			}
208
209			/**
210			* @param $nodes
211			* @return string
212			* @throws \Assert\AssertionFailedException
213			*/
214			public function getFirstValue($nodes)
215			{
216			$val = GetFirstMatchAction::do($nodes);
217
218			return $this->getValue($val);
219			}
220
221			/**
222			* @param $nodes
223			* @return string\|string[]\|null
224			* @throws \Assert\AssertionFailedException
225			*/
226			public function getFirstValueHtml($nodes)
227			{
228			/** @var \DOMElement $val */
229			$val = GetFirstMatchAction::do($nodes);
230			$html = $val->ownerDocument->saveHTML($val);
231			// Удаляем картинки из спарсенного текста
232			$html = preg_replace("/<img[^>]+\>/i", "", $html);
233
234			return $this->proccessValue($html);
235			}
236
237			/**
238			* @param $val
239			* @return string
240			* @throws \Assert\AssertionFailedException
241			*/
242			public function getValue($val)
243			{
244			$result = null;
245			if ($val instanceof \DOMAttr) {
246			$result = ValueStubAction::do($val, 'value');
247			}
248			if ($val instanceof \DOMElement) {
249			$result = ValueStubAction::do($val, 'nodeValue');
250			}
251
252			return $this->proccessValue(trim($result));
253			}
254
255			/**
256			* @param $nodes
257			* @return array
258			*/
259			public function getAllValues($nodes)
260			{
261			$result = [];
262			foreach ($nodes as $node) {
263			$result[] = $this->getValue($node);
264			}
265
266			return $result;
267			}
268
269			/**
270			* Спарсить ссылку, возможно каталог
271			*/
272			public function parseUrl()
273			{
274			$this->zendParser->setSleepAfterRequest($this->sleepAfterRequest);
275			$this->zendParser->setUrl($this->catalogUrl);
276			$this->onBeforeDom();
277			$items = $this->zendParser->dom($this->getEncoding(), $this->getTypeMech())->queryXpath($this->xpathItem);
278			$pageHtml = $this->zendParser->getHtmlBuffer();
279			$result = [];
280			foreach ($items as $index => $item) {
281			$newItem = [];
282			$html = $this->composeHtml($item);
283			$this->zendParser->setRawHtml($html);
284			$link = $this->zendParser->dom($this->getEncoding(), $this->getTypeMech())->queryXpath($this->xpathLink);
285			$link = $this->getFirstValue($link);
286			if (preg_match('/^http(s)?:\/\/.*$/i', $link)) {
287			$newItem['link'] = $link;
288			} else {
289			$newItem['link'] = $this->siteBaseUrl.$link;
290			}
291			if ($this->xpathTitle) {
292			$this->zendParser->setRawHtml($pageHtml);
293			$title = $this->zendParser->dom($this->getEncoding(), $this->getTypeMech())->queryXpath(
294			$this->xpathTitle
295			);
296			$title = $this->getFirstValue($title);
297			$newItem['title'] = $title;
298			}
299			if ($this->goIntoCard && $newItem['link']) {
300			$this->zendParser->setUrl($newItem['link']);
301			foreach ($this->xpathOnCard as $param => $xpath) {
302			$this->onBeforeDom();
303			$temParam = $this->zendParser->dom($this->getEncoding(), $this->getTypeMech())->queryXpath($xpath);
304			if (in_array($param, $this->xpathOnCardMany)) {
305			$newItem[$param] = $this->getAllValues($temParam);
306			} elseif (in_array($param, $this->xpathOnCardHtml)) {
307			$newItem[$param] = $this->getFirstValueHtml($temParam);
308			} else {
309			$newItem[$param] = $this->getFirstValue($temParam);
310			}
311			}
312			}
313			$this->handleCallbacks($newItem);
314			$result[] = $newItem;
315			// Не даем спарсить больше предела если установлен предел
316			if ($this->resultLimit && $this->resultLimit <= ($index + 1)) {
317			break;
318			}
319			}
320
321			return $result;
322			}
323
324			/**
325			* Спарсить результаты поиска
326			*/
327			public function parseSearch($textQuery)
328			{
329			$this->catalogUrl = $this->searchUrl.$textQuery;
330
331			return $this->parseUrl();
332			}
333
334			/**
335			* Спарсить одну карточку
336			*/
337			public function parseCard()
338			{
339			$this->zendParser->setUrl($this->pageUrl);
340			foreach ($this->xpathOnCard as $param => $xpath) {
341			$this->onBeforeDom();
342			$temParam = $this->zendParser->dom($this->getEncoding(), $this->getTypeMech())->queryXpath($xpath);
343			if (in_array($param, $this->xpathOnCardMany)) {
344			$newItem[$param] = $this->getAllValues($temParam);
345			} elseif (in_array($param, $this->xpathOnCardHtml)) {
346			$newItem[$param] = $this->getFirstValueHtml($temParam);
347			} else {
348			$newItem[$param] = $this->getFirstValue($temParam);
349			}
350			}
351			$this->handleCallbacks($newItem);
352
353			return $newItem;
354			}
355
356			/**
357			* @return array
358			*/
359			public function parseGenerator()
360			{
361			$generator = $this->urlGenerator;
362			$urls = $generator();
363			$results = [];
364			foreach ($urls as $url) {
365			$this->pageUrl = $url;
366			$results[] = $this->parseCard();
367			}
368
369			return $results;
370			}
371
372			/**
373			* @param $lastItem
374			* @return HtmlUniParser
375			*/
376			public function handleCallbacks(&$lastItem)
377			{
378			foreach ($this->callbacks as &$callbac) {
379			$callbac($lastItem, $this->pageUrl);
380			}
381
382			return $this;
383			}
384
385			/**
386			* @return $this
387			*/
388			public function onBeforeDom()
389			{
390			$callback = $this->beforeDomCallback;
391			if ($callback) {
392			$callback($this);
393			}
394
395			return $this;
396			}
397
398			/**
399			* @return string
400			*/
401			public function getEncoding(): string
402			{
403			return $this->encoding;
404			}
405
406			/**
407			* @param string $encoding
408			* @return HtmlUniParser
409			*/
410			public function setEncoding(string $encoding)
411			{
412			$this->encoding = $encoding;
413
414			return $this;
415			}
416
417			/**
418			* @param $value
419			* @return false\|string
420			*/
421			private function proccessValue($value)
422			{
423			if ($this->getEncoding() === 'UTF-8') {
424			return $value;
425			}
426			$result = \iconv($this->getEncoding(), 'UTF-8', $value);
427
428			return $result;
429			}
430
431			/**
432			* @return string
433			*/
434			public function getTypeMech(): string
435			{
436			return $this->typeMech ?: 'curl';
437			}
438
439			/**
440			* @return ZendBasedParser
441			*/
442			public function getZendParser(): ZendBasedParser
443			{
444			return $this->zendParser;
445			}
446
447			/**
448			* @param string $typeMech
449			* @return HtmlUniParser
450			*/
451			public function setTypeMech(string $typeMech)
452			{
453			$this->typeMech = $typeMech;
454
455			return $this;
456			}
457
458			/**
459			* @param mixed $catalogUrl
460			* @return HtmlUniParser
461			*/
462			public function setCatalogUrl($catalogUrl)
463			{
464			$this->catalogUrl = $catalogUrl;
465
466			return $this;
467			}
468
469			/**
470			* @param mixed $searchUrl
471			* @return HtmlUniParser
472			*/
473			public function setSearchUrl($searchUrl)
474			{
475			$this->searchUrl = $searchUrl;
476
477			return $this;
478			}
479
480			/**
481			* @param mixed $pageUrl
482			* @return HtmlUniParser
483			*/
484			public function setPageUrl($pageUrl)
485			{
486			$this->pageUrl = $pageUrl;
487
488			return $this;
489			}
490
491			/**
492			* @param bool $forceOuterHtml
493			* @return HtmlUniParser
494			*/
495			public function setForceOuterHtml(bool $forceOuterHtml)
496			{
497			$this->forceOuterHtml = $forceOuterHtml;
498
499			return $this;
500			}
501
502			/**
503			* @return bool
504			*/
505			public function isForceOuterHtml(): bool
506			{
507			return $this->forceOuterHtml;
508			}
509
510			/**
511			* @param mixed $urlGenerator
512			* @return HtmlUniParser
513			*/
514			public function setUrlGenerator($urlGenerator)
515			{
516			$this->urlGenerator = $urlGenerator;
517
518			return $this;
519			}
520
521			/**
522			* @param string $siteBaseUrl
523			* @return HtmlUniParser
524			*/
525			public function setSiteBaseUrl(string $siteBaseUrl)
526			{
527			$this->siteBaseUrl = $siteBaseUrl;
528
529			return $this;
530			}
531
532			/**
533			* @param bool $resultLimit
534			* @return HtmlUniParser
535			*/
536			public function setResultLimit($resultLimit)
537			{
538			$this->resultLimit = $resultLimit;
539
540			return $this;
541			}
542
543			/**
544			* @param int $sleepAfterRequest
545			* @return HtmlUniParser
546			*/
547			public function setSleepAfterRequest(int $sleepAfterRequest)
548			{
549			$this->sleepAfterRequest = $sleepAfterRequest;
550
551			return $this;
552			}
553
554			/**
555			* @param bool $goIntoCard
556			* @return HtmlUniParser
557			*/
558			public function setGoIntoCard(bool $goIntoCard)
559			{
560			$this->goIntoCard = $goIntoCard;
561
562			return $this;
563			}
564
565			/**
566			* @param string $xpathItem
567			* @return HtmlUniParser
568			*/
569			public function setXpathItem(string $xpathItem)
570			{
571			$this->xpathItem = $xpathItem;
572
573			return $this;
574			}
575
576			/**
577			* @param string $xpathLink
578			* @return HtmlUniParser
579			*/
580			public function setXpathLink(string $xpathLink)
581			{
582			$this->xpathLink = $xpathLink;
583
584			return $this;
585			}
586
587			/**
588			* @param array $xpathOnCard
589			* @return HtmlUniParser
590			*/
591			public function setXpathOnCard(array $xpathOnCard)
592			{
593			$this->xpathOnCard = $xpathOnCard;
594
595			return $this;
596			}
597
598			/**
599			* @param array $callbacks
600			* @return HtmlUniParser
601			*/
602			public function setCallbacks(array $callbacks)
603			{
604			$this->callbacks = $callbacks;
605
606			return $this;
607			}
608
609			/**
610			* @param array $xpathOnCardMany
611			* @return HtmlUniParser
612			*/
613			public function setXpathOnCardMany(array $xpathOnCardMany)
614			{
615			$this->xpathOnCardMany = $xpathOnCardMany;
616
617			return $this;
618			}
619
620			/**
621			* @param array $xpathOnCardHtml
622			* @return HtmlUniParser
623			*/
624			public function setXpathOnCardHtml(array $xpathOnCardHtml)
625			{
626			$this->xpathOnCardHtml = $xpathOnCardHtml;
627
628			return $this;
629			}
630
631			/**
632			* @param ZendBasedParser $zendParser
633			* @return HtmlUniParser
634			*/
635			public function setZendParser(ZendBasedParser $zendParser)
636			{
637			$this->zendParser = $zendParser;
638
639			return $this;
640			}
641
642			/**
643			* @param mixed $beforeDomCallback
644			* @return HtmlUniParser
645			*/
646			public function setBeforeDomCallback($beforeDomCallback)
647			{
648			$this->beforeDomCallback = $beforeDomCallback;
649
650			return $this;
651			}
652
653			/**
654			* @return string
655			*/
656			public function getXpathTitle()
657			{
658			return $this->xpathTitle;
659			}
660
661			/**
662			* @param string $xpathTitle
663			* @return HtmlUniParser
664			*/
665			public function setXpathTitle($xpathTitle)
666			{
667			$this->xpathTitle = $xpathTitle;
668
669			return $this;
670			}
671			}
672

kosuha606 / html-uni-parser

HtmlUniParser::setSiteBaseUrl() A last analyzed 2020-03-07 19:38 UTC

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like

HtmlUniParser::setSiteBaseUrl() A
last analyzed 2020-03-07 19:38 UTC