HtmlHandler::walkDomNodesAndReplaceOnlyTextNodes() - Code Metrics - Inspection of "fix: replacement of emoji in headings" - ttskch/esaba - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Pull Request — master (#12)

by Takashi

created 2018-04-15 09:33 UTC

HtmlHandler::walkDomNodesAndReplaceOnlyTextNodes() B

↳ Parent: HtmlHandler

Complexity

Conditions	5
Paths	5

Size

Total Lines	20
Code Lines	10

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	10
CRAP Score	5

Importance

Changes

Metric	Value
cc	5
eloc	10
nc	5
nop	2
dl	0
loc	20
ccs	10
cts	10
cp	1
crap	5
rs	8.8571
c	0
b	0
f	0

<?php

namespace Ttskch\Esa;

use Symfony\Component\DomCrawler\Crawler;
use Symfony\Component\Routing\Generator\UrlGeneratorInterface;

class HtmlHandler
{
    /**
     * @var Crawler
     */
    private $crawler;

    /**
     * @var UrlGeneratorInterface
     */
    private $urlGenerator;

    /**
     * @var EmojiManager
     */
    private $emojiManager;

    /**
     * @var string
     */
    private $teamName;

    /**
     * @param array $replacements
     */
    public function __construct(Crawler $crawler, UrlGeneratorInterface $urlGenerator, EmojiManager $emojiManager, $teamName)
    {
        $this->crawler = $crawler;
        $this->urlGenerator = $urlGenerator;
        $this->emojiManager = $emojiManager;
        $this->teamName = $teamName;
    }

    /**
     * @param string $html
     * @return $this
     */
    public function initialize($html)
    {
        $this->crawler->clear();
        $this->crawler->addHtmlContent($html);

        return $this;
    }

    /**
     * @return string
     */
    public function dumpHtml()
    {
        $this->ensureInitialized();

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

    /**
     * @param array $replacements map of [regexp pattern => replacement].
     */
    public function replaceHtml(array $replacements)
    {
        $this->ensureInitialized();

        $html = $this->crawler->html();

        foreach ($replacements as $pattern => $replacement) {
            $html = preg_replace($pattern, $replacement, $html);
        }

        $this->initialize($html);
    }

    /**
     * @param array $replacements map of [regexp pattern => replacement].
     */
    public function replaceText(array $replacements)
    {
        $this->ensureInitialized();

        $domNode = $this->crawler->getNode(0);

        $this->walkDomNodesAndReplaceOnlyTextNodes($domNode, $replacements);

        $this->crawler->clear();
        $this->crawler->addNode($domNode);
    }

    /**
     * @param \DOMNode $node
     * @param array $replacements map of [regexp pattern => replacement].
     */
    public function walkDomNodesAndReplaceOnlyTextNodes(\DOMNode $node, array $replacements)
    {
        if ($node->nodeType === XML_TEXT_NODE) {
            $text = $node->textContent;

            foreach ($replacements as $pattern => $replacement) {
                $text = preg_replace($pattern, $replacement, $text);
            }

            $node->textContent = $text;

            return;
        }

        if ($node->childNodes->length < 1) {
            return;
        }

        foreach ($node->childNodes as $childNode) {
            $this->walkDomNodesAndReplaceOnlyTextNodes($childNode, $replacements);
        }
    }

    /**
     * Replace links to other post with links to see the post on esaba.
     *
     * @param string $routeName
     * @param string $routeVariableName
     */
    public function replacePostUrls($routeName, $routeVariableName)
    {
        $backReferenceNumberForPostId = null;
        $backReferenceNumberForAnchorHash = null;
        $pattern = $this->getPostUrlPattern($backReferenceNumberForPostId, $backReferenceNumberForAnchorHash);
        $walker = $this->getATagWalkerForPostUrls($pattern, $backReferenceNumberForPostId, $backReferenceNumberForAnchorHash, $routeName, $routeVariableName);

        $this->replaceATagWithWalker($pattern, $walker);
    }

    /**
     * Disable @mention links.
     */
    public function disableMentionLinks()
    {
        $pattern = $this->getMentionLinkPattern();
        $walker = $this->getATagWalkerForMentionLinks($pattern);

        $this->replaceATagWithWalker($pattern, $walker);
    }

    /**
     * Replace <a> tag href values for specified regexp pattern with closure returns map of ['pattern' => regexp pattern, 'replacement' => replacement].
     *
     * @param string $pattern
     * @param \Closure $walker
     */
    public function replaceATagWithWalker($pattern, \Closure $walker)
    {
        $this->ensureInitialized();

        $targetATags = $this->crawler->filter('a')->reduce($this->getATagReducer($pattern));
        $replacements = $targetATags->each($walker);
        $replacements = array_combine(array_column($replacements, 'pattern'), array_column($replacements, 'replacement'));

        $this->replaceHtml($replacements);
    }

    /**
     * @param string $backReferenceNumberForPostId For returning position of post id in regexp pattern.
     * @param string $backReferenceNumberForAnchorHash For returning position of anchor hash regexp pattern.
     * @return string
     */
    public function getPostUrlPattern(&$backReferenceNumberForPostId, &$backReferenceNumberForAnchorHash)
    {
        $backReferenceNumberForPostId = 3;
        $backReferenceNumberForAnchorHash = 5;

        return sprintf('#^((https?:)?//%s\.esa\.io)?/posts/(\d+)(/|/edit/?)?(\#.+)?$#', $this->teamName);
    }

    /**
     * @return string
     */
    public function getMentionLinkPattern()
    {
        return '#/members/([^\'"]+)#';
    }

    /**
     * Return closure reduces ATags Crawler with regexp pattern for href value.
     *
     * @param string $pattern
     * @return \Closure
     */
    public function getATagReducer($pattern)
    {
        $reducer = function (Crawler $node) use ($pattern) {
            preg_match($pattern, $node->attr('href'), $matches);

            return boolval($matches);
        };

        return $reducer;
    }

    /**
     * Return closure returns map of ['pattern' => regexp pattern, 'replacement' => replacement] for href value of post urls.
     *
     * @param string $pattern
     * @param int $backReferenceNumberForPostId
     * @param int $backReferenceNumberForAnchorHash
     * @param string $routeName
     * @param string $routeVariableName
     * @return \Closure
     */
    public function getATagWalkerForPostUrls($pattern, $backReferenceNumberForPostId, $backReferenceNumberForAnchorHash, $routeName, $routeVariableName)
    {
        $that = $this;

        $walker = function (Crawler $node) use ($pattern, $backReferenceNumberForPostId, $backReferenceNumberForAnchorHash, $routeName, $routeVariableName, $that) {
            preg_match($pattern, $node->attr('href'), $matches);
            $href = $matches[0];
            $postId = $matches[$backReferenceNumberForPostId];
            $anchorHash = isset($matches[$backReferenceNumberForAnchorHash]) ? $matches[$backReferenceNumberForAnchorHash] : '';

            $pattern = sprintf('/href=(\'|")%s\1/', str_replace('/', '\/', $href));
            $replacement = sprintf('href="%s%s"', $that->urlGenerator->generate($routeName, [$routeVariableName => $postId]), $anchorHash);

            return [
                'pattern' => $pattern,
                'replacement' => $replacement,
            ];
        };

        return $walker;
    }

    /**
     * Return closure returns map of ['pattern' => regexp pattern, 'replacement' => replacement] for href value of mention links.
     *
     * @param string $pattern
     * @return \Closure
     */
    public function getATagWalkerForMentionLinks($pattern)
    {
        $walker = function (Crawler $node) use ($pattern) {
            preg_match($pattern, $node->attr('href'), $matches);
            $href = $matches[0];

            $pattern = sprintf('/href=(\'|")%s\1/', str_replace('/', '\/', $href));
            $replacement = '';

            return [
                'pattern' => $pattern,
                'replacement' => $replacement,
            ];
        };

        return $walker;
    }

    /**
     * Replace emoji codes only in text content of each nodes with img tags.
     */
    public function replaceEmojiCodes()
    {
        // find emoji codes.
        preg_match_all('/:([^\s:<>\'"]+):/', $this->crawler->text(), $matches);

        $tempReplacements = [];
        foreach (array_unique($matches[1]) as $name) {
            $pattern = sprintf('/:%s:/', preg_quote($name));
            $replacement = sprintf('__ESABA_IMG_TAG__%s__ESABA_IMG_TAG__', $name);

            $tempReplacements[$pattern] = $replacement;
        }

        // set temporarily replaced html content.
        $this->replaceText($tempReplacements);

        $replacements = [];
        foreach (array_values($tempReplacements) as $tempReplacement) {
            preg_match('/__ESABA_IMG_TAG__(.+)__ESABA_IMG_TAG__/', $tempReplacement, $matches);
            $name = $matches[1];

            $pattern = sprintf('/%s/', preg_quote($tempReplacement));
            $replacement = sprintf('<img src="%s" class="emoji" title=":%s:" alt=":%s:">', $this->emojiManager->getImageUrl($name), $name, $name);

            $replacements[$pattern] = $replacement;
        }

        $this->replaceHtml($replacements);
    }

    /**
     * Return map of ['id' => id, 'text' => text] of headings as TOC.
     *
     * @return array
     */
    public function getToc()
    {
        $this->ensureInitialized();

        $toc = $this->crawler->filter('h1, h2, h3')->each($this->getWalkerForToc());

        return $toc;
    }

    /**
     * Return closure returns map of ['id' => id, 'text' => text] of h tags.
     *
     * @return \Closure
     */
    public function getWalkerForToc()
    {
        $walker = function (Crawler $node) {
            return [
                'id' => $node->attr('id'),
                'text' => trim(str_replace($node->filter('a')->text(), '', $node->text())),
            ];
        };

        return $walker;
    }

    private function ensureInitialized()
    {
        if (!$this->crawler->count()) {
            throw new \LogicException('Initialize before using.');
        }
    }
}


1		<?php
2
3		namespace Ttskch\Esa;
4
5		use Symfony\Component\DomCrawler\Crawler;
6		use Symfony\Component\Routing\Generator\UrlGeneratorInterface;
7
8		class HtmlHandler
9		{
10		/**
11		* @var Crawler
12		*/
13		private $crawler;
14
15		/**
16		* @var UrlGeneratorInterface
17		*/
18		private $urlGenerator;
19
20		/**
21		* @var EmojiManager
22		*/
23		private $emojiManager;
24
25		/**
26		* @var string
27		*/
28		private $teamName;
29
30		/**
31		* @param array $replacements
32		*/
33	50	public function __construct(Crawler $crawler, UrlGeneratorInterface $urlGenerator, EmojiManager $emojiManager, $teamName)
34		{
35	50	$this->crawler = $crawler;
36	50	$this->urlGenerator = $urlGenerator;
37	50	$this->emojiManager = $emojiManager;
38	50	$this->teamName = $teamName;
39	50	}
40
41		/**
42		* @param string $html
43		* @return $this
44		*/
45	13	public function initialize($html)
46		{
47	13	$this->crawler->clear();
48	13	$this->crawler->addHtmlContent($html);
49
50	13	return $this;
51		}
52
53		/**
54		* @return string
55		*/
56	3	public function dumpHtml()
57		{
58	3	$this->ensureInitialized();
59
60	2	return $this->crawler->html();
61		}
62
63		/**
64		* @param array $replacements map of [regexp pattern => replacement].
65		*/
66		public function replaceHtml(array $replacements)
67		{
68		$this->ensureInitialized();
69	2
70		$html = $this->crawler->html();
71	2
72	2	foreach ($replacements as $pattern => $replacement) {
73	2	$html = preg_replace($pattern, $replacement, $html);
74	2	}
75
76	2	$this->initialize($html);
77	2	}
78
79		/**
80		* @param array $replacements map of [regexp pattern => replacement].
81		*/
82	2	public function replaceText(array $replacements)
83		{
84	2	$this->ensureInitialized();
85	2
86		$domNode = $this->crawler->getNode(0);
87	2
88	2	$this->walkDomNodesAndReplaceOnlyTextNodes($domNode, $replacements);
89
90		$this->crawler->clear();
91		$this->crawler->addNode($domNode);
92		}
93
94		/**
95		* @param \DOMNode $node
96	4	* @param array $replacements map of [regexp pattern => replacement].
97		*/
98	4	public function walkDomNodesAndReplaceOnlyTextNodes(\DOMNode $node, array $replacements)
99		{
100	3	if ($node->nodeType === XML_TEXT_NODE) {
101	3	$text = $node->textContent;
102	3
103		foreach ($replacements as $pattern => $replacement) {
104	3	$text = preg_replace($pattern, $replacement, $text);
105	3	}
106
107		$node->textContent = $text;
108
109		return;
110		}
111
112	28	if ($node->childNodes->length < 1) {
113		return;
114	28	}
115	28
116		foreach ($node->childNodes as $childNode) {
117	28	$this->walkDomNodesAndReplaceOnlyTextNodes($childNode, $replacements);
118		}
119		}
120
121		/**
122		* Replace links to other post with links to see the post on esaba.
123	3	*
124		* @param string $routeName
125	3	* @param string $routeVariableName
126		*/
127		public function replacePostUrls($routeName, $routeVariableName)
128		{
129		$backReferenceNumberForPostId = null;
130		$backReferenceNumberForAnchorHash = null;
131		$pattern = $this->getPostUrlPattern($backReferenceNumberForPostId, $backReferenceNumberForAnchorHash);
132		$walker = $this->getATagWalkerForPostUrls($pattern, $backReferenceNumberForPostId, $backReferenceNumberForAnchorHash, $routeName, $routeVariableName);
133
134		$this->replaceATagWithWalker($pattern, $walker);
135		}
136	28
137	26	/**
138		* Disable @mention links.
139	26	*/
140	28	public function disableMentionLinks()
141		{
142	28	$pattern = $this->getMentionLinkPattern();
143		$walker = $this->getATagWalkerForMentionLinks($pattern);
144
145		$this->replaceATagWithWalker($pattern, $walker);
146		}
147
148		/**
149		* Replace <a> tag href values for specified regexp pattern with closure returns map of ['pattern' => regexp pattern, 'replacement' => replacement].
150		*
151		* @param string $pattern
152		* @param \Closure $walker
153		*/
154		public function replaceATagWithWalker($pattern, \Closure $walker)
155	3	{
156		$this->ensureInitialized();
157	3
158		$targetATags = $this->crawler->filter('a')->reduce($this->getATagReducer($pattern));
159	3	$replacements = $targetATags->each($walker);
160	1	$replacements = array_combine(array_column($replacements, 'pattern'), array_column($replacements, 'replacement'));
161	1
162	1	$this->replaceHtml($replacements);
163	1	}
164
165	1	/**
166	1	* @param string $backReferenceNumberForPostId For returning position of post id in regexp pattern.
167		* @param string $backReferenceNumberForAnchorHash For returning position of anchor hash regexp pattern.
168		* @return string
169	1	*/
170	1	public function getPostUrlPattern(&$backReferenceNumberForPostId, &$backReferenceNumberForAnchorHash)
171		{
172	3	$backReferenceNumberForPostId = 3;
173		$backReferenceNumberForAnchorHash = 5;
174	3
175		return sprintf('#^((https?:)?//%s\.esa\.io)?/posts/(\d+)(/\|/edit/?)?(\#.+)?$#', $this->teamName);
176		}
177
178		/**
179		* @return string
180		*/
181		public function getMentionLinkPattern()
182		{
183		return '#/members/([^\'"]+)#';
184		}
185	3
186	1	/**
187	1	* Return closure reduces ATags Crawler with regexp pattern for href value.
188		*
189	1	* @param string $pattern
190	1	* @return \Closure
191		*/
192		public function getATagReducer($pattern)
193	1	{
194	1	$reducer = function (Crawler $node) use ($pattern) {
195		preg_match($pattern, $node->attr('href'), $matches);
196	3
197		return boolval($matches);
198	3	};
199
200		return $reducer;
201		}
202
203		/**
204	10	* Return closure returns map of ['pattern' => regexp pattern, 'replacement' => replacement] for href value of post urls.
205		*
206	10	* @param string $pattern
207		* @param int $backReferenceNumberForPostId
208	9	* @param int $backReferenceNumberForAnchorHash
209		* @param string $routeName
210	9	* @param string $routeVariableName
211		* @return \Closure
212	9	*/
213	8	public function getATagWalkerForPostUrls($pattern, $backReferenceNumberForPostId, $backReferenceNumberForAnchorHash, $routeName, $routeVariableName)
214	8	{
215		$that = $this;
216	8
217		$walker = function (Crawler $node) use ($pattern, $backReferenceNumberForPostId, $backReferenceNumberForAnchorHash, $routeName, $routeVariableName, $that) {
218		preg_match($pattern, $node->attr('href'), $matches);
219	9	$href = $matches[0];
220	9	$postId = $matches[$backReferenceNumberForPostId];
221		$anchorHash = isset($matches[$backReferenceNumberForAnchorHash]) ? $matches[$backReferenceNumberForAnchorHash] : '';
222
223		$pattern = sprintf('/href=(\'\|")%s\1/', str_replace('/', '\/', $href));
224		$replacement = sprintf('href="%s%s"', $that->urlGenerator->generate($routeName, [$routeVariableName => $postId]), $anchorHash);
225	5
226		return [
227	5	'pattern' => $pattern,
228		'replacement' => $replacement,
229	4	];
230		};
231	4
232	4	return $walker;
233		}
234
235	4	/**
236	4	* Return closure returns map of ['pattern' => regexp pattern, 'replacement' => replacement] for href value of mention links.
237		*
238		* @param string $pattern
239		* @return \Closure
240		*/
241		public function getATagWalkerForMentionLinks($pattern)
242		{
243	3	$walker = function (Crawler $node) use ($pattern) {
244		preg_match($pattern, $node->attr('href'), $matches);
245	3	$href = $matches[0];
246
247	2	$pattern = sprintf('/href=(\'\|")%s\1/', str_replace('/', '\/', $href));
248		$replacement = '';
249	2
250		return [
251		'pattern' => $pattern,
252		'replacement' => $replacement,
253		];
254		};
255
256		return $walker;
257		}
258
259	3	/**
260		* Replace emoji codes only in text content of each nodes with img tags.
261	2	*/
262	2	public function replaceEmojiCodes()
263		{
264	3	// find emoji codes.
265		preg_match_all('/:([^\s:<>\'"]+):/', $this->crawler->text(), $matches);
266	3
267		$tempReplacements = [];
268		foreach (array_unique($matches[1]) as $name) {
269	19	$pattern = sprintf('/:%s:/', preg_quote($name));
270		$replacement = sprintf('__ESABA_IMG_TAG__%s__ESABA_IMG_TAG__', $name);
271	19
272	5	$tempReplacements[$pattern] = $replacement;
273		}
274	14
275		// set temporarily replaced html content.
276		$this->replaceText($tempReplacements);
277
278		$replacements = [];
279		foreach (array_values($tempReplacements) as $tempReplacement) {
280		preg_match('/__ESABA_IMG_TAG__(.+)__ESABA_IMG_TAG__/', $tempReplacement, $matches);
281		$name = $matches[1];
282
283		$pattern = sprintf('/%s/', preg_quote($tempReplacement));
284		$replacement = sprintf('<img src="%s" class="emoji" title=":%s:" alt=":%s:">', $this->emojiManager->getImageUrl($name), $name, $name);
285
286		$replacements[$pattern] = $replacement;
287		}
288
289		$this->replaceHtml($replacements);
290		}
291
292		/**
293		* Return map of ['id' => id, 'text' => text] of headings as TOC.
294		*
295		* @return array
296		*/
297		public function getToc()
298		{
299		$this->ensureInitialized();
300
301		$toc = $this->crawler->filter('h1, h2, h3')->each($this->getWalkerForToc());
302
303		return $toc;
304		}
305
306		/**
307		* Return closure returns map of ['id' => id, 'text' => text] of h tags.
308		*
309		* @return \Closure
310		*/
311		public function getWalkerForToc()
312		{
313		$walker = function (Crawler $node) {
314		return [
315		'id' => $node->attr('id'),
316		'text' => trim(str_replace($node->filter('a')->text(), '', $node->text())),
317		];
318		};
319
320		return $walker;
321		}
322
323		private function ensureInitialized()
324		{
325		if (!$this->crawler->count()) {
326		throw new \LogicException('Initialize before using.');
327		}
328		}
329		}
330

ttskch / esaba

Pull Request — master (#12)

HtmlHandler::walkDomNodesAndReplaceOnlyTextNodes() B

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like