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

Completed

Pull Request — master (#12)

by Takashi

created 2018-04-15 09:21 UTC

HtmlHandler::replaceText() A

↳ Parent: HtmlHandler

Complexity

Conditions	1
Paths	1

Size

Total Lines	10
Code Lines	5

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	6
CRAP Score	1

Importance

Changes

Metric	Value
cc	1
eloc	5
nc	1
nop	1
dl	0
loc	10
ccs	6
cts	6
cp	1
crap	1
rs	9.4285
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 ($node->childNodes->length > 0) {
            foreach ($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	57	public function __construct(Crawler $crawler, UrlGeneratorInterface $urlGenerator, EmojiManager $emojiManager, $teamName)
34		{
35	57	$this->crawler = $crawler;
36	57	$this->urlGenerator = $urlGenerator;
37	57	$this->emojiManager = $emojiManager;
38	57	$this->teamName = $teamName;
39	57	}
40
41		/**
42		* @param string $html
43		* @return $this
44		*/
45	5	public function initialize($html)
46		{
47	5	$this->crawler->clear();
48	5	$this->crawler->addHtmlContent($html);
49
50	5	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	5	public function replaceHtml(array $replacements)
67		{
68	5	$this->ensureInitialized();
69
70	4	$html = $this->crawler->html();
71
72	4	foreach ($replacements as $pattern => $replacement) {
73	4	$html = preg_replace($pattern, $replacement, $html);
74		}
75
76	4	$this->initialize($html);
77	4	}
78
79		/**
80		* @param array $replacements map of [regexp pattern => replacement].
81		*/
82	13	public function replaceText(array $replacements)
83		{
84	13	$this->ensureInitialized();
85
86	13	$domNode = $this->crawler->getNode(0);
87
88	13	$this->walkDomNodesAndReplaceOnlyTextNodes($domNode, $replacements);
89
90	1	$this->crawler->clear();
91	1	$this->crawler->addNode($domNode);
92	1	}
93
94		/**
95		* @param \DOMNode $node
96		* @param array $replacements map of [regexp pattern => replacement].
97		*/
98	13	public function walkDomNodesAndReplaceOnlyTextNodes(\DOMNode $node, array $replacements)
99		{
100	13	if ($node->nodeType === XML_TEXT_NODE) {
101	1	$text = $node->textContent;
102
103	1	foreach ($replacements as $pattern => $replacement) {
104		$text = preg_replace($pattern, $replacement, $text);
105		}
106
107	1	$node->textContent = $text;
108
109	1	return;
110		}
111
112	13	if ($node->childNodes->length > 0) {
113	13	foreach ($node->childNodes as $childNode) {
114	13	$this->walkDomNodesAndReplaceOnlyTextNodes($childNode, $replacements);
115		}
116		}
117	1	}
118
119		/**
120		* Replace links to other post with links to see the post on esaba.
121		*
122		* @param string $routeName
123		* @param string $routeVariableName
124		*/
125	2	public function replacePostUrls($routeName, $routeVariableName)
126		{
127	2	$backReferenceNumberForPostId = null;
128	2	$backReferenceNumberForAnchorHash = null;
129	2	$pattern = $this->getPostUrlPattern($backReferenceNumberForPostId, $backReferenceNumberForAnchorHash);
130	2	$walker = $this->getATagWalkerForPostUrls($pattern, $backReferenceNumberForPostId, $backReferenceNumberForAnchorHash, $routeName, $routeVariableName);
131
132	2	$this->replaceATagWithWalker($pattern, $walker);
133	2	}
134
135		/**
136		* Disable @mention links.
137		*/
138	2	public function disableMentionLinks()
139		{
140	2	$pattern = $this->getMentionLinkPattern();
141	2	$walker = $this->getATagWalkerForMentionLinks($pattern);
142
143	2	$this->replaceATagWithWalker($pattern, $walker);
144	2	}
145
146		/**
147		* Replace <a> tag href values for specified regexp pattern with closure returns map of ['pattern' => regexp pattern, 'replacement' => replacement].
148		*
149		* @param string $pattern
150		* @param \Closure $walker
151		*/
152	4	public function replaceATagWithWalker($pattern, \Closure $walker)
153		{
154	4	$this->ensureInitialized();
155
156	3	$targetATags = $this->crawler->filter('a')->reduce($this->getATagReducer($pattern));
157	3	$replacements = $targetATags->each($walker);
158	3	$replacements = array_combine(array_column($replacements, 'pattern'), array_column($replacements, 'replacement'));
159
160	3	$this->replaceHtml($replacements);
161	3	}
162
163		/**
164		* @param string $backReferenceNumberForPostId For returning position of post id in regexp pattern.
165		* @param string $backReferenceNumberForAnchorHash For returning position of anchor hash regexp pattern.
166		* @return string
167		*/
168	28	public function getPostUrlPattern(&$backReferenceNumberForPostId, &$backReferenceNumberForAnchorHash)
169		{
170	28	$backReferenceNumberForPostId = 3;
171	28	$backReferenceNumberForAnchorHash = 5;
172
173	28	return sprintf('#^((https?:)?//%s\.esa\.io)?/posts/(\d+)(/\|/edit/?)?(\#.+)?$#', $this->teamName);
174		}
175
176		/**
177		* @return string
178		*/
179	3	public function getMentionLinkPattern()
180		{
181	3	return '#/members/([^\'"]+)#';
182		}
183
184		/**
185		* Return closure reduces ATags Crawler with regexp pattern for href value.
186		*
187		* @param string $pattern
188		* @return \Closure
189		*/
190		public function getATagReducer($pattern)
191		{
192	28	$reducer = function (Crawler $node) use ($pattern) {
193	26	preg_match($pattern, $node->attr('href'), $matches);
194
195	26	return boolval($matches);
196	28	};
197
198	28	return $reducer;
199		}
200
201		/**
202		* Return closure returns map of ['pattern' => regexp pattern, 'replacement' => replacement] for href value of post urls.
203		*
204		* @param string $pattern
205		* @param int $backReferenceNumberForPostId
206		* @param int $backReferenceNumberForAnchorHash
207		* @param string $routeName
208		* @param string $routeVariableName
209		* @return \Closure
210		*/
211	3	public function getATagWalkerForPostUrls($pattern, $backReferenceNumberForPostId, $backReferenceNumberForAnchorHash, $routeName, $routeVariableName)
212		{
213	3	$that = $this;
214
215	3	$walker = function (Crawler $node) use ($pattern, $backReferenceNumberForPostId, $backReferenceNumberForAnchorHash, $routeName, $routeVariableName, $that) {
216	1	preg_match($pattern, $node->attr('href'), $matches);
217	1	$href = $matches[0];
218	1	$postId = $matches[$backReferenceNumberForPostId];
219	1	$anchorHash = isset($matches[$backReferenceNumberForAnchorHash]) ? $matches[$backReferenceNumberForAnchorHash] : '';
220
221	1	$pattern = sprintf('/href=(\'\|")%s\1/', str_replace('/', '\/', $href));
222	1	$replacement = sprintf('href="%s%s"', $that->urlGenerator->generate($routeName, [$routeVariableName => $postId]), $anchorHash);
223
224		return [
225	1	'pattern' => $pattern,
226	1	'replacement' => $replacement,
227		];
228	3	};
229
230	3	return $walker;
231		}
232
233		/**
234		* Return closure returns map of ['pattern' => regexp pattern, 'replacement' => replacement] for href value of mention links.
235		*
236		* @param string $pattern
237		* @return \Closure
238		*/
239		public function getATagWalkerForMentionLinks($pattern)
240		{
241	3	$walker = function (Crawler $node) use ($pattern) {
242	1	preg_match($pattern, $node->attr('href'), $matches);
243	1	$href = $matches[0];
244
245	1	$pattern = sprintf('/href=(\'\|")%s\1/', str_replace('/', '\/', $href));
246	1	$replacement = '';
247
248		return [
249	1	'pattern' => $pattern,
250	1	'replacement' => $replacement,
251		];
252	3	};
253
254	3	return $walker;
255		}
256
257		/**
258		* Replace emoji codes only in text content of each nodes with img tags.
259		*/
260	15	public function replaceEmojiCodes()
261		{
262		// find emoji codes.
263	15	preg_match_all('/:([^\s:<>\'"]+):/', $this->crawler->text(), $matches);
264
265	13	$tempReplacements = [];
266	13	foreach (array_unique($matches[1]) as $name) {
267	12	$pattern = sprintf('/:%s:/', preg_quote($name));
268	12	$replacement = sprintf('__ESABA_IMG_TAG__%s__ESABA_IMG_TAG__', $name);
269
270	12	$tempReplacements[$pattern] = $replacement;
271		}
272
273		// set temporarily replaced html content.
274	13	$this->replaceText($tempReplacements);
275
276	1	$replacements = [];
277	1	foreach (array_values($tempReplacements) as $tempReplacement) {
278		preg_match('/__ESABA_IMG_TAG__(.+)__ESABA_IMG_TAG__/', $tempReplacement, $matches);
279		$name = $matches[1];
280
281		$pattern = sprintf('/%s/', preg_quote($tempReplacement));
282		$replacement = sprintf('<img src="%s" class="emoji" title=":%s:" alt=":%s:">', $this->emojiManager->getImageUrl($name), $name, $name);
283
284		$replacements[$pattern] = $replacement;
285		}
286
287	1	$this->replaceHtml($replacements);
288	1	}
289
290		/**
291		* Return map of ['id' => id, 'text' => text] of headings as TOC.
292		*
293		* @return array
294		*/
295	3	public function getToc()
296		{
297	3	$this->ensureInitialized();
298
299	2	$toc = $this->crawler->filter('h1, h2, h3')->each($this->getWalkerForToc());
300
301	2	return $toc;
302		}
303
304		/**
305		* Return closure returns map of ['id' => id, 'text' => text] of h tags.
306		*
307		* @return \Closure
308		*/
309		public function getWalkerForToc()
310		{
311	3	$walker = function (Crawler $node) {
312		return [
313	2	'id' => $node->attr('id'),
314	2	'text' => trim(str_replace($node->filter('a')->text(), '', $node->text())),
315		];
316	3	};
317
318	3	return $walker;
319		}
320
321	22	private function ensureInitialized()
322		{
323	22	if (!$this->crawler->count()) {
324	4	throw new \LogicException('Initialize before using.');
325		}
326	18	}
327		}
328

ttskch / esaba

Pull Request — master (#12)

HtmlHandler::replaceText() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like