HtmlHandler::replaceText() - Code Metrics - Inspection of "Merge pull request #12 from ttskch/fix-emoji-in-he..." - ttskch/esaba - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Push — master ( a8072d...9d11f5 )

by Takashi

created 2018-04-15 12:57 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) {
            foreach ($replacements as $pattern => $replacement) {
                $node->textContent = preg_replace($pattern, $replacement, $node->textContent);
            }

            return;
        }

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

ttskch / esaba

Push — master ( a8072d...9d11f5 )

HtmlHandler::replaceText() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like