Parser::parse() - Code Metrics - Inspection of "Updated documentation" - paquettg/php-html-parser - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — master ( 668c77...c11634 )

by Gilles

created 2020-07-19 18:29 UTC

Parser::parse() C

↳ Parent: Parser

Complexity

Conditions	15
Paths	25

Size

Total Lines	64
Code Lines	38

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	38
CRAP Score	15

Importance

Changes	1
Bugs	0	Features	0

Metric	Value
cc	15
eloc	38
c	1
b	0
f	0
nc	25
nop	3
dl	0
loc	64
ccs	38
cts	38
cp	1
crap	15
rs	5.9166

How to fix Long Method Complexity

<?php

declare(strict_types=1);

namespace PHPHtmlParser\Dom;

use PHPHtmlParser\Content;
use PHPHtmlParser\Contracts\Dom\ParserInterface;
use PHPHtmlParser\Dom\Node\AbstractNode;
use PHPHtmlParser\Dom\Node\HtmlNode;
use PHPHtmlParser\Dom\Node\TextNode;
use PHPHtmlParser\DTO\TagDTO;
use PHPHtmlParser\Enum\StringToken;
use PHPHtmlParser\Exceptions\ChildNotFoundException;
use PHPHtmlParser\Exceptions\CircularException;
use PHPHtmlParser\Exceptions\ContentLengthException;
use PHPHtmlParser\Exceptions\LogicalException;
use PHPHtmlParser\Exceptions\StrictException;
use PHPHtmlParser\Options;
use stringEncode\Encode;

class Parser implements ParserInterface
{
    /**
     * Attempts to parse the html in content.
     *
     * @throws ChildNotFoundException
     * @throws CircularException
     * @throws ContentLengthException
     * @throws LogicalException
     * @throws StrictException
     */
    public function parse(Options $options, Content $content, int $size): AbstractNode
    {
        // add the root node
        $root = new HtmlNode('root');
        $root->setHtmlSpecialCharsDecode($options->isHtmlSpecialCharsDecode());
        $activeNode = $root;
        while ($activeNode !== null) {
            if ($activeNode && $activeNode->tag->name() === 'script'
                && $options->isCleanupInput() !== true
            ) {
                $str = $content->copyUntil('</');
            } else {
                $str = $content->copyUntil('<');
            }
            if ($str == '') {
                $tagDTO = $this->parseTag($options, $content, $size);
                if (!$tagDTO->isStatus()) {
                    // we are done here
                    $activeNode = null;
                    continue;
                }

                // check if it was a closing tag
                if ($tagDTO->isClosing()) {
                    $foundOpeningTag = true;
                    $originalNode = $activeNode;
                    while ($activeNode->getTag()->name() != $tagDTO->getTag()) {
                        $activeNode = $activeNode->getParent();
                        if ($activeNode === null) {
                            // we could not find opening tag
                            $activeNode = $originalNode;
                            $foundOpeningTag = false;
                            break;
                        }
                    }
                    if ($foundOpeningTag) {
                        $activeNode = $activeNode->getParent();
                    }
                    continue;
                }

                if ($tagDTO->getNode() === null) {
                    continue;
                }

                /** @var AbstractNode $node */
                $node = $tagDTO->getNode();
                $activeNode->addChild($node);

                // check if node is self closing
                if (!$node->getTag()->isSelfClosing()) {
                    $activeNode = $node;
                }
            } elseif ($options->isWhitespaceTextNode() ||
                \trim($str) != ''
            ) {
                // we found text we care about
                $textNode = new TextNode($str, $options->isRemoveDoubleSpace());
                $textNode->setHtmlSpecialCharsDecode($options->isHtmlSpecialCharsDecode());
                $activeNode->addChild($textNode);
            }
        }

        return $root;
    }

    /**
     * Attempts to detect the charset that the html was sent in.
     *
     * @throws ChildNotFoundException
     */
    public function detectCharset(Options $options, string $defaultCharset, AbstractNode $root): bool
    {
        // set the default
        $encode = new Encode();
        $encode->from($defaultCharset);
        $encode->to($defaultCharset);

        $enforceEncoding = $options->getEnforceEncoding();
        if ($enforceEncoding !== null) {
            //  they want to enforce the given encoding
            $encode->from($enforceEncoding);
            $encode->to($enforceEncoding);

            return false;
        }

        /** @var AbstractNode $meta */
        $meta = $root->find('meta[http-equiv=Content-Type]', 0);
        if ($meta == null) {
            if (!$this->detectHTML5Charset($encode, $root)) {
                // could not find meta tag
                $root->propagateEncoding($encode);

                return false;
            }

            return true;
        }
        $content = $meta->getAttribute('content');
        if (\is_null($content)) {
            // could not find content
            $root->propagateEncoding($encode);

            return false;
        }
        $matches = [];
        if (\preg_match('/charset=([^;]+)/', $content, $matches)) {
            $encode->from(\trim($matches[1]));
            $root->propagateEncoding($encode);

            return true;
        }

        // no charset found
        $root->propagateEncoding($encode);

        return false;
    }

    /**
     * Attempt to parse a tag out of the content.
     *
     * @throws StrictException
     * @throws ContentLengthException
     * @throws LogicalException
     * @throws StrictException
     */
    private function parseTag(Options $options, Content $content, int $size): TagDTO
    {
        $return = [];
        if ($content->char() != '<') {
            // we are not at the beginning of a tag
            return new TagDTO();
        }

        // check if this is a closing tag
        try {
            $content->fastForward(1);
        } catch (ContentLengthException $exception) {
            // we are at the end of the file
            return new TagDTO();
        }
        if ($content->char() == '/') {
            return $this->makeEndTag($content, $options);
        }
        if ($content->char() == '?') {
            // special setting tag
            $tag = $content->fastForward(1)
                ->copyByToken(StringToken::SLASH(), true);
            $tag = (new Tag($tag))
                ->setOpening('<?')
                ->setClosing(' ?>')
                ->selfClosing();
        } else {
            $tag = \strtolower($content->copyByToken(StringToken::SLASH(), true));
            if (\trim($tag) == '') {
                // no tag found, invalid < found
                return new TagDTO();
            }
        }
        $node = new HtmlNode($tag);
        $node->setHtmlSpecialCharsDecode($options->isHtmlSpecialCharsDecode());
        $this->setUpAttributes($content, $size, $node, $options, $tag);

        $content->skipByToken(StringToken::BLANK());
        if ($content->char() == '/') {
            // self closing tag
            $node->getTag()->selfClosing();
            $content->fastForward(1);
        } elseif (\in_array($node->getTag()->name(), $options->getSelfClosing(), true)) {
            // Should be a self closing tag, check if we are strict
            if ($options->isStrict()) {
                $character = $content->getPosition();
                throw new StrictException("Tag '" . $node->getTag()->name() . "' is not self closing! (character #$character)");
            }

            // We force self closing on this tag.
            $node->getTag()->selfClosing();

            // Should this tag use a trailing slash?
            if (\in_array($node->getTag()->name(), $options->getNoSlash(), true)) {
                $node->getTag()->noTrailingSlash();
            }
        }

        if ($content->canFastForward(1)) {
            $content->fastForward(1);
        }

        $return['status'] = true;
        $return['node'] = $node;

        return new TagDTO($return);
    }

    /**
     * @throws ChildNotFoundException
     */
    private function detectHTML5Charset(Encode $encode, AbstractNode $root): bool
    {
        /** @var AbstractNode|null $meta */
        $meta = $root->find('meta[charset]', 0);
        if ($meta == null) {
            return false;
        }

        $encode->from(\trim($meta->getAttribute('charset')));
        $root->propagateEncoding($encode);

        return true;
    }

    /**
     * @throws ContentLengthException
     * @throws LogicalException
     */
    private function makeEndTag(Content $content, Options $options): TagDTO
    {
        $return = [];
        $tag = $content->fastForward(1)
            ->copyByToken(StringToken::SLASH(), true);
        // move to end of tag
        $content->copyUntil('>');
        $content->fastForward(1);

        // check if this closing tag counts
        $tag = \strtolower($tag);
        if (\in_array($tag, $options->getSelfClosing(), true)) {
            $return['status'] = true;

            return new TagDTO($return);
        }
        $return['status'] = true;
        $return['closing'] = true;
        $return['tag'] = \strtolower($tag);

        return new TagDTO($return);
    }

    /**
     * @param string|Tag $tag
     *
     * @throws ContentLengthException
     * @throws LogicalException
     * @throws StrictException
     */
    private function setUpAttributes(Content $content, int $size, HtmlNode $node, Options $options, $tag): void
    {
        while (
            $content->char() != '>' &&
            $content->char() != '/'
        ) {
            $space = $content->skipByToken(StringToken::BLANK(), true);
            if (empty($space)) {
                try {
                    $content->fastForward(1);
                } catch (ContentLengthException $exception) {
                    // reached the end of the content
                    break;
                }
                continue;
            }

            $name = $content->copyByToken(StringToken::EQUAL(), true);
            if ($name == '/') {
                break;
            }

            if (empty($name)) {
                $content->skipByToken(StringToken::BLANK());
                continue;
            }

            $content->skipByToken(StringToken::BLANK());
            if ($content->char() == '=') {
                $content->fastForward(1)
                    ->skipByToken(StringToken::BLANK());
                switch ($content->char()) {
                    case '"':
                        $content->fastForward(1);
                        $string = $content->copyUntil('"', true);
                        do {
                            $moreString = $content->copyUntilUnless('"', '=>');
                            $string .= $moreString;
                        } while (\strlen($moreString) > 0 && $content->getPosition() < $size);
                        $content->fastForward(1);
                        $node->getTag()->setAttribute($name, $string);
                        break;
                    case "'":
                        $content->fastForward(1);
                        $string = $content->copyUntil("'", true);
                        do {
                            $moreString = $content->copyUntilUnless("'", '=>');
                            $string .= $moreString;
                        } while (\strlen($moreString) > 0 && $content->getPosition() < $size);
                        $content->fastForward(1);
                        $node->getTag()->setAttribute($name, $string, false);
                        break;
                    default:
                        $node->getTag()->setAttribute($name, $content->copyByToken(StringToken::ATTR(), true));
                        break;
                }
            } else {
                // no value attribute
                if ($options->isStrict()) {
                    // can't have this in strict html
                    $character = $content->getPosition();
                    throw new StrictException("Tag '$tag' has an attribute '$name' with out a value! (character #$character)");
                }
                $node->getTag()->setAttribute($name, null);
                if ($content->char() != '>') {
                    $content->rewind(1);
                }
            }
        }
    }
}


1		<?php
2
3		declare(strict_types=1);
4
5		namespace PHPHtmlParser\Dom;
6
7		use PHPHtmlParser\Content;
8		use PHPHtmlParser\Contracts\Dom\ParserInterface;
9		use PHPHtmlParser\Dom\Node\AbstractNode;
10		use PHPHtmlParser\Dom\Node\HtmlNode;
11		use PHPHtmlParser\Dom\Node\TextNode;
12		use PHPHtmlParser\DTO\TagDTO;
13		use PHPHtmlParser\Enum\StringToken;
14		use PHPHtmlParser\Exceptions\ChildNotFoundException;
15		use PHPHtmlParser\Exceptions\CircularException;
16		use PHPHtmlParser\Exceptions\ContentLengthException;
17		use PHPHtmlParser\Exceptions\LogicalException;
18		use PHPHtmlParser\Exceptions\StrictException;
19		use PHPHtmlParser\Options;
20		use stringEncode\Encode;
21
22		class Parser implements ParserInterface
23		{
24		/**
25		* Attempts to parse the html in content.
26		*
27		* @throws ChildNotFoundException
28		* @throws CircularException
29		* @throws ContentLengthException
30		* @throws LogicalException
31		* @throws StrictException
32		*/
33	288	public function parse(Options $options, Content $content, int $size): AbstractNode
34		{
35		// add the root node
36	288	$root = new HtmlNode('root');
37	288	$root->setHtmlSpecialCharsDecode($options->isHtmlSpecialCharsDecode());
38	288	$activeNode = $root;
39	288	while ($activeNode !== null) {
40	288	if ($activeNode && $activeNode->tag->name() === 'script'
41	288	&& $options->isCleanupInput() !== true
42		) {
43	9	$str = $content->copyUntil('</');
44		} else {
45	288	$str = $content->copyUntil('<');
46		}
47	288	if ($str == '') {
48	288	$tagDTO = $this->parseTag($options, $content, $size);
49	288	if (!$tagDTO->isStatus()) {
50		// we are done here
51	282	$activeNode = null;
52	282	continue;
53		}
54
55		// check if it was a closing tag
56	282	if ($tagDTO->isClosing()) {
57	252	$foundOpeningTag = true;
58	252	$originalNode = $activeNode;
59	252	while ($activeNode->getTag()->name() != $tagDTO->getTag()) {
60	105	$activeNode = $activeNode->getParent();
61	105	if ($activeNode === null) {
62		// we could not find opening tag
63	39	$activeNode = $originalNode;
64	39	$foundOpeningTag = false;
65	39	break;
66		}
67		}
68	252	if ($foundOpeningTag) {
69	252	$activeNode = $activeNode->getParent();
70		}
71	252	continue;
72		}
73
74	282	if ($tagDTO->getNode() === null) {
75	57	continue;
76		}
77
78		/** @var AbstractNode $node */
79	282	$node = $tagDTO->getNode();
80	282	$activeNode->addChild($node);
81
82		// check if node is self closing
83	282	if (!$node->getTag()->isSelfClosing()) {
84	282	$activeNode = $node;
85		}
86	249	} elseif ($options->isWhitespaceTextNode() \|\|
87	249	\trim($str) != ''
88		) {
89		// we found text we care about
90	246	$textNode = new TextNode($str, $options->isRemoveDoubleSpace());
91	246	$textNode->setHtmlSpecialCharsDecode($options->isHtmlSpecialCharsDecode());
92	246	$activeNode->addChild($textNode);
93		}
94		}
95
96	282	return $root;
97		}
98
99		/**
100		* Attempts to detect the charset that the html was sent in.
101		*
102		* @throws ChildNotFoundException
103		*/
104	282	public function detectCharset(Options $options, string $defaultCharset, AbstractNode $root): bool
105		{
106		// set the default
107	282	$encode = new Encode();
108	282	$encode->from($defaultCharset);
109	282	$encode->to($defaultCharset);
110
111	282	$enforceEncoding = $options->getEnforceEncoding();
112	282	if ($enforceEncoding !== null) {
113		// they want to enforce the given encoding
114		$encode->from($enforceEncoding);
115		$encode->to($enforceEncoding);
116
117		return false;
118		}
119
120		/** @var AbstractNode $meta */
121	282	$meta = $root->find('meta[http-equiv=Content-Type]', 0);
122	282	if ($meta == null) {
123	252	if (!$this->detectHTML5Charset($encode, $root)) {
124		// could not find meta tag
125	249	$root->propagateEncoding($encode);
126
127	249	return false;
128		}
129
130	3	return true;
131		}
132	30	$content = $meta->getAttribute('content');
133	30	if (\is_null($content)) {
134		// could not find content
135		$root->propagateEncoding($encode);
136
137		return false;
138		}
139	30	$matches = [];
140	30	if (\preg_match('/charset=([^;]+)/', $content, $matches)) {
141	30	$encode->from(\trim($matches[1]));
142	30	$root->propagateEncoding($encode);
143
144	30	return true;
145		}
146
147		// no charset found
148		$root->propagateEncoding($encode);
149
150		return false;
151		}
152
153		/**
154		* Attempt to parse a tag out of the content.
155		*
156		* @throws StrictException
157		* @throws ContentLengthException
158		* @throws LogicalException
159		* @throws StrictException
160		*/
161	288	private function parseTag(Options $options, Content $content, int $size): TagDTO
162		{
163	288	$return = [];
164	288	if ($content->char() != '<') {
165		// we are not at the beginning of a tag
166	279	return new TagDTO();
167		}
168
169		// check if this is a closing tag
170		try {
171	282	$content->fastForward(1);
172		} catch (ContentLengthException $exception) {
173		// we are at the end of the file
174		return new TagDTO();
175		}
176	282	if ($content->char() == '/') {
177	252	return $this->makeEndTag($content, $options);
178		}
179	282	if ($content->char() == '?') {
180		// special setting tag
181	3	$tag = $content->fastForward(1)
182	3	->copyByToken(StringToken::SLASH(), true);
183	3	$tag = (new Tag($tag))
184	3	->setOpening('<?')
185	3	->setClosing(' ?>')
186	3	->selfClosing();
187		} else {
188	282	$tag = \strtolower($content->copyByToken(StringToken::SLASH(), true));
189	282	if (\trim($tag) == '') {
190		// no tag found, invalid < found
191	3	return new TagDTO();
192		}
193		}
194	282	$node = new HtmlNode($tag);
195	282	$node->setHtmlSpecialCharsDecode($options->isHtmlSpecialCharsDecode());
196	282	$this->setUpAttributes($content, $size, $node, $options, $tag);
197
198	282	$content->skipByToken(StringToken::BLANK());
199	282	if ($content->char() == '/') {
200		// self closing tag
201	117	$node->getTag()->selfClosing();
202	117	$content->fastForward(1);
203	279	} elseif (\in_array($node->getTag()->name(), $options->getSelfClosing(), true)) {
204		// Should be a self closing tag, check if we are strict
205	144	if ($options->isStrict()) {
206	3	$character = $content->getPosition();
207	3	throw new StrictException("Tag '" . $node->getTag()->name() . "' is not self closing! (character #$character)");
208		}
209
210		// We force self closing on this tag.
211	141	$node->getTag()->selfClosing();
212
213		// Should this tag use a trailing slash?
214	141	if (\in_array($node->getTag()->name(), $options->getNoSlash(), true)) {
215	3	$node->getTag()->noTrailingSlash();
216		}
217		}
218
219	282	if ($content->canFastForward(1)) {
220	279	$content->fastForward(1);
221		}
222
223	282	$return['status'] = true;
224	282	$return['node'] = $node;
225
226	282	return new TagDTO($return);
227		}
228
229		/**
230		* @throws ChildNotFoundException
231		*/
232	252	private function detectHTML5Charset(Encode $encode, AbstractNode $root): bool
233		{
234		/** @var AbstractNode\|null $meta */
235	252	$meta = $root->find('meta[charset]', 0);
236	252	if ($meta == null) {
237	249	return false;
238		}
239
240	3	$encode->from(\trim($meta->getAttribute('charset')));
241	3	$root->propagateEncoding($encode);
242
243	3	return true;
244		}
245
246		/**
247		* @throws ContentLengthException
248		* @throws LogicalException
249		*/
250	252	private function makeEndTag(Content $content, Options $options): TagDTO
251		{
252	252	$return = [];
253	252	$tag = $content->fastForward(1)
254	252	->copyByToken(StringToken::SLASH(), true);
255		// move to end of tag
256	252	$content->copyUntil('>');
257	252	$content->fastForward(1);
258
259		// check if this closing tag counts
260	252	$tag = \strtolower($tag);
261	252	if (\in_array($tag, $options->getSelfClosing(), true)) {
262	57	$return['status'] = true;
263
264	57	return new TagDTO($return);
265		}
266	252	$return['status'] = true;
267	252	$return['closing'] = true;
268	252	$return['tag'] = \strtolower($tag);
269
270	252	return new TagDTO($return);
271		}
272
273		/**
274		* @param string\|Tag $tag
275		*
276		* @throws ContentLengthException
277		* @throws LogicalException
278		* @throws StrictException
279		*/
280	282	private function setUpAttributes(Content $content, int $size, HtmlNode $node, Options $options, $tag): void
281		{
282		while (
283	282	$content->char() != '>' &&
284	282	$content->char() != '/'
285		) {
286	264	$space = $content->skipByToken(StringToken::BLANK(), true);
287	264	if (empty($space)) {
288		try {
289	12	$content->fastForward(1);
290	3	} catch (ContentLengthException $exception) {
291		// reached the end of the content
292	3	break;
293		}
294	12	continue;
295		}
296
297	264	$name = $content->copyByToken(StringToken::EQUAL(), true);
298	264	if ($name == '/') {
299		break;
300		}
301
302	264	if (empty($name)) {
303	120	$content->skipByToken(StringToken::BLANK());
304	120	continue;
305		}
306
307	261	$content->skipByToken(StringToken::BLANK());
308	261	if ($content->char() == '=') {
309	255	$content->fastForward(1)
310	255	->skipByToken(StringToken::BLANK());
311	255	switch ($content->char()) {
312	255	case '"':
313	240	$content->fastForward(1);
314	240	$string = $content->copyUntil('"', true);
315		do {
316	240	$moreString = $content->copyUntilUnless('"', '=>');
317	240	$string .= $moreString;
318	240	} while (\strlen($moreString) > 0 && $content->getPosition() < $size);
319	240	$content->fastForward(1);
320	240	$node->getTag()->setAttribute($name, $string);
321	240	break;
322	21	case "'":
323	18	$content->fastForward(1);
324	18	$string = $content->copyUntil("'", true);
325		do {
326	18	$moreString = $content->copyUntilUnless("'", '=>');
327	18	$string .= $moreString;
328	18	} while (\strlen($moreString) > 0 && $content->getPosition() < $size);
329	18	$content->fastForward(1);
330	18	$node->getTag()->setAttribute($name, $string, false);
331	18	break;
332		default:
333	3	$node->getTag()->setAttribute($name, $content->copyByToken(StringToken::ATTR(), true));
334	255	break;
335		}
336		} else {
337		// no value attribute
338	81	if ($options->isStrict()) {
339		// can't have this in strict html
340	3	$character = $content->getPosition();
341	3	throw new StrictException("Tag '$tag' has an attribute '$name' with out a value! (character #$character)");
342		}
343	78	$node->getTag()->setAttribute($name, null);
344	78	if ($content->char() != '>') {
345	21	$content->rewind(1);
346		}
347		}
348		}
349	282	}
350		}
351

paquettg / php-html-parser

Push — master ( 668c77...c11634 )

Parser::parse() C

Complexity

Size

Duplication

Code Coverage

Importance

How to fix Long Method Complexity

Long Method

Duplication Side-by-Side

Filter issues like