Xml::findNextRealNode() - Code Metrics - Inspection of "Feature/entity tests (#56)" - simplepie/simplepie-ng - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Push — master ( f4efa7...b29027 )

by Ryan

created 2018-01-27 23:35 UTC

Xml::findNextRealNode() A

↳ Parent: Xml

Complexity

Conditions	4
Paths	2

Size

Total Lines	9
Code Lines	4

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
eloc	4
nc	2
nop	1
dl	0
loc	9
c	0
b	0
f	0
cc	4
rs	9.2

<?php
/**
 * Copyright (c) 2017–2018 Ryan Parman <http://ryanparman.com>.
 * Copyright (c) 2017–2018 Contributors.
 *
 * http://opensource.org/licenses/Apache2.0
 */

declare(strict_types=1);

namespace SimplePie\Parser;

use DOMComment;
use DOMDocument;
use DOMNode;
use DOMText;
use DOMXPath;
use Psr\Http\Message\StreamInterface;
use Psr\Log\LoggerInterface;
use SimplePie\Enum as E;
use SimplePie\HandlerStackInterface;
use SimplePie\Mixin as Tr;
use SimplePie\SimplePie;
use SimplePie\Type\Feed;
use SimplePie\Util\Ns;

/**
 * The core parser for all XML content.
 */
class Xml extends AbstractParser
{
    use Tr\DomDocumentTrait;
    use Tr\LoggerTrait;
    use Tr\RawDocumentTrait;

    /**
     * The object which contains the parsed results.
     *
     * @var Feed
     */
    protected $feed;

    /**
     * Bitwise libxml options to use for parsing XML.
     *
     * @var int
     */
    protected $libxml;

    /**
     * The handler stack which contains registered middleware.
     *
     * @var HandlerStackInterface
     */
    protected $middleware;

    /**
     * The XML namespace handler.
     *
     * @var Ns
     */
    protected $ns;

    /**
     * Constructs a new instance of this class.
     *
     * @param StreamInterface       $stream                  A PSR-7 `StreamInterface` which is typically returned by
     *                                                       the `getBody()` method of a `ResponseInterface` class.
     * @param LoggerInterface       $logger                  The PSR-3 logger.
     * @param HandlerStackInterface $handlerStack            The handler stack which contains registered middleware.
     * @param int                   $libxml                  The libxml value to use for parsing XML.
     * @param bool                  $handleHtmlEntitiesInXml Whether or not SimplePie should pre-parse the XML as HTML
     *                                                       to resolve the entities. A value of `true` means that
     *                                                       SimplePie should inject the entity definitions. A value of
     *                                                       `false` means that SimplePie should NOT inject the entity
     *                                                       definitions. The default value is `false`.
     *
     * @throws Error
     * @throws TypeError
     *
     * @phpcs:disable Generic.Functions.OpeningFunctionBraceBsdAllman.BraceOnSameLine
     */
    public function __construct(
        StreamInterface $stream,
        LoggerInterface $logger,
        HandlerStackInterface $handlerStack,
        int $libxml,
        bool $handleHtmlEntitiesInXml
    ) {
        // @phpcs:enable

        // Logger
        $this->logger = $logger;

        // Middleware
        $this->middleware = $handlerStack;

        // Libxml2
        $this->libxml = $libxml;

        // Raw stream
        $this->rawDocument = $this->readStream($stream);

        // DOMDocument
        $this->domDocument = new DOMDocument('1.0', 'utf-8');

        // Don't barf errors all over the output
        \libxml_use_internal_errors(true);

        // DOMDocument configuration
        $this->domDocument->recover             = true;
        $this->domDocument->formatOutput        = false;
        $this->domDocument->preserveWhiteSpace  = false;
        $this->domDocument->resolveExternals    = true;
        $this->domDocument->substituteEntities  = true;
        $this->domDocument->strictErrorChecking = false;
        $this->domDocument->validateOnParse     = false;

        // If enabled, force-inject the contents of `entities.dtd` into the feed.
        if ($handleHtmlEntitiesInXml) {
            $this->getLogger()->debug('Enabled handing HTML entities in XML.');
            $this->domDocument->loadXML($this->rawDocument, $this->libxml);

            // Make sure this is an XML element instead of a comment or text.
            $firstElement = $this->findNextRealNode($this->domDocument->firstChild);

            // <feed, <rss, etc.
            $rootElementStart = \sprintf('<%s', (string) $firstElement->nodeName);

            // Read the entity definition file, and force-inject it into the XML document
            $this->rawDocument = \str_replace(
                $rootElementStart,
                \sprintf(
                    '%s%s',
                    \trim(
                        \file_get_contents(\dirname(SIMPLEPIE_ROOT) . '/resources/entities.dtd')
                    ),
                    $rootElementStart
                ),
                $this->rawDocument
            );
        }

        // Parse the XML document with the configured libxml options
        $this->domDocument->loadXML($this->rawDocument, $this->libxml);

        // Register the namespace handler.
        $this->ns = (new Ns($this->domDocument))
            ->setLogger($this->getLogger());

        // Look at which namespaces the registered middleware understands.
        $this->middleware->registerNamespaces($this->ns);

        // Instantiate a new write-to feed object.
        $this->feed = (new Feed($this->getNamespaceAlias() ?? ''))
            ->setLogger($this->getLogger());

        // Invoke the registered middleware.
        $this->middleware->invoke(
            E\FeedType::XML,
            $this->getFeed()->getRoot(),
            $this->getNamespaceAlias(),
            $this->xpath()
        );

        // Clear the libxml errors to avoid excessive memory usage
        \libxml_clear_errors();
    }

    /**
     * Get the XML namespace handler.
     *
     * @return Ns
     */
    public function getNs(): Ns
    {
        return $this->ns;
    }

    /**
     * Get the preferred namespace alias.
     *
     * @return string|null
     */
    public function getNamespaceAlias(): ?string
    {
        $namespace = $this->getNs();

        return $namespace->getPreferredNamespaceAlias(
            $this->domDocument->documentElement->namespaceURI
        );
    }

    /**
     * Gets a reference to the `DOMXPath` object, with the default namespace
     * already registered.
     *
     * @return DOMXPath
     */
    public function xpath()
    {
        $ns    = $this->getNamespaceAlias();
        $xpath = new DOMXPath($this->domDocument);

        // Register the namespace alias with the XPath instance
        if (null !== $ns) {
            $xpath->registerNamespace(
                $ns,
                $this->domDocument->documentElement->namespaceURI ?? ''
            );
        }

        return $xpath;
    }

    /**
     * Some DOMNode names are `#comment` or `#text`. This method will move the
     * pointer to the next node, then the next until it finds a real XML node.
     *
     * @param DOMNode $node The `DOMNode` element to evaluate.
     *
     * @return DOMNode
     */
    public function findNextRealNode(DOMNode $node): DOMNode
    {
        $n = $node;

        while (($n instanceof DOMComment || $n instanceof DOMText) && null !== $n) {
            $n = $n->nextSibling;
        }

        return $n;
    }

    /**
     * Retrieves the object which represents the top-level feed.
     *
     * @return Feed
     */
    public function getFeed(): Feed
    {
        return $this->feed;
    }
}


1			<?php
2			/**
3			* Copyright (c) 2017–2018 Ryan Parman <http://ryanparman.com>.
4			* Copyright (c) 2017–2018 Contributors.
5			*
6			* http://opensource.org/licenses/Apache2.0
7			*/
8
9			declare(strict_types=1);
10
11			namespace SimplePie\Parser;
12
13			use DOMComment;
14			use DOMDocument;
15			use DOMNode;
16			use DOMText;
17			use DOMXPath;
18			use Psr\Http\Message\StreamInterface;
19			use Psr\Log\LoggerInterface;
20			use SimplePie\Enum as E;
21			use SimplePie\HandlerStackInterface;
22			use SimplePie\Mixin as Tr;
23			use SimplePie\SimplePie;
24			use SimplePie\Type\Feed;
25			use SimplePie\Util\Ns;
26
27			/**
28			* The core parser for all XML content.
29			*/
30			class Xml extends AbstractParser
31			{
32			use Tr\DomDocumentTrait;
33			use Tr\LoggerTrait;
34			use Tr\RawDocumentTrait;
35
36			/**
37			* The object which contains the parsed results.
38			*
39			* @var Feed
40			*/
41			protected $feed;
42
43			/**
44			* Bitwise libxml options to use for parsing XML.
45			*
46			* @var int
47			*/
48			protected $libxml;
49
50			/**
51			* The handler stack which contains registered middleware.
52			*
53			* @var HandlerStackInterface
54			*/
55			protected $middleware;
56
57			/**
58			* The XML namespace handler.
59			*
60			* @var Ns
61			*/
62			protected $ns;
63
64			/**
65			* Constructs a new instance of this class.
66			*
67			* @param StreamInterface $stream A PSR-7 `StreamInterface` which is typically returned by
68			* the `getBody()` method of a `ResponseInterface` class.
69			* @param LoggerInterface $logger The PSR-3 logger.
70			* @param HandlerStackInterface $handlerStack The handler stack which contains registered middleware.
71			* @param int $libxml The libxml value to use for parsing XML.
72			* @param bool $handleHtmlEntitiesInXml Whether or not SimplePie should pre-parse the XML as HTML
73			* to resolve the entities. A value of `true` means that
74			* SimplePie should inject the entity definitions. A value of
75			* `false` means that SimplePie should NOT inject the entity
76			* definitions. The default value is `false`.
77			*
78			* @throws Error
79			* @throws TypeError
80			*
81			* @phpcs:disable Generic.Functions.OpeningFunctionBraceBsdAllman.BraceOnSameLine
82			*/
83			public function __construct(
84			StreamInterface $stream,
85			LoggerInterface $logger,
86			HandlerStackInterface $handlerStack,
87			int $libxml,
88			bool $handleHtmlEntitiesInXml
89			) {
90			// @phpcs:enable
91
92			// Logger
93			$this->logger = $logger;
94
95			// Middleware
96			$this->middleware = $handlerStack;
97
98			// Libxml2
99			$this->libxml = $libxml;
100
101			// Raw stream
102			$this->rawDocument = $this->readStream($stream);
103
104			// DOMDocument
105			$this->domDocument = new DOMDocument('1.0', 'utf-8');
106
107			// Don't barf errors all over the output
108			\libxml_use_internal_errors(true);
109
110			// DOMDocument configuration
111			$this->domDocument->recover = true;
112			$this->domDocument->formatOutput = false;
113			$this->domDocument->preserveWhiteSpace = false;
114			$this->domDocument->resolveExternals = true;
115			$this->domDocument->substituteEntities = true;
116			$this->domDocument->strictErrorChecking = false;
117			$this->domDocument->validateOnParse = false;
118
119			// If enabled, force-inject the contents of `entities.dtd` into the feed.
120			if ($handleHtmlEntitiesInXml) {
121			$this->getLogger()->debug('Enabled handing HTML entities in XML.');
122			$this->domDocument->loadXML($this->rawDocument, $this->libxml);
123
124			// Make sure this is an XML element instead of a comment or text.
125			$firstElement = $this->findNextRealNode($this->domDocument->firstChild);
126
127			// <feed, <rss, etc.
128			$rootElementStart = \sprintf('<%s', (string) $firstElement->nodeName);
129
130			// Read the entity definition file, and force-inject it into the XML document
131			$this->rawDocument = \str_replace(
132			$rootElementStart,
133			\sprintf(
134			'%s%s',
135			\trim(
136			\file_get_contents(\dirname(SIMPLEPIE_ROOT) . '/resources/entities.dtd')
137			),
138			$rootElementStart
139			),
140			$this->rawDocument
141			);
142			}
143
144			// Parse the XML document with the configured libxml options
145			$this->domDocument->loadXML($this->rawDocument, $this->libxml);
146
147			// Register the namespace handler.
148			$this->ns = (new Ns($this->domDocument))
149			->setLogger($this->getLogger());
150
151			// Look at which namespaces the registered middleware understands.
152			$this->middleware->registerNamespaces($this->ns);
153
154			// Instantiate a new write-to feed object.
155			$this->feed = (new Feed($this->getNamespaceAlias() ?? ''))
156			->setLogger($this->getLogger());
157
158			// Invoke the registered middleware.
159			$this->middleware->invoke(
160			E\FeedType::XML,
161			$this->getFeed()->getRoot(),
162			$this->getNamespaceAlias(),
163			$this->xpath()
164			);
165
166			// Clear the libxml errors to avoid excessive memory usage
167			\libxml_clear_errors();
168			}
169
170			/**
171			* Get the XML namespace handler.
172			*
173			* @return Ns
174			*/
175			public function getNs(): Ns
176			{
177			return $this->ns;
178			}
179
180			/**
181			* Get the preferred namespace alias.
182			*
183			* @return string\|null
184			*/
185			public function getNamespaceAlias(): ?string
186			{
187			$namespace = $this->getNs();
188
189			return $namespace->getPreferredNamespaceAlias(
190			$this->domDocument->documentElement->namespaceURI
191			);
192			}
193
194			/**
195			* Gets a reference to the `DOMXPath` object, with the default namespace
196			* already registered.
197			*
198			* @return DOMXPath
199			*/
200			public function xpath()
201			{
202			$ns = $this->getNamespaceAlias();
203			$xpath = new DOMXPath($this->domDocument);
204
205			// Register the namespace alias with the XPath instance
206			if (null !== $ns) {
207			$xpath->registerNamespace(
208			$ns,
209			$this->domDocument->documentElement->namespaceURI ?? ''
210			);
211			}
212
213			return $xpath;
214			}
215
216			/**
217			* Some DOMNode names are `#comment` or `#text`. This method will move the
218			* pointer to the next node, then the next until it finds a real XML node.
219			*
220			* @param DOMNode $node The `DOMNode` element to evaluate.
221			*
222			* @return DOMNode
223			*/
224			public function findNextRealNode(DOMNode $node): DOMNode
225			{
226			$n = $node;
227
228			while (($n instanceof DOMComment \|\| $n instanceof DOMText) && null !== $n) {
229			$n = $n->nextSibling;
230			}
231
232			return $n;
233			}
234
235			/**
236			* Retrieves the object which represents the top-level feed.
237			*
238			* @return Feed
239			*/
240			public function getFeed(): Feed
241			{
242			return $this->feed;
243			}
244			}
245

simplepie / simplepie-ng

Push — master ( f4efa7...b29027 )

Xml::findNextRealNode() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like