1<?php declare(strict_types=1);
2
3namespace DOMWrap;
4
5use DOMWrap\Traits\{
6    CommonTrait,
7    TraversalTrait,
8    ManipulationTrait
9};
10
11/**
12 * Document Node
13 *
14 * @package DOMWrap
15 * @license http://opensource.org/licenses/BSD-3-Clause BSD 3 Clause
16 */
17class Document extends \DOMDocument
18{
19    use CommonTrait;
20    use TraversalTrait;
21    use ManipulationTrait;
22
23    /** @var int */
24    protected $libxmlOptions = 0;
25
26    /** @var string|null */
27    protected $documentEncoding = null;
28
29    public function __construct(string $version = '1.0', string $encoding = 'UTF-8') {
30        parent::__construct($version, $encoding);
31
32        $this->registerNodeClass('DOMText', 'DOMWrap\\Text');
33        $this->registerNodeClass('DOMElement', 'DOMWrap\\Element');
34        $this->registerNodeClass('DOMComment', 'DOMWrap\\Comment');
35        $this->registerNodeClass('DOMDocument', 'DOMWrap\\Document');
36        $this->registerNodeClass('DOMDocumentType', 'DOMWrap\\DocumentType');
37        $this->registerNodeClass('DOMProcessingInstruction', 'DOMWrap\\ProcessingInstruction');
38    }
39
40    /**
41     * Set libxml options.
42     *
43     * Multiple values must use bitwise OR.
44     * eg: LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD
45     *
46     * @link http://php.net/manual/en/libxml.constants.php
47     *
48     * @param int $libxmlOptions
49     */
50    public function setLibxmlOptions(int $libxmlOptions): void {
51        $this->libxmlOptions = $libxmlOptions;
52    }
53
54    /**
55     * {@inheritdoc}
56     */
57    public function document(): ?\DOMDocument {
58        return $this;
59    }
60
61    /**
62     * {@inheritdoc}
63     */
64    public function collection(): NodeList {
65        return $this->newNodeList([$this]);
66    }
67
68    /**
69     * {@inheritdoc}
70     */
71    public function result(NodeList $nodeList) {
72        if ($nodeList->count()) {
73            return $nodeList->first();
74        }
75
76        return null;
77    }
78
79    /**
80     * {@inheritdoc}
81     */
82    public function parent() {
83        return null;
84    }
85
86    /**
87     * {@inheritdoc}
88     */
89    public function parents() {
90        return $this->newNodeList();
91    }
92
93    /**
94     * {@inheritdoc}
95     */
96    public function substituteWith($newNode): self {
97        $this->replaceChild($newNode, $this);
98
99        return $this;
100    }
101
102    /**
103     * {@inheritdoc}
104     */
105    public function _clone() {
106        return null;
107    }
108
109    /**
110     * {@inheritdoc}
111     */
112    public function getHtml(): string {
113        return $this->getOuterHtml();
114    }
115
116    /**
117     * {@inheritdoc}
118     */
119    public function setHtml($html): self {
120        if (!is_string($html) || trim($html) == '') {
121            return $this;
122        }
123
124        $internalErrors = libxml_use_internal_errors(true);
125        if (\PHP_VERSION_ID < 80000) {
126            $disableEntities = libxml_disable_entity_loader(true);
127            $this->composeXmlNode($html);
128            libxml_use_internal_errors($internalErrors);
129            libxml_disable_entity_loader($disableEntities);
130        } else {
131            $this->composeXmlNode($html);
132            libxml_use_internal_errors($internalErrors);
133        }
134
135        return $this;
136    }
137
138    /**
139     * @param string $html
140     * @param int $options
141     *
142     * @return bool
143     */
144    public function loadHTML($html, $options = 0): bool {
145        // Fix LibXML's crazy-ness RE root nodes
146        // While importing HTML using the LIBXML_HTML_NOIMPLIED option LibXML insists
147        //  on having one root node. All subsequent nodes are appended to this first node.
148        // To counter this we will create a fake element, allow LibXML to 'do its thing'
149        //  then undo it by taking the contents of the fake element, placing it back into
150        //  the root and then remove our fake element.
151        if ($options & LIBXML_HTML_NOIMPLIED) {
152            $html = '<domwrap></domwrap>' . $html;
153        }
154
155        $html = '<?xml encoding="' . ($this->getEncoding() ?? 'UTF-8') . '">' . $html;
156
157        $result = parent::loadHTML($html, $options);
158
159        // Do our re-shuffling of nodes.
160        if ($this->libxmlOptions & LIBXML_HTML_NOIMPLIED) {
161            $this->children()->first()->contents()->each(function($node){
162                $this->appendWith($node);
163            });
164
165            $this->removeChild($this->children()->first());
166        }
167
168        return $result;
169    }
170
171    /*
172     * @param $encoding string|null
173     */
174    public function setEncoding(string $encoding = null) {
175        $this->documentEncoding = $encoding;
176    }
177
178    /*
179     * @return string|null
180     */
181    public function getEncoding(): ?string {
182        return $this->documentEncoding;
183    }
184
185    /*
186     * @param $html string
187     *
188     * @return string|null
189     */
190    private function getCharset(string $html): ?string {
191        $charset = null;
192
193        if (preg_match('@<meta[^>]*?charset=["\']?([^"\'\s>]+)@im', $html, $matches)) {
194            $charset = mb_strtoupper($matches[1]);
195        }
196
197        return $charset;
198    }
199
200    /*
201     * @param $html string
202     */
203    private function detectEncoding(string $html) {
204        $charset = $this->getEncoding();
205
206        if (is_null($charset)) {
207            $charset = $this->getCharset($html);
208        }
209
210        $detectedCharset = mb_detect_encoding($html, mb_detect_order(), true);
211
212        if ($charset === null && $detectedCharset == 'UTF-8') {
213            $charset = $detectedCharset;
214        }
215
216        $this->setEncoding($charset);
217    }
218
219    /*
220     * @param $html string
221     *
222     * @return string
223     */
224    private function convertToUtf8(string $html): string {
225        $charset = $this->getEncoding();
226
227        if ($charset !== null) {
228            $html = preg_replace('@(charset=["]?)([^"\s]+)([^"]*["]?)@im', '$1UTF-8$3', $html);
229            $mbHasCharset = in_array($charset, array_map('mb_strtoupper', mb_list_encodings()));
230
231            if ($mbHasCharset) {
232                $html = mb_convert_encoding($html, 'UTF-8', $charset);
233
234            // Fallback to iconv if available.
235            } elseif (extension_loaded('iconv')) {
236                $htmlIconv = iconv($charset, 'UTF-8', $html);
237
238                if ($htmlIconv !== false) {
239                    $html = $htmlIconv;
240                } else {
241                    $charset = null;
242                }
243            }
244        }
245
246        if ($charset === null) {
247            $html = htmlspecialchars_decode(mb_encode_numericentity(htmlentities($html, ENT_QUOTES, 'UTF-8'), [0x80, 0x10FFFF, 0, ~0], 'UTF-8'));
248        }
249
250        return $html;
251    }
252
253    /**
254     * @param $html
255     */
256    private function composeXmlNode($html)
257    {
258        $this->detectEncoding($html);
259
260        $html = $this->convertToUtf8($html);
261
262        $this->loadHTML($html, $this->libxmlOptions);
263
264        // Remove <?xml ...> processing instruction.
265        $this->contents()->each(function($node) {
266            if ($node instanceof ProcessingInstruction && $node->nodeName == 'xml') {
267                $node->destroy();
268            }
269        });
270    }
271}
272