Tocifier::setAugmentCallback() - Code Metrics - Inspection of "Merge branch 'dev'" - ntd/silverstripe-autotoc - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Push — master ( 94943b...ea0a0b )

by Nicola

created 2016-08-04 08:18 UTC

Tocifier::setAugmentCallback() A

↳ Parent: Tocifier

Complexity

Conditions	1
Paths	1

Size

Total Lines	4
Code Lines	2

Duplication

Lines	0
Ratio	0 %

Importance

Changes	1
Bugs	0	Features	0

Metric	Value
c	1
b	0
f	0
dl	0
loc	4
rs	10
cc	1
eloc	2
nc	1
nop	1

<?php

/*
 * Tocifier is intentionally decoupled from SilverStripe to be able to
 * test it without needing to put all the test infrastructure up.
 */
class Tocifier
namespace YourVendor;

class YourClass { }
{
    // Prefix to prepend to every URL fragment
    public static $prefix = 'TOC-';

    // The original HTML
    private $_raw_html = '';

    // $_raw_html augmented for proper navigation
    private $_html = '';

    // The most recently generated TOC tree.
    private $_tree;

    // Array of references to the potential parents
    private $_dangling = array();

    // Callback for augmenting a single DOMElement
    private $_augment_callback;


    /**
     * Get the TOC node closest to a given nesting level.
     *
     * @param  int $level  The requested nesting level.
     * @return array
     */
    private function &_getParent($level)
    {
        while (--$level >= 0) {
            if (isset($this->_dangling[$level])) {
                return $this->_dangling[$level];
            }
        }
        // This should never be reached
        assert(false);
    }

    /**
     * Get the plain text content from a DOM element.
     *
     * @param  DOMElement $tag  The DOM element to inspect.
     * @return string
     */
    private function _getPlainText(DOMElement $tag)
    {
        // Work on a copy
        $clone = $tag->cloneNode(true);

        // Strip unneded tags (<small>)
        while (($tag = $clone->getElementsByTagName('small')) && $tag->length) {
            $tag->item(0)->parentNode->removeChild($tag->item(0));
        }

        return $clone->textContent;
    }

    /**
     * Create a new TOC node.
     *
     * @param  string $id     Node id, used for anchoring
     * @param  string $text   Title text
     * @param  int    $level  The nesting level of the node
     * @return array
     */
    private function &_newNode($id, $text, $level)
    {
        $node = array(
            'id'    => $id,
            'title' => $text
        );

        // Clear the trailing dangling parents after level, if any
        end($this->_dangling);
        $last = key($this->_dangling);
        for ($n = $level+1; $n <= $last; ++$n) {
            unset($this->_dangling[$n]);
        }

        // Consider this node a potential dangling parent
        $this->_dangling[$level] =& $node;

        return $node;
    }

    /**
     * Process the specific document.
     *
     * @param  DOMDocument $doc  The document to process.
     */
    private function _processDocument($doc)
    {
        $this->_tree =& $this->_newNode(self::$prefix, '', 0);
        $n = 1;

        $xpath = new DOMXPath($doc);
        $query = '//*[translate(name(), "123456", "......") = "h."][not(@data-hide-from-toc)]';

        foreach ($xpath->query($query) as $h) {
            $text = $this->_getPlainText($h);
            $level = (int) substr($h->tagName, 1);
            $id = self::$prefix . $n;
            ++$n;

            // Build the tree
            $parent =& $this->_getParent($level);
            $node =& $this->_newNode($id, $text, $level);
            if (! isset($parent['children'])) {
                $parent['children'] = array();
            }
            $parent['children'][] =& $node;

            call_user_func($this->_augment_callback, $doc, $h, $id);
        }

        $body = $doc->getElementsByTagName('body')->item(0);
        $this->_html = str_replace(array("<body>\n", '</body>'), '',
                                   $doc->saveHTML($body));
    }

    /**
     * Debug function for dumping a TOC node and its children.
     *
     * @param  array  $node    The TOC node to dump
     * @param  string $indent  Indentation string.
     */
    private function _dumpBranch($node, $indent = '')
    {
        echo $indent . $node['title'] . "\n";
        if (isset($node['children'])) {
            foreach ($node['children'] as &$child) {
                $this->_dumpBranch($child, "$indent\t");
            }
        }
    }


    /**
     * Create a new TOCifier instance.
     *
     * A string containing the HTML to parse for TOC must be passed
     * in. The real processing will be triggered by the process()
     * method.
     *
     * Parsing a file can be easily performed by using
     * file_get_contents():
     *
     * <code>
     * $tocifier = new Tocifier(@file_get_content($file));
     * </code>
     *
     * @param string $html A chunk of valid HTML (UTF-8 encoded).
     */
    public function __construct($html)
    {
        $this->_raw_html = $html;
        // Default augmenting method (kept for backward compatibility)
        $this->setAugmentCallback(array(__CLASS__, 'prependAnchor'));
    }

    /**
     * Change the augment method used by this Tocifier instance.
     *
     * By default the HTML is augmented prepending an anchor before
     * every valid destination. This behavior can be changed by using
     * Tocifier::setId() (that directly sets the ID on the destination
     * elements) or by providing your own callback.
     *
     * The signature of the callback to pass in should be compatible
     * with:
     *
     *     function callback(DOMDocument $dom, DOMElement $element, $id)
     *
     * @param callable $callback  The new function to call for
     *                            augmenting DOMElement
     */
    public function setAugmentCallback($callback)
    {
        $this->_augment_callback = $callback;
    }

    /**
     * Parse and process the HTML chunk.
     *
     * The parsing phase involves picking up all the HTML header
     * elements (from <h1> to <h6>), so if the HTML is not well formed
     * or any other error is encountered this function will fail.
     *
     * @return boolean true on success, false on errors.
     */
    public function process()
    {
        // Check if $this->_raw_html is valid
        if (! is_string($this->_raw_html) || empty($this->_raw_html)) {
            return false;
        }

        // DOMDocument sucks ass (welcome to PHP, you poor shit). I
        // really don't understand why it is so difficult for loadHTML()
        // to read a chunk of text in UTF-8...
        $html = mb_convert_encoding($this->_raw_html, 'HTML-ENTITIES', 'UTF-8');

        // Parse the HTML into a DOMDocument tree
        $doc = new DOMDocument();
        if (! @$doc->loadHTML($html)) {
            return false;
        }

        // Process the doc
        $this->_processDocument($doc);
        return true;
    }

    /**
     * Get the TOC (Table Of Contents) from the provided HTML.
     *
     * The HTML must be provided throught the constructor.
     *
     * The TOC is represented in the form of:
     *
     * <code>
     * array(
     *     array('id'       => 'TOC-1',
     *           'title'    => 'Item 1',
     *           'children' => array(
     *               array('id'       => 'TOC-2',
     *                     'title'    => 'Subitem 1.1'
     *               ),
     *               array('id'       => 'TOC-3',
     *                     'title'    => 'Subitem 1.2',
     *                     'children' => array(
     *                         array('id'      => 'TOC-4',
     *                               'title    => 'Subsubitem 1.2.1'
     *     ))))),
     *     array('id'       => 'TOC-5,
     *           'title'    => 'Item 2',
     *           'children' => array(
     *               array('id'       => 'TOC-6',
     *                     'title'    => 'Subitem 2.1'
     *               ),
     *               array('id'       => 'TOC-7',
     *                     'title'    => 'Subitem 2.2'
     * ))));
     * </code>
     *
     * The TOC is cached, so subsequent calls will return the same tree.
     *
     * @return Array An array representing the TOC. A valid array is

     *               always returned.
     */
    public function getTOC()
    {
        return isset($this->_tree['children']) ? $this->_tree['children'] : array();
    }

    /**
     * Get the HTML augmented for proper navigation.
     *
     * The HTML must be provided throught the feedHtml() method.
     * The returned string is cached, so subsequent calls will return
     * the same string without further processing.
     *
     * @return String The augmented HTML.
     */
    public function getHtml()
    {
        return $this->_html;
    }

    /**
     * Dump the TOC to stdout for debugging purpose.
     */
    public function dumpTOC()
    {
        $this->_dumpBranch($this->_tree);
    }

    /**
     * Augment a DOMElement by prepending an anchor.
     *
     * An HTML fragment such as:
     *
     *     <h1>First</h2>
     *     <h2>Second</h1>
     *
     * will become:
     *
     *     <a id="TOC-1" class="anchor"></a><h1>First</h2>
     *     <a id="TOC-2" class="anchor"></a><h2>Second</h1>
     *
     * @param DOMDocument $dom      The DOM owning $element
     * @param DOMElement  $element  The element to augment
     * @param string      $id       The destination ID
     */
    public static function prependAnchor(DOMDocument $dom, DOMElement $element, $id)
    {
        $anchor = $dom->createElement('a');
        $anchor->setAttribute('id', $id);
        $anchor->setAttribute('class', 'anchor');
        $element->parentNode->insertBefore($anchor, $element);
    }

    /**
     * Augment a DOMElement by setting its ID.
     *
     * An HTML fragment such as:
     *
     *     <h1>First</h2>
     *     <h2>Second</h1>
     *
     * will become:
     *
     *     <h1 id="TOC-1" class="anchor">First</h2>
     *     <h2 id="TOC-2" class="anchor">Second</h1>
     *
     * @param DOMDocument $dom      The DOM owning $element
     * @param DOMElement  $element  The element to augment
     * @param string      $id       The destination ID
     */
    public static function setId(DOMDocument $dom, DOMElement $element, $id)
    {
        $element->setAttribute('id', $id);
        $element->setAttribute('class', 'anchor');
    }
}


1			<?php
2
3			/*
4			* Tocifier is intentionally decoupled from SilverStripe to be able to
5			* test it without needing to put all the test infrastructure up.
6			*/
7			class Tocifier
			0 ignored issues – show Coding Style Compatibility introduced 2016-02-08 21:00 UTC by Report Bug Copy Issue Report PSR1 recommends that each class must be in a namespace of at least one level to avoid collisions. You can fix this by adding a namespace to your class: namespace YourVendor; class YourClass { } When choosing a vendor namespace, try to pick something that is not too generic to avoid conflicts with other libraries. Loading history...
8			{
9			// Prefix to prepend to every URL fragment
10			public static $prefix = 'TOC-';
11
12			// The original HTML
13			private $_raw_html = '';
14
15			// $_raw_html augmented for proper navigation
16			private $_html = '';
17
18			// The most recently generated TOC tree.
19			private $_tree;
20
21			// Array of references to the potential parents
22			private $_dangling = array();
23
24			// Callback for augmenting a single DOMElement
25			private $_augment_callback;
26
27
28			/**
29			* Get the TOC node closest to a given nesting level.
30			*
31			* @param int $level The requested nesting level.
32			* @return array
33			*/
34			private function &_getParent($level)
35			{
36			while (--$level >= 0) {
37			if (isset($this->_dangling[$level])) {
38			return $this->_dangling[$level];
39			}
40			}
41			// This should never be reached
42			assert(false);
43			}
44
45			/**
46			* Get the plain text content from a DOM element.
47			*
48			* @param DOMElement $tag The DOM element to inspect.
49			* @return string
50			*/
51			private function _getPlainText(DOMElement $tag)
52			{
53			// Work on a copy
54			$clone = $tag->cloneNode(true);
55
56			// Strip unneded tags (<small>)
57			while (($tag = $clone->getElementsByTagName('small')) && $tag->length) {
58			$tag->item(0)->parentNode->removeChild($tag->item(0));
59			}
60
61			return $clone->textContent;
62			}
63
64			/**
65			* Create a new TOC node.
66			*
67			* @param string $id Node id, used for anchoring
68			* @param string $text Title text
69			* @param int $level The nesting level of the node
70			* @return array
71			*/
72			private function &_newNode($id, $text, $level)
73			{
74			$node = array(
75			'id' => $id,
76			'title' => $text
77			);
78
79			// Clear the trailing dangling parents after level, if any
80			end($this->_dangling);
81			$last = key($this->_dangling);
82			for ($n = $level+1; $n <= $last; ++$n) {
83			unset($this->_dangling[$n]);
84			}
85
86			// Consider this node a potential dangling parent
87			$this->_dangling[$level] =& $node;
88
89			return $node;
90			}
91
92			/**
93			* Process the specific document.
94			*
95			* @param DOMDocument $doc The document to process.
96			*/
97			private function _processDocument($doc)
98			{
99			$this->_tree =& $this->_newNode(self::$prefix, '', 0);
100			$n = 1;
101
102			$xpath = new DOMXPath($doc);
103			$query = '//*[translate(name(), "123456", "......") = "h."][not(@data-hide-from-toc)]';
104
105			foreach ($xpath->query($query) as $h) {
106			$text = $this->_getPlainText($h);
107			$level = (int) substr($h->tagName, 1);
108			$id = self::$prefix . $n;
109			++$n;
110
111			// Build the tree
112			$parent =& $this->_getParent($level);
113			$node =& $this->_newNode($id, $text, $level);
114			if (! isset($parent['children'])) {
115			$parent['children'] = array();
116			}
117			$parent['children'][] =& $node;
118
119			call_user_func($this->_augment_callback, $doc, $h, $id);
120			}
121
122			$body = $doc->getElementsByTagName('body')->item(0);
123			$this->_html = str_replace(array("<body>\n", '</body>'), '',
124			$doc->saveHTML($body));
125			}
126
127			/**
128			* Debug function for dumping a TOC node and its children.
129			*
130			* @param array $node The TOC node to dump
131			* @param string $indent Indentation string.
132			*/
133			private function _dumpBranch($node, $indent = '')
134			{
135			echo $indent . $node['title'] . "\n";
136			if (isset($node['children'])) {
137			foreach ($node['children'] as &$child) {
138			$this->_dumpBranch($child, "$indent\t");
139			}
140			}
141			}
142
143
144			/**
145			* Create a new TOCifier instance.
146			*
147			* A string containing the HTML to parse for TOC must be passed
148			* in. The real processing will be triggered by the process()
149			* method.
150			*
151			* Parsing a file can be easily performed by using
152			* file_get_contents():
153			*
154			* <code>
155			* $tocifier = new Tocifier(@file_get_content($file));
156			* </code>
157			*
158			* @param string $html A chunk of valid HTML (UTF-8 encoded).
159			*/
160			public function __construct($html)
161			{
162			$this->_raw_html = $html;
163			// Default augmenting method (kept for backward compatibility)
164			$this->setAugmentCallback(array(__CLASS__, 'prependAnchor'));
165			}
166
167			/**
168			* Change the augment method used by this Tocifier instance.
169			*
170			* By default the HTML is augmented prepending an anchor before
171			* every valid destination. This behavior can be changed by using
172			* Tocifier::setId() (that directly sets the ID on the destination
173			* elements) or by providing your own callback.
174			*
175			* The signature of the callback to pass in should be compatible
176			* with:
177			*
178			* function callback(DOMDocument $dom, DOMElement $element, $id)
179			*
180			* @param callable $callback The new function to call for
181			* augmenting DOMElement
182			*/
183			public function setAugmentCallback($callback)
184			{
185			$this->_augment_callback = $callback;
186			}
187
188			/**
189			* Parse and process the HTML chunk.
190			*
191			* The parsing phase involves picking up all the HTML header
192			* elements (from <h1> to <h6>), so if the HTML is not well formed
193			* or any other error is encountered this function will fail.
194			*
195			* @return boolean true on success, false on errors.
196			*/
197			public function process()
198			{
199			// Check if $this->_raw_html is valid
200			if (! is_string($this->_raw_html) \|\| empty($this->_raw_html)) {
201			return false;
202			}
203
204			// DOMDocument sucks ass (welcome to PHP, you poor shit). I
205			// really don't understand why it is so difficult for loadHTML()
206			// to read a chunk of text in UTF-8...
207			$html = mb_convert_encoding($this->_raw_html, 'HTML-ENTITIES', 'UTF-8');
208
209			// Parse the HTML into a DOMDocument tree
210			$doc = new DOMDocument();
211			if (! @$doc->loadHTML($html)) {
212			return false;
213			}
214
215			// Process the doc
216			$this->_processDocument($doc);
217			return true;
218			}
219
220			/**
221			* Get the TOC (Table Of Contents) from the provided HTML.
222			*
223			* The HTML must be provided throught the constructor.
224			*
225			* The TOC is represented in the form of:
226			*
227			* <code>
228			* array(
229			* array('id' => 'TOC-1',
230			* 'title' => 'Item 1',
231			* 'children' => array(
232			* array('id' => 'TOC-2',
233			* 'title' => 'Subitem 1.1'
234			* ),
235			* array('id' => 'TOC-3',
236			* 'title' => 'Subitem 1.2',
237			* 'children' => array(
238			* array('id' => 'TOC-4',
239			* 'title => 'Subsubitem 1.2.1'
240			* ))))),
241			* array('id' => 'TOC-5,
242			* 'title' => 'Item 2',
243			* 'children' => array(
244			* array('id' => 'TOC-6',
245			* 'title' => 'Subitem 2.1'
246			* ),
247			* array('id' => 'TOC-7',
248			* 'title' => 'Subitem 2.2'
249			* ))));
250			* </code>
251			*
252			* The TOC is cached, so subsequent calls will return the same tree.
253			*
254			* @return Array An array representing the TOC. A valid array is
			0 ignored issues – show Documentation introduced 2016-05-26 19:47 UTC by Report Bug Copy Issue Report Should the return type not be `string\|array`? This check compares the return type specified in the `@return` annotation of a function or method doc comment with the types returned by the function and raises an issue if they mismatch. Loading history...
255			* always returned.
256			*/
257			public function getTOC()
258			{
259			return isset($this->_tree['children']) ? $this->_tree['children'] : array();
260			}
261
262			/**
263			* Get the HTML augmented for proper navigation.
264			*
265			* The HTML must be provided throught the feedHtml() method.
266			* The returned string is cached, so subsequent calls will return
267			* the same string without further processing.
268			*
269			* @return String The augmented HTML.
270			*/
271			public function getHtml()
272			{
273			return $this->_html;
274			}
275
276			/**
277			* Dump the TOC to stdout for debugging purpose.
278			*/
279			public function dumpTOC()
280			{
281			$this->_dumpBranch($this->_tree);
282			}
283
284			/**
285			* Augment a DOMElement by prepending an anchor.
286			*
287			* An HTML fragment such as:
288			*
289			* <h1>First</h2>
290			* <h2>Second</h1>
291			*
292			* will become:
293			*
294			* <a id="TOC-1" class="anchor"></a><h1>First</h2>
295			* <a id="TOC-2" class="anchor"></a><h2>Second</h1>
296			*
297			* @param DOMDocument $dom The DOM owning $element
298			* @param DOMElement $element The element to augment
299			* @param string $id The destination ID
300			*/
301			public static function prependAnchor(DOMDocument $dom, DOMElement $element, $id)
302			{
303			$anchor = $dom->createElement('a');
304			$anchor->setAttribute('id', $id);
305			$anchor->setAttribute('class', 'anchor');
306			$element->parentNode->insertBefore($anchor, $element);
307			}
308
309			/**
310			* Augment a DOMElement by setting its ID.
311			*
312			* An HTML fragment such as:
313			*
314			* <h1>First</h2>
315			* <h2>Second</h1>
316			*
317			* will become:
318			*
319			* <h1 id="TOC-1" class="anchor">First</h2>
320			* <h2 id="TOC-2" class="anchor">Second</h1>
321			*
322			* @param DOMDocument $dom The DOM owning $element
323			* @param DOMElement $element The element to augment
324			* @param string $id The destination ID
325			*/
326			public static function setId(DOMDocument $dom, DOMElement $element, $id)
327			{
328			$element->setAttribute('id', $id);
329			$element->setAttribute('class', 'anchor');
330			}
331			}
332

ntd / silverstripe-autotoc

Push — master ( 94943b...ea0a0b )

Tocifier::setAugmentCallback() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like