Parser::characterData() - Code Metrics - PurpleBooth/htmlstrip - Measure and Improve Code Quality continuously with Scrutinizer

Parser::characterData() A
last analyzed 2020-06-03 06:40 UTC

↳ Parent: Parser

Complexity

Conditions	1
Paths	1

Size

Total Lines

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	3
CRAP Score	1

Importance

Changes

Metric	Value
cc	1
nc	1
nop	2
dl	0
loc	4
ccs	3
cts	3
cp	1
crap	1
rs	10
c	0
b	0
f	0

<?php

/*
 * Copyright (C) 2016 Billie Thompson
 *
 * This software may be modified and distributed under the terms
 * of the MIT license.  See the LICENSE file for details.
 */

namespace PurpleBooth;

/**
 * This class is designed to by hooked into an XML parser to convert HTML to text.
 *
 * @see     xml_parser_create()
 * @see     HtmlStripperExtension
 */
class Parser
{
    /**
     * @var string
     */
    private $text = '';

    /**
     * @var \SplStack
     */
    private $transformedTextStack;

    /**
     * @var \SplStack
     */
    private $blockTypeStack;

    /**
     * @var \SplStack
     */
    private $blockAttributesStack;

    /**
     * Parser constructor.
     *
     * Sets up the stacks
     */
    public function __construct()
    {
        $this->transformedTextStack = new \SplStack();
        $this->blockTypeStack = new \SplStack();
        $this->blockAttributesStack = new \SplStack();
    }

    /**
     * Function called on the start of an element.
     *
     * Mostly used to prepend text to things, like the "*"s on LIs
     *
     * @see xml_set_element_handler()
     *
     * @param resource $parser
     * @param string   $name
     * @param array    $attrs
     */
    public function startElement($parser, $name, $attrs)
    {
        $this->blockBegin($name, $attrs);

        switch ($name) {
            case 'LI':
                $this->appendBlockText('* ');

                break;
        }
    }

    /**
     * Called when we begin a block.
     *
     * We build a series of stacks to represent the tree of the document
     *
     * We operate at this level of the stack when we're editing content
     *
     * @param string $name
     * @param array  $attributes
     */
    private function blockBegin($name, $attributes)
    {
        $this->transformedTextStack->push('');
        $this->blockTypeStack->push($name);
        $this->blockAttributesStack->push($attributes);
    }

    /**
     * Append some text to the current level of the stack.
     *
     * @param string $value
     */
    private function appendBlockText($value)
    {
        $this->setBlockText($this->getBlockText().$value);
    }

    /**
     * Set the text for a block.
     *
     * @param string $value
     */
    private function setBlockText($value)
    {
        $this->transformedTextStack->pop();
        $this->transformedTextStack->push($value);
    }

    /**
     * Get the current text that's in this block.
     *
     * @return string
     */
    private function getBlockText()
    {
        return $this->transformedTextStack->top();
    }

    /**
     * When we reach a closing element do something.
     *
     * This is mostly used to add stuff to the end of a statement, like putting new lines where div tags close
     *
     * @see xml_set_element_handler()
     *
     * @param resource $parser
     * @param string   $name
     */
    public function endElement($parser, $name)
    {
        switch ($name) {
            case 'P':
                $this->appendBlockText("\n\n");

                break;
            case 'UL':
                $this->appendBlockText("\n\n");

                break;
            case 'LI':
                $this->appendBlockText("\n");

                break;
            case 'DIV':
                $this->appendBlockText("\n\n\n");

                break;
            case 'A':
                $attrs = $this->blockAttributesStack->top();

                if (isset($attrs['HREF'])) {
                    $this->appendBlockText(" ({$attrs['HREF']})");
                }
        }

        $blockContent = $this->blockFinished();

        if (count($this->transformedTextStack)) {
            $this->appendBlockText($blockContent);
        } else {
            $this->text .= $blockContent;
        }
    }

    /**
     * Get the transformed text off the stack, and clear down the other stacks.
     *
     * @return string
     */
    private function blockFinished()
    {
        $transformedText = $this->transformedTextStack->pop();
        $this->blockTypeStack->pop();
        $this->blockAttributesStack->pop();

        return $transformedText;
    }

    /**
     * This converts character data to human character data.
     *
     * Primarily this is used for removing newlines and replacing them with spaces.
     *
     * @see xml_set_character_data_handler()
     *
     * @param resource $parser
     * @param string   $data
     */
    public function characterData($parser, $data)
    {
        $this->appendBlockText(str_replace("\n", ' ', $data));
    }

    /**
     * This gets the text that has been parsed and returns it.
     *
     * @return string
     */
    public function getText()
    {
        $text = trim($this->text);
        $lines = explode("\n", $text);

        return implode("\n", array_map('trim', $lines));
    }
}


1		<?php
2
3		/*
4		* Copyright (C) 2016 Billie Thompson
5		*
6		* This software may be modified and distributed under the terms
7		* of the MIT license. See the LICENSE file for details.
8		*/
9
10		namespace PurpleBooth;
11
12		/**
13		* This class is designed to by hooked into an XML parser to convert HTML to text.
14		*
15		* @see xml_parser_create()
16		* @see HtmlStripperExtension
17		*/
18		class Parser
19		{
20		/**
21		* @var string
22		*/
23		private $text = '';
24
25		/**
26		* @var \SplStack
27		*/
28		private $transformedTextStack;
29
30		/**
31		* @var \SplStack
32		*/
33		private $blockTypeStack;
34
35		/**
36		* @var \SplStack
37		*/
38		private $blockAttributesStack;
39
40		/**
41		* Parser constructor.
42		*
43		* Sets up the stacks
44		*/
45	6	public function __construct()
46		{
47	6	$this->transformedTextStack = new \SplStack();
48	6	$this->blockTypeStack = new \SplStack();
49	6	$this->blockAttributesStack = new \SplStack();
50	6	}
51
52		/**
53		* Function called on the start of an element.
54		*
55		* Mostly used to prepend text to things, like the "*"s on LIs
56		*
57		* @see xml_set_element_handler()
58		*
59		* @param resource $parser
60		* @param string $name
61		* @param array $attrs
62		*/
63	6	public function startElement($parser, $name, $attrs)
64		{
65	6	$this->blockBegin($name, $attrs);
66
67		switch ($name) {
68	6	case 'LI':
69	2	$this->appendBlockText('* ');
70
71	2	break;
72		}
73	6	}
74
75		/**
76		* Called when we begin a block.
77		*
78		* We build a series of stacks to represent the tree of the document
79		*
80		* We operate at this level of the stack when we're editing content
81		*
82		* @param string $name
83		* @param array $attributes
84		*/
85	6	private function blockBegin($name, $attributes)
86		{
87	6	$this->transformedTextStack->push('');
88	6	$this->blockTypeStack->push($name);
89	6	$this->blockAttributesStack->push($attributes);
90	6	}
91
92		/**
93		* Append some text to the current level of the stack.
94		*
95		* @param string $value
96		*/
97	6	private function appendBlockText($value)
98		{
99	6	$this->setBlockText($this->getBlockText().$value);
100	6	}
101
102		/**
103		* Set the text for a block.
104		*
105		* @param string $value
106		*/
107	6	private function setBlockText($value)
108		{
109	6	$this->transformedTextStack->pop();
110	6	$this->transformedTextStack->push($value);
111	6	}
112
113		/**
114		* Get the current text that's in this block.
115		*
116		* @return string
117		*/
118	6	private function getBlockText()
119		{
120	6	return $this->transformedTextStack->top();
121		}
122
123		/**
124		* When we reach a closing element do something.
125		*
126		* This is mostly used to add stuff to the end of a statement, like putting new lines where div tags close
127		*
128		* @see xml_set_element_handler()
129		*
130		* @param resource $parser
131		* @param string $name
132		*/
133	6	public function endElement($parser, $name)
134		{
135		switch ($name) {
136	6	case 'P':
137	2	$this->appendBlockText("\n\n");
138
139	2	break;
140	6	case 'UL':
141	2	$this->appendBlockText("\n\n");
142
143	2	break;
144	6	case 'LI':
145	2	$this->appendBlockText("\n");
146
147	2	break;
148	6	case 'DIV':
149	2	$this->appendBlockText("\n\n\n");
150
151	2	break;
152	6	case 'A':
153	1	$attrs = $this->blockAttributesStack->top();
154
155	1	if (isset($attrs['HREF'])) {
156	1	$this->appendBlockText(" ({$attrs['HREF']})");
157		}
158		}
159
160	6	$blockContent = $this->blockFinished();
161
162	6	if (count($this->transformedTextStack)) {
163	6	$this->appendBlockText($blockContent);
164		} else {
165	6	$this->text .= $blockContent;
166		}
167	6	}
168
169		/**
170		* Get the transformed text off the stack, and clear down the other stacks.
171		*
172		* @return string
173		*/
174	6	private function blockFinished()
175		{
176	6	$transformedText = $this->transformedTextStack->pop();
177	6	$this->blockTypeStack->pop();
178	6	$this->blockAttributesStack->pop();
179
180	6	return $transformedText;
181		}
182
183		/**
184		* This converts character data to human character data.
185		*
186		* Primarily this is used for removing newlines and replacing them with spaces.
187		*
188		* @see xml_set_character_data_handler()
189		*
190		* @param resource $parser
191		* @param string $data
192		*/
193	6	public function characterData($parser, $data)
194		{
195	6	$this->appendBlockText(str_replace("\n", ' ', $data));
196	6	}
197
198		/**
199		* This gets the text that has been parsed and returns it.
200		*
201		* @return string
202		*/
203	6	public function getText()
204		{
205	6	$text = trim($this->text);
206	6	$lines = explode("\n", $text);
207
208	6	return implode("\n", array_map('trim', $lines));
209		}
210		}
211

PurpleBooth / htmlstrip

Parser::characterData() A last analyzed 2020-06-03 06:40 UTC

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like

Parser::characterData() A
last analyzed 2020-06-03 06:40 UTC