Parser::getText() - Code Metrics - Inspection of "Initial commit of twig free htmlstrip" - PurpleBooth/htmlstrip - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Push — master ( 17f425...1f2f65 )

by Billie

created 2016-07-24 18:13 UTC

Parser::getText() A

↳ Parent: Parser

Complexity

Conditions	1
Paths	1

Size

Total Lines	7
Code Lines	4

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	4
CRAP Score	1

Importance

Changes	1
Bugs	0	Features	0

Metric	Value
c	1
b	0
f	0
dl	0
loc	7
ccs	4
cts	4
cp	1
rs	9.4285
cc	1
eloc	4
nc	1
nop	0
crap	1

<?php

namespace PurpleBooth;

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

PurpleBooth / htmlstrip

Push — master ( 17f425...1f2f65 )

Parser::getText() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like