Utf8TokenMatcher - Code Metrics - Inspection of "Token matchers use built over minimized DFA" - remorhaz/php-unilex - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — master ( c01dfe...f39ce2 )

by Edward

created 2020-04-07 11:27 UTC

Utf8TokenMatcher A

↳ Parent: Project

Complexity

Total Complexity

Size/Duplication

Total Lines	211
Duplicated Lines	0 %

Importance

Changes	1
Bugs	0	Features	0

Metric	Value
eloc	163
dl	0
loc	211
rs	9.52
c	1
b	0
f	0
wmc	36

1 Method

Rating	Name	Duplication	Size	Complexity
F	match()	0	208	36

<?php

/**
 * Unicode UTF-8 token matcher.
 *
 * Auto-generated file, please don't edit manually.
 * Generated by UniLex.
 */

declare(strict_types=1);

namespace Remorhaz\UniLex\Unicode\Grammar;

use Remorhaz\UniLex\IO\CharBufferInterface;
use Remorhaz\UniLex\Lexer\TokenFactoryInterface;
use Remorhaz\UniLex\Lexer\TokenMatcherTemplate;

class Utf8TokenMatcher extends TokenMatcherTemplate
{

    public function match(CharBufferInterface $buffer, TokenFactoryInterface $tokenFactory): bool
    {
        $context = $this->createContext($buffer, $tokenFactory);
        $context->setRegExps(
            '[\\x00-\\x7F]',
            '[\\xC0-\\xDF][\\x80-\\xBF]',
            '[\\xE0-\\xEF][\\x80-\\xBF]{2}',
            '[\\xF0-\\xF7][\\x80-\\xBF]{3}',
            '[\\xF8-\\xFB][\\x80-\\xBF]{4}',
            '[\\xFC-\\xFD][\\x80-\\xBF]{5}'
        );
        goto state1;

        state1:
        if ($context->getBuffer()->isEnd()) {
            goto error;
        }
        $char = $context->getBuffer()->getSymbol();
        if (0x00 <= $char && $char <= 0x7F) {
            $context->getBuffer()->nextSymbol();
            $context->allowRegExps('[\\x00-\\x7F]');
            goto state2;
        }
        if (0xC0 <= $char && $char <= 0xDF) {
            $context->getBuffer()->nextSymbol();
            $context->allowRegExps('[\\xC0-\\xDF][\\x80-\\xBF]');
            goto state3;
        }
        if (0xE0 <= $char && $char <= 0xEF) {
            $context->getBuffer()->nextSymbol();
            $context->allowRegExps('[\\xE0-\\xEF][\\x80-\\xBF]{2}');
            goto state4;
        }
        if (0xF0 <= $char && $char <= 0xF7) {
            $context->getBuffer()->nextSymbol();
            $context->allowRegExps('[\\xF0-\\xF7][\\x80-\\xBF]{3}');
            goto state5;
        }
        if (0xF8 <= $char && $char <= 0xFB) {
            $context->getBuffer()->nextSymbol();
            $context->allowRegExps('[\\xF8-\\xFB][\\x80-\\xBF]{4}');
            goto state6;
        }
        if (0xFC == $char || 0xFD == $char) {
            $context->getBuffer()->nextSymbol();
            $context->allowRegExps('[\\xFC-\\xFD][\\x80-\\xBF]{5}');
            goto state7;
        }
        goto error;

        state2:
        switch ($context->getRegExp()) {
            case '[\\x00-\\x7F]':
                $context
                    ->setNewToken(TokenType::SYMBOL)
                    ->setTokenAttribute(TokenAttribute::UNICODE_CHAR, $char);

                return true;

            case '[\\xC0-\\xDF][\\x80-\\xBF]':
                $charList = array_slice($context->getSymbolList(), -2);
                $symbol = ($charList[0] & 0x1F) << 6;
                $symbol |= ($charList[1] & 0x3F);
                $context
                    ->setNewToken(TokenType::SYMBOL)
                    ->setTokenAttribute(TokenAttribute::UNICODE_CHAR, $symbol);

                return true;

            case '[\\xE0-\\xEF][\\x80-\\xBF]{2}':
                $charList = array_slice($context->getSymbolList(), -3);
                $symbol = ($charList[0] & 0x0F) << 12;
                $symbol |= ($charList[1] & 0x3F) << 6;
                $symbol |= ($charList[2] & 0x3F);
                $context
                    ->setNewToken(TokenType::SYMBOL)
                    ->setTokenAttribute(TokenAttribute::UNICODE_CHAR, $symbol);

                return true;

            case '[\\xF0-\\xF7][\\x80-\\xBF]{3}':
                $charList = array_slice($context->getSymbolList(), -4);
                $symbol = ($charList[0] & 0x07) << 18;
                $symbol |= ($charList[1] & 0x3F) << 12;
                $symbol |= ($charList[2] & 0x3F) << 6;
                $symbol |= ($charList[3] & 0x3F);
                $context
                    ->setNewToken(TokenType::SYMBOL)
                    ->setTokenAttribute(TokenAttribute::UNICODE_CHAR, $symbol);

                return true;

            case '[\\xF8-\\xFB][\\x80-\\xBF]{4}':
                $charList = array_slice($context->getSymbolList(), -5);
                $symbol = ($charList[0] & 0x03) << 24;
                $symbol |= ($charList[1] & 0x3F) << 18;
                $symbol |= ($charList[2] & 0x3F) << 12;
                $symbol |= ($charList[3] & 0x3F) << 6;
                $symbol |= ($charList[4] & 0x3F);
                $context
                    ->setNewToken(TokenType::SYMBOL)
                    ->setTokenAttribute(TokenAttribute::UNICODE_CHAR, $symbol);

                return true;

            case '[\\xFC-\\xFD][\\x80-\\xBF]{5}':
                $charList = array_slice($context->getSymbolList(), -6);
                $symbol = ($charList[0] & 0x01) << 30;
                $symbol |= ($charList[1] & 0x03) << 24;
                $symbol |= ($charList[2] & 0x3F) << 18;
                $symbol |= ($charList[3] & 0x3F) << 12;
                $symbol |= ($charList[4] & 0x3F) << 6;
                $symbol |= ($charList[5] & 0x3F);
                $context
                    ->setNewToken(TokenType::SYMBOL)
                    ->setTokenAttribute(TokenAttribute::UNICODE_CHAR, $symbol);

                return true;

            default:
                goto error;
        }

        state3:
        if ($context->getBuffer()->isEnd()) {
            goto error;
        }
        $char = $context->getBuffer()->getSymbol();
        if (0x80 <= $char && $char <= 0xBF) {
            $context->getBuffer()->nextSymbol();
            $context->allowRegExps(
                '[\\xC0-\\xDF][\\x80-\\xBF]',
                '[\\xE0-\\xEF][\\x80-\\xBF]{2}',
                '[\\xF0-\\xF7][\\x80-\\xBF]{3}',
                '[\\xF8-\\xFB][\\x80-\\xBF]{4}',
                '[\\xFC-\\xFD][\\x80-\\xBF]{5}'
            );
            goto state2;
        }
        goto error;

        state4:
        if ($context->getBuffer()->isEnd()) {
            goto error;
        }
        $char = $context->getBuffer()->getSymbol();
        if (0x80 <= $char && $char <= 0xBF) {
            $context->getBuffer()->nextSymbol();
            $context->allowRegExps(
                '[\\xE0-\\xEF][\\x80-\\xBF]{2}',
                '[\\xF0-\\xF7][\\x80-\\xBF]{3}',
                '[\\xF8-\\xFB][\\x80-\\xBF]{4}',
                '[\\xFC-\\xFD][\\x80-\\xBF]{5}'
            );
            goto state3;
        }
        goto error;

        state5:
        if ($context->getBuffer()->isEnd()) {
            goto error;
        }
        $char = $context->getBuffer()->getSymbol();
        if (0x80 <= $char && $char <= 0xBF) {
            $context->getBuffer()->nextSymbol();
            $context->allowRegExps(
                '[\\xF0-\\xF7][\\x80-\\xBF]{3}',
                '[\\xF8-\\xFB][\\x80-\\xBF]{4}',
                '[\\xFC-\\xFD][\\x80-\\xBF]{5}'
            );
            goto state4;
        }
        goto error;

        state6:
        if ($context->getBuffer()->isEnd()) {
            goto error;
        }
        $char = $context->getBuffer()->getSymbol();
        if (0x80 <= $char && $char <= 0xBF) {
            $context->getBuffer()->nextSymbol();
            $context->allowRegExps(
                '[\\xF8-\\xFB][\\x80-\\xBF]{4}',
                '[\\xFC-\\xFD][\\x80-\\xBF]{5}'
            );
            goto state5;
        }
        goto error;

        state7:
        if ($context->getBuffer()->isEnd()) {
            goto error;
        }
        $char = $context->getBuffer()->getSymbol();
        if (0x80 <= $char && $char <= 0xBF) {
            $context->getBuffer()->nextSymbol();
            $context->allowRegExps('[\\xFC-\\xFD][\\x80-\\xBF]{5}');
            goto state6;
        }
        goto error;

        error:
        if ($context->getBuffer()->isEnd()) {
            return false;
        }
        $context->getBuffer()->nextSymbol();
        $context->setNewToken(TokenType::INVALID_BYTES);
        return true;
    }
}


1			<?php
2
3			/**
4			* Unicode UTF-8 token matcher.
5			*
6			* Auto-generated file, please don't edit manually.
7			* Generated by UniLex.
8			*/
9
10			declare(strict_types=1);
11
12			namespace Remorhaz\UniLex\Unicode\Grammar;
13
14			use Remorhaz\UniLex\IO\CharBufferInterface;
15			use Remorhaz\UniLex\Lexer\TokenFactoryInterface;
16			use Remorhaz\UniLex\Lexer\TokenMatcherTemplate;
17
18			class Utf8TokenMatcher extends TokenMatcherTemplate
19			{
20
21			public function match(CharBufferInterface $buffer, TokenFactoryInterface $tokenFactory): bool
22			{
23			$context = $this->createContext($buffer, $tokenFactory);
24			$context->setRegExps(
25			'[\\x00-\\x7F]',
26			'[\\xC0-\\xDF][\\x80-\\xBF]',
27			'[\\xE0-\\xEF][\\x80-\\xBF]{2}',
28			'[\\xF0-\\xF7][\\x80-\\xBF]{3}',
29			'[\\xF8-\\xFB][\\x80-\\xBF]{4}',
30			'[\\xFC-\\xFD][\\x80-\\xBF]{5}'
31			);
32			goto state1;
33
34			state1:
35			if ($context->getBuffer()->isEnd()) {
36			goto error;
37			}
38			$char = $context->getBuffer()->getSymbol();
39			if (0x00 <= $char && $char <= 0x7F) {
40			$context->getBuffer()->nextSymbol();
41			$context->allowRegExps('[\\x00-\\x7F]');
42			goto state2;
43			}
44			if (0xC0 <= $char && $char <= 0xDF) {
45			$context->getBuffer()->nextSymbol();
46			$context->allowRegExps('[\\xC0-\\xDF][\\x80-\\xBF]');
47			goto state3;
48			}
49			if (0xE0 <= $char && $char <= 0xEF) {
50			$context->getBuffer()->nextSymbol();
51			$context->allowRegExps('[\\xE0-\\xEF][\\x80-\\xBF]{2}');
52			goto state4;
53			}
54			if (0xF0 <= $char && $char <= 0xF7) {
55			$context->getBuffer()->nextSymbol();
56			$context->allowRegExps('[\\xF0-\\xF7][\\x80-\\xBF]{3}');
57			goto state5;
58			}
59			if (0xF8 <= $char && $char <= 0xFB) {
60			$context->getBuffer()->nextSymbol();
61			$context->allowRegExps('[\\xF8-\\xFB][\\x80-\\xBF]{4}');
62			goto state6;
63			}
64			if (0xFC == $char \|\| 0xFD == $char) {
65			$context->getBuffer()->nextSymbol();
66			$context->allowRegExps('[\\xFC-\\xFD][\\x80-\\xBF]{5}');
67			goto state7;
68			}
69			goto error;
70
71			state2:
72			switch ($context->getRegExp()) {
73			case '[\\x00-\\x7F]':
74			$context
75			->setNewToken(TokenType::SYMBOL)
76			->setTokenAttribute(TokenAttribute::UNICODE_CHAR, $char);
77
78			return true;
79
80			case '[\\xC0-\\xDF][\\x80-\\xBF]':
81			$charList = array_slice($context->getSymbolList(), -2);
82			$symbol = ($charList[0] & 0x1F) << 6;
83			$symbol \|= ($charList[1] & 0x3F);
84			$context
85			->setNewToken(TokenType::SYMBOL)
86			->setTokenAttribute(TokenAttribute::UNICODE_CHAR, $symbol);
87
88			return true;
89
90			case '[\\xE0-\\xEF][\\x80-\\xBF]{2}':
91			$charList = array_slice($context->getSymbolList(), -3);
92			$symbol = ($charList[0] & 0x0F) << 12;
93			$symbol \|= ($charList[1] & 0x3F) << 6;
94			$symbol \|= ($charList[2] & 0x3F);
95			$context
96			->setNewToken(TokenType::SYMBOL)
97			->setTokenAttribute(TokenAttribute::UNICODE_CHAR, $symbol);
98
99			return true;
100
101			case '[\\xF0-\\xF7][\\x80-\\xBF]{3}':
102			$charList = array_slice($context->getSymbolList(), -4);
103			$symbol = ($charList[0] & 0x07) << 18;
104			$symbol \|= ($charList[1] & 0x3F) << 12;
105			$symbol \|= ($charList[2] & 0x3F) << 6;
106			$symbol \|= ($charList[3] & 0x3F);
107			$context
108			->setNewToken(TokenType::SYMBOL)
109			->setTokenAttribute(TokenAttribute::UNICODE_CHAR, $symbol);
110
111			return true;
112
113			case '[\\xF8-\\xFB][\\x80-\\xBF]{4}':
114			$charList = array_slice($context->getSymbolList(), -5);
115			$symbol = ($charList[0] & 0x03) << 24;
116			$symbol \|= ($charList[1] & 0x3F) << 18;
117			$symbol \|= ($charList[2] & 0x3F) << 12;
118			$symbol \|= ($charList[3] & 0x3F) << 6;
119			$symbol \|= ($charList[4] & 0x3F);
120			$context
121			->setNewToken(TokenType::SYMBOL)
122			->setTokenAttribute(TokenAttribute::UNICODE_CHAR, $symbol);
123
124			return true;
125
126			case '[\\xFC-\\xFD][\\x80-\\xBF]{5}':
127			$charList = array_slice($context->getSymbolList(), -6);
128			$symbol = ($charList[0] & 0x01) << 30;
129			$symbol \|= ($charList[1] & 0x03) << 24;
130			$symbol \|= ($charList[2] & 0x3F) << 18;
131			$symbol \|= ($charList[3] & 0x3F) << 12;
132			$symbol \|= ($charList[4] & 0x3F) << 6;
133			$symbol \|= ($charList[5] & 0x3F);
134			$context
135			->setNewToken(TokenType::SYMBOL)
136			->setTokenAttribute(TokenAttribute::UNICODE_CHAR, $symbol);
137
138			return true;
139
140			default:
141			goto error;
142			}
143
144			state3:
145			if ($context->getBuffer()->isEnd()) {
146			goto error;
147			}
148			$char = $context->getBuffer()->getSymbol();
149			if (0x80 <= $char && $char <= 0xBF) {
150			$context->getBuffer()->nextSymbol();
151			$context->allowRegExps(
152			'[\\xC0-\\xDF][\\x80-\\xBF]',
153			'[\\xE0-\\xEF][\\x80-\\xBF]{2}',
154			'[\\xF0-\\xF7][\\x80-\\xBF]{3}',
155			'[\\xF8-\\xFB][\\x80-\\xBF]{4}',
156			'[\\xFC-\\xFD][\\x80-\\xBF]{5}'
157			);
158			goto state2;
159			}
160			goto error;
161
162			state4:
163			if ($context->getBuffer()->isEnd()) {
164			goto error;
165			}
166			$char = $context->getBuffer()->getSymbol();
167			if (0x80 <= $char && $char <= 0xBF) {
168			$context->getBuffer()->nextSymbol();
169			$context->allowRegExps(
170			'[\\xE0-\\xEF][\\x80-\\xBF]{2}',
171			'[\\xF0-\\xF7][\\x80-\\xBF]{3}',
172			'[\\xF8-\\xFB][\\x80-\\xBF]{4}',
173			'[\\xFC-\\xFD][\\x80-\\xBF]{5}'
174			);
175			goto state3;
176			}
177			goto error;
178
179			state5:
180			if ($context->getBuffer()->isEnd()) {
181			goto error;
182			}
183			$char = $context->getBuffer()->getSymbol();
184			if (0x80 <= $char && $char <= 0xBF) {
185			$context->getBuffer()->nextSymbol();
186			$context->allowRegExps(
187			'[\\xF0-\\xF7][\\x80-\\xBF]{3}',
188			'[\\xF8-\\xFB][\\x80-\\xBF]{4}',
189			'[\\xFC-\\xFD][\\x80-\\xBF]{5}'
190			);
191			goto state4;
192			}
193			goto error;
194
195			state6:
196			if ($context->getBuffer()->isEnd()) {
197			goto error;
198			}
199			$char = $context->getBuffer()->getSymbol();
200			if (0x80 <= $char && $char <= 0xBF) {
201			$context->getBuffer()->nextSymbol();
202			$context->allowRegExps(
203			'[\\xF8-\\xFB][\\x80-\\xBF]{4}',
204			'[\\xFC-\\xFD][\\x80-\\xBF]{5}'
205			);
206			goto state5;
207			}
208			goto error;
209
210			state7:
211			if ($context->getBuffer()->isEnd()) {
212			goto error;
213			}
214			$char = $context->getBuffer()->getSymbol();
215			if (0x80 <= $char && $char <= 0xBF) {
216			$context->getBuffer()->nextSymbol();
217			$context->allowRegExps('[\\xFC-\\xFD][\\x80-\\xBF]{5}');
218			goto state6;
219			}
220			goto error;
221
222			error:
223			if ($context->getBuffer()->isEnd()) {
224			return false;
225			}
226			$context->getBuffer()->nextSymbol();
227			$context->setNewToken(TokenType::INVALID_BYTES);
228			return true;
229			}
230			}
231

remorhaz / php-unilex

Push — master ( c01dfe...f39ce2 )

Utf8TokenMatcher A

Complexity

Size/Duplication

Importance

1 Method

Duplication Side-by-Side

Filter issues like