Utf8TokenMatcher::match() - Code Metrics - remorhaz/php-unilex - Measure and Improve Code Quality continuously with Scrutinizer

Utf8TokenMatcher::match() F
last analyzed 2023-06-12 18:01 UTC

↳ Parent: Utf8TokenMatcher

Complexity

Conditions	60
Paths	70

Size

Total Lines	263
Code Lines	195

Duplication

Lines	0
Ratio	0 %

Importance

Changes	3
Bugs	0	Features	0

Metric	Value
cc	60
eloc	195
nc	70
nop	2
dl	0
loc	263
rs	3.3333
c	3
b	0
f	0

How to fix Long Method Complexity

<?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);
        goto state1;

        state1:
        if ($context->getBuffer()->isEnd()) {
            goto error;
        }
        $char = $context->getBuffer()->getSymbol();
        if (0x00 <= $char && $char <= 0x7F) {
            $context->getBuffer()->nextSymbol();
            // [\x00-\x7F]
            $context
                ->setNewToken(TokenType::SYMBOL)
                ->setTokenAttribute(TokenAttribute::UNICODE_CHAR, $char);

            return true;
        }
        if (0xC0 <= $char && $char <= 0xDF) {
            $context->getBuffer()->nextSymbol();
            goto state3;
        }
        if (0xE0 <= $char && $char <= 0xEF) {
            $context->getBuffer()->nextSymbol();
            goto state4;
        }
        if (0xF0 <= $char && $char <= 0xF7) {
            $context->getBuffer()->nextSymbol();
            goto state5;
        }
        if (0xF8 <= $char && $char <= 0xFB) {
            $context->getBuffer()->nextSymbol();
            goto state6;
        }
        if (0xFC == $char || 0xFD == $char) {
            $context->getBuffer()->nextSymbol();
            goto state7;
        }
        goto error;

        state3:
        if ($context->getBuffer()->isEnd()) {
            goto error;
        }
        $char = $context->getBuffer()->getSymbol();
        if (0x80 <= $char && $char <= 0xBF) {
            $context->getBuffer()->nextSymbol();
            // [\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;
        }
        goto error;

        state4:
        if ($context->getBuffer()->isEnd()) {
            goto error;
        }
        $char = $context->getBuffer()->getSymbol();
        if (0x80 <= $char && $char <= 0xBF) {
            $context->getBuffer()->nextSymbol();
            goto state20;
        }
        goto error;

        state5:
        if ($context->getBuffer()->isEnd()) {
            goto error;
        }
        $char = $context->getBuffer()->getSymbol();
        if (0x80 <= $char && $char <= 0xBF) {
            $context->getBuffer()->nextSymbol();
            goto state17;
        }
        goto error;

        state6:
        if ($context->getBuffer()->isEnd()) {
            goto error;
        }
        $char = $context->getBuffer()->getSymbol();
        if (0x80 <= $char && $char <= 0xBF) {
            $context->getBuffer()->nextSymbol();
            goto state13;
        }
        goto error;

        state7:
        if ($context->getBuffer()->isEnd()) {
            goto error;
        }
        $char = $context->getBuffer()->getSymbol();
        if (0x80 <= $char && $char <= 0xBF) {
            $context->getBuffer()->nextSymbol();
            goto state8;
        }
        goto error;

        state8:
        if ($context->getBuffer()->isEnd()) {
            goto error;
        }
        $char = $context->getBuffer()->getSymbol();
        if (0x80 <= $char && $char <= 0xBF) {
            $context->getBuffer()->nextSymbol();
            goto state9;
        }
        goto error;

        state9:
        if ($context->getBuffer()->isEnd()) {
            goto error;
        }
        $char = $context->getBuffer()->getSymbol();
        if (0x80 <= $char && $char <= 0xBF) {
            $context->getBuffer()->nextSymbol();
            goto state10;
        }
        goto error;

        state10:
        if ($context->getBuffer()->isEnd()) {
            goto error;
        }
        $char = $context->getBuffer()->getSymbol();
        if (0x80 <= $char && $char <= 0xBF) {
            $context->getBuffer()->nextSymbol();
            goto state11;
        }
        goto error;

        state11:
        if ($context->getBuffer()->isEnd()) {
            goto error;
        }
        $char = $context->getBuffer()->getSymbol();
        if (0x80 <= $char && $char <= 0xBF) {
            $context->getBuffer()->nextSymbol();
            // [\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;
        }
        goto error;

        state13:
        if ($context->getBuffer()->isEnd()) {
            goto error;
        }
        $char = $context->getBuffer()->getSymbol();
        if (0x80 <= $char && $char <= 0xBF) {
            $context->getBuffer()->nextSymbol();
            goto state14;
        }
        goto error;

        state14:
        if ($context->getBuffer()->isEnd()) {
            goto error;
        }
        $char = $context->getBuffer()->getSymbol();
        if (0x80 <= $char && $char <= 0xBF) {
            $context->getBuffer()->nextSymbol();
            goto state15;
        }
        goto error;

        state15:
        if ($context->getBuffer()->isEnd()) {
            goto error;
        }
        $char = $context->getBuffer()->getSymbol();
        if (0x80 <= $char && $char <= 0xBF) {
            $context->getBuffer()->nextSymbol();
            // [\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;
        }
        goto error;

        state17:
        if ($context->getBuffer()->isEnd()) {
            goto error;
        }
        $char = $context->getBuffer()->getSymbol();
        if (0x80 <= $char && $char <= 0xBF) {
            $context->getBuffer()->nextSymbol();
            goto state18;
        }
        goto error;

        state18:
        if ($context->getBuffer()->isEnd()) {
            goto error;
        }
        $char = $context->getBuffer()->getSymbol();
        if (0x80 <= $char && $char <= 0xBF) {
            $context->getBuffer()->nextSymbol();
            // [\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;
        }
        goto error;

        state20:
        if ($context->getBuffer()->isEnd()) {
            goto error;
        }
        $char = $context->getBuffer()->getSymbol();
        if (0x80 <= $char && $char <= 0xBF) {
            $context->getBuffer()->nextSymbol();
            // [\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;
        }
        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			public function match(CharBufferInterface $buffer, TokenFactoryInterface $tokenFactory): bool
21			{
22			$context = $this->createContext($buffer, $tokenFactory);
23			goto state1;
24
25			state1:
26			if ($context->getBuffer()->isEnd()) {
27			goto error;
28			}
29			$char = $context->getBuffer()->getSymbol();
30			if (0x00 <= $char && $char <= 0x7F) {
31			$context->getBuffer()->nextSymbol();
32			// [\x00-\x7F]
33			$context
34			->setNewToken(TokenType::SYMBOL)
35			->setTokenAttribute(TokenAttribute::UNICODE_CHAR, $char);
36
37			return true;
38			}
39			if (0xC0 <= $char && $char <= 0xDF) {
40			$context->getBuffer()->nextSymbol();
41			goto state3;
42			}
43			if (0xE0 <= $char && $char <= 0xEF) {
44			$context->getBuffer()->nextSymbol();
45			goto state4;
46			}
47			if (0xF0 <= $char && $char <= 0xF7) {
48			$context->getBuffer()->nextSymbol();
49			goto state5;
50			}
51			if (0xF8 <= $char && $char <= 0xFB) {
52			$context->getBuffer()->nextSymbol();
53			goto state6;
54			}
55			if (0xFC == $char \|\| 0xFD == $char) {
56			$context->getBuffer()->nextSymbol();
57			goto state7;
58			}
59			goto error;
60
61			state3:
62			if ($context->getBuffer()->isEnd()) {
63			goto error;
64			}
65			$char = $context->getBuffer()->getSymbol();
66			if (0x80 <= $char && $char <= 0xBF) {
67			$context->getBuffer()->nextSymbol();
68			// [\xC0-\xDF][\x80-\xBF]
69			$charList = array_slice($context->getSymbolList(), -2);
70			$symbol = ($charList[0] & 0x1F) << 6;
71			$symbol \|= ($charList[1] & 0x3F);
72			$context
73			->setNewToken(TokenType::SYMBOL)
74			->setTokenAttribute(TokenAttribute::UNICODE_CHAR, $symbol);
75
76			return true;
77			}
78			goto error;
79
80			state4:
81			if ($context->getBuffer()->isEnd()) {
82			goto error;
83			}
84			$char = $context->getBuffer()->getSymbol();
85			if (0x80 <= $char && $char <= 0xBF) {
86			$context->getBuffer()->nextSymbol();
87			goto state20;
88			}
89			goto error;
90
91			state5:
92			if ($context->getBuffer()->isEnd()) {
93			goto error;
94			}
95			$char = $context->getBuffer()->getSymbol();
96			if (0x80 <= $char && $char <= 0xBF) {
97			$context->getBuffer()->nextSymbol();
98			goto state17;
99			}
100			goto error;
101
102			state6:
103			if ($context->getBuffer()->isEnd()) {
104			goto error;
105			}
106			$char = $context->getBuffer()->getSymbol();
107			if (0x80 <= $char && $char <= 0xBF) {
108			$context->getBuffer()->nextSymbol();
109			goto state13;
110			}
111			goto error;
112
113			state7:
114			if ($context->getBuffer()->isEnd()) {
115			goto error;
116			}
117			$char = $context->getBuffer()->getSymbol();
118			if (0x80 <= $char && $char <= 0xBF) {
119			$context->getBuffer()->nextSymbol();
120			goto state8;
121			}
122			goto error;
123
124			state8:
125			if ($context->getBuffer()->isEnd()) {
126			goto error;
127			}
128			$char = $context->getBuffer()->getSymbol();
129			if (0x80 <= $char && $char <= 0xBF) {
130			$context->getBuffer()->nextSymbol();
131			goto state9;
132			}
133			goto error;
134
135			state9:
136			if ($context->getBuffer()->isEnd()) {
137			goto error;
138			}
139			$char = $context->getBuffer()->getSymbol();
140			if (0x80 <= $char && $char <= 0xBF) {
141			$context->getBuffer()->nextSymbol();
142			goto state10;
143			}
144			goto error;
145
146			state10:
147			if ($context->getBuffer()->isEnd()) {
148			goto error;
149			}
150			$char = $context->getBuffer()->getSymbol();
151			if (0x80 <= $char && $char <= 0xBF) {
152			$context->getBuffer()->nextSymbol();
153			goto state11;
154			}
155			goto error;
156
157			state11:
158			if ($context->getBuffer()->isEnd()) {
159			goto error;
160			}
161			$char = $context->getBuffer()->getSymbol();
162			if (0x80 <= $char && $char <= 0xBF) {
163			$context->getBuffer()->nextSymbol();
164			// [\xFC-\xFD][\x80-\xBF]{5}
165			$charList = array_slice($context->getSymbolList(), -6);
166			$symbol = ($charList[0] & 0x01) << 30;
167			$symbol \|= ($charList[1] & 0x03) << 24;
168			$symbol \|= ($charList[2] & 0x3F) << 18;
169			$symbol \|= ($charList[3] & 0x3F) << 12;
170			$symbol \|= ($charList[4] & 0x3F) << 6;
171			$symbol \|= ($charList[5] & 0x3F);
172			$context
173			->setNewToken(TokenType::SYMBOL)
174			->setTokenAttribute(TokenAttribute::UNICODE_CHAR, $symbol);
175
176			return true;
177			}
178			goto error;
179
180			state13:
181			if ($context->getBuffer()->isEnd()) {
182			goto error;
183			}
184			$char = $context->getBuffer()->getSymbol();
185			if (0x80 <= $char && $char <= 0xBF) {
186			$context->getBuffer()->nextSymbol();
187			goto state14;
188			}
189			goto error;
190
191			state14:
192			if ($context->getBuffer()->isEnd()) {
193			goto error;
194			}
195			$char = $context->getBuffer()->getSymbol();
196			if (0x80 <= $char && $char <= 0xBF) {
197			$context->getBuffer()->nextSymbol();
198			goto state15;
199			}
200			goto error;
201
202			state15:
203			if ($context->getBuffer()->isEnd()) {
204			goto error;
205			}
206			$char = $context->getBuffer()->getSymbol();
207			if (0x80 <= $char && $char <= 0xBF) {
208			$context->getBuffer()->nextSymbol();
209			// [\xF8-\xFB][\x80-\xBF]{4}
210			$charList = array_slice($context->getSymbolList(), -5);
211			$symbol = ($charList[0] & 0x03) << 24;
212			$symbol \|= ($charList[1] & 0x3F) << 18;
213			$symbol \|= ($charList[2] & 0x3F) << 12;
214			$symbol \|= ($charList[3] & 0x3F) << 6;
215			$symbol \|= ($charList[4] & 0x3F);
216			$context
217			->setNewToken(TokenType::SYMBOL)
218			->setTokenAttribute(TokenAttribute::UNICODE_CHAR, $symbol);
219
220			return true;
221			}
222			goto error;
223
224			state17:
225			if ($context->getBuffer()->isEnd()) {
226			goto error;
227			}
228			$char = $context->getBuffer()->getSymbol();
229			if (0x80 <= $char && $char <= 0xBF) {
230			$context->getBuffer()->nextSymbol();
231			goto state18;
232			}
233			goto error;
234
235			state18:
236			if ($context->getBuffer()->isEnd()) {
237			goto error;
238			}
239			$char = $context->getBuffer()->getSymbol();
240			if (0x80 <= $char && $char <= 0xBF) {
241			$context->getBuffer()->nextSymbol();
242			// [\xF0-\xF7][\x80-\xBF]{3}
243			$charList = array_slice($context->getSymbolList(), -4);
244			$symbol = ($charList[0] & 0x07) << 18;
245			$symbol \|= ($charList[1] & 0x3F) << 12;
246			$symbol \|= ($charList[2] & 0x3F) << 6;
247			$symbol \|= ($charList[3] & 0x3F);
248			$context
249			->setNewToken(TokenType::SYMBOL)
250			->setTokenAttribute(TokenAttribute::UNICODE_CHAR, $symbol);
251
252			return true;
253			}
254			goto error;
255
256			state20:
257			if ($context->getBuffer()->isEnd()) {
258			goto error;
259			}
260			$char = $context->getBuffer()->getSymbol();
261			if (0x80 <= $char && $char <= 0xBF) {
262			$context->getBuffer()->nextSymbol();
263			// [\xE0-\xEF][\x80-\xBF]{2}
264			$charList = array_slice($context->getSymbolList(), -3);
265			$symbol = ($charList[0] & 0x0F) << 12;
266			$symbol \|= ($charList[1] & 0x3F) << 6;
267			$symbol \|= ($charList[2] & 0x3F);
268			$context
269			->setNewToken(TokenType::SYMBOL)
270			->setTokenAttribute(TokenAttribute::UNICODE_CHAR, $symbol);
271
272			return true;
273			}
274			goto error;
275
276			error:
277			if ($context->getBuffer()->isEnd()) {
278			return false;
279			}
280			$context->getBuffer()->nextSymbol();
281			$context->setNewToken(TokenType::INVALID_BYTES);
282			return true;
283			}
284			}
285

remorhaz / php-unilex

Utf8TokenMatcher::match() F last analyzed 2023-06-12 18:01 UTC

Complexity

Size

Duplication

Importance

How to fix Long Method Complexity

Long Method

Duplication Side-by-Side

Filter issues like

Utf8TokenMatcher::match() F
last analyzed 2023-06-12 18:01 UTC