Lexer::isWhitespace() - Code Metrics - Inspection of "Improved Lexer" - stenin-nikita/PHP-Yacc - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — master ( 0853ff...bcb7ae )

by Nikita

created 2017-12-27 16:06 UTC

Lexer::isWhitespace() A

↳ Parent: Lexer

Complexity

Conditions	3
Paths	3

Size

Total Lines	14
Code Lines	8

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
cc	3
eloc	8
nc	3
nop	0
dl	0
loc	14
rs	9.4285
c	0
b	0
f	0

<?php
/**
 * This file is part of PHP-Yacc package.
 *
 * For the full copyright and license information, please view the LICENSE
 * file that was distributed with this source code.
 */
declare(strict_types=1);

namespace PhpYacc\Yacc;

use PhpYacc\Exception\LexingException;
use PhpYacc\Exception\ParseException;
use PhpYacc\Support\Utils;

/**
 * Class Lexer.
 */
class Lexer
{
    /**
     * Whitespace tokens.
     */
    private const SPACE_TOKENS = [
        Token::SPACE,
        Token::COMMENT,
        Token::NEWLINE,
    ];

    /**
     * Tag map.
     */
    private const TAG_MAP = [
        '%%'            => Token::MARK,
        '%{'            => Token::BEGININC,
        '%}'            => Token::ENDINC,
        '%token'        => Token::TOKEN,
        '%term'         => Token::TOKEN,
        '%left'         => Token::LEFT,
        '%right'        => Token::RIGHT,
        '%nonassoc'     => Token::NONASSOC,
        '%prec'         => Token::PRECTOK,
        '%type'         => Token::TYPE,
        '%union'        => Token::UNION,
        '%start'        => Token::START,
        '%expect'       => Token::EXPECT,
        '%pure_parser'  => Token::PURE_PARSER,
    ];

    /**
     * @var string
     */
    protected $buffer;

    /**
     * @var string
     */
    protected $filename;

    /**
     * @var int
     */
    protected $line;

    /**
     * @var int
     */
    protected $offset;

    /**
     * @var Token
     */
    protected $currentToken;

    /**
     * @var Token
     */
    protected $backToken;

    /**
     * @var string
     */
    protected $backChar;

    /**
     * @var bool
     */
    protected $prevIsDollar;

    /**
     * @var string
     */
    protected $char;

    /**
     * @var string
     */
    protected $value;

    /**
     * @param string $code
     * @param string $filename
     */
    public function startLexing(string $code, string $filename = '')
    {
        $this->buffer = $code;
        $this->filename = $filename;

        $this->reset();
    }

    /**
     * @return void
     */
    protected function reset()
    {
        $this->line = 1;
        $this->offset = 0;
        $this->backChar = null;
        $this->backToken = null;
        $this->prevIsDollar = false;
    }

    /**
     * @throws LexingException
     * @throws ParseException
     *
     * @return Token
     */
    public function getToken(): Token
    {
        $this->currentToken = $this->getRawToken();

        while (in_array($this->currentToken->getType(), self::SPACE_TOKENS)) {
            $this->currentToken = $this->getRawToken();
        }

        return $this->currentToken;
    }

    /**
     * @throws LexingException
     */
    public function ungetToken()
    {
        if ($this->backToken !== null) {
            throw new LexingException('Too many ungetToken calls');
        }

        $this->backToken = $this->currentToken;
    }

    /**
     * @throws LexingException
     * @throws ParseException
     *
     * @return Token
     */
    public function peek(): Token
    {
        $result = $this->getToken();
        $this->ungetToken();

        return $result;
    }

    /**
     * @throws LexingException
     * @throws ParseException
     *
     * @return Token
     */
    public function getRawToken()
    {
        if ($this->backToken !== null) {
            $this->currentToken = $this->backToken;
            $this->backToken = null;

            return $this->currentToken;
        }

        $this->char = $this->getChar();
        $this->value = '';

        switch (true) {
            case $this->isWhitespace():
                return $this->token(Token::SPACE, $this->value);
            case $this->isNewline():
                return $this->token(Token::NEWLINE, $this->value);
            case $this->isComment():
                return $this->token(Token::COMMENT, $this->value);
            case $this->isEof():
                return $this->token(Token::EOF, $this->value);
        }

        $tag = $this->detectToken();

        switch (true) {
            case isset(self::TAG_MAP[$this->value]):
                return $this->token(self::TAG_MAP[$this->value], $this->value);
            case $this->value === ':':
                return $this->token(Token::COLON, $this->value);
            case $this->value === ';':
                return $this->token(Token::SEMICOLON, $this->value);
            case $this->value === '$':
                return $this->token(Token::DOLLAR, $this->value);
            default:
                return $this->token($tag, $this->value);
        }
    }

    /**
     * @return bool
     */
    protected function isWhitespace(): bool
    {
        if (Utils::isWhite($this->char)) {
            while (Utils::isWhite($this->char)) {
                $this->value .= $this->char;
                $this->char = $this->getChar();
            }
            $this->ungetChar($this->char);

            return true;
        }

        return false;
    }

    /**
     * @return bool
     */
    protected function isNewline(): bool
    {
        if ($this->char === "\n") {
            $this->value = $this->char;
            return true;
        }

        return false;
    }

    /**
     * @throws ParseException
     *
     * @return bool
     */
    protected function isComment(): bool
    {
        if ($this->char === '/') {
            if (($this->char = $this->getChar()) === '*') {
                $this->value = '/*';

                while (true) {
                    if (($this->char = $this->getChar()) === '*') {
                        if (($this->char = $this->getChar()) === '/') {
                            break;
                        }
                        $this->ungetChar($this->char);
                    }

                    if ($this->char === "\0") {
                        throw ParseException::unexpected($this->token(Token::EOF, "\0"), '*/');
                    }

                    $this->value .= $this->char;
                }

                $this->value .= '*/';

                return true;
            } elseif ($this->char === '/') {
                $this->value = '//';

                do {
                    $this->char = $this->getChar();
                    if ($this->char !== "\0") {
                        $this->value .= $this->char;
                    }
                } while ($this->char !== "\n" && $this->char !== "\0");

                return true;
            }

            $this->ungetChar($this->char);
            $this->char = '/';
        }

        return false;
    }

    /**
     * @return bool
     */
    protected function isEof(): bool
    {
        if ($this->char === "\0") {
            $this->value = $this->char;
            return true;
        }

        return false;
    }

    /**
     * @throws ParseException
     *
     * @return int
     */
    protected function detectToken()
    {
        $tag = Token::UNKNOW;

        if ($this->char === '%') {
            $this->char = $this->getChar();
            if ($this->char === '%' || $this->char === '{' | $this->char === '}' || Utils::isSymChar($this->char)) {
// Returns always int(0).
return 0 === $foo & 4;
return (0 === $foo) & 4;

// More likely intended return: true/false
return 0 === ($foo & 4);
                $this->value .= '%';
            } else {
                $this->ungetChar($this->char);
                $this->char = '%';
            }
        }

        if ($this->char === '$') {
            if (!$this->prevIsDollar) {
                $this->value .= '$';
                $this->char = $this->getChar();

                if ($this->char === '$') {
                    $this->ungetChar($this->char);
                    $this->prevIsDollar = true;
                } elseif (!\ctype_digit($this->char) && Utils::isSymChar($this->char)) {
                    do {
                        $this->value .= $this->char;
                        $this->char = $this->getChar();
                    } while (Utils::isSymChar($this->char));
                    $this->ungetChar($this->char);
                    $tag = Token::NAME;
                } else {
                    $this->ungetChar($this->char);
                }
            } else {
                $this->value .= '$';
                $this->prevIsDollar = false;
            }
        } elseif (Utils::isSymChar($this->char)) {
            do {
                $this->value .= $this->char;
                $this->char = $this->getChar();
            } while ($this->char !== "\0" && Utils::isSymChar($this->char));

            $this->ungetChar($this->char);
            $tag = \ctype_digit($this->value) ? Token::NUMBER : Token::NAME;
        } elseif ($this->char === '\'' || $this->char === '"') {
            $quote = $this->char;
            $this->value .= $this->char;

            while (($this->char = $this->getChar()) !== $quote) {
                if ($this->char === "\0") {
                    throw ParseException::unexpected($this->token(Token::EOF, "\0"), $quote);
                }

                if ($this->char === "\n") {
                    throw ParseException::unexpected($this->token(Token::NEWLINE, "\n"), $quote);
                }

                $this->value .= $this->char;
                if ($this->char === '\\') {
                    $this->char = $this->getChar();

                    if ($this->char === "\0") {
                        break;
                    }

                    if ($this->char === "\n") {
                        continue;
                    }

                    $this->value .= $this->char;
                }
            }
            $this->value .= $this->char;
            $tag = Token::STRING;
        } else {
            $this->value .= $this->char;
        }

        return $tag;
    }

    /**
     * @return string
     */
    protected function getChar(): string
    {
        if ($this->backChar !== null) {
            $result = $this->backChar;
            $this->backChar = null;

            return $result;
        }

        if ($this->offset >= \mb_strlen($this->buffer)) {
            return "\0";
        }

        $char = $this->buffer[$this->offset++];

        if ($char === "\n") {
            $this->line++;
        }

        return $char;
    }

    /**
     * @param string $char
     *
     * @throws LexingException
     */
    protected function ungetChar(string $char)
    {
        if ($char == "\0") {
            return;
        }

        if ($this->backChar !== null) {
            throw new LexingException('To many ungetChar calls');
        }

        $this->backChar = $char;
    }

    /**
     * @param int    $type
     * @param string $value
     *
     * @return Token
     */
    protected function token(int $type, string $value): Token
    {
        return new Token($type, $value, $this->line, $this->filename);
    }
}


1			<?php
2			/**
3			* This file is part of PHP-Yacc package.
4			*
5			* For the full copyright and license information, please view the LICENSE
6			* file that was distributed with this source code.
7			*/
8			declare(strict_types=1);
9
10			namespace PhpYacc\Yacc;
11
12			use PhpYacc\Exception\LexingException;
13			use PhpYacc\Exception\ParseException;
14			use PhpYacc\Support\Utils;
15
16			/**
17			* Class Lexer.
18			*/
19			class Lexer
20			{
21			/**
22			* Whitespace tokens.
23			*/
24			private const SPACE_TOKENS = [
25			Token::SPACE,
26			Token::COMMENT,
27			Token::NEWLINE,
28			];
29
30			/**
31			* Tag map.
32			*/
33			private const TAG_MAP = [
34			'%%' => Token::MARK,
35			'%{' => Token::BEGININC,
36			'%}' => Token::ENDINC,
37			'%token' => Token::TOKEN,
38			'%term' => Token::TOKEN,
39			'%left' => Token::LEFT,
40			'%right' => Token::RIGHT,
41			'%nonassoc' => Token::NONASSOC,
42			'%prec' => Token::PRECTOK,
43			'%type' => Token::TYPE,
44			'%union' => Token::UNION,
45			'%start' => Token::START,
46			'%expect' => Token::EXPECT,
47			'%pure_parser' => Token::PURE_PARSER,
48			];
49
50			/**
51			* @var string
52			*/
53			protected $buffer;
54
55			/**
56			* @var string
57			*/
58			protected $filename;
59
60			/**
61			* @var int
62			*/
63			protected $line;
64
65			/**
66			* @var int
67			*/
68			protected $offset;
69
70			/**
71			* @var Token
72			*/
73			protected $currentToken;
74
75			/**
76			* @var Token
77			*/
78			protected $backToken;
79
80			/**
81			* @var string
82			*/
83			protected $backChar;
84
85			/**
86			* @var bool
87			*/
88			protected $prevIsDollar;
89
90			/**
91			* @var string
92			*/
93			protected $char;
94
95			/**
96			* @var string
97			*/
98			protected $value;
99
100			/**
101			* @param string $code
102			* @param string $filename
103			*/
104			public function startLexing(string $code, string $filename = '')
105			{
106			$this->buffer = $code;
107			$this->filename = $filename;
108
109			$this->reset();
110			}
111
112			/**
113			* @return void
114			*/
115			protected function reset()
116			{
117			$this->line = 1;
118			$this->offset = 0;
119			$this->backChar = null;
120			$this->backToken = null;
121			$this->prevIsDollar = false;
122			}
123
124			/**
125			* @throws LexingException
126			* @throws ParseException
127			*
128			* @return Token
129			*/
130			public function getToken(): Token
131			{
132			$this->currentToken = $this->getRawToken();
133
134			while (in_array($this->currentToken->getType(), self::SPACE_TOKENS)) {
135			$this->currentToken = $this->getRawToken();
136			}
137
138			return $this->currentToken;
139			}
140
141			/**
142			* @throws LexingException
143			*/
144			public function ungetToken()
145			{
146			if ($this->backToken !== null) {
147			throw new LexingException('Too many ungetToken calls');
148			}
149
150			$this->backToken = $this->currentToken;
151			}
152
153			/**
154			* @throws LexingException
155			* @throws ParseException
156			*
157			* @return Token
158			*/
159			public function peek(): Token
160			{
161			$result = $this->getToken();
162			$this->ungetToken();
163
164			return $result;
165			}
166
167			/**
168			* @throws LexingException
169			* @throws ParseException
170			*
171			* @return Token
172			*/
173			public function getRawToken()
174			{
175			if ($this->backToken !== null) {
176			$this->currentToken = $this->backToken;
177			$this->backToken = null;
178
179			return $this->currentToken;
180			}
181
182			$this->char = $this->getChar();
183			$this->value = '';
184
185			switch (true) {
186			case $this->isWhitespace():
187			return $this->token(Token::SPACE, $this->value);
188			case $this->isNewline():
189			return $this->token(Token::NEWLINE, $this->value);
190			case $this->isComment():
191			return $this->token(Token::COMMENT, $this->value);
192			case $this->isEof():
193			return $this->token(Token::EOF, $this->value);
194			}
195
196			$tag = $this->detectToken();
197
198			switch (true) {
199			case isset(self::TAG_MAP[$this->value]):
200			return $this->token(self::TAG_MAP[$this->value], $this->value);
201			case $this->value === ':':
202			return $this->token(Token::COLON, $this->value);
203			case $this->value === ';':
204			return $this->token(Token::SEMICOLON, $this->value);
205			case $this->value === '$':
206			return $this->token(Token::DOLLAR, $this->value);
207			default:
208			return $this->token($tag, $this->value);
209			}
210			}
211
212			/**
213			* @return bool
214			*/
215			protected function isWhitespace(): bool
216			{
217			if (Utils::isWhite($this->char)) {
218			while (Utils::isWhite($this->char)) {
219			$this->value .= $this->char;
220			$this->char = $this->getChar();
221			}
222			$this->ungetChar($this->char);
223
224			return true;
225			}
226
227			return false;
228			}
229
230			/**
231			* @return bool
232			*/
233			protected function isNewline(): bool
234			{
235			if ($this->char === "\n") {
236			$this->value = $this->char;
237			return true;
238			}
239
240			return false;
241			}
242
243			/**
244			* @throws ParseException
245			*
246			* @return bool
247			*/
248			protected function isComment(): bool
249			{
250			if ($this->char === '/') {
251			if (($this->char = $this->getChar()) === '*') {
252			$this->value = '/*';
253
254			while (true) {
255			if (($this->char = $this->getChar()) === '*') {
256			if (($this->char = $this->getChar()) === '/') {
257			break;
258			}
259			$this->ungetChar($this->char);
260			}
261
262			if ($this->char === "\0") {
263			throw ParseException::unexpected($this->token(Token::EOF, "\0"), '*/');
264			}
265
266			$this->value .= $this->char;
267			}
268
269			$this->value .= '*/';
270
271			return true;
272			} elseif ($this->char === '/') {
273			$this->value = '//';
274
275			do {
276			$this->char = $this->getChar();
277			if ($this->char !== "\0") {
278			$this->value .= $this->char;
279			}
280			} while ($this->char !== "\n" && $this->char !== "\0");
281
282			return true;
283			}
284
285			$this->ungetChar($this->char);
286			$this->char = '/';
287			}
288
289			return false;
290			}
291
292			/**
293			* @return bool
294			*/
295			protected function isEof(): bool
296			{
297			if ($this->char === "\0") {
298			$this->value = $this->char;
299			return true;
300			}
301
302			return false;
303			}
304
305			/**
306			* @throws ParseException
307			*
308			* @return int
309			*/
310			protected function detectToken()
311			{
312			$tag = Token::UNKNOW;
313
314			if ($this->char === '%') {
315			$this->char = $this->getChar();
316			if ($this->char === '%' \|\| $this->char === '{' \| $this->char === '}' \|\| Utils::isSymChar($this->char)) {
			0 ignored issues – show Comprehensibility introduced 2017-12-27 16:11 UTC by Report Bug Copy Issue Report Consider adding parentheses for clarity. Current Interpretation: `($this->char === '{') \| $this->char === '}'`, Probably Intended Meaning: `$this->char === ('{' \| $this->char === '}')` When comparing the result of a bit operation, we suggest to add explicit parenthesis and not to rely on PHP’s built-in operator precedence to ensure the code behaves as intended and to make it more readable. Let’s take a look at these examples: // Returns always int(0). return 0 === $foo & 4; return (0 === $foo) & 4; // More likely intended return: true/false return 0 === ($foo & 4); Loading history...
317			$this->value .= '%';
318			} else {
319			$this->ungetChar($this->char);
320			$this->char = '%';
321			}
322			}
323
324			if ($this->char === '$') {
325			if (!$this->prevIsDollar) {
326			$this->value .= '$';
327			$this->char = $this->getChar();
328
329			if ($this->char === '$') {
330			$this->ungetChar($this->char);
331			$this->prevIsDollar = true;
332			} elseif (!\ctype_digit($this->char) && Utils::isSymChar($this->char)) {
333			do {
334			$this->value .= $this->char;
335			$this->char = $this->getChar();
336			} while (Utils::isSymChar($this->char));
337			$this->ungetChar($this->char);
338			$tag = Token::NAME;
339			} else {
340			$this->ungetChar($this->char);
341			}
342			} else {
343			$this->value .= '$';
344			$this->prevIsDollar = false;
345			}
346			} elseif (Utils::isSymChar($this->char)) {
347			do {
348			$this->value .= $this->char;
349			$this->char = $this->getChar();
350			} while ($this->char !== "\0" && Utils::isSymChar($this->char));
351
352			$this->ungetChar($this->char);
353			$tag = \ctype_digit($this->value) ? Token::NUMBER : Token::NAME;
354			} elseif ($this->char === '\'' \|\| $this->char === '"') {
355			$quote = $this->char;
356			$this->value .= $this->char;
357
358			while (($this->char = $this->getChar()) !== $quote) {
359			if ($this->char === "\0") {
360			throw ParseException::unexpected($this->token(Token::EOF, "\0"), $quote);
361			}
362
363			if ($this->char === "\n") {
364			throw ParseException::unexpected($this->token(Token::NEWLINE, "\n"), $quote);
365			}
366
367			$this->value .= $this->char;
368			if ($this->char === '\\') {
369			$this->char = $this->getChar();
370
371			if ($this->char === "\0") {
372			break;
373			}
374
375			if ($this->char === "\n") {
376			continue;
377			}
378
379			$this->value .= $this->char;
380			}
381			}
382			$this->value .= $this->char;
383			$tag = Token::STRING;
384			} else {
385			$this->value .= $this->char;
386			}
387
388			return $tag;
389			}
390
391			/**
392			* @return string
393			*/
394			protected function getChar(): string
395			{
396			if ($this->backChar !== null) {
397			$result = $this->backChar;
398			$this->backChar = null;
399
400			return $result;
401			}
402
403			if ($this->offset >= \mb_strlen($this->buffer)) {
404			return "\0";
405			}
406
407			$char = $this->buffer[$this->offset++];
408
409			if ($char === "\n") {
410			$this->line++;
411			}
412
413			return $char;
414			}
415
416			/**
417			* @param string $char
418			*
419			* @throws LexingException
420			*/
421			protected function ungetChar(string $char)
422			{
423			if ($char == "\0") {
424			return;
425			}
426
427			if ($this->backChar !== null) {
428			throw new LexingException('To many ungetChar calls');
429			}
430
431			$this->backChar = $char;
432			}
433
434			/**
435			* @param int $type
436			* @param string $value
437			*
438			* @return Token
439			*/
440			protected function token(int $type, string $value): Token
441			{
442			return new Token($type, $value, $this->line, $this->filename);
443			}
444			}
445

stenin-nikita / PHP-Yacc

Push — master ( 0853ff...bcb7ae )

Lexer::isWhitespace() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like