Lexer::readToken() - Code Metrics - Inspection of "Lexer optimizations" - digiaonline/graphql-php - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Pull Request — master (#196)

by Christoffer

created 2018-04-11 12:26 UTC

Lexer::readToken() C

↳ Parent: Lexer

Complexity

Conditions	24
Paths	9

Size

Total Lines	53
Code Lines	23

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
dl	0
loc	53
rs	6.1349
c	0
b	0
f	0
cc	24
eloc	23
nc	9
nop	1

How to fix Long Method Complexity

<?php

namespace Digia\GraphQL\Language;

use Digia\GraphQL\Error\SyntaxErrorException;

class Lexer implements LexerInterface
{
    protected const ENCODING = 'UTF-8';

    /**
     * A map between punctuation character code and the corresponding token kind.
     *
     * @var array
     */
    protected static $codeTokenKindMap = [
        33  => TokenKindEnum::BANG,
        36  => TokenKindEnum::DOLLAR,
        38  => TokenKindEnum::AMP,
        40  => TokenKindEnum::PAREN_L,
        41  => TokenKindEnum::PAREN_R,
        58  => TokenKindEnum::COLON,
        61  => TokenKindEnum::EQUALS,
        64  => TokenKindEnum::AT,
        91  => TokenKindEnum::BRACKET_L,
        93  => TokenKindEnum::BRACKET_R,
        123 => TokenKindEnum::BRACE_L,
        124 => TokenKindEnum::PIPE,
        125 => TokenKindEnum::BRACE_R,
    ];

    /**
     * The source file for this lexer.
     *
     * @var Source
     */
    protected $source;

    /**
     * The contents of the source file.
     *
     * @var string
     */
    protected $body;

    /**
     * The total number of characters in the source file.
     *
     * @var int
     */
    protected $bodyLength;

    /**
     * The options for this lexer.
     *
     * @var array
     */
    protected $options = [];

    /**
     * The previously focused non-ignored token.
     *
     * @var Token
     */
    protected $lastToken;

    /**
     * The currently focused non-ignored token.
     *
     * @var Token
     */
    protected $token;

    /**
     * The current position.
     *
     * @var int
     */
    protected $pos;

    /**
     * The (1-indexed) line containing the current token.
     *
     * @var int
     */
    protected $line;

    /**
     * The character offset at which the current line begins.
     *
     * @var int
     */
    protected $lineStart;

    /**
     * @var array
     */
    protected static $charCodeCache = [];

    /**
     * Lexer constructor.
     * @param Source|null $source
     * @param array       $options
     */
    public function __construct(Source $source, array $options)
    {
        $startOfFileToken = $this->createStartOfFileToken();

        $this->lastToken  = $startOfFileToken;
        $this->token      = $startOfFileToken;
        $this->line       = 1;
        $this->lineStart  = 0;
        $this->body       = $source->getBody();
        $this->bodyLength = \strlen($this->body);
        $this->source     = $source;
        $this->options    = $options;
    }

    /**
     * @inheritdoc
     * @throws SyntaxErrorException
     */
    public function advance(): Token
    {
        $this->lastToken = $this->token;
        return $this->token = $this->lookahead();
    }

    /**
     * @inheritdoc
     * @throws SyntaxErrorException
     */
    public function lookahead(): Token
    {
        $token = $this->token;

        if (TokenKindEnum::EOF !== $token->kind) {
            do {
                $next  = $token->next = $this->readToken($token);
                $token = $next;
            } while (TokenKindEnum::COMMENT === $token->kind);
        }

        return $token;
    }

    /**
     * @inheritdoc
     */
    public function getOption(string $name, $default = null)
    {
        return $this->options[$name] ?? $default;
    }

    /**
     * @inheritdoc
     */
    public function getSource(): Source
    {
        return $this->source;
    }

    /**
     * @inheritdoc
     */
    public function getToken(): Token
    {
        return $this->token;
    }

    /**
     * @inheritdoc
     */
    public function getLastToken(): Token
    {
        return $this->lastToken;
    }

    /**
     * @inheritdoc
     */
    public function createSyntaxErrorException(?string $description = null): SyntaxErrorException
    {
        return new SyntaxErrorException(
            $this->source,
            $this->pos,
            $description ?? $this->unexpectedCharacterMessage($this->readCharCode($this->pos))
        );
    }

    /**
     * Reads the token after the given token.
     *
     * @param Token $prev
     * @return Token
     * @throws SyntaxErrorException
     */
    protected function readToken(Token $prev): Token
    {
        $this->pos = $prev->end;

        $this->skipWhitespace();

        $line = $this->line;
        $col  = (1 + $this->pos) - $this->lineStart;

        if ($this->pos >= $this->bodyLength) {
            return $this->createEndOfFileToken($line, $col, $prev);
        }

        $code = $this->readCharCode($this->pos);

        // Punctuation: [!$&:=@|()\[\]{}]{1}
        if (33 === $code || 36 === $code || 38 === $code || 58 === $code || 61 === $code || 64 === $code || 124 === $code ||
            40 === $code || 41 === $code || 91 === $code || 93 === $code || 123 === $code || 125 === $code) {
            return $this->lexPunctuation($code, $line, $col, $prev);
        }

        // Comment: #[\u0009\u0020-\uFFFF]*
        if (35 === $code) {
            return $this->lexComment($line, $col, $prev);
        }

        // Int:   -?(0|[1-9][0-9]*)
        // Float: -?(0|[1-9][0-9]*)(\.[0-9]+)?((E|e)(+|-)?[0-9]+)?
        if (45 === $code || isNumber($code)) {
            return $this->lexNumber($code, $line, $col, $prev);
        }

        // Name: [_A-Za-z][_0-9A-Za-z]*
        if (isAlphaNumeric($code)) {
            return $this->lexName($line, $col, $prev);
        }

        // Spread: ...
        if ($this->bodyLength >= 3 && $this->isSpread($code)) {
            return $this->lexSpread($line, $col, $prev);
        }

        // String: "([^"\\\u000A\u000D]|(\\(u[0-9a-fA-F]{4}|["\\/bfnrt])))*"
        if ($this->isString($code)) {
            return $this->lexString($line, $col, $prev);
        }

        // Block String: """("?"?(\\"""|\\(?!=""")|[^"\\]))*"""
        if ($this->bodyLength >= 3 && $this->isTripleQuote($code)) {
            return $this->lexBlockString($line, $col, $prev);
        }

        throw $this->createSyntaxErrorException();
    }

    /**
     * @return Token
     */
    protected function createStartOfFileToken(): Token
    {
        return new Token(TokenKindEnum::SOF);
    }

    /**
     * Creates an End Of File (EOF) token.
     *
     * @param int   $line
     * @param int   $col
     * @param Token $prev
     * @return Token
     */
    protected function createEndOfFileToken(int $line, int $col, Token $prev): Token
    {
        return new Token(TokenKindEnum::EOF, $this->bodyLength, $this->bodyLength, $line, $col, $prev);
    }

    /**
     * Reads a punctuation token from the source file.
     *
     * @param int   $code
     * @param int   $line
     * @param int   $col
     * @param Token $prev
     * @return Token
     * @throws SyntaxErrorException
     */
    protected function lexPunctuation(int $code, int $line, int $col, Token $prev): ?Token
    {
        if (!isset(self::$codeTokenKindMap[$code])) {
            throw $this->createSyntaxErrorException();
        }

        return new Token(self::$codeTokenKindMap[$code], $this->pos, $this->pos + 1, $line, $col, $prev);
    }

    /**
     * Reads a name token from the source file.
     *
     * @param int   $line
     * @param int   $col
     * @param Token $prev
     * @return Token
     */
    protected function lexName(int $line, int $col, Token $prev): Token
    {
        $start = $this->pos;

        ++$this->pos;

        while ($this->pos !== $this->bodyLength &&
            ($code = $this->readCharCode($this->pos)) !== null &&
            isAlphaNumeric($code)) {
            ++$this->pos;
        }

        $value = sliceString($this->body, $start, $this->pos);

        return new Token(TokenKindEnum::NAME, $start, $this->pos, $line, $col, $prev, $value);
    }

    /**
     * Reads a number (int or float) token from the source file.
     *
     * @param int   $code
     * @param int   $line
     * @param int   $col
     * @param Token $prev
     * @return Token
     * @throws SyntaxErrorException
     */
    protected function lexNumber(int $code, int $line, int $col, Token $prev): Token
    {
        $start   = $this->pos;
        $isFloat = false;

        if (45 === $code) {
            // -
            $code = $this->readCharCode(++$this->pos);
        }

        if (48 === $code) {
            // 0
            $code = $this->readCharCode(++$this->pos);

            if (isNumber($code)) {
                throw $this->createSyntaxErrorException(
                    \sprintf('Invalid number, unexpected digit after 0: %s.', printCharCode($code))
                );
            }
        } else {
            $this->skipDigits($code);
            $code = $this->readCharCode($this->pos);
        }

        if (46 === $code) {
            // .
            $isFloat = true;

            $code = $this->readCharCode(++$this->pos);
            $this->skipDigits($code);
            $code = $this->readCharCode($this->pos);
        }

        if (69 === $code || 101 === $code) {
            // e or E
            $isFloat = true;

            $code = $this->readCharCode(++$this->pos);

            if (43 === $code || 45 === $code) {
                // + or -
                $code = $this->readCharCode(++$this->pos);
            }

            $this->skipDigits($code);
        }

        return new Token(
            $isFloat ? TokenKindEnum::FLOAT : TokenKindEnum::INT,
            $start,
            $this->pos,
            $line,
            $col,
            $prev,
            sliceString($this->body, $start, $this->pos)
        );
    }

    /**
     * Skips digits at the current position.
     *
     * @param int $code
     * @throws SyntaxErrorException
     */
    protected function skipDigits(int $code): void
    {
        if (isNumber($code)) {
            do {
                $code = $this->readCharCode(++$this->pos);
            } while (isNumber($code));

            return;
        }

        throw $this->createSyntaxErrorException(
            \sprintf('Invalid number, expected digit but got: %s.', printCharCode($code))
        );
    }

    /**
     * Reads a comment token from the source file.
     *
     * @param int   $line
     * @param int   $col
     * @param Token $prev
     * @return Token
     */
    protected function lexComment(int $line, int $col, Token $prev): Token
    {
        $start = $this->pos;

        do {
            $code = $this->readCharCode(++$this->pos);
        } while ($code !== null && ($code > 0x001f || 0x0009 === $code)); // SourceCharacter but not LineTerminator

        return new Token(
            TokenKindEnum::COMMENT,
            $start,
            $this->pos,
            $line,
            $col,
            $prev,
            sliceString($this->body, $start + 1, $this->pos)
        );
    }

    /**
     * Reads a spread token from the source.
     *
     * @param int   $line
     * @param int   $col
     * @param Token $prev
     * @return Token
     */
    protected function lexSpread(int $line, int $col, Token $prev): Token
    {
        return new Token(TokenKindEnum::SPREAD, $this->pos, $this->pos + 3, $line, $col, $prev);
    }

    /**
     * Reads a string token from the source.
     *
     * @param int   $line
     * @param int   $col
     * @param Token $prev
     * @return Token
     * @throws SyntaxErrorException
     */
    protected function lexString(int $line, int $col, Token $prev): Token
    {
        $start      = $this->pos;
        $chunkStart = ++$this->pos; // skip the quote
        $value      = '';

        while ($this->pos < $this->bodyLength &&
            ($code = $this->readCharCode($this->pos)) !== null && !isLineTerminator($code)) {
            // Closing Quote (")
            if (34 === $code) {
                $value .= sliceString($this->body, $chunkStart, $this->pos);
                return new Token(TokenKindEnum::STRING, $start, $this->pos + 1, $line, $col, $prev, $value);
            }

            if (isSourceCharacter($code)) {
                throw $this->createSyntaxErrorException(
                    \sprintf('Invalid character within String: %s.', printCharCode($code))
                );
            }

            ++$this->pos;

            if (92 === $code) {
                // \
                $value .= sliceString($this->body, $chunkStart, $this->pos - 1);

                $code = $this->readCharCode($this->pos);

                switch ($code) {
                    case 34: // "
                        $value .= '"';
                        break;
                    case 47: // /
                        $value .= '/';
                        break;
                    case 92: // \
                        $value .= '\\';
                        break;
                    case 98: // b
                        $value .= '\b';
                        break;
                    case 102: // f
                        $value .= '\f';
                        break;
                    case 110: // n
                        $value .= '\n';
                        break;
                    case 114: // r
                        $value .= '\r';
                        break;
                    case 116: // t
                        $value .= '\t';
                        break;
                    case 117: // u
                        $unicodeString = sliceString($this->body, $this->pos + 1, $this->pos + 5);

                        if (!\preg_match('/[0-9A-Fa-f]{4}/', $unicodeString)) {
                            throw $this->createSyntaxErrorException(
                                \sprintf('Invalid character escape sequence: \\u%s.', $unicodeString)
                            );
                        }

                        $value     .= '\\u' . $unicodeString;
                        $this->pos += 4;

                        break;
                    default:
                        throw $this->createSyntaxErrorException(
                            \sprintf('Invalid character escape sequence: \\%s.', \chr($code))
                        );
                }

                ++$this->pos;

                $chunkStart = $this->pos;
            }
        }

        throw $this->createSyntaxErrorException('Unterminated string.');
    }

    /**
     * Reads a block string token from the source file.
     *
     * @param int   $line
     * @param int   $col
     * @param Token $prev
     * @return Token
     * @throws SyntaxErrorException
     */
    protected function lexBlockString(int $line, int $col, Token $prev): Token
    {
        $start      = $this->pos;
        $this->pos  = $start + 3; // skip the triple-quote
        $chunkStart = $this->pos;
        $rawValue   = '';

        while ($this->pos < $this->bodyLength && ($code = $this->readCharCode($this->pos)) !== null) {
            // Closing Triple-Quote (""")
            if ($this->isTripleQuote($code)) {
                $rawValue .= sliceString($this->body, $chunkStart, $this->pos);
                return new Token(
                    TokenKindEnum::BLOCK_STRING,
                    $start,
                    $this->pos + 3,
                    $line,
                    $col,
                    $prev,
                    blockStringValue($rawValue)
                );
            }

            if (isSourceCharacter($code) && !isLineTerminator($code)) {
                throw $this->createSyntaxErrorException(
                    \sprintf('Invalid character within String: %s.', printCharCode($code))
                );
            }

            if ($this->isEscapedTripleQuote($code)) {
                $rawValue   .= sliceString($this->body, $chunkStart, $this->pos) . '"""';
                $this->pos  += 4;
                $chunkStart = $this->pos;
            } else {
                ++$this->pos;
            }
        }

        throw $this->createSyntaxErrorException('Unterminated string.');
    }

    /**
     * Skips whitespace at the current position.
     */
    protected function skipWhitespace(): void
    {
        while ($this->pos < $this->bodyLength) {
            $code = $this->readCharCode($this->pos);

            if (9 === $code || 32 === $code || 44 === $code || 0xfeff === $code) {
                // tab | space | comma | BOM
                ++$this->pos;
            } elseif (10 === $code) {
                // new line (\n)
                ++$this->pos;
                ++$this->line;
                $this->lineStart = $this->pos;
            } elseif (13 === $code) {
                // carriage return (\r)
                if (10 === $this->readCharCode($this->pos + 1)) {
                    // carriage return and new line (\r\n)
                    $this->pos += 2;
                } else {
                    ++$this->pos;
                }
                ++$this->line;
                $this->lineStart = $this->pos;
            } else {
                break;
            }
        }
    }

    /**
     * @param int $pos
     * @return int
     */
    protected function readCharCode(int $pos): int
    {
        $char = \mb_substr($this->body, $pos, 1, self::ENCODING);

        if ('' === $char) {
            return 0;
        }

        if (!isset(self::$charCodeCache[$char])) {
            $code = \ord($char);

            if ($code >= 128) {
                $code = \mb_ord($char, self::ENCODING);
            }

            self::$charCodeCache[$char] = $code;
        }

        return self::$charCodeCache[$char];
    }

    /**
     * Report a message that an unexpected character was encountered.
     *
     * @param int $code
     * @return string
     */
    protected function unexpectedCharacterMessage(int $code): string
    {
        if (isSourceCharacter($code) && !isLineTerminator($code)) {
            return \sprintf('Cannot contain the invalid character %s.', printCharCode($code));
        }

        if ($code === 39) {
            // '
            return 'Unexpected single quote character (\'), did you mean to use a double quote (")?';
        }

        return \sprintf('Cannot parse the unexpected character %s.', printCharCode($code));
    }

    /**
     * @param int $code
     * @return bool
     */
    protected function isSpread(int $code): bool
    {
        return 46 === $code &&
            $this->readCharCode($this->pos + 1) === 46 &&
            $this->readCharCode($this->pos + 2) === 46; // ...
    }

    /**
     * @param int $code
     * @return bool
     */
    protected function isString(int $code): bool
    {
        return 34 === $code && $this->readCharCode($this->pos + 1) !== 34;
    }

    /**
     * @param int $code
     * @return bool
     */
    protected function isTripleQuote(int $code): bool
    {
        return 34 === $code &&
            34 === $this->readCharCode($this->pos + 1) &&
            34 === $this->readCharCode($this->pos + 2); // """
    }

    /**
     * @param int $code
     * @return bool
     */
    protected function isEscapedTripleQuote(int $code): bool
    {
        return $code === 92 &&
            34 === $this->readCharCode($this->pos + 1) &&
            34 === $this->readCharCode($this->pos + 2) &&
            34 === $this->readCharCode($this->pos + 3); // \"""
    }
}


1			<?php
2
3			namespace Digia\GraphQL\Language;
4
5			use Digia\GraphQL\Error\SyntaxErrorException;
6
7			class Lexer implements LexerInterface
8			{
9			protected const ENCODING = 'UTF-8';
10
11			/**
12			* A map between punctuation character code and the corresponding token kind.
13			*
14			* @var array
15			*/
16			protected static $codeTokenKindMap = [
17			33 => TokenKindEnum::BANG,
18			36 => TokenKindEnum::DOLLAR,
19			38 => TokenKindEnum::AMP,
20			40 => TokenKindEnum::PAREN_L,
21			41 => TokenKindEnum::PAREN_R,
22			58 => TokenKindEnum::COLON,
23			61 => TokenKindEnum::EQUALS,
24			64 => TokenKindEnum::AT,
25			91 => TokenKindEnum::BRACKET_L,
26			93 => TokenKindEnum::BRACKET_R,
27			123 => TokenKindEnum::BRACE_L,
28			124 => TokenKindEnum::PIPE,
29			125 => TokenKindEnum::BRACE_R,
30			];
31
32			/**
33			* The source file for this lexer.
34			*
35			* @var Source
36			*/
37			protected $source;
38
39			/**
40			* The contents of the source file.
41			*
42			* @var string
43			*/
44			protected $body;
45
46			/**
47			* The total number of characters in the source file.
48			*
49			* @var int
50			*/
51			protected $bodyLength;
52
53			/**
54			* The options for this lexer.
55			*
56			* @var array
57			*/
58			protected $options = [];
59
60			/**
61			* The previously focused non-ignored token.
62			*
63			* @var Token
64			*/
65			protected $lastToken;
66
67			/**
68			* The currently focused non-ignored token.
69			*
70			* @var Token
71			*/
72			protected $token;
73
74			/**
75			* The current position.
76			*
77			* @var int
78			*/
79			protected $pos;
80
81			/**
82			* The (1-indexed) line containing the current token.
83			*
84			* @var int
85			*/
86			protected $line;
87
88			/**
89			* The character offset at which the current line begins.
90			*
91			* @var int
92			*/
93			protected $lineStart;
94
95			/**
96			* @var array
97			*/
98			protected static $charCodeCache = [];
99
100			/**
101			* Lexer constructor.
102			* @param Source\|null $source
103			* @param array $options
104			*/
105			public function __construct(Source $source, array $options)
106			{
107			$startOfFileToken = $this->createStartOfFileToken();
108
109			$this->lastToken = $startOfFileToken;
110			$this->token = $startOfFileToken;
111			$this->line = 1;
112			$this->lineStart = 0;
113			$this->body = $source->getBody();
114			$this->bodyLength = \strlen($this->body);
115			$this->source = $source;
116			$this->options = $options;
117			}
118
119			/**
120			* @inheritdoc
121			* @throws SyntaxErrorException
122			*/
123			public function advance(): Token
124			{
125			$this->lastToken = $this->token;
126			return $this->token = $this->lookahead();
127			}
128
129			/**
130			* @inheritdoc
131			* @throws SyntaxErrorException
132			*/
133			public function lookahead(): Token
134			{
135			$token = $this->token;
136
137			if (TokenKindEnum::EOF !== $token->kind) {
138			do {
139			$next = $token->next = $this->readToken($token);
140			$token = $next;
141			} while (TokenKindEnum::COMMENT === $token->kind);
142			}
143
144			return $token;
145			}
146
147			/**
148			* @inheritdoc
149			*/
150			public function getOption(string $name, $default = null)
151			{
152			return $this->options[$name] ?? $default;
153			}
154
155			/**
156			* @inheritdoc
157			*/
158			public function getSource(): Source
159			{
160			return $this->source;
161			}
162
163			/**
164			* @inheritdoc
165			*/
166			public function getToken(): Token
167			{
168			return $this->token;
169			}
170
171			/**
172			* @inheritdoc
173			*/
174			public function getLastToken(): Token
175			{
176			return $this->lastToken;
177			}
178
179			/**
180			* @inheritdoc
181			*/
182			public function createSyntaxErrorException(?string $description = null): SyntaxErrorException
183			{
184			return new SyntaxErrorException(
185			$this->source,
186			$this->pos,
187			$description ?? $this->unexpectedCharacterMessage($this->readCharCode($this->pos))
188			);
189			}
190
191			/**
192			* Reads the token after the given token.
193			*
194			* @param Token $prev
195			* @return Token
196			* @throws SyntaxErrorException
197			*/
198			protected function readToken(Token $prev): Token
199			{
200			$this->pos = $prev->end;
201
202			$this->skipWhitespace();
203
204			$line = $this->line;
205			$col = (1 + $this->pos) - $this->lineStart;
206
207			if ($this->pos >= $this->bodyLength) {
208			return $this->createEndOfFileToken($line, $col, $prev);
209			}
210
211			$code = $this->readCharCode($this->pos);
212
213			// Punctuation: [!$&:=@\|()\[\]{}]{1}
214			if (33 === $code \|\| 36 === $code \|\| 38 === $code \|\| 58 === $code \|\| 61 === $code \|\| 64 === $code \|\| 124 === $code \|\|
215			40 === $code \|\| 41 === $code \|\| 91 === $code \|\| 93 === $code \|\| 123 === $code \|\| 125 === $code) {
216			return $this->lexPunctuation($code, $line, $col, $prev);
217			}
218
219			// Comment: #[\u0009\u0020-\uFFFF]*
220			if (35 === $code) {
221			return $this->lexComment($line, $col, $prev);
222			}
223
224			// Int: -?(0\|[1-9][0-9]*)
225			// Float: -?(0\|[1-9][0-9]*)(\.[0-9]+)?((E\|e)(+\|-)?[0-9]+)?
226			if (45 === $code \|\| isNumber($code)) {
227			return $this->lexNumber($code, $line, $col, $prev);
228			}
229
230			// Name: [_A-Za-z][_0-9A-Za-z]*
231			if (isAlphaNumeric($code)) {
232			return $this->lexName($line, $col, $prev);
233			}
234
235			// Spread: ...
236			if ($this->bodyLength >= 3 && $this->isSpread($code)) {
237			return $this->lexSpread($line, $col, $prev);
238			}
239
240			// String: "([^"\\\u000A\u000D]\|(\\(u[0-9a-fA-F]{4}\|["\\/bfnrt])))*"
241			if ($this->isString($code)) {
242			return $this->lexString($line, $col, $prev);
243			}
244
245			// Block String: """("?"?(\\"""\|\\(?!=""")\|[^"\\]))*"""
246			if ($this->bodyLength >= 3 && $this->isTripleQuote($code)) {
247			return $this->lexBlockString($line, $col, $prev);
248			}
249
250			throw $this->createSyntaxErrorException();
251			}
252
253			/**
254			* @return Token
255			*/
256			protected function createStartOfFileToken(): Token
257			{
258			return new Token(TokenKindEnum::SOF);
259			}
260
261			/**
262			* Creates an End Of File (EOF) token.
263			*
264			* @param int $line
265			* @param int $col
266			* @param Token $prev
267			* @return Token
268			*/
269			protected function createEndOfFileToken(int $line, int $col, Token $prev): Token
270			{
271			return new Token(TokenKindEnum::EOF, $this->bodyLength, $this->bodyLength, $line, $col, $prev);
272			}
273
274			/**
275			* Reads a punctuation token from the source file.
276			*
277			* @param int $code
278			* @param int $line
279			* @param int $col
280			* @param Token $prev
281			* @return Token
282			* @throws SyntaxErrorException
283			*/
284			protected function lexPunctuation(int $code, int $line, int $col, Token $prev): ?Token
285			{
286			if (!isset(self::$codeTokenKindMap[$code])) {
287			throw $this->createSyntaxErrorException();
288			}
289
290			return new Token(self::$codeTokenKindMap[$code], $this->pos, $this->pos + 1, $line, $col, $prev);
291			}
292
293			/**
294			* Reads a name token from the source file.
295			*
296			* @param int $line
297			* @param int $col
298			* @param Token $prev
299			* @return Token
300			*/
301			protected function lexName(int $line, int $col, Token $prev): Token
302			{
303			$start = $this->pos;
304
305			++$this->pos;
306
307			while ($this->pos !== $this->bodyLength &&
308			($code = $this->readCharCode($this->pos)) !== null &&
309			isAlphaNumeric($code)) {
310			++$this->pos;
311			}
312
313			$value = sliceString($this->body, $start, $this->pos);
314
315			return new Token(TokenKindEnum::NAME, $start, $this->pos, $line, $col, $prev, $value);
316			}
317
318			/**
319			* Reads a number (int or float) token from the source file.
320			*
321			* @param int $code
322			* @param int $line
323			* @param int $col
324			* @param Token $prev
325			* @return Token
326			* @throws SyntaxErrorException
327			*/
328			protected function lexNumber(int $code, int $line, int $col, Token $prev): Token
329			{
330			$start = $this->pos;
331			$isFloat = false;
332
333			if (45 === $code) {
334			// -
335			$code = $this->readCharCode(++$this->pos);
336			}
337
338			if (48 === $code) {
339			// 0
340			$code = $this->readCharCode(++$this->pos);
341
342			if (isNumber($code)) {
343			throw $this->createSyntaxErrorException(
344			\sprintf('Invalid number, unexpected digit after 0: %s.', printCharCode($code))
345			);
346			}
347			} else {
348			$this->skipDigits($code);
349			$code = $this->readCharCode($this->pos);
350			}
351
352			if (46 === $code) {
353			// .
354			$isFloat = true;
355
356			$code = $this->readCharCode(++$this->pos);
357			$this->skipDigits($code);
358			$code = $this->readCharCode($this->pos);
359			}
360
361			if (69 === $code \|\| 101 === $code) {
362			// e or E
363			$isFloat = true;
364
365			$code = $this->readCharCode(++$this->pos);
366
367			if (43 === $code \|\| 45 === $code) {
368			// + or -
369			$code = $this->readCharCode(++$this->pos);
370			}
371
372			$this->skipDigits($code);
373			}
374
375			return new Token(
376			$isFloat ? TokenKindEnum::FLOAT : TokenKindEnum::INT,
377			$start,
378			$this->pos,
379			$line,
380			$col,
381			$prev,
382			sliceString($this->body, $start, $this->pos)
383			);
384			}
385
386			/**
387			* Skips digits at the current position.
388			*
389			* @param int $code
390			* @throws SyntaxErrorException
391			*/
392			protected function skipDigits(int $code): void
393			{
394			if (isNumber($code)) {
395			do {
396			$code = $this->readCharCode(++$this->pos);
397			} while (isNumber($code));
398
399			return;
400			}
401
402			throw $this->createSyntaxErrorException(
403			\sprintf('Invalid number, expected digit but got: %s.', printCharCode($code))
404			);
405			}
406
407			/**
408			* Reads a comment token from the source file.
409			*
410			* @param int $line
411			* @param int $col
412			* @param Token $prev
413			* @return Token
414			*/
415			protected function lexComment(int $line, int $col, Token $prev): Token
416			{
417			$start = $this->pos;
418
419			do {
420			$code = $this->readCharCode(++$this->pos);
421			} while ($code !== null && ($code > 0x001f \|\| 0x0009 === $code)); // SourceCharacter but not LineTerminator
422
423			return new Token(
424			TokenKindEnum::COMMENT,
425			$start,
426			$this->pos,
427			$line,
428			$col,
429			$prev,
430			sliceString($this->body, $start + 1, $this->pos)
431			);
432			}
433
434			/**
435			* Reads a spread token from the source.
436			*
437			* @param int $line
438			* @param int $col
439			* @param Token $prev
440			* @return Token
441			*/
442			protected function lexSpread(int $line, int $col, Token $prev): Token
443			{
444			return new Token(TokenKindEnum::SPREAD, $this->pos, $this->pos + 3, $line, $col, $prev);
445			}
446
447			/**
448			* Reads a string token from the source.
449			*
450			* @param int $line
451			* @param int $col
452			* @param Token $prev
453			* @return Token
454			* @throws SyntaxErrorException
455			*/
456			protected function lexString(int $line, int $col, Token $prev): Token
457			{
458			$start = $this->pos;
459			$chunkStart = ++$this->pos; // skip the quote
460			$value = '';
461
462			while ($this->pos < $this->bodyLength &&
463			($code = $this->readCharCode($this->pos)) !== null && !isLineTerminator($code)) {
464			// Closing Quote (")
465			if (34 === $code) {
466			$value .= sliceString($this->body, $chunkStart, $this->pos);
467			return new Token(TokenKindEnum::STRING, $start, $this->pos + 1, $line, $col, $prev, $value);
468			}
469
470			if (isSourceCharacter($code)) {
471			throw $this->createSyntaxErrorException(
472			\sprintf('Invalid character within String: %s.', printCharCode($code))
473			);
474			}
475
476			++$this->pos;
477
478			if (92 === $code) {
479			// \
480			$value .= sliceString($this->body, $chunkStart, $this->pos - 1);
481
482			$code = $this->readCharCode($this->pos);
483
484			switch ($code) {
485			case 34: // "
486			$value .= '"';
487			break;
488			case 47: // /
489			$value .= '/';
490			break;
491			case 92: // \
492			$value .= '\\';
493			break;
494			case 98: // b
495			$value .= '\b';
496			break;
497			case 102: // f
498			$value .= '\f';
499			break;
500			case 110: // n
501			$value .= '\n';
502			break;
503			case 114: // r
504			$value .= '\r';
505			break;
506			case 116: // t
507			$value .= '\t';
508			break;
509			case 117: // u
510			$unicodeString = sliceString($this->body, $this->pos + 1, $this->pos + 5);
511
512			if (!\preg_match('/[0-9A-Fa-f]{4}/', $unicodeString)) {
513			throw $this->createSyntaxErrorException(
514			\sprintf('Invalid character escape sequence: \\u%s.', $unicodeString)
515			);
516			}
517
518			$value .= '\\u' . $unicodeString;
519			$this->pos += 4;
520
521			break;
522			default:
523			throw $this->createSyntaxErrorException(
524			\sprintf('Invalid character escape sequence: \\%s.', \chr($code))
525			);
526			}
527
528			++$this->pos;
529
530			$chunkStart = $this->pos;
531			}
532			}
533
534			throw $this->createSyntaxErrorException('Unterminated string.');
535			}
536
537			/**
538			* Reads a block string token from the source file.
539			*
540			* @param int $line
541			* @param int $col
542			* @param Token $prev
543			* @return Token
544			* @throws SyntaxErrorException
545			*/
546			protected function lexBlockString(int $line, int $col, Token $prev): Token
547			{
548			$start = $this->pos;
549			$this->pos = $start + 3; // skip the triple-quote
550			$chunkStart = $this->pos;
551			$rawValue = '';
552
553			while ($this->pos < $this->bodyLength && ($code = $this->readCharCode($this->pos)) !== null) {
554			// Closing Triple-Quote (""")
555			if ($this->isTripleQuote($code)) {
556			$rawValue .= sliceString($this->body, $chunkStart, $this->pos);
557			return new Token(
558			TokenKindEnum::BLOCK_STRING,
559			$start,
560			$this->pos + 3,
561			$line,
562			$col,
563			$prev,
564			blockStringValue($rawValue)
565			);
566			}
567
568			if (isSourceCharacter($code) && !isLineTerminator($code)) {
569			throw $this->createSyntaxErrorException(
570			\sprintf('Invalid character within String: %s.', printCharCode($code))
571			);
572			}
573
574			if ($this->isEscapedTripleQuote($code)) {
575			$rawValue .= sliceString($this->body, $chunkStart, $this->pos) . '"""';
576			$this->pos += 4;
577			$chunkStart = $this->pos;
578			} else {
579			++$this->pos;
580			}
581			}
582
583			throw $this->createSyntaxErrorException('Unterminated string.');
584			}
585
586			/**
587			* Skips whitespace at the current position.
588			*/
589			protected function skipWhitespace(): void
590			{
591			while ($this->pos < $this->bodyLength) {
592			$code = $this->readCharCode($this->pos);
593
594			if (9 === $code \|\| 32 === $code \|\| 44 === $code \|\| 0xfeff === $code) {
595			// tab \| space \| comma \| BOM
596			++$this->pos;
597			} elseif (10 === $code) {
598			// new line (\n)
599			++$this->pos;
600			++$this->line;
601			$this->lineStart = $this->pos;
602			} elseif (13 === $code) {
603			// carriage return (\r)
604			if (10 === $this->readCharCode($this->pos + 1)) {
605			// carriage return and new line (\r\n)
606			$this->pos += 2;
607			} else {
608			++$this->pos;
609			}
610			++$this->line;
611			$this->lineStart = $this->pos;
612			} else {
613			break;
614			}
615			}
616			}
617
618			/**
619			* @param int $pos
620			* @return int
621			*/
622			protected function readCharCode(int $pos): int
623			{
624			$char = \mb_substr($this->body, $pos, 1, self::ENCODING);
625
626			if ('' === $char) {
627			return 0;
628			}
629
630			if (!isset(self::$charCodeCache[$char])) {
631			$code = \ord($char);
632
633			if ($code >= 128) {
634			$code = \mb_ord($char, self::ENCODING);
635			}
636
637			self::$charCodeCache[$char] = $code;
638			}
639
640			return self::$charCodeCache[$char];
641			}
642
643			/**
644			* Report a message that an unexpected character was encountered.
645			*
646			* @param int $code
647			* @return string
648			*/
649			protected function unexpectedCharacterMessage(int $code): string
650			{
651			if (isSourceCharacter($code) && !isLineTerminator($code)) {
652			return \sprintf('Cannot contain the invalid character %s.', printCharCode($code));
653			}
654
655			if ($code === 39) {
656			// '
657			return 'Unexpected single quote character (\'), did you mean to use a double quote (")?';
658			}
659
660			return \sprintf('Cannot parse the unexpected character %s.', printCharCode($code));
661			}
662
663			/**
664			* @param int $code
665			* @return bool
666			*/
667			protected function isSpread(int $code): bool
668			{
669			return 46 === $code &&
670			$this->readCharCode($this->pos + 1) === 46 &&
671			$this->readCharCode($this->pos + 2) === 46; // ...
672			}
673
674			/**
675			* @param int $code
676			* @return bool
677			*/
678			protected function isString(int $code): bool
679			{
680			return 34 === $code && $this->readCharCode($this->pos + 1) !== 34;
681			}
682
683			/**
684			* @param int $code
685			* @return bool
686			*/
687			protected function isTripleQuote(int $code): bool
688			{
689			return 34 === $code &&
690			34 === $this->readCharCode($this->pos + 1) &&
691			34 === $this->readCharCode($this->pos + 2); // """
692			}
693
694			/**
695			* @param int $code
696			* @return bool
697			*/
698			protected function isEscapedTripleQuote(int $code): bool
699			{
700			return $code === 92 &&
701			34 === $this->readCharCode($this->pos + 1) &&
702			34 === $this->readCharCode($this->pos + 2) &&
703			34 === $this->readCharCode($this->pos + 3); // \"""
704			}
705			}
706

digiaonline / graphql-php

Pull Request — master (#196)

Lexer::readToken() C

Complexity

Size

Duplication

Importance

How to fix Long Method Complexity

Long Method

Duplication Side-by-Side

Filter issues like