Emulative::preprocessCode() - Code Metrics - nikic/PHP-Parser - Measure and Improve Code Quality continuously with Scrutinizer

Emulative::preprocessCode() A
last analyzed 2016-02-20 20:57 UTC

↳ Parent: Emulative

Complexity

Conditions	3
Paths	3

Size

Total Lines	22
Code Lines	14

Duplication

Lines	0
Ratio	0 %

Importance

Changes	4
Bugs	0	Features	3

Metric	Value
c	4
b	0
f	3
dl	0
loc	22
rs	9.2
cc	3
eloc	14
nc	3
nop	1

<?php

namespace PhpParser\Lexer;

use PhpParser\Parser\Tokens;

/**
 * ATTENTION: This code is WRITE-ONLY. Do not try to read it.
 */
class Emulative extends \PhpParser\Lexer
{
    protected $newKeywords;
    protected $inObjectAccess;

    const T_ELLIPSIS   = 1001;
    const T_POW        = 1002;
    const T_POW_EQUAL  = 1003;
    const T_COALESCE   = 1004;
    const T_SPACESHIP  = 1005;
    const T_YIELD_FROM = 1006;

    const PHP_7_0 = '7.0.0dev';
    const PHP_5_6 = '5.6.0rc1';
    const PHP_5_5 = '5.5.0beta1';

    public function __construct(array $options = array()) {
        parent::__construct($options);

        $newKeywordsPerVersion = array(
            self::PHP_5_5 => array(
                'finally'       => Tokens::T_FINALLY,
                'yield'         => Tokens::T_YIELD,
            ),
        );

        $this->newKeywords = array();
        foreach ($newKeywordsPerVersion as $version => $newKeywords) {
            if (version_compare(PHP_VERSION, $version, '>=')) {
                break;
            }

            $this->newKeywords += $newKeywords;
        }

        if (version_compare(PHP_VERSION, self::PHP_7_0, '>=')) {
            return;
        }
        $this->tokenMap[self::T_COALESCE]   = Tokens::T_COALESCE;
        $this->tokenMap[self::T_SPACESHIP]  = Tokens::T_SPACESHIP;
        $this->tokenMap[self::T_YIELD_FROM] = Tokens::T_YIELD_FROM;

        if (version_compare(PHP_VERSION, self::PHP_5_6, '>=')) {
            return;
        }
        $this->tokenMap[self::T_ELLIPSIS]  = Tokens::T_ELLIPSIS;
        $this->tokenMap[self::T_POW]       = Tokens::T_POW;
        $this->tokenMap[self::T_POW_EQUAL] = Tokens::T_POW_EQUAL;
    }

    public function startLexing($code) {
        $this->inObjectAccess = false;

        $preprocessedCode = $this->preprocessCode($code);
        parent::startLexing($preprocessedCode);
        if ($preprocessedCode !== $code) {
            $this->postprocessTokens();
        }

        // Set code property back to the original code, so __halt_compiler()
        // handling and (start|end)FilePos attributes use the correct offsets
        $this->code = $code;
    }

    /*
     * Replaces new features in the code by ~__EMU__{NAME}__{DATA}__~ sequences.
     * ~LABEL~ is never valid PHP code, that's why we can (to some degree) safely
     * use it here.
     * Later when preprocessing the tokens these sequences will either be replaced
     * by real tokens or replaced with their original content (e.g. if they occurred
     * inside a string, i.e. a place where they don't have a special meaning).
     */
    protected function preprocessCode($code) {
        if (version_compare(PHP_VERSION, self::PHP_7_0, '>=')) {
            return $code;
        }

        $code = str_replace('??', '~__EMU__COALESCE__~', $code);
        $code = str_replace('<=>', '~__EMU__SPACESHIP__~', $code);
        $code = preg_replace_callback('(yield[ \n\r\t]+from)', function($matches) {
            // Encoding $0 in order to preserve exact whitespace
            return '~__EMU__YIELDFROM__' . bin2hex($matches[0]) . '__~';
        }, $code);

        if (version_compare(PHP_VERSION, self::PHP_5_6, '>=')) {
            return $code;
        }

        $code = str_replace('...', '~__EMU__ELLIPSIS__~', $code);
        $code = preg_replace('((?<!/)\*\*=)', '~__EMU__POWEQUAL__~', $code);
        $code = preg_replace('((?<!/)\*\*(?!/))', '~__EMU__POW__~', $code);

        return $code;
    }

    /*
     * Replaces the ~__EMU__...~ sequences with real tokens or their original
     * value.
     */
    protected function postprocessTokens() {
        // we need to manually iterate and manage a count because we'll change
        // the tokens array on the way
        for ($i = 0, $c = count($this->tokens); $i < $c; ++$i) {
            // first check that the following tokens are of form ~LABEL~,
            // then match the __EMU__... sequence.
            if ('~' === $this->tokens[$i]
                && isset($this->tokens[$i + 2])
                && '~' === $this->tokens[$i + 2]
                && T_STRING === $this->tokens[$i + 1][0]
                && preg_match('(^__EMU__([A-Z]++)__(?:([A-Za-z0-9]++)__)?$)', $this->tokens[$i + 1][1], $matches)
            ) {
                if ('ELLIPSIS' === $matches[1]) {
                    $replace = array(
                        array(self::T_ELLIPSIS, '...', $this->tokens[$i + 1][2])
                    );
                } else if ('POW' === $matches[1]) {
                    $replace = array(
                        array(self::T_POW, '**', $this->tokens[$i + 1][2])
                    );
                } else if ('POWEQUAL' === $matches[1]) {
                    $replace = array(
                        array(self::T_POW_EQUAL, '**=', $this->tokens[$i + 1][2])
                    );
                } else if ('COALESCE' === $matches[1]) {
                    $replace = array(
                        array(self::T_COALESCE, '??', $this->tokens[$i + 1][2])
                    );
                } else if ('SPACESHIP' === $matches[1]) {
                    $replace = array(
                        array(self::T_SPACESHIP, '<=>', $this->tokens[$i + 1][2]),
                    );
                } else if ('YIELDFROM' === $matches[1]) {
                    $content = hex2bin($matches[2]);
                    $replace = array(
                        array(self::T_YIELD_FROM, $content, $this->tokens[$i + 1][2] - substr_count($content, "\n"))
                    );
                } else {
                    throw new \RuntimeException('Invalid __EMU__ sequence');
                }

                array_splice($this->tokens, $i, 3, $replace);
                $c -= 3 - count($replace);
            // for multichar tokens (e.g. strings) replace any ~__EMU__...~ sequences
            // in their content with the original character sequence
            } elseif (is_array($this->tokens[$i])
                      && 0 !== strpos($this->tokens[$i][1], '__EMU__')
            ) {
                $this->tokens[$i][1] = preg_replace_callback(
                    '(~__EMU__([A-Z]++)__(?:([A-Za-z0-9]++)__)?~)',
                    array($this, 'restoreContentCallback'),
                    $this->tokens[$i][1]
                );
            }
        }
    }

    /*
     * This method is a callback for restoring EMU sequences in
     * multichar tokens (like strings) to their original value.
     */
    public function restoreContentCallback(array $matches) {
        if ('ELLIPSIS' === $matches[1]) {
            return '...';
        } else if ('POW' === $matches[1]) {
            return '**';
        } else if ('POWEQUAL' === $matches[1]) {
            return '**=';
        } else if ('COALESCE' === $matches[1]) {
            return '??';
        } else if ('SPACESHIP' === $matches[1]) {
            return '<=>';
        } else if ('YIELDFROM' === $matches[1]) {
            return hex2bin($matches[2]);
        } else {
            return $matches[0];
        }
    }

    public function getNextToken(&$value = null, &$startAttributes = null, &$endAttributes = null) {
        $token = parent::getNextToken($value, $startAttributes, $endAttributes);

        // replace new keywords by their respective tokens. This is not done
        // if we currently are in an object access (e.g. in $obj->namespace
        // "namespace" stays a T_STRING tokens and isn't converted to T_NAMESPACE)
        if (Tokens::T_STRING === $token && !$this->inObjectAccess) {
            if (isset($this->newKeywords[strtolower($value)])) {
                return $this->newKeywords[strtolower($value)];
            }
        } else {
            // keep track of whether we currently are in an object access (after ->)
            $this->inObjectAccess = Tokens::T_OBJECT_OPERATOR === $token;
        }

        return $token;
    }
}


1			<?php
2
3			namespace PhpParser\Lexer;
4
5			use PhpParser\Parser\Tokens;
6
7			/**
8			* ATTENTION: This code is WRITE-ONLY. Do not try to read it.
9			*/
10			class Emulative extends \PhpParser\Lexer
11			{
12			protected $newKeywords;
13			protected $inObjectAccess;
14
15			const T_ELLIPSIS = 1001;
16			const T_POW = 1002;
17			const T_POW_EQUAL = 1003;
18			const T_COALESCE = 1004;
19			const T_SPACESHIP = 1005;
20			const T_YIELD_FROM = 1006;
21
22			const PHP_7_0 = '7.0.0dev';
23			const PHP_5_6 = '5.6.0rc1';
24			const PHP_5_5 = '5.5.0beta1';
25
26			public function __construct(array $options = array()) {
27			parent::__construct($options);
28
29			$newKeywordsPerVersion = array(
30			self::PHP_5_5 => array(
31			'finally' => Tokens::T_FINALLY,
32			'yield' => Tokens::T_YIELD,
33			),
34			);
35
36			$this->newKeywords = array();
37			foreach ($newKeywordsPerVersion as $version => $newKeywords) {
38			if (version_compare(PHP_VERSION, $version, '>=')) {
39			break;
40			}
41
42			$this->newKeywords += $newKeywords;
43			}
44
45			if (version_compare(PHP_VERSION, self::PHP_7_0, '>=')) {
46			return;
47			}
48			$this->tokenMap[self::T_COALESCE] = Tokens::T_COALESCE;
49			$this->tokenMap[self::T_SPACESHIP] = Tokens::T_SPACESHIP;
50			$this->tokenMap[self::T_YIELD_FROM] = Tokens::T_YIELD_FROM;
51
52			if (version_compare(PHP_VERSION, self::PHP_5_6, '>=')) {
53			return;
54			}
55			$this->tokenMap[self::T_ELLIPSIS] = Tokens::T_ELLIPSIS;
56			$this->tokenMap[self::T_POW] = Tokens::T_POW;
57			$this->tokenMap[self::T_POW_EQUAL] = Tokens::T_POW_EQUAL;
58			}
59
60			public function startLexing($code) {
61			$this->inObjectAccess = false;
62
63			$preprocessedCode = $this->preprocessCode($code);
64			parent::startLexing($preprocessedCode);
65			if ($preprocessedCode !== $code) {
66			$this->postprocessTokens();
67			}
68
69			// Set code property back to the original code, so __halt_compiler()
70			// handling and (start\|end)FilePos attributes use the correct offsets
71			$this->code = $code;
72			}
73
74			/*
75			* Replaces new features in the code by ~__EMU__{NAME}__{DATA}__~ sequences.
76			* ~LABEL~ is never valid PHP code, that's why we can (to some degree) safely
77			* use it here.
78			* Later when preprocessing the tokens these sequences will either be replaced
79			* by real tokens or replaced with their original content (e.g. if they occurred
80			* inside a string, i.e. a place where they don't have a special meaning).
81			*/
82			protected function preprocessCode($code) {
83			if (version_compare(PHP_VERSION, self::PHP_7_0, '>=')) {
84			return $code;
85			}
86
87			$code = str_replace('??', '~__EMU__COALESCE__~', $code);
88			$code = str_replace('<=>', '~__EMU__SPACESHIP__~', $code);
89			$code = preg_replace_callback('(yield[ \n\r\t]+from)', function($matches) {
90			// Encoding $0 in order to preserve exact whitespace
91			return '~__EMU__YIELDFROM__' . bin2hex($matches[0]) . '__~';
92			}, $code);
93
94			if (version_compare(PHP_VERSION, self::PHP_5_6, '>=')) {
95			return $code;
96			}
97
98			$code = str_replace('...', '~__EMU__ELLIPSIS__~', $code);
99			$code = preg_replace('((?<!/)\\=)', '~__EMU__POWEQUAL__~', $code);
100			$code = preg_replace('((?<!/)\\(?!/))', '~__EMU__POW__~', $code);
101
102			return $code;
103			}
104
105			/*
106			* Replaces the ~__EMU__...~ sequences with real tokens or their original
107			* value.
108			*/
109			protected function postprocessTokens() {
110			// we need to manually iterate and manage a count because we'll change
111			// the tokens array on the way
112			for ($i = 0, $c = count($this->tokens); $i < $c; ++$i) {
113			// first check that the following tokens are of form ~LABEL~,
114			// then match the __EMU__... sequence.
115			if ('~' === $this->tokens[$i]
116			&& isset($this->tokens[$i + 2])
117			&& '~' === $this->tokens[$i + 2]
118			&& T_STRING === $this->tokens[$i + 1][0]
119			&& preg_match('(^__EMU__([A-Z]++)__(?:([A-Za-z0-9]++)__)?$)', $this->tokens[$i + 1][1], $matches)
120			) {
121			if ('ELLIPSIS' === $matches[1]) {
122			$replace = array(
123			array(self::T_ELLIPSIS, '...', $this->tokens[$i + 1][2])
124			);
125			} else if ('POW' === $matches[1]) {
126			$replace = array(
127			array(self::T_POW, '**', $this->tokens[$i + 1][2])
128			);
129			} else if ('POWEQUAL' === $matches[1]) {
130			$replace = array(
131			array(self::T_POW_EQUAL, '**=', $this->tokens[$i + 1][2])
132			);
133			} else if ('COALESCE' === $matches[1]) {
134			$replace = array(
135			array(self::T_COALESCE, '??', $this->tokens[$i + 1][2])
136			);
137			} else if ('SPACESHIP' === $matches[1]) {
138			$replace = array(
139			array(self::T_SPACESHIP, '<=>', $this->tokens[$i + 1][2]),
140			);
141			} else if ('YIELDFROM' === $matches[1]) {
142			$content = hex2bin($matches[2]);
143			$replace = array(
144			array(self::T_YIELD_FROM, $content, $this->tokens[$i + 1][2] - substr_count($content, "\n"))
145			);
146			} else {
147			throw new \RuntimeException('Invalid __EMU__ sequence');
148			}
149
150			array_splice($this->tokens, $i, 3, $replace);
151			$c -= 3 - count($replace);
152			// for multichar tokens (e.g. strings) replace any ~__EMU__...~ sequences
153			// in their content with the original character sequence
154			} elseif (is_array($this->tokens[$i])
155			&& 0 !== strpos($this->tokens[$i][1], '__EMU__')
156			) {
157			$this->tokens[$i][1] = preg_replace_callback(
158			'(~__EMU__([A-Z]++)__(?:([A-Za-z0-9]++)__)?~)',
159			array($this, 'restoreContentCallback'),
160			$this->tokens[$i][1]
161			);
162			}
163			}
164			}
165
166			/*
167			* This method is a callback for restoring EMU sequences in
168			* multichar tokens (like strings) to their original value.
169			*/
170			public function restoreContentCallback(array $matches) {
171			if ('ELLIPSIS' === $matches[1]) {
172			return '...';
173			} else if ('POW' === $matches[1]) {
174			return '**';
175			} else if ('POWEQUAL' === $matches[1]) {
176			return '**=';
177			} else if ('COALESCE' === $matches[1]) {
178			return '??';
179			} else if ('SPACESHIP' === $matches[1]) {
180			return '<=>';
181			} else if ('YIELDFROM' === $matches[1]) {
182			return hex2bin($matches[2]);
183			} else {
184			return $matches[0];
185			}
186			}
187
188			public function getNextToken(&$value = null, &$startAttributes = null, &$endAttributes = null) {
189			$token = parent::getNextToken($value, $startAttributes, $endAttributes);
190
191			// replace new keywords by their respective tokens. This is not done
192			// if we currently are in an object access (e.g. in $obj->namespace
193			// "namespace" stays a T_STRING tokens and isn't converted to T_NAMESPACE)
194			if (Tokens::T_STRING === $token && !$this->inObjectAccess) {
195			if (isset($this->newKeywords[strtolower($value)])) {
196			return $this->newKeywords[strtolower($value)];
197			}
198			} else {
199			// keep track of whether we currently are in an object access (after ->)
200			$this->inObjectAccess = Tokens::T_OBJECT_OPERATOR === $token;
201			}
202
203			return $token;
204			}
205			}
206

nikic / PHP-Parser

Emulative::preprocessCode() A last analyzed 2016-02-20 20:57 UTC

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like

Emulative::preprocessCode() A
last analyzed 2016-02-20 20:57 UTC