Formatter::getMergedOptions() - Code Metrics - Inspection of "Simplify rules merging" - phpmyadmin/sql-parser - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Push — master ( 7eebd4...4aadd2 )

by Michal

created 2017-01-03 12:31 UTC

Formatter::getMergedOptions() B

↳ Parent: Formatter

Complexity

Conditions	4
Paths	6

Size

Total Lines	22
Code Lines	12

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	3
CRAP Score	4

Importance

Changes

Metric	Value
cc	4
eloc	12
nc	6
nop	1
dl	0
loc	22
ccs	3
cts	3
cp	1
crap	4
rs	8.9197
c	0
b	0
f	0

<?php

/**
 * Utilities that are used for formatting queries.
 *
 * @package    SqlParser
 * @subpackage Utils
 */
namespace SqlParser\Utils;

use SqlParser\Lexer;
use SqlParser\Parser;
use SqlParser\Token;
use SqlParser\TokensList;

/**
 * Utilities that are used for formatting queries.
 *
 * @category   Misc
 * @package    SqlParser
 * @subpackage Utils
 * @license    https://www.gnu.org/licenses/gpl-2.0.txt GPL-2.0+
 */
class Formatter
{

    /**
     * The formatting options.
     *
     * @var array
     */
    public $options;

    /**
     * Clauses that must be inlined.
     *
     * These clauses usually are short and it's nicer to have them inline.
     *
     * @var array
     */
    public static $INLINE_CLAUSES = array(
        'CREATE'                        => true,
        'LIMIT'                         => true,
        'PARTITION BY'                  => true,
        'PARTITION'                     => true,
        'PROCEDURE'                     => true,
        'SUBPARTITION BY'               => true,
        'VALUES'                        => true,
    );

    /**
     * Constructor.
     *
     * @param array $options The formatting options.
     */
    public function __construct(array $options = array())
    {
        $this->options = $this->getMergedOptions($options);
    }

    /**
     * The specified formatting options are merged with the default values.
     *
     * @param array $options
     * @return array
     */
    private function getMergedOptions(array $options)
    {
        $options = array_merge(
            $this->getDefaultOptions(),
            $options
        );

        if (isset($options['formats'])) {
            $options['formats'] = self::mergeFormats($this->getDefaultFormats(), $options['formats']);
        } else {
            $options['formats'] = $this->getDefaultFormats();
        }

        if (is_null($options['line_ending'])) {
            $options['line_ending'] = $options['type'] === 'html' ? '<br/>' : "\n";
        }

        // `parts_newline` requires `clause_newline`
        $options['parts_newline'] &= $options['clause_newline'];

        return $options;
    }

    /**
     * The default formatting options.
     *
     * @return array
     */
    protected function getDefaultOptions()
    {
        return array(
            /**
             * The format of the result.
             *
             * @var string The type ('text', 'cli' or 'html')
             */
            'type' => php_sapi_name() === 'cli' ? 'cli' : 'text',

            /**
             * The line ending used.
             * By default, for text this is "\n" and for HTML this is "<br/>".
             *
             * @var string
             */
            'line_ending' => NULL,

            /**
             * The string used for indentation.
             *
             * @var string
             */
            'indentation' => '  ',

            /**
             * Whether comments should be removed or not.
             *
             * @var bool
             */
            'remove_comments' => false,

            /**
             * Whether each clause should be on a new line.
             *
             * @var bool
             */
            'clause_newline' => true,

            /**
             * Whether each part should be on a new line.
             * Parts are delimited by brackets and commas.
             *
             * @var bool
             */
            'parts_newline' => true,

            /**
             * Whether each part of each clause should be indented.
             *
             * @var bool
             */
            'indent_parts' => true,
        );
    }

    /**
     * The styles used for HTML formatting.
     * array($type, $flags, $span, $callback)
     *
     * @return array
     */
    protected function getDefaultFormats()
    {
        return array(
            array(
                'type' => Token::TYPE_KEYWORD,
                'flags' => Token::FLAG_KEYWORD_RESERVED,
                'html' => 'class="sql-reserved"',
                'cli' => "\x1b[35m",
                'function' => 'strtoupper',
            ),
            array(
                'type' => Token::TYPE_KEYWORD,
                'flags' => 0,
                'html' => 'class="sql-keyword"',
                'cli' => "\x1b[95m",
                'function' => 'strtoupper',
            ),
            array(
                'type' => Token::TYPE_COMMENT,
                'flags' => 0,
                'html' => 'class="sql-comment"',
                'cli' => "\x1b[37m",
                'function' => '',
            ),
            array(
                'type' => Token::TYPE_BOOL,
                'flags' => 0,
                'html' => 'class="sql-atom"',
                'cli' => "\x1b[36m",
                'function' => 'strtoupper',
            ),
            array(
                'type' => Token::TYPE_NUMBER,
                'flags' => 0,
                'html' => 'class="sql-number"',
                'cli' => "\x1b[92m",
                'function' => 'strtolower',
            ),
            array(
                'type' => Token::TYPE_STRING,
                'flags' => 0,
                'html' => 'class="sql-string"',
                'cli' => "\x1b[91m",
                'function' => '',
            ),
            array(
                'type' => Token::TYPE_SYMBOL,
                'flags' => 0,
                'html' => 'class="sql-variable"',
                'cli' => "\x1b[36m",
                'function' => '',
            ),
        );
    }

    private static function mergeFormats(array $formats, array $newFormats)
    {
        $added = array();
        $integers = array('flags', 'type');
        $strings = array('html', 'cli', 'function');

        /* Sanitize the array so that we do not have to care later */
        foreach ($newFormats as $j => $new) {
            foreach ($integers as $name) {
                if (! isset($new[$name])) {
                    $newFormats[$j][$name] = 0;
                }
            }
            foreach ($strings as $name) {
                if (! isset($new[$name])) {
                    $newFormats[$j][$name] = '';
                }
            }
        }

        /* Process changes to existing formats */
        foreach ($formats as $i => $original) {
            foreach ($newFormats as $j => $new) {
                if ($new['type'] === $original['type']
                    && $original['flags'] === $new['flags']
                ) {
                    $formats[$i] = $new;
                    $added[] = $j;
                }
            }
        }

        /* Add not already handled formats */
        foreach ($newFormats as $j => $new) {
            if (! in_array($j, $added)) {
                $formats[] = $new;
            }
        }

        return $formats;
    }

    /**
     * Formats the given list of tokens.
     *
     * @param TokensList $list The list of tokens.
     *
     * @return string
     */
    public function formatList($list)
    {

        /**
         * The query to be returned.
         *
         * @var string $ret
         */
        $ret = '';

        /**
         * The indentation level.
         *
         * @var int $indent
         */
        $indent = 0;

        /**
         * Whether the line ended.
         *
         * @var bool $lineEnded
         */
        $lineEnded = false;

        /**
         * Whether current group is short (no linebreaks)
         *
         * @var bool $shortGroup
         */
        $shortGroup = false;

        /**
         * The name of the last clause.
         *
         * @var string $lastClause
         */
        $lastClause = '';

        /**
         * A stack that keeps track of the indentation level every time a new
         * block is found.
         *
         * @var array $blocksIndentation
         */
        $blocksIndentation = array();

        /**
         * A stack that keeps track of the line endings every time a new block
         * is found.
         *
         * @var array $blocksLineEndings
         */
        $blocksLineEndings = array();

        /**
         * Whether clause's options were formatted.
         *
         * @var bool $formattedOptions
         */
        $formattedOptions = false;

        /**
         * Previously parsed token.
         *
         * @var Token $prev
         */
        $prev = null;

        /**
         * Comments are being formatted separately to maintain the whitespaces
         * before and after them.
         *
         * @var string $comment
         */
        $comment = '';

        // In order to be able to format the queries correctly, the next token
        // must be taken into consideration. The loop below uses two pointers,
        // `$prev` and `$curr` which store two consecutive tokens.
        // Actually, at every iteration the previous token is being used.
        for ($list->idx = 0; $list->idx < $list->count; ++$list->idx) {
            /**
             * Token parsed at this moment.
             *
             * @var Token $curr
             */
            $curr = $list->tokens[$list->idx];

            if ($curr->type === Token::TYPE_WHITESPACE) {
                // Whitespaces are skipped because the formatter adds its own.
                continue;
            } elseif ($curr->type === Token::TYPE_COMMENT) {
                // Whether the comments should be parsed.
                if (!empty($this->options['remove_comments'])) {
                    continue;
                }

                if ($list->tokens[$list->idx - 1]->type === Token::TYPE_WHITESPACE) {
                    // The whitespaces before and after are preserved for
                    // formatting reasons.
                    $comment .= $list->tokens[$list->idx - 1]->token;
                }
                $comment .= $this->toString($curr);
                if (($list->tokens[$list->idx + 1]->type === Token::TYPE_WHITESPACE)
                    && ($list->tokens[$list->idx + 2]->type !== Token::TYPE_COMMENT)
                ) {
                    // Adding the next whitespace only there is no comment that
                    // follows it immediately which may cause adding a
                    // whitespace twice.
                    $comment .= $list->tokens[$list->idx + 1]->token;
                }

                // Everything was handled here, no need to continue.
                continue;
            }

            // Checking if pointers were initialized.
            if ($prev !== null) {
                // Checking if a new clause started.
                if (static::isClause($prev) !== false) {
                    $lastClause = $prev->value;
                    $formattedOptions = false;
                }

                // The options of a clause should stay on the same line and everything that follows.
                if (($this->options['parts_newline'])
                    && (!$formattedOptions)
                    && (empty(self::$INLINE_CLAUSES[$lastClause]))
                    && (($curr->type !== Token::TYPE_KEYWORD)
                    || (($curr->type === Token::TYPE_KEYWORD)
                    && ($curr->flags & Token::FLAG_KEYWORD_FUNCTION)))
                ) {
                    $formattedOptions = true;
                    $lineEnded = true;
                    ++$indent;
                }

                // Checking if this clause ended.
                if ($tmp = static::isClause($curr)) {
                    if (($tmp == 2) || ($this->options['clause_newline'])) {
                        $lineEnded = true;
                        if ($this->options['parts_newline']) {
                            --$indent;
                        }
                    }
                }

                // Indenting BEGIN ... END blocks.
                if (($prev->type === Token::TYPE_KEYWORD) && ($prev->value === 'BEGIN')) {
                    $lineEnded = true;
                    array_push($blocksIndentation, $indent);
                    ++$indent;
                } elseif (($curr->type === Token::TYPE_KEYWORD) && ($curr->value === 'END')) {
                    $lineEnded = true;
                    $indent = array_pop($blocksIndentation);
                }

                // Formatting fragments delimited by comma.
                if (($prev->type === Token::TYPE_OPERATOR) && ($prev->value === ',')) {
                    // Fragments delimited by a comma are broken into multiple
                    // pieces only if the clause is not inlined or this fragment
                    // is between brackets that are on new line.
                    if (((empty(self::$INLINE_CLAUSES[$lastClause]))
                        && ! $shortGroup
                        && ($this->options['parts_newline']))
                        || (end($blocksLineEndings) === true)
                    ) {
                        $lineEnded = true;
                    }
                }

                // Handling brackets.
                // Brackets are indented only if the length of the fragment between
                // them is longer than 30 characters.
                if (($prev->type === Token::TYPE_OPERATOR) && ($prev->value === '(')) {
                    array_push($blocksIndentation, $indent);
                    $shortGroup = true;
                    if (static::getGroupLength($list) > 30) {
                        ++$indent;
                        $lineEnded = true;
                        $shortGroup = false;
                    }
                    array_push($blocksLineEndings, $lineEnded);
                } elseif (($curr->type === Token::TYPE_OPERATOR) && ($curr->value === ')')) {
                    $indent = array_pop($blocksIndentation);
                    $lineEnded |= array_pop($blocksLineEndings);
                    $shortGroup = false;
                }

                // Delimiter must be placed on the same line with the last
                // clause.
                if ($curr->type === Token::TYPE_DELIMITER) {
                    $lineEnded = false;
                }

                // Adding the token.
                $ret .= $this->toString($prev);

                // Finishing the line.
                if ($lineEnded) {
                    if ($indent < 0) {
                        // TODO: Make sure this never occurs and delete it.

                        $indent = 0;
                    }

                    if ($curr->type !== Token::TYPE_COMMENT) {
                        $ret .= $this->options['line_ending']
                            . str_repeat($this->options['indentation'], $indent);
                    }
                    $lineEnded = false;
                } else {
                    // If the line ended there is no point in adding whitespaces.
                    // Also, some tokens do not have spaces before or after them.
                    if (!((($prev->type === Token::TYPE_OPERATOR) && (($prev->value === '.') || ($prev->value === '(')))
                        // No space after . (
                        || (($curr->type === Token::TYPE_OPERATOR) && (($curr->value === '.') || ($curr->value === ',')
                        || ($curr->value === '(') || ($curr->value === ')')))
                        // No space before . , ( )
                        || (($curr->type === Token::TYPE_DELIMITER)) && (mb_strlen($curr->value, 'UTF-8') < 2))
                        // A space after delimiters that are longer than 2 characters.
                        || ($prev->value === 'DELIMITER')
                    ) {
                        $ret .= ' ';
                    }
                }
            }

            if (!empty($comment)) {
                $ret .= $comment;
                $comment = '';
            }

            // Iteration finished, consider current token as previous.
            $prev = $curr;
        }

        if ($this->options['type'] === 'cli') {
            return $ret . "\x1b[0m";
        }

        return $ret;
    }

    public function escapeConsole($string)
    {
        return str_replace(
            array(
                "\x00", "\x01", "\x02", "\x03", "\x04",
                "\x05", "\x06", "\x07", "\x08", "\x09", "\x0A",
                "\x0B","\x0C","\x0D", "\x0E", "\x0F", "\x10", "\x11",
                "\x12","\x13","\x14","\x15", "\x16", "\x17", "\x18",
                "\x19","\x1A","\x1B","\x1C","\x1D", "\x1E", "\x1F"
            ),
            array(
                '\x00', '\x01', '\x02', '\x03', '\x04',
                '\x05', '\x06', '\x07', '\x08', '\x09', '\x0A',
                '\x0B', '\x0C', '\x0D', '\x0E', '\x0F', '\x10', '\x11',
                '\x12', '\x13', '\x14', '\x15', '\x16', '\x17', '\x18',
                '\x19', '\x1A', '\x1B', '\x1C', '\x1D', '\x1E', '\x1F'
            ),
            $string
        );
    }

    /**
     * Tries to print the query and returns the result.
     *
     * @param Token $token The token to be printed.
     *
     * @return string
     */
    public function toString($token)
    {
        $text = $token->token;

        foreach ($this->options['formats'] as $format) {
            if (($token->type === $format['type'])
                && (($token->flags & $format['flags']) === $format['flags'])
            ) {
                // Running transformation function.
                if (!empty($format['function'])) {
                    $func = $format['function'];
                    $text = $func($text);
                }

                // Formatting HTML.
                if ($this->options['type'] === 'html') {
                    return '<span ' . $format['html'] . '>' . htmlspecialchars($text, ENT_NOQUOTES) . '</span>';
                } elseif ($this->options['type'] === 'cli') {
                    return $format['cli'] . $this->escapeConsole($text);
                }

                break;
            }
        }

        if ($this->options['type'] === 'cli') {
            return "\x1b[39m" . $this->escapeConsole($text);
        } elseif ($this->options['type'] === 'html') {
            return htmlspecialchars($text, ENT_NOQUOTES);
        }
        return $text;
    }

    /**
     * Formats a query.
     *
     * @param string $query   The query to be formatted
     * @param array  $options The formatting options.
     *
     * @return string          The formatted string.
     */
    public static function format($query, array $options = array())
    {
        $lexer = new Lexer($query);
        $formatter = new Formatter($options);
        return $formatter->formatList($lexer->list);
    }

    /**
     * Computes the length of a group.
     *
     * A group is delimited by a pair of brackets.
     *
     * @param TokensList $list The list of tokens.
     *
     * @return int
     */
    public static function getGroupLength($list)
    {
        /**
         * The number of opening brackets found.
         * This counter starts at one because by the time this function called,
         * the list already advanced one position and the opening bracket was
         * already parsed.
         *
         * @var int $count
         */
        $count = 1;

        /**
         * The length of this group.
         *
         * @var int $length
         */
        $length = 0;

        for ($idx = $list->idx; $idx < $list->count; ++$idx) {
            // Counting the brackets.
            if ($list->tokens[$idx]->type === Token::TYPE_OPERATOR) {
                if ($list->tokens[$idx]->value === '(') {
                    ++$count;
                } elseif ($list->tokens[$idx]->value === ')') {
                    --$count;
                    if ($count == 0) {
                        break;
                    }
                }
            }

            // Keeping track of this group's length.
            $length += mb_strlen($list->tokens[$idx]->value, 'UTF-8');
        }

        return $length;
    }

    /**
     * Checks if a token is a statement or a clause inside a statement.
     *
     * @param Token $token The token to be checked.
     *
     * @return int|bool
     */
    public static function isClause($token)
    {
        if ((($token->type === Token::TYPE_NONE) && (strtoupper($token->token) === 'DELIMITER'))
            || (($token->type === Token::TYPE_KEYWORD) && (isset(Parser::$STATEMENT_PARSERS[$token->value])))
        ) {
            return 2;
        } elseif (($token->type === Token::TYPE_KEYWORD) && (isset(Parser::$KEYWORD_PARSERS[$token->value]))) {
            return 1;
        }
        return false;
    }
}


1			<?php
2
3			/**
4			* Utilities that are used for formatting queries.
5			*
6			* @package SqlParser
7			* @subpackage Utils
8			*/
9			namespace SqlParser\Utils;
10
11			use SqlParser\Lexer;
12			use SqlParser\Parser;
13			use SqlParser\Token;
14			use SqlParser\TokensList;
15
16			/**
17			* Utilities that are used for formatting queries.
18			*
19			* @category Misc
20			* @package SqlParser
21			* @subpackage Utils
22			* @license https://www.gnu.org/licenses/gpl-2.0.txt GPL-2.0+
23			*/
24			class Formatter
25			{
26
27			/**
28			* The formatting options.
29			*
30			* @var array
31			*/
32			public $options;
33
34			/**
35			* Clauses that must be inlined.
36			*
37			* These clauses usually are short and it's nicer to have them inline.
38			*
39			* @var array
40			*/
41			public static $INLINE_CLAUSES = array(
42			'CREATE' => true,
43			'LIMIT' => true,
44			'PARTITION BY' => true,
45			'PARTITION' => true,
46			'PROCEDURE' => true,
47			'SUBPARTITION BY' => true,
48			'VALUES' => true,
49			);
50
51			/**
52			* Constructor.
53			*
54			* @param array $options The formatting options.
55			*/
56			public function __construct(array $options = array())
57	8		{
58			$this->options = $this->getMergedOptions($options);
59			}
60	8
61			/**
62			* The specified formatting options are merged with the default values.
63			*
64			* @param array $options
65			* @return array
66			*/
67			private function getMergedOptions(array $options)
68	8		{
69			$options = array_merge(
70			$this->getDefaultOptions(),
71			$options
72			);
73
74			if (isset($options['formats'])) {
75			$options['formats'] = self::mergeFormats($this->getDefaultFormats(), $options['formats']);
76	8		} else {
77			$options['formats'] = $this->getDefaultFormats();
78			}
79
80			if (is_null($options['line_ending'])) {
81			$options['line_ending'] = $options['type'] === 'html' ? '<br/>' : "\n";
82			}
83	8
84			// `parts_newline` requires `clause_newline`
85			$options['parts_newline'] &= $options['clause_newline'];
86
87			return $options;
88			}
89
90	8		/**
91			* The default formatting options.
92			*
93			* @return array
94			*/
95			protected function getDefaultOptions()
96			{
97	8		return array(
98			/**
99			* The format of the result.
100			*
101			* @var string The type ('text', 'cli' or 'html')
102			*/
103			'type' => php_sapi_name() === 'cli' ? 'cli' : 'text',
104
105	8		/**
106			* The line ending used.
107			* By default, for text this is "\n" and for HTML this is "<br/>".
108			*
109			* @var string
110			*/
111			'line_ending' => NULL,
112	8
113			/**
114			* The string used for indentation.
115			*
116			* @var string
117			*/
118			'indentation' => ' ',
119
120			/**
121			* Whether comments should be removed or not.
122	8		*
123	8		* @var bool
124	8		*/
125	8		'remove_comments' => false,
126	8
127	8		/**
128			* Whether each clause should be on a new line.
129	8		*
130	8		* @var bool
131	8		*/
132	8		'clause_newline' => true,
133	8
134	8		/**
135			* Whether each part should be on a new line.
136	8		* Parts are delimited by brackets and commas.
137	8		*
138	8		* @var bool
139	8		*/
140	8		'parts_newline' => true,
141	8
142			/**
143	8		* Whether each part of each clause should be indented.
144	8		*
145	8		* @var bool
146	8		*/
147	8		'indent_parts' => true,
148	8		);
149			}
150	8
151	8		/**
152	8		* The styles used for HTML formatting.
153	8		* array($type, $flags, $span, $callback)
154	8		*
155	8		* @return array
156			*/
157	8		protected function getDefaultFormats()
158	8		{
159	8		return array(
160	8		array(
161	8		'type' => Token::TYPE_KEYWORD,
162	8		'flags' => Token::FLAG_KEYWORD_RESERVED,
163			'html' => 'class="sql-reserved"',
164	8		'cli' => "\x1b[35m",
165	8		'function' => 'strtoupper',
166	8		),
167	8		array(
168	8		'type' => Token::TYPE_KEYWORD,
169	8		'flags' => 0,
170			'html' => 'class="sql-keyword"',
171	8		'cli' => "\x1b[95m",
172			'function' => 'strtoupper',
173	8		),
174			array(
175			'type' => Token::TYPE_COMMENT,
176	8		'flags' => 0,
177	8		'html' => 'class="sql-comment"',
178			'cli' => "\x1b[37m",
179			'function' => '',
180			),
181			array(
182			'type' => Token::TYPE_BOOL,
183			'flags' => 0,
184			'html' => 'class="sql-atom"',
185			'cli' => "\x1b[36m",
186	8		'function' => 'strtoupper',
187			),
188			array(
189			'type' => Token::TYPE_NUMBER,
190			'flags' => 0,
191			'html' => 'class="sql-number"',
192			'cli' => "\x1b[92m",
193			'function' => 'strtolower',
194	8		),
195			array(
196			'type' => Token::TYPE_STRING,
197			'flags' => 0,
198			'html' => 'class="sql-string"',
199			'cli' => "\x1b[91m",
200			'function' => '',
201	8		),
202			array(
203			'type' => Token::TYPE_SYMBOL,
204			'flags' => 0,
205			'html' => 'class="sql-variable"',
206			'cli' => "\x1b[36m",
207			'function' => '',
208	8		),
209			);
210			}
211
212			private static function mergeFormats(array $formats, array $newFormats)
213			{
214			$added = array();
215	8		$integers = array('flags', 'type');
216			$strings = array('html', 'cli', 'function');
217
218			/* Sanitize the array so that we do not have to care later */
219			foreach ($newFormats as $j => $new) {
220			foreach ($integers as $name) {
221			if (! isset($new[$name])) {
222	8		$newFormats[$j][$name] = 0;
223			}
224			}
225			foreach ($strings as $name) {
226			if (! isset($new[$name])) {
227			$newFormats[$j][$name] = '';
228			}
229			}
230	8		}
231
232			/* Process changes to existing formats */
233			foreach ($formats as $i => $original) {
234			foreach ($newFormats as $j => $new) {
235			if ($new['type'] === $original['type']
236			&& $original['flags'] === $new['flags']
237			) {
238	8		$formats[$i] = $new;
239			$added[] = $j;
240			}
241			}
242			}
243
244			/* Add not already handled formats */
245	8		foreach ($newFormats as $j => $new) {
246			if (! in_array($j, $added)) {
247			$formats[] = $new;
248			}
249			}
250
251			return $formats;
252	8		}
253
254			/**
255			* Formats the given list of tokens.
256			*
257			* @param TokensList $list The list of tokens.
258			*
259			* @return string
260	8		*/
261			public function formatList($list)
262			{
263
264			/**
265			* The query to be returned.
266	8		*
267			* @var string $ret
268			*/
269			$ret = '';
270
271			/**
272	8		* The indentation level.
273			*
274	8		* @var int $indent
275			*/
276	8		$indent = 0;
277	8
278			/**
279	1		* Whether the line ended.
280			*
281			* @var bool $lineEnded
282			*/
283	1		$lineEnded = false;
284
285			/**
286	1		* Whether current group is short (no linebreaks)
287	1		*
288	1		* @var bool $shortGroup
289	1		*/
290	1		$shortGroup = false;
291	1
292			/**
293			* The name of the last clause.
294			*
295			* @var string $lastClause
296			*/
297			$lastClause = '';
298
299	1		/**
300			* A stack that keeps track of the indentation level every time a new
301			* block is found.
302			*
303	8		* @var array $blocksIndentation
304			*/
305	8		$blocksIndentation = array();
306	8
307	8		/**
308	8		* A stack that keeps track of the line endings every time a new block
309			* is found.
310			*
311	8		* @var array $blocksLineEndings
312	8		*/
313	8		$blocksLineEndings = array();
314	8
315	3		/**
316	3		* Whether clause's options were formatted.
317	8		*
318	8		* @var bool $formattedOptions
319	8		*/
320	8		$formattedOptions = false;
321	8
322			/**
323			* Previously parsed token.
324	8		*
325	4		* @var Token $prev
326	4		*/
327	4		$prev = null;
328	4
329	4		/**
330	4		* Comments are being formatted separately to maintain the whitespaces
331	4		* before and after them.
332			*
333			* @var string $comment
334	8		*/
335	1		$comment = '';
336	1
337	1		// In order to be able to format the queries correctly, the next token
338	8		// must be taken into consideration. The loop below uses two pointers,
339	1		// `$prev` and `$curr` which store two consecutive tokens.
340	1		// Actually, at every iteration the previous token is being used.
341	1		for ($list->idx = 0; $list->idx < $list->count; ++$list->idx) {
342			/**
343			* Token parsed at this moment.
344	8		*
345			* @var Token $curr
346			*/
347			$curr = $list->tokens[$list->idx];
348	2
349	2		if ($curr->type === Token::TYPE_WHITESPACE) {
350	2		// Whitespaces are skipped because the formatter adds its own.
351	2		continue;
352	2		} elseif ($curr->type === Token::TYPE_COMMENT) {
353	1		// Whether the comments should be parsed.
354	1		if (!empty($this->options['remove_comments'])) {
355	2		continue;
356			}
357
358			if ($list->tokens[$list->idx - 1]->type === Token::TYPE_WHITESPACE) {
359			// The whitespaces before and after are preserved for
360	8		// formatting reasons.
361	4		$comment .= $list->tokens[$list->idx - 1]->token;
362	4		}
363	4		$comment .= $this->toString($curr);
364			if (($list->tokens[$list->idx + 1]->type === Token::TYPE_WHITESPACE)
365			&& ($list->tokens[$list->idx + 2]->type !== Token::TYPE_COMMENT)
366			) {
367			// Adding the next whitespace only there is no comment that
368	4		// follows it immediately which may cause adding a
369	8		// whitespace twice.
370	3		$comment .= $list->tokens[$list->idx + 1]->token;
371	3		}
372	3
373	3		// Everything was handled here, no need to continue.
374			continue;
375			}
376
377	8		// Checking if pointers were initialized.
378	8		if ($prev !== null) {
379	8		// Checking if a new clause started.
380			if (static::isClause($prev) !== false) {
381			$lastClause = $prev->value;
382	8		$formattedOptions = false;
383			}
384
385	8		// The options of a clause should stay on the same line and everything that follows.
386	8		if (($this->options['parts_newline'])
387			&& (!$formattedOptions)
388	2		&& (empty(self::$INLINE_CLAUSES[$lastClause]))
389	2		&& (($curr->type !== Token::TYPE_KEYWORD)
390			\|\| (($curr->type === Token::TYPE_KEYWORD)
391	8		&& ($curr->flags & Token::FLAG_KEYWORD_FUNCTION)))
392	8		) {
393	8		$formattedOptions = true;
394	8		$lineEnded = true;
395	8		++$indent;
396	8		}
397
398			// Checking if this clause ended.
399	8		if ($tmp = static::isClause($curr)) {
400			if (($tmp == 2) \|\| ($this->options['clause_newline'])) {
401	8		$lineEnded = true;
402	5		if ($this->options['parts_newline']) {
403			--$indent;
404	8		}
405			}
406	8		}
407	8
408	4		// Indenting BEGIN ... END blocks.
409	4		if (($prev->type === Token::TYPE_KEYWORD) && ($prev->value === 'BEGIN')) {
410			$lineEnded = true;
411	8		array_push($blocksIndentation, $indent);
412			++$indent;
413	8	View Code Duplication	} elseif (($curr->type === Token::TYPE_KEYWORD) && ($curr->value === 'END')) {
414	1		$lineEnded = true;
415	1		$indent = array_pop($blocksIndentation);
416	1		}
417
418			// Formatting fragments delimited by comma.
419	8		if (($prev->type === Token::TYPE_OPERATOR) && ($prev->value === ',')) {
420	8		// Fragments delimited by a comma are broken into multiple
421			// pieces only if the clause is not inlined or this fragment
422	8		// is between brackets that are on new line.
423	1		if (((empty(self::$INLINE_CLAUSES[$lastClause]))
424			&& ! $shortGroup
425			&& ($this->options['parts_newline']))
426	7		\|\| (end($blocksLineEndings) === true)
427			) {
428			$lineEnded = true;
429			}
430			}
431
432			// Handling brackets.
433			// Brackets are indented only if the length of the fragment between
434			// them is longer than 30 characters.
435			if (($prev->type === Token::TYPE_OPERATOR) && ($prev->value === '(')) {
436	8		array_push($blocksIndentation, $indent);
437			$shortGroup = true;
438	8		if (static::getGroupLength($list) > 30) {
439			++$indent;
440	8		$lineEnded = true;
441	8		$shortGroup = false;
442	8		}
443	8		array_push($blocksLineEndings, $lineEnded);
444		View Code Duplication	} elseif (($curr->type === Token::TYPE_OPERATOR) && ($curr->value === ')')) {
445	8		$indent = array_pop($blocksIndentation);
446	8		$lineEnded \|= array_pop($blocksLineEndings);
447	8		$shortGroup = false;
448	8		}
449
450			// Delimiter must be placed on the same line with the last
451	8		// clause.
452	7		if ($curr->type === Token::TYPE_DELIMITER) {
453	1		$lineEnded = false;
454	1		}
455
456			// Adding the token.
457			$ret .= $this->toString($prev);
458
459	8		// Finishing the line.
460			if ($lineEnded) {
461	5		if ($indent < 0) {
462			// TODO: Make sure this never occurs and delete it.
			0 ignored issues – show Coding Style Best Practice introduced 2016-02-12 21:51 UTC by Report Bug Copy Issue Report Comments for TODO tasks are often forgotten in the code; it might be better to use a dedicated issue tracker. Loading history...
463			$indent = 0;
464	5		}
465
466			if ($curr->type !== Token::TYPE_COMMENT) {
467			$ret .= $this->options['line_ending']
468			. str_repeat($this->options['indentation'], $indent);
469			}
470			$lineEnded = false;
471			} else {
472			// If the line ended there is no point in adding whitespaces.
473			// Also, some tokens do not have spaces before or after them.
474			if (!((($prev->type === Token::TYPE_OPERATOR) && (($prev->value === '.') \|\| ($prev->value === '(')))
475	8		// No space after . (
476			\|\| (($curr->type === Token::TYPE_OPERATOR) && (($curr->value === '.') \|\| ($curr->value === ',')
477	8		\|\| ($curr->value === '(') \|\| ($curr->value === ')')))
478	8		// No space before . , ( )
479	8		\|\| (($curr->type === Token::TYPE_DELIMITER)) && (mb_strlen($curr->value, 'UTF-8') < 2))
480			// A space after delimiters that are longer than 2 characters.
481			\|\| ($prev->value === 'DELIMITER')
482			) {
483			$ret .= ' ';
484			}
485			}
486			}
487
488			if (!empty($comment)) {
489			$ret .= $comment;
490			$comment = '';
491	4		}
492
493			// Iteration finished, consider current token as previous.
494			$prev = $curr;
495			}
496
497			if ($this->options['type'] === 'cli') {
498			return $ret . "\x1b[0m";
499			}
500
501	4		return $ret;
502			}
503
504			public function escapeConsole($string)
505			{
506			return str_replace(
507			array(
508	4		"\x00", "\x01", "\x02", "\x03", "\x04",
509			"\x05", "\x06", "\x07", "\x08", "\x09", "\x0A",
510	4		"\x0B","\x0C","\x0D", "\x0E", "\x0F", "\x10", "\x11",
511			"\x12","\x13","\x14","\x15", "\x16", "\x17", "\x18",
512	4		"\x19","\x1A","\x1B","\x1C","\x1D", "\x1E", "\x1F"
513	4		),
514			array(
515	4		'\x00', '\x01', '\x02', '\x03', '\x04',
516	4		'\x05', '\x06', '\x07', '\x08', '\x09', '\x0A',
517	4		'\x0B', '\x0C', '\x0D', '\x0E', '\x0F', '\x10', '\x11',
518	4		'\x12', '\x13', '\x14', '\x15', '\x16', '\x17', '\x18',
519			'\x19', '\x1A', '\x1B', '\x1C', '\x1D', '\x1E', '\x1F'
520			),
521	2		$string
522			);
523			}
524	3
525	3		/**
526			* Tries to print the query and returns the result.
527	4		*
528			* @param Token $token The token to be printed.
529			*
530			* @return string
531			*/
532			public function toString($token)
533			{
534			$text = $token->token;
535
536			foreach ($this->options['formats'] as $format) {
537	8		if (($token->type === $format['type'])
538			&& (($token->flags & $format['flags']) === $format['flags'])
539	8		) {
540	8		// Running transformation function.
541	8		if (!empty($format['function'])) {
542	8		$func = $format['function'];
543	8		$text = $func($text);
544	4		}
545
546	8		// Formatting HTML.
547			if ($this->options['type'] === 'html') {
548			return '<span ' . $format['html'] . '>' . htmlspecialchars($text, ENT_NOQUOTES) . '</span>';
549			} elseif ($this->options['type'] === 'cli') {
550			return $format['cli'] . $this->escapeConsole($text);
551			}
552
553			break;
554			}
555			}
556
557			if ($this->options['type'] === 'cli') {
558			return "\x1b[39m" . $this->escapeConsole($text);
559			} elseif ($this->options['type'] === 'html') {
560			return htmlspecialchars($text, ENT_NOQUOTES);
561			}
562			return $text;
563			}
564
565			/**
566			* Formats a query.
567			*
568			* @param string $query The query to be formatted
569			* @param array $options The formatting options.
570			*
571			* @return string The formatted string.
572			*/
573			public static function format($query, array $options = array())
574			{
575			$lexer = new Lexer($query);
576			$formatter = new Formatter($options);
577			return $formatter->formatList($lexer->list);
578			}
579
580			/**
581			* Computes the length of a group.
582			*
583			* A group is delimited by a pair of brackets.
584			*
585			* @param TokensList $list The list of tokens.
586			*
587			* @return int
588			*/
589			public static function getGroupLength($list)
590			{
591			/**
592			* The number of opening brackets found.
593			* This counter starts at one because by the time this function called,
594			* the list already advanced one position and the opening bracket was
595			* already parsed.
596			*
597			* @var int $count
598			*/
599			$count = 1;
600
601			/**
602			* The length of this group.
603			*
604			* @var int $length
605			*/
606			$length = 0;
607
608			for ($idx = $list->idx; $idx < $list->count; ++$idx) {
609			// Counting the brackets.
610			if ($list->tokens[$idx]->type === Token::TYPE_OPERATOR) {
611			if ($list->tokens[$idx]->value === '(') {
612			++$count;
613			} elseif ($list->tokens[$idx]->value === ')') {
614			--$count;
615			if ($count == 0) {
616			break;
617			}
618			}
619			}
620
621			// Keeping track of this group's length.
622			$length += mb_strlen($list->tokens[$idx]->value, 'UTF-8');
623			}
624
625			return $length;
626			}
627
628			/**
629			* Checks if a token is a statement or a clause inside a statement.
630			*
631			* @param Token $token The token to be checked.
632			*
633			* @return int\|bool
634			*/
635			public static function isClause($token)
636			{
637			if ((($token->type === Token::TYPE_NONE) && (strtoupper($token->token) === 'DELIMITER'))
638			\|\| (($token->type === Token::TYPE_KEYWORD) && (isset(Parser::$STATEMENT_PARSERS[$token->value])))
639			) {
640			return 2;
641			} elseif (($token->type === Token::TYPE_KEYWORD) && (isset(Parser::$KEYWORD_PARSERS[$token->value]))) {
642			return 1;
643			}
644			return false;
645			}
646			}
647

phpmyadmin / sql-parser

Push — master ( 7eebd4...4aadd2 )

Formatter::getMergedOptions() B

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like