Formatter::toString() - Code Metrics - Inspection of "Don't print duplicated cli formatting characters" - phpmyadmin/sql-parser - Measure and Improve Code Quality continuously with Scrutinizer

Test Failed

Pull Request — master (#153)

by Michal

created 2017-05-17 06:10 UTC

Formatter::toString() C

↳ Parent: Formatter

Complexity

Conditions	11
Paths	22

Size

Total Lines	44
Code Lines	25

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	19
CRAP Score	11

Importance

Changes

Metric	Value
dl	0
loc	44
ccs	19
cts	19
cp	1
rs	5.2653
c	0
b	0
f	0
cc	11
eloc	25
nc	22
nop	1
crap	11

How to fix Complexity

<?php

/**
 * Utilities that are used for formatting queries.
 */

namespace PhpMyAdmin\SqlParser\Utils;

use PhpMyAdmin\SqlParser\Components\JoinKeyword;
use PhpMyAdmin\SqlParser\Lexer;
use PhpMyAdmin\SqlParser\Parser;
use PhpMyAdmin\SqlParser\Token;
use PhpMyAdmin\SqlParser\TokensList;

/**
 * Utilities that are used for formatting queries.
 *
 * @category   Misc
 *
 * @license    https://www.gnu.org/licenses/gpl-2.0.txt GPL-2.0+
 */
class Formatter
{
    /**
     * The formatting options.
     *
     * @var array
     */
    public $options;

    /**
     * Clauses that are usually short.
     *
     * These clauses share the line with the next clause.
     *
     * E.g. if INSERT was not here, the formatter would produce:
     *
     *      INSERT
     *      INTO foo
     *      VALUES(0, 0, 0),(1, 1, 1);
     *
     * Instead of:
     *
     *      INSERT INTO foo
     *      VALUES(0, 0, 0),(1, 1, 1)
     *
     * @var array
     */
    public static $SHORT_CLAUSES = array(
        'CREATE' => true,
        'INSERT' => true,
    );

    /**
     * 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,
        'INTO' => 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";
        }

        if (is_null($options['indentation'])) {
            $options['indentation'] = $options['type'] === 'html' ? '&nbsp;&nbsp;&nbsp;&nbsp;' : '    ';
        }

        // `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' => null,

            /*
             * 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 = '';

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

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

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

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

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

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

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

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

        // 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 = $list->tokens[$list->idx];

            if ($curr->type === Token::TYPE_WHITESPACE) {
                // Keep linebreaks after comments
                if (strpos($curr->token, "\n") !== false && $prev !== null && $prev->type === Token::TYPE_COMMENT) {
                    $lineEnded = true;
                }
                // Whitespaces are skipped because the formatter adds its own.
                continue;
            }

            if ($curr->type === Token::TYPE_COMMENT && $this->options['remove_comments']) {
                // Skip Comments if option `remove_comments` is enabled
                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 || $this->options['clause_newline']) && empty(self::$SHORT_CLAUSES[$lastClause])) {
                        $lineEnded = true;
                        if ($this->options['parts_newline'] && $indent > 0) {
                            --$indent;
                        }
                    }
                }

                // Inline JOINs
                if (($prev->type === Token::TYPE_KEYWORD && isset(JoinKeyword::$JOINS[$prev->value]))
                    || (in_array($curr->value, array('ON', 'USING'), true) && isset(JoinKeyword::$JOINS[$list->tokens[$list->idx - 2]->value]))

                    || (isset($list->tokens[$list->idx - 4]) && isset(JoinKeyword::$JOINS[$list->tokens[$list->idx - 4]->value]))

                    || (isset($list->tokens[$list->idx - 6]) && isset(JoinKeyword::$JOINS[$list->tokens[$list->idx - 6]->value]))

                ) {
                    $lineEnded = false;
                }

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

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

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

                // Finishing the line.
                if ($lineEnded) {
                    $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 (
                        // A space after delimiters that are longer than 2 characters.
                        $prev->keyword === 'DELIMITER'
                        || !(
                            ($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
                        )
                    ) {
                        $ret .= ' ';
                    }
                }
            }

            // 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;
        static $prev;

        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 ($this->options['type'] === 'cli') {
                    if ($prev != $format['cli']) {
                        $prev = $format['cli'];
                        return $format['cli'] . $this->escapeConsole($text);
                    }

                    return $this->escapeConsole($text);
                }

                break;
            }
        }

        if ($this->options['type'] === 'cli') {
            if ($prev != "\x1b[39m") {
                $prev = "\x1b[39m";
                return "\x1b[39m" . $this->escapeConsole($text);
            }

            return $this->escapeConsole($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);
        $formatter = new self($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 = 1;

        /**
         * The length of this group.
         *
         * @var int
         */
        $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 (
            ($token->type === Token::TYPE_KEYWORD && isset(Parser::$STATEMENT_PARSERS[$token->keyword]))
            || ($token->type === Token::TYPE_NONE && strtoupper($token->token) === 'DELIMITER')
        ) {
            return 2;
        } elseif (
            $token->type === Token::TYPE_KEYWORD && isset(Parser::$KEYWORD_PARSERS[$token->keyword])
        ) {
            return 1;
        }

        return false;
    }
}


1		<?php
2
3		/**
4		* Utilities that are used for formatting queries.
5		*/
6
7		namespace PhpMyAdmin\SqlParser\Utils;
8
9		use PhpMyAdmin\SqlParser\Components\JoinKeyword;
10		use PhpMyAdmin\SqlParser\Lexer;
11		use PhpMyAdmin\SqlParser\Parser;
12		use PhpMyAdmin\SqlParser\Token;
13		use PhpMyAdmin\SqlParser\TokensList;
14
15		/**
16		* Utilities that are used for formatting queries.
17		*
18		* @category Misc
19		*
20		* @license https://www.gnu.org/licenses/gpl-2.0.txt GPL-2.0+
21		*/
22		class Formatter
23		{
24		/**
25		* The formatting options.
26		*
27		* @var array
28		*/
29		public $options;
30
31		/**
32		* Clauses that are usually short.
33		*
34		* These clauses share the line with the next clause.
35		*
36		* E.g. if INSERT was not here, the formatter would produce:
37		*
38		* INSERT
39		* INTO foo
40		* VALUES(0, 0, 0),(1, 1, 1);
41		*
42		* Instead of:
43		*
44		* INSERT INTO foo
45		* VALUES(0, 0, 0),(1, 1, 1)
46		*
47		* @var array
48		*/
49		public static $SHORT_CLAUSES = array(
50		'CREATE' => true,
51		'INSERT' => true,
52		);
53
54		/**
55		* Clauses that must be inlined.
56		*
57		* These clauses usually are short and it's nicer to have them inline.
58		*
59		* @var array
60		*/
61		public static $INLINE_CLAUSES = array(
62		'CREATE' => true,
63		'INTO' => true,
64		'LIMIT' => true,
65		'PARTITION BY' => true,
66		'PARTITION' => true,
67		'PROCEDURE' => true,
68		'SUBPARTITION BY' => true,
69		'VALUES' => true,
70		);
71
72		/**
73		* Constructor.
74		*
75		* @param array $options the formatting options
76		*/
77	19	public function __construct(array $options = array())
78		{
79	19	$this->options = $this->getMergedOptions($options);
80	19	}
81
82		/**
83		* The specified formatting options are merged with the default values.
84		*
85		* @param array $options
86		*
87		* @return array
88		*/
89	23	private function getMergedOptions(array $options)
90		{
91	23	$options = array_merge(
92	23	$this->getDefaultOptions(),
93		$options
94		);
95
96	23	if (isset($options['formats'])) {
97	4	$options['formats'] = self::mergeFormats($this->getDefaultFormats(), $options['formats']);
98		} else {
99	19	$options['formats'] = $this->getDefaultFormats();
100		}
101
102	23	if (is_null($options['line_ending'])) {
103	19	$options['line_ending'] = $options['type'] === 'html' ? '<br/>' : "\n";
104		}
105
106	23	if (is_null($options['indentation'])) {
107	23	$options['indentation'] = $options['type'] === 'html' ? '    ' : ' ';
108		}
109
110		// `parts_newline` requires `clause_newline`
111	23	$options['parts_newline'] &= $options['clause_newline'];
112
113	23	return $options;
114		}
115
116		/**
117		* The default formatting options.
118		*
119		* @return array
120		*/
121	19	protected function getDefaultOptions()
122		{
123		return array(
124		/*
125		* The format of the result.
126		*
127		* @var string The type ('text', 'cli' or 'html')
128		*/
129	19	'type' => php_sapi_name() === 'cli' ? 'cli' : 'text',
130
131		/*
132		* The line ending used.
133		* By default, for text this is "\n" and for HTML this is "<br/>".
134		*
135		* @var string
136		*/
137		'line_ending' => null,
138
139		/*
140		* The string used for indentation.
141		*
142		* @var string
143		*/
144		'indentation' => null,
145
146		/*
147		* Whether comments should be removed or not.
148		*
149		* @var bool
150		*/
151		'remove_comments' => false,
152
153		/*
154		* Whether each clause should be on a new line.
155		*
156		* @var bool
157		*/
158		'clause_newline' => true,
159
160		/*
161		* Whether each part should be on a new line.
162		* Parts are delimited by brackets and commas.
163		*
164		* @var bool
165		*/
166		'parts_newline' => true,
167
168		/*
169		* Whether each part of each clause should be indented.
170		*
171		* @var bool
172		*/
173		'indent_parts' => true,
174		);
175		}
176
177		/**
178		* The styles used for HTML formatting.
179		* array($type, $flags, $span, $callback).
180		*
181		* @return array
182		*/
183	19	protected function getDefaultFormats()
184		{
185		return array(
186		array(
187		'type' => Token::TYPE_KEYWORD,
188		'flags' => Token::FLAG_KEYWORD_RESERVED,
189		'html' => 'class="sql-reserved"',
190		'cli' => "\x1b[35m",
191		'function' => 'strtoupper',
192	19	),
193		array(
194		'type' => Token::TYPE_KEYWORD,
195		'flags' => 0,
196		'html' => 'class="sql-keyword"',
197		'cli' => "\x1b[95m",
198		'function' => 'strtoupper',
199		),
200		array(
201		'type' => Token::TYPE_COMMENT,
202		'flags' => 0,
203		'html' => 'class="sql-comment"',
204		'cli' => "\x1b[37m",
205		'function' => '',
206		),
207		array(
208		'type' => Token::TYPE_BOOL,
209		'flags' => 0,
210		'html' => 'class="sql-atom"',
211		'cli' => "\x1b[36m",
212		'function' => 'strtoupper',
213		),
214		array(
215		'type' => Token::TYPE_NUMBER,
216		'flags' => 0,
217		'html' => 'class="sql-number"',
218		'cli' => "\x1b[92m",
219		'function' => 'strtolower',
220		),
221		array(
222		'type' => Token::TYPE_STRING,
223		'flags' => 0,
224		'html' => 'class="sql-string"',
225		'cli' => "\x1b[91m",
226		'function' => '',
227		),
228		array(
229		'type' => Token::TYPE_SYMBOL,
230		'flags' => 0,
231		'html' => 'class="sql-variable"',
232		'cli' => "\x1b[36m",
233		'function' => '',
234		),
235		);
236		}
237
238	4	private static function mergeFormats(array $formats, array $newFormats)
239		{
240	4	$added = array();
241	4	$integers = array('flags', 'type');
242	4	$strings = array('html', 'cli', 'function');
243
244		/* Sanitize the array so that we do not have to care later */
245	4	foreach ($newFormats as $j => $new) {
246	4	foreach ($integers as $name) {
247	4	if (!isset($new[$name])) {
248	4	$newFormats[$j][$name] = 0;
249		}
250		}
251	4	foreach ($strings as $name) {
252	4	if (!isset($new[$name])) {
253	4	$newFormats[$j][$name] = '';
254		}
255		}
256		}
257
258		/* Process changes to existing formats */
259	4	foreach ($formats as $i => $original) {
260	4	foreach ($newFormats as $j => $new) {
261	4	if ($new['type'] === $original['type']
262	4	&& $original['flags'] === $new['flags']
263		) {
264	3	$formats[$i] = $new;
265	4	$added[] = $j;
266		}
267		}
268		}
269
270		/* Add not already handled formats */
271	4	foreach ($newFormats as $j => $new) {
272	4	if (!in_array($j, $added)) {
273	4	$formats[] = $new;
274		}
275		}
276
277	4	return $formats;
278		}
279
280		/**
281		* Formats the given list of tokens.
282		*
283		* @param TokensList $list the list of tokens
284		*
285		* @return string
286		*/
287	19	public function formatList($list)
288		{
289		/**
290		* The query to be returned.
291		*
292		* @var string
293		*/
294	19	$ret = '';
295
296		/**
297		* The indentation level.
298		*
299		* @var int
300		*/
301	19	$indent = 0;
302
303		/**
304		* Whether the line ended.
305		*
306		* @var bool
307		*/
308	19	$lineEnded = false;
309
310		/**
311		* Whether current group is short (no linebreaks).
312		*
313		* @var bool
314		*/
315	19	$shortGroup = false;
316
317		/**
318		* The name of the last clause.
319		*
320		* @var string
321		*/
322	19	$lastClause = '';
323
324		/**
325		* A stack that keeps track of the indentation level every time a new
326		* block is found.
327		*
328		* @var array
329		*/
330	19	$blocksIndentation = array();
331
332		/**
333		* A stack that keeps track of the line endings every time a new block
334		* is found.
335		*
336		* @var array
337		*/
338	19	$blocksLineEndings = array();
339
340		/**
341		* Whether clause's options were formatted.
342		*
343		* @var bool
344		*/
345	19	$formattedOptions = false;
346
347		/**
348		* Previously parsed token.
349		*
350		* @var Token\|null
351		*/
352	19	$prev = null;
353
354		// In order to be able to format the queries correctly, the next token
355		// must be taken into consideration. The loop below uses two pointers,
356		// `$prev` and `$curr` which store two consecutive tokens.
357		// Actually, at every iteration the previous token is being used.
358	19	for ($list->idx = 0; $list->idx < $list->count; ++$list->idx) {
359		/**
360		* Token parsed at this moment.
361		*
362		* @var Token
363		*/
364	19	$curr = $list->tokens[$list->idx];
365
366	19	if ($curr->type === Token::TYPE_WHITESPACE) {
367		// Keep linebreaks after comments
368	18	if (strpos($curr->token, "\n") !== false && $prev !== null && $prev->type === Token::TYPE_COMMENT) {
369	1	$lineEnded = true;
370		}
371		// Whitespaces are skipped because the formatter adds its own.
372	18	continue;
373		}
374
375	19	if ($curr->type === Token::TYPE_COMMENT && $this->options['remove_comments']) {
376		// Skip Comments if option `remove_comments` is enabled
377	1	continue;
378		}
379
380		// Checking if pointers were initialized.
381	19	if ($prev !== null) {
382		// Checking if a new clause started.
383	18	if (static::isClause($prev) !== false) {
384	18	$lastClause = $prev->value;
385	18	$formattedOptions = false;
386		}
387
388		// The options of a clause should stay on the same line and everything that follows.
389	18	if ($this->options['parts_newline']
390	18	&& !$formattedOptions
391	18	&& empty(self::$INLINE_CLAUSES[$lastClause])
392		&& (
393	17	$curr->type !== Token::TYPE_KEYWORD
394		\|\| (
395	5	$curr->type === Token::TYPE_KEYWORD
396	18	&& $curr->flags & Token::FLAG_KEYWORD_FUNCTION
397		)
398		)
399		) {
400	16	$formattedOptions = true;
401	16	$lineEnded = true;
402	16	++$indent;
403		}
404
405		// Checking if this clause ended.
406	18	if ($tmp = static::isClause($curr)) {
407	8	if (($tmp == 2 \|\| $this->options['clause_newline']) && empty(self::$SHORT_CLAUSES[$lastClause])) {
408	8	$lineEnded = true;
409	8	if ($this->options['parts_newline'] && $indent > 0) {
410	7	--$indent;
411		}
412		}
413		}
414
415		// Inline JOINs
416	18	if (($prev->type === Token::TYPE_KEYWORD && isset(JoinKeyword::$JOINS[$prev->value]))
417	18	\|\| (in_array($curr->value, array('ON', 'USING'), true) && isset(JoinKeyword::$JOINS[$list->tokens[$list->idx - 2]->value]))
		0 ignored issues – show Coding Style introduced 2017-02-16 21:16 UTC by Report Bug Copy Issue Report This line exceeds maximum limit of 120 characters; contains 143 characters Overly long lines are hard to read on any screen. Most code styles therefor impose a maximum limit on the number of characters in a line. Loading history...
418	18	\|\| (isset($list->tokens[$list->idx - 4]) && isset(JoinKeyword::$JOINS[$list->tokens[$list->idx - 4]->value]))
		0 ignored issues – show Coding Style introduced 2016-02-12 21:51 UTC by Report Bug Copy Issue Report This line exceeds maximum limit of 120 characters; contains 129 characters Overly long lines are hard to read on any screen. Most code styles therefor impose a maximum limit on the number of characters in a line. Loading history...
419	18	\|\| (isset($list->tokens[$list->idx - 6]) && isset(JoinKeyword::$JOINS[$list->tokens[$list->idx - 6]->value]))
		0 ignored issues – show Coding Style introduced 2017-02-16 21:16 UTC by Report Bug Copy Issue Report This line exceeds maximum limit of 120 characters; contains 129 characters Overly long lines are hard to read on any screen. Most code styles therefor impose a maximum limit on the number of characters in a line. Loading history...
420		) {
421	1	$lineEnded = false;
422		}
423
424		// Indenting BEGIN ... END blocks.
425	18	if ($prev->type === Token::TYPE_KEYWORD && $prev->keyword === 'BEGIN') {
426	1	$lineEnded = true;
427	1	array_push($blocksIndentation, $indent);
428	1	++$indent;
429	18	} elseif ($curr->type === Token::TYPE_KEYWORD && $curr->keyword === 'END') {
430	1	$lineEnded = true;
431	1	$indent = array_pop($blocksIndentation);
432		}
433
434		// Formatting fragments delimited by comma.
435	18	if ($prev->type === Token::TYPE_OPERATOR && $prev->value === ',') {
436		// Fragments delimited by a comma are broken into multiple
437		// pieces only if the clause is not inlined or this fragment
438		// is between brackets that are on new line.
439	4	if (end($blocksLineEndings) === true
440		\|\| (
441	3	empty(self::$INLINE_CLAUSES[$lastClause])
442	3	&& !$shortGroup
443	4	&& $this->options['parts_newline']
444		)
445		) {
446	3	$lineEnded = true;
447		}
448		}
449
450		// Handling brackets.
451		// Brackets are indented only if the length of the fragment between
452		// them is longer than 30 characters.
453	18	if ($prev->type === Token::TYPE_OPERATOR && $prev->value === '(') {
454	6	array_push($blocksIndentation, $indent);
455	6	$shortGroup = true;
456	6	if (static::getGroupLength($list) > 30) {
457	1	++$indent;
458	1	$lineEnded = true;
459	1	$shortGroup = false;
460		}
461	6	array_push($blocksLineEndings, $lineEnded);
462	18	} elseif ($curr->type === Token::TYPE_OPERATOR && $curr->value === ')') {
463	5	$indent = array_pop($blocksIndentation);
464	5	$lineEnded \|= array_pop($blocksLineEndings);
465	5	$shortGroup = false;
466		}
467
468		// Adding the token.
469	18	$ret .= $this->toString($prev);
470
471		// Finishing the line.
472	18	if ($lineEnded) {
473	17	$ret .= $this->options['line_ending']
474	17	. str_repeat($this->options['indentation'], $indent);
475
476	17	$lineEnded = false;
477		} else {
478		// If the line ended there is no point in adding whitespaces.
479		// Also, some tokens do not have spaces before or after them.
480		if (
481		// A space after delimiters that are longer than 2 characters.
482	18	$prev->keyword === 'DELIMITER'
483		\|\| !(
484	18	($prev->type === Token::TYPE_OPERATOR && ($prev->value === '.' \|\| $prev->value === '('))
485		// No space after . (
486	18	\|\| ($curr->type === Token::TYPE_OPERATOR && ($curr->value === '.' \|\| $curr->value === ',' \|\| $curr->value === '(' \|\| $curr->value === ')'))
		0 ignored issues – show Coding Style introduced 2017-01-06 17:15 UTC by Report Bug Copy Issue Report This line exceeds maximum limit of 120 characters; contains 167 characters Overly long lines are hard to read on any screen. Most code styles therefor impose a maximum limit on the number of characters in a line. Loading history...
487		// No space before . , ( )
488	18	\|\| $curr->type === Token::TYPE_DELIMITER && mb_strlen($curr->value, 'UTF-8') < 2
489		)
490		) {
491	10	$ret .= ' ';
492		}
493		}
494		}
495
496		// Iteration finished, consider current token as previous.
497	19	$prev = $curr;
498		}
499
500	19	if ($this->options['type'] === 'cli') {
501	17	return $ret . "\x1b[0m";
502		}
503
504	17	return $ret;
505		}
506
507	16	public function escapeConsole($string)
508		{
509	16	return str_replace(
510		array(
511	16	"\x00", "\x01", "\x02", "\x03", "\x04", "\x05", "\x06", "\x07", "\x08", "\x09", "\x0A", "\x0B", "\x0C", "\x0D", "\x0E", "\x0F",
		0 ignored issues – show Coding Style introduced 2017-01-04 05:52 UTC by Report Bug Copy Issue Report This line exceeds maximum limit of 120 characters; contains 143 characters Overly long lines are hard to read on any screen. Most code styles therefor impose a maximum limit on the number of characters in a line. Loading history...
512		"\x10", "\x11", "\x12", "\x13", "\x14", "\x15", "\x16", "\x17", "\x18", "\x19", "\x1A", "\x1B", "\x1C", "\x1D", "\x1E", "\x1F",
		0 ignored issues – show Coding Style introduced 2017-01-04 05:52 UTC by Report Bug Copy Issue Report This line exceeds maximum limit of 120 characters; contains 143 characters Overly long lines are hard to read on any screen. Most code styles therefor impose a maximum limit on the number of characters in a line. Loading history...
513		),
514		array(
515	16	'\x00', '\x01', '\x02', '\x03', '\x04', '\x05', '\x06', '\x07', '\x08', '\x09', '\x0A', '\x0B', '\x0C', '\x0D', '\x0E', '\x0F',
		0 ignored issues – show Coding Style introduced 2017-01-04 05:52 UTC by Report Bug Copy Issue Report This line exceeds maximum limit of 120 characters; contains 143 characters Overly long lines are hard to read on any screen. Most code styles therefor impose a maximum limit on the number of characters in a line. Loading history...
516		'\x10', '\x11', '\x12', '\x13', '\x14', '\x15', '\x16', '\x17', '\x18', '\x19', '\x1A', '\x1B', '\x1C', '\x1D', '\x1E', '\x1F',
		0 ignored issues – show Coding Style introduced 2017-01-04 05:52 UTC by Report Bug Copy Issue Report This line exceeds maximum limit of 120 characters; contains 143 characters Overly long lines are hard to read on any screen. Most code styles therefor impose a maximum limit on the number of characters in a line. Loading history...
517		),
518		$string
519		);
520		}
521
522		/**
523		* Tries to print the query and returns the result.
524		*
525		* @param Token $token the token to be printed
526		*
527		* @return string
528		*/
529	18	public function toString($token)
530		{
531	18	$text = $token->token;
532		static $prev;
533	18
534	18	foreach ($this->options['formats'] as $format) {
535	18	if ($token->type === $format['type']
536		&& ($token->flags & $format['flags']) === $format['flags']
537		) {
538	18	// Running transformation function.
539	18	if (!empty($format['function'])) {
540	18	$func = $format['function'];
541		$text = $func($text);
542		}
543
544	18	// Formatting HTML.
545	15	if ($this->options['type'] === 'html') {
546	17	return '<span ' . $format['html'] . '>' . htmlspecialchars($text, ENT_NOQUOTES) . '</span>';
547	16	} elseif ($this->options['type'] === 'cli') {
548		if ($prev != $format['cli']) {
549		$prev = $format['cli'];
550	18	return $format['cli'] . $this->escapeConsole($text);
551		}
552
553		return $this->escapeConsole($text);
554	15	}
555	12
556	15	break;
557	12	}
558		}
559
560	15	if ($this->options['type'] === 'cli') {
561		if ($prev != "\x1b[39m") {
562		$prev = "\x1b[39m";
563		return "\x1b[39m" . $this->escapeConsole($text);
564		}
565
566		return $this->escapeConsole($text);
567		} elseif ($this->options['type'] === 'html') {
568		return htmlspecialchars($text, ENT_NOQUOTES);
569		}
570
571	19	return $text;
572		}
573	19
574	19	/**
575		* Formats a query.
576	19	*
577		* @param string $query The query to be formatted
578		* @param array $options the formatting options
579		*
580		* @return string the formatted string
581		*/
582		public static function format($query, array $options = array())
583		{
584		$lexer = new Lexer($query);
585		$formatter = new self($options);
586
587		return $formatter->formatList($lexer->list);
588	6	}
589
590		/**
591		* Computes the length of a group.
592		*
593		* A group is delimited by a pair of brackets.
594		*
595		* @param TokensList $list the list of tokens
596		*
597		* @return int
598	6	*/
599		public static function getGroupLength($list)
600		{
601		/**
602		* The number of opening brackets found.
603		* This counter starts at one because by the time this function called,
604		* the list already advanced one position and the opening bracket was
605	6	* already parsed.
606		*
607	6	* @var int
608		*/
609	6	$count = 1;
610	6
611	1	/**
612	6	* The length of this group.
613	6	*
614	6	* @var int
615	6	*/
616		$length = 0;
617
618		for ($idx = $list->idx; $idx < $list->count; ++$idx) {
619		// Counting the brackets.
620		if ($list->tokens[$idx]->type === Token::TYPE_OPERATOR) {
621	5	if ($list->tokens[$idx]->value === '(') {
622		++$count;
623		} elseif ($list->tokens[$idx]->value === ')') {
624	6	--$count;
625		if ($count == 0) {
626		break;
627		}
628		}
629		}
630
631		// Keeping track of this group's length.
632		$length += mb_strlen($list->tokens[$idx]->value, 'UTF-8');
633		}
634	18
635		return $length;
636		}
637	18
638	18	/**
639		* Checks if a token is a statement or a clause inside a statement.
640	17	*
641		* @param Token $token the token to be checked
642	18	*
643		* @return int\|bool
644	8	*/
645		public static function isClause($token)
646		{
647	18	if (
648		($token->type === Token::TYPE_KEYWORD && isset(Parser::$STATEMENT_PARSERS[$token->keyword]))
649		\|\| ($token->type === Token::TYPE_NONE && strtoupper($token->token) === 'DELIMITER')
650		) {
651		return 2;
652		} elseif (
653		$token->type === Token::TYPE_KEYWORD && isset(Parser::$KEYWORD_PARSERS[$token->keyword])
654		) {
655		return 1;
656		}
657
658		return false;
659		}
660		}
661

phpmyadmin / sql-parser

Pull Request — master (#153)

Formatter::toString() C

Complexity

Size

Duplication

Code Coverage

Importance

How to fix Complexity

Long Method

Duplication Side-by-Side

Filter issues like