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

Passed

Pull Request — master (#153)

unknown

created 2017-05-16 22:32 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	25
CRAP Score	11

Importance

Changes

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

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