Formatter::format() - Code Metrics - Inspection of "Better Formatter tests" - phpmyadmin/sql-parser - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Pull Request — master (#120)

unknown

created 2017-02-05 22:00 UTC

Formatter::format() A

↳ Parent: Formatter

Complexity

Conditions	1
Paths	1

Size

Total Lines	7
Code Lines	4

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	4
CRAP Score	1

Importance

Changes

Metric	Value
cc	1
eloc	4
nc	1
nop	2
dl	0
loc	7
ccs	4
cts	4
cp	1
crap	1
rs	9.4285
c	0
b	0
f	0

<?php

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

namespace PhpMyAdmin\SqlParser\Utils;

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 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";
        }

        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) {
                // 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']) {
                        $lineEnded = true;
                        if ($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 && $prev->value === ',') {
                    // Fragments delimited by a comma are broken into multiple
                    // pieces only if the clause is not inlined or this fragment
                    // is between brackets that are on new line.
                    if (end($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 ($lineEnded) {
                    if ($indent < 0) {
                        // TODO: Make sure this never occurs and delete it.

                        $indent = 0;
                    }

                    $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.
                        $prev->value === 'DELIMITER'
                        || !(
                            ($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;

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

phpmyadmin / sql-parser

Pull Request — master (#120)

Formatter::format() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like