BufferedQuery::setDelimiter() - Code Metrics - Inspection of "Fixes striping of inline comments #11717" - phpmyadmin/sql-parser - Measure and Improve Code Quality continuously with Scrutinizer

Test Failed

Pull Request — master (#146)

by Raghuram

created 2017-03-31 17:06 UTC

BufferedQuery::setDelimiter() A

↳ Parent: BufferedQuery

Complexity

Conditions	1
Paths	1

Size

Total Lines	5
Code Lines	3

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	4
CRAP Score	1

Importance

Changes

Metric	Value
cc	1
eloc	3
nc	1
nop	1
dl	0
loc	5
ccs	4
cts	4
cp	1
crap	1
rs	9.4285
c	0
b	0
f	0

<?php

/**
 * Buffered query utilities.
 */

namespace PhpMyAdmin\SqlParser\Utils;

use PhpMyAdmin\SqlParser\Context;

/**
 * Buffer query utilities.
 *
 * Implements a specialized lexer used to extract statements from large inputs
 * that are being buffered. After each statement has been extracted, a lexer or
 * a parser may be used.
 *
 * All comments are skipped, with one exception: MySQL commands inside `/*!`.
 *
 * @category   Lexer
 *
 * @license    https://www.gnu.org/licenses/gpl-2.0.txt GPL-2.0+
 */
class BufferedQuery
{
    // Constants that describe the current status of the parser.

    // A string is being parsed.
    const STATUS_STRING = 16; // 0001 0000
    const STATUS_STRING_SINGLE_QUOTES = 17; // 0001 0001
    const STATUS_STRING_DOUBLE_QUOTES = 18; // 0001 0010
    const STATUS_STRING_BACKTICK = 20; // 0001 0100

    // A comment is being parsed.
    const STATUS_COMMENT = 32; // 0010 0000
    const STATUS_COMMENT_BASH = 33; // 0010 0001
    const STATUS_COMMENT_C = 34; // 0010 0010
    const STATUS_COMMENT_SQL = 36; // 0010 0100

    /**
     * The query that is being processed.
     *
     * This field can be modified just by appending to it!
     *
     * @var string
     */
    public $query = '';

    /**
     * The options of this parser.
     *
     * @var array
     */
    public $options = array();

    /**
     * The last delimiter used.
     *
     * @var string
     */
    public $delimiter;

    /**
     * The length of the delimiter.
     *
     * @var int
     */
    public $delimiterLen;

    /**
     * The current status of the parser.
     *
     * @var int
     */
    public $status;

    /**
     * The last incomplete query that was extracted.
     *
     * @var string
     */
    public $current = '';

    /**
     * Constructor.
     *
     * @param string $query   the query to be parsed
     * @param array  $options the options of this parser
     */
    public function __construct($query = '', array $options = array())
    {
        // Merges specified options with defaults.
        $this->options = array_merge(
            array(
                /*
                 * The starting delimiter.
                 *
                 * @var string
                 */
                'delimiter' => ';',

                /*
                 * Whether `DELIMITER` statements should be parsed.
                 *
                 * @var bool
                 */
                'parse_delimiter' => false,

                /*
                 * Whether a delimiter should be added at the end of the
                 * statement.
                 *
                 * @var bool
                 */
                'add_delimiter' => false,
            ),
            $options
        );

        $this->query = $query;
        $this->setDelimiter($this->options['delimiter']);
    }

    /**
     * Sets the delimiter.
     *
     * Used to update the length of it too.
     *
     * @param string $delimiter
     */
    public function setDelimiter($delimiter)
    {
        $this->delimiter = $delimiter;
        $this->delimiterLen = strlen($delimiter);
    }

    /**
     * Extracts a statement from the buffer.
     *
     * @param bool $end whether the end of the buffer was reached
     *
     * @return string

     */
    public function extract($end = false)
    {
        /**
         * The last parsed position.
         *
         * This is statically defined because it is not used outside anywhere
         * outside this method and there is probably a (minor) performance
         * improvement to it.
         *
         * @var int
         */
        static $i = 0;

        if (empty($this->query)) {
            return false;
        }

        /**
         * The length of the buffer.
         *
         * @var int
         */
        $len = strlen($this->query);

        /**
         * The last index of the string that is going to be parsed.
         *
         * There must be a few characters left in the buffer so the parser can
         * avoid confusing some symbols that may have multiple meanings.
         *
         * For example, if the buffer ends in `-` that may be an operator or the
         * beginning of a comment.
         *
         * Another example if the buffer ends in `DELIMITE`. The parser is going
         * to require a few more characters because that may be a part of the
         * `DELIMITER` keyword or just a column named `DELIMITE`.
         *
         * Those extra characters are required only if there is more data
         * expected (the end of the buffer was not reached).
         *
         * @var int
         */
        $loopLen = $end ? $len : $len - 16;

        for (; $i < $loopLen; ++$i) {
            /*
             * Handling backslash.
             *
             * Even if the next character is a special character that should be
             * treated differently, because of the preceding backslash, it will
             * be ignored.
             */
            if ((($this->status & static::STATUS_COMMENT) == 0) && ($this->query[$i] === '\\')) {
                $this->current .= $this->query[$i] . $this->query[++$i];
                continue;
            }

            /*
             * Handling special parses statuses.
             */
            if ($this->status === static::STATUS_STRING_SINGLE_QUOTES) {
                // Single-quoted strings like 'foo'.
                if ($this->query[$i] === '\'') {
                    $this->status = 0;
                }
                $this->current .= $this->query[$i];
                continue;
            } elseif ($this->status === static::STATUS_STRING_DOUBLE_QUOTES) {
                // Double-quoted strings like "bar".
                if ($this->query[$i] === '"') {
                    $this->status = 0;
                }
                $this->current .= $this->query[$i];
                continue;
            } elseif ($this->status === static::STATUS_STRING_BACKTICK) {
                if ($this->query[$i] === '`') {
                    $this->status = 0;
                }
                $this->current .= $this->query[$i];
                continue;
            } elseif (($this->status === static::STATUS_COMMENT_BASH)
                || ($this->status === static::STATUS_COMMENT_SQL)
            ) {
                // Bash-like (#) or SQL-like (-- ) comments end in new line.
                if ($this->query[$i] === "\n") {
                    $this->status = 0;
                }
            } elseif ($this->status === static::STATUS_COMMENT_C) {
                // C-like comments end in */.
                if (($this->query[$i - 1] === '*') && ($this->query[$i] === '/')) {
                    $this->status = 0;
                }
            }

            /*
             * Checking if a string started.
             */
            if ($this->query[$i] === '\'') {
                $this->status = static::STATUS_STRING_SINGLE_QUOTES;
                $this->current .= $this->query[$i];
                continue;
            } elseif ($this->query[$i] === '"') {
                $this->status = static::STATUS_STRING_DOUBLE_QUOTES;
                $this->current .= $this->query[$i];
                continue;
            } elseif ($this->query[$i] === '`') {
                $this->status = static::STATUS_STRING_BACKTICK;
                $this->current .= $this->query[$i];
                continue;
            }

            /*
             * Checking if a comment started.
             */
            if ($this->query[$i] === '#') {
                $this->status = static::STATUS_COMMENT_BASH;
            } elseif (($i + 2 < $len)
                && ($this->query[$i] === '-')
                && ($this->query[$i + 1] === '-')
                && (Context::isWhitespace($this->query[$i + 2]))
            ) {
                $this->status = static::STATUS_COMMENT_SQL;
            } elseif (($i + 2 < $len)
                && ($this->query[$i] === '/')
                && ($this->query[$i + 1] === '*')
                && ($this->query[$i + 2] !== '!')
            ) {
                $this->status = static::STATUS_COMMENT_C;
            }

            /*
             * Handling `DELIMITER` statement.
             *
             * The code below basically checks for
             *     `strtoupper(substr($this->query, $i, 9)) === 'DELIMITER'`
             *
             * This optimization makes the code about 3 times faster.
             *
             * `DELIMITER` is not being considered a keyword. The only context
             * it has a special meaning is when it is the beginning of a
             * statement. This is the reason for the last condition.
             */
            if (($i + 9 < $len)
                && (($this->query[$i] === 'D') || ($this->query[$i] === 'd'))
                && (($this->query[$i + 1] === 'E') || ($this->query[$i + 1] === 'e'))
                && (($this->query[$i + 2] === 'L') || ($this->query[$i + 2] === 'l'))
                && (($this->query[$i + 3] === 'I') || ($this->query[$i + 3] === 'i'))
                && (($this->query[$i + 4] === 'M') || ($this->query[$i + 4] === 'm'))
                && (($this->query[$i + 5] === 'I') || ($this->query[$i + 5] === 'i'))
                && (($this->query[$i + 6] === 'T') || ($this->query[$i + 6] === 't'))
                && (($this->query[$i + 7] === 'E') || ($this->query[$i + 7] === 'e'))
                && (($this->query[$i + 8] === 'R') || ($this->query[$i + 8] === 'r'))
                && (Context::isWhitespace($this->query[$i + 9]))
                && (trim($this->current) === '')
            ) {
                // Saving the current index to be able to revert any parsing
                // done in this block.
                $iBak = $i;
                $i += 9; // Skipping `DELIMITER`.

                // Skipping whitespaces.
                while (($i < $len) && (Context::isWhitespace($this->query[$i]))) {
                    ++$i;
                }

                // Parsing the delimiter.
                $delimiter = '';
                while (($i < $len) && (!Context::isWhitespace($this->query[$i]))) {
                    $delimiter .= $this->query[$i++];
                }

                // Checking if the delimiter definition ended.
                if (($delimiter != '')
                    && ((($i < $len) && (Context::isWhitespace($this->query[$i])))
                    || (($i === $len) && ($end)))
                ) {
                    // Saving the delimiter.
                    $this->setDelimiter($delimiter);

                    // Whether this statement should be returned or not.
                    $ret = '';
                    if (!empty($this->options['parse_delimiter'])) {
                        // Appending the `DELIMITER` statement that was just
                        // found to the current statement.
                        $ret = trim(
                            $this->current . ' ' . substr($this->query, $iBak, $i - $iBak)
                        );
                    }

                    // Removing the statement that was just extracted from the
                    // query.
                    $this->query = substr($this->query, $i);
                    $i = 0;

                    // Resetting the current statement.
                    $this->current = '';

                    return $ret;
                }

                // Incomplete statement. Reverting
                $i = $iBak;

                return false;
            }

            /*
             * Checking if the current statement finished.
             *
             * The first letter of the delimiter is being checked as an
             * optimization. This code is almost as fast as the one above.
             *
             * There is no point in checking if two strings match if not even
             * the first letter matches.
             */
            if (($this->query[$i] === $this->delimiter[0])
                && (($this->delimiterLen === 1)
                || (substr($this->query, $i, $this->delimiterLen) === $this->delimiter))
            ) {
                // Saving the statement that just ended.
                $ret = $this->current;

                // If needed, adds a delimiter at the end of the statement.
                if (!empty($this->options['add_delimiter'])) {
                    $ret .= $this->delimiter;
                }

                // Removing the statement that was just extracted from the
                // query.
                $this->query = substr($this->query, $i + $this->delimiterLen);
                $i = 0;

                // Resetting the current statement.
                $this->current = '';

                // Returning the statement.
                return trim($ret);
            }

            /*
             * Appending current character to current statement.
             */
            $this->current .= $this->query[$i];
        }

        if (($end) && ($i === $len)) {
            // If the end of the buffer was reached, the buffer is emptied and
            // the current statement that was extracted is returned.
            $ret = $this->current;

            // Emptying the buffer.
            $this->query = '';
            $i = 0;

            // Resetting the current statement.
            $this->current = '';

            // Returning the statement.
            return trim($ret);
        }

        return '';
    }
}


1			<?php
2
3			/**
4			* Buffered query utilities.
5			*/
6
7			namespace PhpMyAdmin\SqlParser\Utils;
8
9			use PhpMyAdmin\SqlParser\Context;
10
11			/**
12			* Buffer query utilities.
13			*
14			* Implements a specialized lexer used to extract statements from large inputs
15			* that are being buffered. After each statement has been extracted, a lexer or
16			* a parser may be used.
17			*
18			* All comments are skipped, with one exception: MySQL commands inside `/*!`.
19			*
20			* @category Lexer
21			*
22			* @license https://www.gnu.org/licenses/gpl-2.0.txt GPL-2.0+
23			*/
24			class BufferedQuery
25			{
26			// Constants that describe the current status of the parser.
27
28			// A string is being parsed.
29			const STATUS_STRING = 16; // 0001 0000
30			const STATUS_STRING_SINGLE_QUOTES = 17; // 0001 0001
31			const STATUS_STRING_DOUBLE_QUOTES = 18; // 0001 0010
32			const STATUS_STRING_BACKTICK = 20; // 0001 0100
33
34			// A comment is being parsed.
35			const STATUS_COMMENT = 32; // 0010 0000
36			const STATUS_COMMENT_BASH = 33; // 0010 0001
37			const STATUS_COMMENT_C = 34; // 0010 0010
38			const STATUS_COMMENT_SQL = 36; // 0010 0100
39
40			/**
41			* The query that is being processed.
42			*
43			* This field can be modified just by appending to it!
44			*
45			* @var string
46			*/
47			public $query = '';
48
49			/**
50			* The options of this parser.
51			*
52			* @var array
53			*/
54			public $options = array();
55
56			/**
57			* The last delimiter used.
58			*
59			* @var string
60			*/
61			public $delimiter;
62
63			/**
64			* The length of the delimiter.
65			*
66			* @var int
67			*/
68			public $delimiterLen;
69
70			/**
71			* The current status of the parser.
72			*
73			* @var int
74			*/
75			public $status;
76
77			/**
78			* The last incomplete query that was extracted.
79			*
80			* @var string
81			*/
82			public $current = '';
83
84			/**
85			* Constructor.
86			*
87			* @param string $query the query to be parsed
88			* @param array $options the options of this parser
89			*/
90	7		public function __construct($query = '', array $options = array())
91			{
92			// Merges specified options with defaults.
93	7		$this->options = array_merge(
94			array(
95			/*
96			* The starting delimiter.
97			*
98			* @var string
99			*/
100	7		'delimiter' => ';',
101
102			/*
103			* Whether `DELIMITER` statements should be parsed.
104			*
105			* @var bool
106			*/
107			'parse_delimiter' => false,
108
109			/*
110			* Whether a delimiter should be added at the end of the
111			* statement.
112			*
113			* @var bool
114			*/
115			'add_delimiter' => false,
116			),
117			$options
118			);
119
120	7		$this->query = $query;
121	7		$this->setDelimiter($this->options['delimiter']);
122	7		}
123
124			/**
125			* Sets the delimiter.
126			*
127			* Used to update the length of it too.
128			*
129			* @param string $delimiter
130			*/
131	7		public function setDelimiter($delimiter)
132			{
133	7		$this->delimiter = $delimiter;
134	7		$this->delimiterLen = strlen($delimiter);
135	7		}
136
137			/**
138			* Extracts a statement from the buffer.
139			*
140			* @param bool $end whether the end of the buffer was reached
141			*
142			* @return string
			0 ignored issues – show Documentation introduced 2016-02-12 21:51 UTC by Report Bug Copy Issue Report Should the return type not be `false\|string`? This check compares the return type specified in the `@return` annotation of a function or method doc comment with the types returned by the function and raises an issue if they mismatch. Loading history...
143			*/
144	7		public function extract($end = false)
145			{
146			/**
147			* The last parsed position.
148			*
149			* This is statically defined because it is not used outside anywhere
150			* outside this method and there is probably a (minor) performance
151			* improvement to it.
152			*
153			* @var int
154			*/
155	7		static $i = 0;
156
157	7		if (empty($this->query)) {
158	7		return false;
159			}
160
161			/**
162			* The length of the buffer.
163			*
164			* @var int
165			*/
166	7		$len = strlen($this->query);
167
168			/**
169			* The last index of the string that is going to be parsed.
170			*
171			* There must be a few characters left in the buffer so the parser can
172			* avoid confusing some symbols that may have multiple meanings.
173			*
174			* For example, if the buffer ends in `-` that may be an operator or the
175			* beginning of a comment.
176			*
177			* Another example if the buffer ends in `DELIMITE`. The parser is going
178			* to require a few more characters because that may be a part of the
179			* `DELIMITER` keyword or just a column named `DELIMITE`.
180			*
181			* Those extra characters are required only if there is more data
182			* expected (the end of the buffer was not reached).
183			*
184			* @var int
185			*/
186	7		$loopLen = $end ? $len : $len - 16;
187
188	7		for (; $i < $loopLen; ++$i) {
189			/*
190			* Handling backslash.
191			*
192			* Even if the next character is a special character that should be
193			* treated differently, because of the preceding backslash, it will
194			* be ignored.
195			*/
196	7		if ((($this->status & static::STATUS_COMMENT) == 0) && ($this->query[$i] === '\\')) {
197	3		$this->current .= $this->query[$i] . $this->query[++$i];
198	3		continue;
199			}
200
201			/*
202			* Handling special parses statuses.
203			*/
204	7		if ($this->status === static::STATUS_STRING_SINGLE_QUOTES) {
205			// Single-quoted strings like 'foo'.
206	5		if ($this->query[$i] === '\'') {
207	5		$this->status = 0;
208			}
209	5		$this->current .= $this->query[$i];
210	5		continue;
211	7	View Code Duplication	} elseif ($this->status === static::STATUS_STRING_DOUBLE_QUOTES) {
212			// Double-quoted strings like "bar".
213	4		if ($this->query[$i] === '"') {
214	4		$this->status = 0;
215			}
216	4		$this->current .= $this->query[$i];
217	4		continue;
218	7		} elseif ($this->status === static::STATUS_STRING_BACKTICK) {
219	4		if ($this->query[$i] === '`') {
220	4		$this->status = 0;
221			}
222	4		$this->current .= $this->query[$i];
223	4		continue;
224	7		} elseif (($this->status === static::STATUS_COMMENT_BASH)
225	7		\|\| ($this->status === static::STATUS_COMMENT_SQL)
226			) {
227			// Bash-like (#) or SQL-like (-- ) comments end in new line.
228	3		if ($this->query[$i] === "\n") {
229	3		$this->status = 0;
230			}
231	3		} elseif ($this->status === static::STATUS_COMMENT_C) {
232	7		// C-like comments end in */.
233			if (($this->query[$i - 1] === '*') && ($this->query[$i] === '/')) {
234	3		$this->status = 0;
235	3		}
236			}
237	3
238			/*
239			* Checking if a string started.
240			*/
241			if ($this->query[$i] === '\'') {
242			$this->status = static::STATUS_STRING_SINGLE_QUOTES;
243	7		$this->current .= $this->query[$i];
244	5		continue;
245	5	View Code Duplication	} elseif ($this->query[$i] === '"') {
246	5		$this->status = static::STATUS_STRING_DOUBLE_QUOTES;
247	7		$this->current .= $this->query[$i];
248	4		continue;
249	4		} elseif ($this->query[$i] === '`') {
250	4		$this->status = static::STATUS_STRING_BACKTICK;
251	7		$this->current .= $this->query[$i];
252	4		continue;
253	4		}
254	4
255			/*
256			* Checking if a comment started.
257			*/
258			if ($this->query[$i] === '#') {
259			$this->status = static::STATUS_COMMENT_BASH;
260	7		} elseif (($i + 2 < $len)
261	3		&& ($this->query[$i] === '-')
262	3		&& ($this->query[$i + 1] === '-')
263	7		&& (Context::isWhitespace($this->query[$i + 2]))
264	7		) {
265	7		$this->status = static::STATUS_COMMENT_SQL;
266	7		} elseif (($i + 2 < $len)
267			&& ($this->query[$i] === '/')
268	3		&& ($this->query[$i + 1] === '*')
269	3		&& ($this->query[$i + 2] !== '!')
270	7		) {
271	7		$this->status = static::STATUS_COMMENT_C;
272	7		}
273	7
274			/*
275	3		* Handling `DELIMITER` statement.
276	3		*
277			* The code below basically checks for
278			* `strtoupper(substr($this->query, $i, 9)) === 'DELIMITER'`
279			*
280			* This optimization makes the code about 3 times faster.
281			*
282			* `DELIMITER` is not being considered a keyword. The only context
283			* it has a special meaning is when it is the beginning of a
284			* statement. This is the reason for the last condition.
285			*/
286			if (($i + 9 < $len)
287			&& (($this->query[$i] === 'D') \|\| ($this->query[$i] === 'd'))
288			&& (($this->query[$i + 1] === 'E') \|\| ($this->query[$i + 1] === 'e'))
289			&& (($this->query[$i + 2] === 'L') \|\| ($this->query[$i + 2] === 'l'))
290			&& (($this->query[$i + 3] === 'I') \|\| ($this->query[$i + 3] === 'i'))
291	7		&& (($this->query[$i + 4] === 'M') \|\| ($this->query[$i + 4] === 'm'))
292	7		&& (($this->query[$i + 5] === 'I') \|\| ($this->query[$i + 5] === 'i'))
293	7		&& (($this->query[$i + 6] === 'T') \|\| ($this->query[$i + 6] === 't'))
294	7		&& (($this->query[$i + 7] === 'E') \|\| ($this->query[$i + 7] === 'e'))
295	7		&& (($this->query[$i + 8] === 'R') \|\| ($this->query[$i + 8] === 'r'))
296	7		&& (Context::isWhitespace($this->query[$i + 9]))
297	7		&& (trim($this->current) === '')
298	7		) {
299	7		// Saving the current index to be able to revert any parsing
300	7		// done in this block.
301	7		$iBak = $i;
302	7		$i += 9; // Skipping `DELIMITER`.
303
304			// Skipping whitespaces.
305			while (($i < $len) && (Context::isWhitespace($this->query[$i]))) {
306	4		++$i;
307	4		}
308
309			// Parsing the delimiter.
310	4		$delimiter = '';
311	4		while (($i < $len) && (!Context::isWhitespace($this->query[$i]))) {
312			$delimiter .= $this->query[$i++];
313			}
314
315	4		// Checking if the delimiter definition ended.
316	4		if (($delimiter != '')
317	4		&& ((($i < $len) && (Context::isWhitespace($this->query[$i])))
318			\|\| (($i === $len) && ($end)))
319			) {
320			// Saving the delimiter.
321	4		$this->setDelimiter($delimiter);
322	4
323	4		// Whether this statement should be returned or not.
324			$ret = '';
325			if (!empty($this->options['parse_delimiter'])) {
326	4		// Appending the `DELIMITER` statement that was just
327			// found to the current statement.
328			$ret = trim(
329	4		$this->current . ' ' . substr($this->query, $iBak, $i - $iBak)
330	4		);
331			}
332
333	2		// Removing the statement that was just extracted from the
334	2		// query.
335			$this->query = substr($this->query, $i);
336			$i = 0;
337
338			// Resetting the current statement.
339			$this->current = '';
340	4
341	4		return $ret;
342			}
343
344	4		// Incomplete statement. Reverting
345			$i = $iBak;
346	4
347			return false;
348			}
349
350	1		/*
351			* Checking if the current statement finished.
352	1		*
353			* The first letter of the delimiter is being checked as an
354			* optimization. This code is almost as fast as the one above.
355			*
356			* There is no point in checking if two strings match if not even
357			* the first letter matches.
358			*/
359			if (($this->query[$i] === $this->delimiter[0])
360			&& (($this->delimiterLen === 1)
361			\|\| (substr($this->query, $i, $this->delimiterLen) === $this->delimiter))
362			) {
363			// Saving the statement that just ended.
364	7		$ret = $this->current;
365	7
366	7		// If needed, adds a delimiter at the end of the statement.
367			if (!empty($this->options['add_delimiter'])) {
368			$ret .= $this->delimiter;
369	7		}
370
371			// Removing the statement that was just extracted from the
372	7		// query.
373	5		$this->query = substr($this->query, $i + $this->delimiterLen);
374			$i = 0;
375
376			// Resetting the current statement.
377			$this->current = '';
378	7
379	7		// Returning the statement.
380			return trim($ret);
381			}
382	7
383			/*
384			* Appending current character to current statement.
385	7		*/
386			$this->current .= $this->query[$i];
387			}
388
389			if (($end) && ($i === $len)) {
390			// If the end of the buffer was reached, the buffer is emptied and
391	7		// the current statement that was extracted is returned.
392			$ret = $this->current;
393
394	7		// Emptying the buffer.
395			$this->query = '';
396			$i = 0;
397	5
398			// Resetting the current statement.
399			$this->current = '';
400	5
401	5		// Returning the statement.
402			return trim($ret);
403			}
404	5
405			return '';
406			}
407			}
408

phpmyadmin / sql-parser

Pull Request — master (#146)

BufferedQuery::setDelimiter() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like