Token::extract() - Code Metrics - Inspection of "Apply php-cs-fixer" - phpmyadmin/sql-parser - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — master ( 6e1435...eac7d2 )

by Michal

created 2017-05-05 12:02 UTC

Token::extract() D

↳ Parent: Token

Complexity

Conditions	20
Paths	14

Size

Total Lines	81
Code Lines	50

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	46
CRAP Score	20

Importance

Changes	1
Bugs	0	Features	0

Metric	Value
c	1
b	0
f	0
dl	0
loc	81
ccs	46
cts	46
cp	1
rs	4.9645
cc	20
eloc	50
nc	14
nop	0
crap	20

How to fix Long Method Complexity

<?php

/**
 * Defines a token along with a set of types and flags and utility functions.
 *
 * An array of tokens will result after parsing the query.
 */

namespace PhpMyAdmin\SqlParser;

/**
 * A structure representing a lexeme that explicitly indicates its
 * categorization for the purpose of parsing.
 *
 * @category Tokens
 *
 * @license  https://www.gnu.org/licenses/gpl-2.0.txt GPL-2.0+
 */
class Token
{
    // Types of tokens (a vague description of a token's purpose).

    /**
     * This type is used when the token is invalid or its type cannot be
     * determined because of the ambiguous context. Further analysis might be
     * required to detect its type.
     *
     * @var int
     */
    const TYPE_NONE = 0;

    /**
     * SQL specific keywords: SELECT, UPDATE, INSERT, etc.
     *
     * @var int
     */
    const TYPE_KEYWORD = 1;

    /**
     * Any type of legal operator.
     *
     * Arithmetic operators: +, -, *, /, etc.
     * Logical operators: ===, <>, !==, etc.
     * Bitwise operators: &, |, ^, etc.
     * Assignment operators: =, +=, -=, etc.
     * SQL specific operators: . (e.g. .. WHERE database.table ..),
     *                         * (e.g. SELECT * FROM ..)
     *
     * @var int
     */
    const TYPE_OPERATOR = 2;

    /**
     * Spaces, tabs, new lines, etc.
     *
     * @var int
     */
    const TYPE_WHITESPACE = 3;

    /**
     * Any type of legal comment.
     *
     * Bash (#), C (/* *\/) or SQL (--) comments:
     *
     *      -- SQL-comment
     *
     *      #Bash-like comment
     *
     *      /*C-like comment*\/
     *
     * or:
     *
     *      /*C-like
     *        comment*\/
     *
     * Backslashes were added to respect PHP's comments syntax.
     *
     * @var int
     */
    const TYPE_COMMENT = 4;

    /**
     * Boolean values: true or false.
     *
     * @var int
     */
    const TYPE_BOOL = 5;

    /**
     * Numbers: 4, 0x8, 15.16, 23e42, etc.
     *
     * @var int
     */
    const TYPE_NUMBER = 6;

    /**
     * Literal strings: 'string', "test".
     * Some of these strings are actually symbols.
     *
     * @var int
     */
    const TYPE_STRING = 7;

    /**
     * Database, table names, variables, etc.
     * For example: ```SELECT `foo`, `bar` FROM `database`.`table`;```.
     *
     * @var int
     */
    const TYPE_SYMBOL = 8;

    /**
     * Delimits an unknown string.
     * For example: ```SELECT * FROM test;```, `test` is a delimiter.
     *
     * @var int
     */
    const TYPE_DELIMITER = 9;

    /**
     * Labels in LOOP statement, ITERATE statement etc.
     * For example (only for begin label):
     *  begin_label: BEGIN [statement_list] END [end_label]
     *  begin_label: LOOP [statement_list] END LOOP [end_label]
     *  begin_label: REPEAT [statement_list] ... END REPEAT [end_label]
     *  begin_label: WHILE ... DO [statement_list] END WHILE [end_label].
     *
     * @var int
     */
    const TYPE_LABEL = 10;

    // Flags that describe the tokens in more detail.
    // All keywords must have flag 1 so `Context::isKeyword` method doesn't
    // require strict comparison.
    const FLAG_KEYWORD_RESERVED = 2;
    const FLAG_KEYWORD_COMPOSED = 4;
    const FLAG_KEYWORD_DATA_TYPE = 8;
    const FLAG_KEYWORD_KEY = 16;
    const FLAG_KEYWORD_FUNCTION = 32;

    // Numbers related flags.
    const FLAG_NUMBER_HEX = 1;
    const FLAG_NUMBER_FLOAT = 2;
    const FLAG_NUMBER_APPROXIMATE = 4;
    const FLAG_NUMBER_NEGATIVE = 8;
    const FLAG_NUMBER_BINARY = 16;

    // Strings related flags.
    const FLAG_STRING_SINGLE_QUOTES = 1;
    const FLAG_STRING_DOUBLE_QUOTES = 2;

    // Comments related flags.
    const FLAG_COMMENT_BASH = 1;
    const FLAG_COMMENT_C = 2;
    const FLAG_COMMENT_SQL = 4;
    const FLAG_COMMENT_MYSQL_CMD = 8;

    // Operators related flags.
    const FLAG_OPERATOR_ARITHMETIC = 1;
    const FLAG_OPERATOR_LOGICAL = 2;
    const FLAG_OPERATOR_BITWISE = 4;
    const FLAG_OPERATOR_ASSIGNMENT = 8;
    const FLAG_OPERATOR_SQL = 16;

    // Symbols related flags.
    const FLAG_SYMBOL_VARIABLE = 1;
    const FLAG_SYMBOL_BACKTICK = 2;
    const FLAG_SYMBOL_USER = 4;
    const FLAG_SYMBOL_SYSTEM = 8;

    /**
     * The token it its raw string representation.
     *
     * @var string
     */
    public $token;

    /**
     * The value this token contains (i.e. token after some evaluation).
     *
     * @var mixed
     */
    public $value;

    /**
     * The keyword value this token contains, always uppercase.
     *
     * @var mixed
     */
    public $keyword;

    /**
     * The type of this token.
     *
     * @var int
     */
    public $type;

    /**
     * The flags of this token.
     *
     * @var int
     */
    public $flags;

    /**
     * The position in the initial string where this token started.
     *
     * @var int
     */
    public $position;

    /**
     * Constructor.
     *
     * @param string $token the value of the token
     * @param int    $type  the type of the token
     * @param int    $flags the flags of the token
     */
    public function __construct($token, $type = 0, $flags = 0)
    {
        $this->token = $token;
        $this->type = $type;
        $this->flags = $flags;
        $this->keyword = null;
        $this->value = $this->extract();
    }

    /**
     * Does little processing to the token to extract a value.
     *
     * If no processing can be done it will return the initial string.
     *
     * @return mixed
     */
    public function extract()
    {
        switch ($this->type) {
            case self::TYPE_KEYWORD:
                $this->keyword = strtoupper($this->token);
                if (!($this->flags & self::FLAG_KEYWORD_RESERVED)) {
                    // Unreserved keywords should stay the way they are because they
                    // might represent field names.
                    return $this->token;
                }

                return $this->keyword;
            case self::TYPE_WHITESPACE:
                return ' ';
            case self::TYPE_BOOL:
                return strtoupper($this->token) === 'TRUE';
            case self::TYPE_NUMBER:
                $ret = str_replace('--', '', $this->token); // e.g. ---42 === -42
                if ($this->flags & self::FLAG_NUMBER_HEX) {
                    if ($this->flags & self::FLAG_NUMBER_NEGATIVE) {
                        $ret = str_replace('-', '', $this->token);
                        sscanf($ret, '%x', $ret);
                        $ret = -$ret;
                    } else {
                        sscanf($ret, '%x', $ret);
                    }
                } elseif (($this->flags & self::FLAG_NUMBER_APPROXIMATE)
                || ($this->flags & self::FLAG_NUMBER_FLOAT)
                ) {
                    sscanf($ret, '%f', $ret);
                } else {
                    sscanf($ret, '%d', $ret);
                }

                return $ret;
            case self::TYPE_STRING:
                // Trims quotes.
                $str = $this->token;
                $str = mb_substr($str, 1, -1, 'UTF-8');

                // Removes surrounding quotes.
                $quote = $this->token[0];
                $str = str_replace($quote . $quote, $quote, $str);

                // Finally unescapes the string.
                //
                // `stripcslashes` replaces escape sequences with their
                // representation.
                //
                // NOTE: In MySQL, `\f` and `\v` have no representation,
                // even they usually represent: form-feed and vertical tab.
                $str = str_replace('\f', 'f', $str);
                $str = str_replace('\v', 'v', $str);
                $str = stripcslashes($str);

                return $str;
            case self::TYPE_SYMBOL:
                $str = $this->token;
                if ((isset($str[0])) && ($str[0] === '@')) {
                    // `mb_strlen($str)` must be used instead of `null` because
                    // in PHP 5.3- the `null` parameter isn't handled correctly.
                    $str = mb_substr(
                        $str,
                        ((!empty($str[1])) && ($str[1] === '@')) ? 2 : 1,
                        mb_strlen($str),
                        'UTF-8'
                    );
                }
                if ((isset($str[0])) && (($str[0] === '`')
                || ($str[0] === '"') || ($str[0] === '\''))
                ) {
                    $quote = $str[0];
                    $str = str_replace($quote . $quote, $quote, $str);
                    $str = mb_substr($str, 1, -1, 'UTF-8');
                }

                return $str;
        }

        return $this->token;
    }

    /**
     * Converts the token into an inline token by replacing tabs and new lines.
     *
     * @return string
     */
    public function getInlineToken()
    {
        return str_replace(
            array("\r", "\n", "\t"),
            array('\r', '\n', '\t'),
            $this->token
        );
    }
}


1		<?php
2
3		/**
4		* Defines a token along with a set of types and flags and utility functions.
5		*
6		* An array of tokens will result after parsing the query.
7		*/
8
9		namespace PhpMyAdmin\SqlParser;
10
11		/**
12		* A structure representing a lexeme that explicitly indicates its
13		* categorization for the purpose of parsing.
14		*
15		* @category Tokens
16		*
17		* @license https://www.gnu.org/licenses/gpl-2.0.txt GPL-2.0+
18		*/
19		class Token
20		{
21		// Types of tokens (a vague description of a token's purpose).
22
23		/**
24		* This type is used when the token is invalid or its type cannot be
25		* determined because of the ambiguous context. Further analysis might be
26		* required to detect its type.
27		*
28		* @var int
29		*/
30		const TYPE_NONE = 0;
31
32		/**
33		* SQL specific keywords: SELECT, UPDATE, INSERT, etc.
34		*
35		* @var int
36		*/
37		const TYPE_KEYWORD = 1;
38
39		/**
40		* Any type of legal operator.
41		*
42		* Arithmetic operators: +, -, *, /, etc.
43		* Logical operators: ===, <>, !==, etc.
44		* Bitwise operators: &, \|, ^, etc.
45		* Assignment operators: =, +=, -=, etc.
46		* SQL specific operators: . (e.g. .. WHERE database.table ..),
47		* * (e.g. SELECT * FROM ..)
48		*
49		* @var int
50		*/
51		const TYPE_OPERATOR = 2;
52
53		/**
54		* Spaces, tabs, new lines, etc.
55		*
56		* @var int
57		*/
58		const TYPE_WHITESPACE = 3;
59
60		/**
61		* Any type of legal comment.
62		*
63		* Bash (#), C (/* *\/) or SQL (--) comments:
64		*
65		* -- SQL-comment
66		*
67		* #Bash-like comment
68		*
69		* /C-like comment\/
70		*
71		* or:
72		*
73		* /*C-like
74		* comment*\/
75		*
76		* Backslashes were added to respect PHP's comments syntax.
77		*
78		* @var int
79		*/
80		const TYPE_COMMENT = 4;
81
82		/**
83		* Boolean values: true or false.
84		*
85		* @var int
86		*/
87		const TYPE_BOOL = 5;
88
89		/**
90		* Numbers: 4, 0x8, 15.16, 23e42, etc.
91		*
92		* @var int
93		*/
94		const TYPE_NUMBER = 6;
95
96		/**
97		* Literal strings: 'string', "test".
98		* Some of these strings are actually symbols.
99		*
100		* @var int
101		*/
102		const TYPE_STRING = 7;
103
104		/**
105		* Database, table names, variables, etc.
106		* For example: ```SELECT `foo`, `bar` FROM `database`.`table`;```.
107		*
108		* @var int
109		*/
110		const TYPE_SYMBOL = 8;
111
112		/**
113		* Delimits an unknown string.
114		* For example: ```SELECT * FROM test;```, `test` is a delimiter.
115		*
116		* @var int
117		*/
118		const TYPE_DELIMITER = 9;
119
120		/**
121		* Labels in LOOP statement, ITERATE statement etc.
122		* For example (only for begin label):
123		* begin_label: BEGIN [statement_list] END [end_label]
124		* begin_label: LOOP [statement_list] END LOOP [end_label]
125		* begin_label: REPEAT [statement_list] ... END REPEAT [end_label]
126		* begin_label: WHILE ... DO [statement_list] END WHILE [end_label].
127		*
128		* @var int
129		*/
130		const TYPE_LABEL = 10;
131
132		// Flags that describe the tokens in more detail.
133		// All keywords must have flag 1 so `Context::isKeyword` method doesn't
134		// require strict comparison.
135		const FLAG_KEYWORD_RESERVED = 2;
136		const FLAG_KEYWORD_COMPOSED = 4;
137		const FLAG_KEYWORD_DATA_TYPE = 8;
138		const FLAG_KEYWORD_KEY = 16;
139		const FLAG_KEYWORD_FUNCTION = 32;
140
141		// Numbers related flags.
142		const FLAG_NUMBER_HEX = 1;
143		const FLAG_NUMBER_FLOAT = 2;
144		const FLAG_NUMBER_APPROXIMATE = 4;
145		const FLAG_NUMBER_NEGATIVE = 8;
146		const FLAG_NUMBER_BINARY = 16;
147
148		// Strings related flags.
149		const FLAG_STRING_SINGLE_QUOTES = 1;
150		const FLAG_STRING_DOUBLE_QUOTES = 2;
151
152		// Comments related flags.
153		const FLAG_COMMENT_BASH = 1;
154		const FLAG_COMMENT_C = 2;
155		const FLAG_COMMENT_SQL = 4;
156		const FLAG_COMMENT_MYSQL_CMD = 8;
157
158		// Operators related flags.
159		const FLAG_OPERATOR_ARITHMETIC = 1;
160		const FLAG_OPERATOR_LOGICAL = 2;
161		const FLAG_OPERATOR_BITWISE = 4;
162		const FLAG_OPERATOR_ASSIGNMENT = 8;
163		const FLAG_OPERATOR_SQL = 16;
164
165		// Symbols related flags.
166		const FLAG_SYMBOL_VARIABLE = 1;
167		const FLAG_SYMBOL_BACKTICK = 2;
168		const FLAG_SYMBOL_USER = 4;
169		const FLAG_SYMBOL_SYSTEM = 8;
170
171		/**
172		* The token it its raw string representation.
173		*
174		* @var string
175		*/
176		public $token;
177
178		/**
179		* The value this token contains (i.e. token after some evaluation).
180		*
181		* @var mixed
182		*/
183		public $value;
184
185		/**
186		* The keyword value this token contains, always uppercase.
187		*
188		* @var mixed
189		*/
190		public $keyword;
191
192		/**
193		* The type of this token.
194		*
195		* @var int
196		*/
197		public $type;
198
199		/**
200		* The flags of this token.
201		*
202		* @var int
203		*/
204		public $flags;
205
206		/**
207		* The position in the initial string where this token started.
208		*
209		* @var int
210		*/
211		public $position;
212
213		/**
214		* Constructor.
215		*
216		* @param string $token the value of the token
217		* @param int $type the type of the token
218		* @param int $flags the flags of the token
219		*/
220	381	public function __construct($token, $type = 0, $flags = 0)
221		{
222	381	$this->token = $token;
223	381	$this->type = $type;
224	381	$this->flags = $flags;
225	381	$this->keyword = null;
226	381	$this->value = $this->extract();
227	381	}
228
229		/**
230		* Does little processing to the token to extract a value.
231		*
232		* If no processing can be done it will return the initial string.
233		*
234		* @return mixed
235		*/
236	381	public function extract()
237		{
238	381	switch ($this->type) {
239	381	case self::TYPE_KEYWORD:
240	342	$this->keyword = strtoupper($this->token);
241	342	if (!($this->flags & self::FLAG_KEYWORD_RESERVED)) {
242		// Unreserved keywords should stay the way they are because they
243		// might represent field names.
244	155	return $this->token;
245		}
246
247	335	return $this->keyword;
248	380	case self::TYPE_WHITESPACE:
249	354	return ' ';
250	379	case self::TYPE_BOOL:
251	2	return strtoupper($this->token) === 'TRUE';
252	378	case self::TYPE_NUMBER:
253	174	$ret = str_replace('--', '', $this->token); // e.g. ---42 === -42
254	174	if ($this->flags & self::FLAG_NUMBER_HEX) {
255	2	if ($this->flags & self::FLAG_NUMBER_NEGATIVE) {
256	1	$ret = str_replace('-', '', $this->token);
257	1	sscanf($ret, '%x', $ret);
258	1	$ret = -$ret;
259		} else {
260	2	sscanf($ret, '%x', $ret);
261		}
262	174	} elseif (($this->flags & self::FLAG_NUMBER_APPROXIMATE)
263	174	\|\| ($this->flags & self::FLAG_NUMBER_FLOAT)
264		) {
265	5	sscanf($ret, '%f', $ret);
266		} else {
267	174	sscanf($ret, '%d', $ret);
268		}
269
270	174	return $ret;
271	377	case self::TYPE_STRING:
272		// Trims quotes.
273	157	$str = $this->token;
274	157	$str = mb_substr($str, 1, -1, 'UTF-8');
275
276		// Removes surrounding quotes.
277	157	$quote = $this->token[0];
278	157	$str = str_replace($quote . $quote, $quote, $str);
279
280		// Finally unescapes the string.
281		//
282		// `stripcslashes` replaces escape sequences with their
283		// representation.
284		//
285		// NOTE: In MySQL, `\f` and `\v` have no representation,
286		// even they usually represent: form-feed and vertical tab.
287	157	$str = str_replace('\f', 'f', $str);
288	157	$str = str_replace('\v', 'v', $str);
289	157	$str = stripcslashes($str);
290
291	157	return $str;
292	376	case self::TYPE_SYMBOL:
293	98	$str = $this->token;
294	98	if ((isset($str[0])) && ($str[0] === '@')) {
295		// `mb_strlen($str)` must be used instead of `null` because
296		// in PHP 5.3- the `null` parameter isn't handled correctly.
297	29	$str = mb_substr(
298		$str,
299	29	((!empty($str[1])) && ($str[1] === '@')) ? 2 : 1,
300		mb_strlen($str),
301	29	'UTF-8'
302		);
303		}
304	98	if ((isset($str[0])) && (($str[0] === '`')
305	98	\|\| ($str[0] === '"') \|\| ($str[0] === '\''))
306		) {
307	82	$quote = $str[0];
308	82	$str = str_replace($quote . $quote, $quote, $str);
309	82	$str = mb_substr($str, 1, -1, 'UTF-8');
310		}
311
312	98	return $str;
313		}
314
315	375	return $this->token;
316		}
317
318		/**
319		* Converts the token into an inline token by replacing tabs and new lines.
320		*
321		* @return string
322		*/
323	1	public function getInlineToken()
324		{
325	1	return str_replace(
326	1	array("\r", "\n", "\t"),
327	1	array('\r', '\n', '\t'),
328	1	$this->token
329		);
330		}
331		}
332

phpmyadmin / sql-parser

Push — master ( 6e1435...eac7d2 )

Token::extract() D

Complexity

Size

Duplication

Code Coverage

Importance

How to fix Long Method Complexity

Long Method

Duplication Side-by-Side

Filter issues like