TokenStream::readEscaped() - Code Metrics - koesie10/LuaSerializer - Measure and Improve Code Quality continuously with Scrutinizer

TokenStream::readEscaped() D
last analyzed 2016-10-20 19:27 UTC

↳ Parent: TokenStream

Complexity

Conditions	9
Paths	9

Size

Total Lines	39
Code Lines	33

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	33
CRAP Score	9

Importance

Changes

Metric	Value
dl	0
loc	39
ccs	33
cts	33
cp	1
rs	4.909
c	0
b	0
f	0
cc	9
eloc	33
nc	9
nop	1
crap	9

<?php

namespace Vlaswinkel\Lua;

/**
 * Class TokenStream
 *
 * @see     http://lisperator.net/pltut/parser/token-stream
 *
 * @author  Koen Vlaswinkel <[email protected]>
 * @package Vlaswinkel\Lua
 */
class TokenStream {
    private $current = null;
    /**
     * @var InputStream
     */
    private $input;

    /**
     * TokenStream constructor.
     *
     * @param InputStream $input
     */
    public function __construct(InputStream $input) {
        $this->input = $input;
    }

    /**
     * @return Token
     */
    public function next() {
        $token         = $this->current;
        $this->current = null;
        if ($token) {
            return $token;
        }
        return $this->readNext();
    }

    /**
     * @return bool
     */
    public function eof() {
        return $this->peek() == null;
    }

    /**
     * @return Token

     */
    public function peek() {
        if ($this->current) {
            return $this->current;
        }
        $this->current = $this->readNext();
        return $this->current;
    }

    /**
     * @param string $msg
     *
     * @throws ParseException
     */
    public function error($msg) {
        $this->input->error($msg);
    }

    /**
     * @return Token

     * @throws ParseException
     */
    protected function readNext() {
        $this->readWhile([$this, 'isWhitespace']);
        if ($this->input->eof()) {
            return null;
        }
        $char = $this->input->peek();
        if ($this->isComment()) {
            $this->skipComment();
            return $this->readNext();
        }
        if ($char == '"') {
            return $this->readDoubleQuotedString();
        }
        if ($char == '\'') {
            return $this->readSingleQuotedString();
        }
        if ($this->isDoubleBracketString()) {
            return $this->readDoubleBracketString();
        }
        if ($this->isDigit($char)) {
            return $this->readNumber();
        }
        if ($this->isStartIdentifierCharacter($char)) {
            return $this->readIdentifier();
        }
        if ($this->isPunctuation($char)) {
            return $this->readPunctuation();
        }
        if ($char == ';') {
            $this->input->next(); // skip the semi-colon
            return $this->readNext(); // just move on to the next
        }
        $this->input->error('Cannot handle character: ' . $char . ' (ord: ' . ord($char) . ')');
    }

    protected function skipComment() {
        $this->readWhile(
            function ($char) {
                return $char != "\n";
            }
        );
        if (!$this->input->eof()) {
            $this->input->next();
        }
    }

    /**
     * @return Token
     */
    protected function readDoubleQuotedString() {
        return new Token(Token::TYPE_STRING, $this->readEscaped('"'));
    }

    /**
     * @return Token
     */
    protected function readSingleQuotedString() {
        return new Token(Token::TYPE_STRING, $this->readEscaped('\''));
    }

    /**
     * @return Token
     */
    protected function readDoubleBracketString() {
        // we cannot use readEscaped because it only supports a single char as $end
        // and we do not support escaping in double bracke strings
        $str                      = "";
        $startNumberOfEqualsSigns = 0;
        // skip both
        $this->input->next();
        while ($this->input->peek() == '=') {
            $startNumberOfEqualsSigns++;
            $this->input->next();
        }
        if ($this->input->peek() != '[') {
            $this->error('Unexpected character \'' . $this->input->peek() . '\', expected \'[\'');
        }
        $this->input->next();
        while (!$this->input->eof()) {
            $char = $this->input->next();
            if ($char == ']') { // we might have reached the end
                if ($startNumberOfEqualsSigns != 0) {
                    if ($this->input->peek() == '=') {
                        $endNumberOfEqualsSigns = 0;
                        while ($this->input->peek() == '=') {
                            $endNumberOfEqualsSigns++;
                            $this->input->next();
                        }

                        // we have an equal number of equal signs
                        if ($endNumberOfEqualsSigns == $startNumberOfEqualsSigns) {
                            if ($this->input->peek() != ']') {
                                $this->error('Unexpected character \'' . $this->input->peek() . '\', expected \'[\'');
                            }
                            $this->input->next();
                            break;
                        } else {
                            $str .= $char . str_repeat('=', $endNumberOfEqualsSigns);
                        }
                    } else {
                        $str .= $char;
                    }
                } else {
                    if ($this->input->peek() == ']') {
                        $this->input->next();
                        break;
                    }
                }
            } else {
                $str .= $char;
            }
        }
        return new Token(Token::TYPE_STRING, $str);
    }

    /**
     * @param string $end
     *
     * @return string
     */
    protected function readEscaped($end) {
        $escaped = false;
        $str     = "";
        $this->input->next();
        while (!$this->input->eof()) {
            $char = $this->input->next();
            if ($escaped) {
                switch ($char) {
                    case 'n':
                        $str .= "\n";
                        break;
                    case 'r':
                        $str .= "\r";
                        break;
                    case 't':
                        $str .= "\t";
                        break;
                    case 'v':
                        $str .= "\v";
                        break;
                    default:
                        $str .= $char;
                        break;
                }
                $escaped = false;
            } else {
                if ($char == "\\") {
                    $escaped = true;
                } else {
                    if ($char == $end) {
                        break;
                    } else {
                        $str .= $char;
                    }
                }
            }
        }
        return $str;
    }

    /**
     * @return Token
     */
    protected function readNumber() {
        $hasDot = false;
        $number = $this->readWhile(
            function ($char) use (&$hasDot) {
                if ($char == '.') {
                    if ($hasDot) {
                        return false;
                    }
                    $hasDot = true;
                    return true;
                }
                return $this->isDigit($char);
            }
        );
        return new Token(Token::TYPE_NUMBER, $hasDot ? floatval($number) : intval($number));
    }

    /**
     * @return Token
     */
    protected function readIdentifier() {
        $first      = false;
        $identifier = $this->readWhile(
            function ($char) use (&$first) {
                if ($first) {
                    $first = false;
                    return $this->isStartIdentifierCharacter($char);
                }
                return $this->isIdentifierCharacter($char);
            }
        );
        if ($this->isKeyword($identifier)) {
            return new Token(Token::TYPE_KEYWORD, $identifier);
        }
        return new Token(Token::TYPE_IDENTIFIER, $identifier);
    }

    /**
     * @return Token
     */
    protected function readPunctuation() {
        return new Token(Token::TYPE_PUNCTUATION, $this->input->next());
    }

    /**
     * @param callable $predicate
     *
     * @return string
     */
    protected function readWhile(callable $predicate) {
        $str = "";
        while (!$this->input->eof() && call_user_func($predicate, $this->input->peek())) {
            $str .= $this->input->next();
        }
        return $str;
    }

    /**
     * @param string $char
     *
     * @return bool
     */
    protected function isWhitespace($char) {
        return strpos(" \t\n\r", $char) !== false;
    }

    /**
     * @param string $char
     *
     * @return bool
     */
    protected function isDigit($char) {
        return is_numeric($char);
    }

    /**
     * @return bool
     */
    protected function isDoubleBracketString() {
        return $this->input->peek() == '[' && !$this->input->eof(1) && ($this->input->peek(1) == '[' || $this->input->peek(1) == '=');
    }

    /**
     * @return bool
     */
    protected function isComment() {
        return $this->input->peek() == '-' && !$this->input->eof(1) && $this->input->peek(1) == '-';
    }

    /**
     * @param string $char
     *
     * @return bool
     */
    protected function isStartIdentifierCharacter($char) {
        return preg_match('/[a-zA-Z_]/', $char) === 1;
    }

    /**
     * @param string $char
     *
     * @return bool
     */
    protected function isIdentifierCharacter($char) {
        return preg_match('/[a-zA-Z0-9_]/', $char) === 1;
    }

    /**
     * @param string $char
     *
     * @return bool
     */
    protected function isPunctuation($char) {
        return strpos(",{}=[]", $char) !== false;
    }

    /**
     * @param string $text
     *
     * @return bool
     */
    protected function isKeyword($text) {
        return in_array($text, Lua::$luaKeywords);
    }
}

1		<?php
2
3		namespace Vlaswinkel\Lua;
4
5		/**
6		* Class TokenStream
7		*
8		* @see http://lisperator.net/pltut/parser/token-stream
9		*
10		* @author Koen Vlaswinkel <[email protected]>
11		* @package Vlaswinkel\Lua
12		*/
13		class TokenStream {
14		private $current = null;
15		/**
16		* @var InputStream
17		*/
18		private $input;
19
20		/**
21		* TokenStream constructor.
22		*
23		* @param InputStream $input
24		*/
25	40	public function __construct(InputStream $input) {
26	40	$this->input = $input;
27	40	}
28
29		/**
30		* @return Token
31		*/
32	40	public function next() {
33	40	$token = $this->current;
34	40	$this->current = null;
35	40	if ($token) {
36	24	return $token;
37		}
38	16	return $this->readNext();
39		}
40
41		/**
42		* @return bool
43		*/
44	23	public function eof() {
45	23	return $this->peek() == null;
46		}
47
48		/**
49		* @return Token
		0 ignored issues – show Documentation introduced 2015-12-22 12:03 UTC by Report Bug Copy Issue Report Should the return type not be `Token\|null`? This check compares the return type specified in the `@return` annotation of a function or method doc comment with the types returned by the function and raises an issue if they mismatch. Loading history...
50		*/
51	24	public function peek() {
52	24	if ($this->current) {
53	24	return $this->current;
54		}
55	24	$this->current = $this->readNext();
56	24	return $this->current;
57		}
58
59		/**
60		* @param string $msg
61		*
62		* @throws ParseException
63		*/
64	4	public function error($msg) {
65	4	$this->input->error($msg);
66		}
67
68		/**
69		* @return Token
		0 ignored issues – show Documentation introduced 2015-12-22 12:03 UTC by Report Bug Copy Issue Report Should the return type not be `Token\|null`? 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...
70		* @throws ParseException
71		*/
72	40	protected function readNext() {
73	40	$this->readWhile([$this, 'isWhitespace']);
74	40	if ($this->input->eof()) {
75	20	return null;
76		}
77	40	$char = $this->input->peek();
78	40	if ($this->isComment()) {
79	6	$this->skipComment();
80	6	return $this->readNext();
81		}
82	40	if ($char == '"') {
83	14	return $this->readDoubleQuotedString();
84		}
85	35	if ($char == '\'') {
86	1	return $this->readSingleQuotedString();
87		}
88	34	if ($this->isDoubleBracketString()) {
89	13	return $this->readDoubleBracketString();
90		}
91	25	if ($this->isDigit($char)) {
92	10	return $this->readNumber();
93		}
94	21	if ($this->isStartIdentifierCharacter($char)) {
95	17	return $this->readIdentifier();
96		}
97	16	if ($this->isPunctuation($char)) {
98	15	return $this->readPunctuation();
99		}
100	2	if ($char == ';') {
101	1	$this->input->next(); // skip the semi-colon
102	1	return $this->readNext(); // just move on to the next
103		}
104	1	$this->input->error('Cannot handle character: ' . $char . ' (ord: ' . ord($char) . ')');
105		}
106
107	6	protected function skipComment() {
108	6	$this->readWhile(
109		function ($char) {
110	6	return $char != "\n";
111		}
112	6	);
113	6	if (!$this->input->eof()) {
114	5	$this->input->next();
115	5	}
116	6	}
117
118		/**
119		* @return Token
120		*/
121	14	protected function readDoubleQuotedString() {
122	14	return new Token(Token::TYPE_STRING, $this->readEscaped('"'));
123		}
124
125		/**
126		* @return Token
127		*/
128	1	protected function readSingleQuotedString() {
129	1	return new Token(Token::TYPE_STRING, $this->readEscaped('\''));
130		}
131
132		/**
133		* @return Token
134		*/
135	13	protected function readDoubleBracketString() {
136		// we cannot use readEscaped because it only supports a single char as $end
137		// and we do not support escaping in double bracke strings
138	13	$str = "";
139	13	$startNumberOfEqualsSigns = 0;
140		// skip both
141	13	$this->input->next();
142	13	while ($this->input->peek() == '=') {
143	8	$startNumberOfEqualsSigns++;
144	8	$this->input->next();
145	8	}
146	13	if ($this->input->peek() != '[') {
147	1	$this->error('Unexpected character \'' . $this->input->peek() . '\', expected \'[\'');
148		}
149	12	$this->input->next();
150	12	while (!$this->input->eof()) {
151	12	$char = $this->input->next();
152	12	if ($char == ']') { // we might have reached the end
153	12	if ($startNumberOfEqualsSigns != 0) {
154	7	if ($this->input->peek() == '=') {
155	6	$endNumberOfEqualsSigns = 0;
156	6	while ($this->input->peek() == '=') {
157	6	$endNumberOfEqualsSigns++;
158	6	$this->input->next();
159	6	}
160
161		// we have an equal number of equal signs
162	6	if ($endNumberOfEqualsSigns == $startNumberOfEqualsSigns) {
163	6	if ($this->input->peek() != ']') {
164	1	$this->error('Unexpected character \'' . $this->input->peek() . '\', expected \'[\'');
165		}
166	5	$this->input->next();
167	5	break;
168		} else {
169	1	$str .= $char . str_repeat('=', $endNumberOfEqualsSigns);
170		}
171	1	} else {
172	3	$str .= $char;
173		}
174	3	} else {
175	5	if ($this->input->peek() == ']') {
176	5	$this->input->next();
177	5	break;
178		}
179		}
180	3	} else {
181	12	$str .= $char;
182		}
183	12	}
184	10	return new Token(Token::TYPE_STRING, $str);
185		}
186
187		/**
188		* @param string $end
189		*
190		* @return string
191		*/
192	15	protected function readEscaped($end) {
193	15	$escaped = false;
194	15	$str = "";
195	15	$this->input->next();
196	15	while (!$this->input->eof()) {
197	15	$char = $this->input->next();
198	15	if ($escaped) {
199		switch ($char) {
200	1	case 'n':
201	1	$str .= "\n";
202	1	break;
203	1	case 'r':
204	1	$str .= "\r";
205	1	break;
206	1	case 't':
207	1	$str .= "\t";
208	1	break;
209	1	case 'v':
210	1	$str .= "\v";
211	1	break;
212	1	default:
213	1	$str .= $char;
214	1	break;
215	1	}
216	1	$escaped = false;
217	1	} else {
218	15	if ($char == "\\") {
219	1	$escaped = true;
220	1	} else {
221	15	if ($char == $end) {
222	15	break;
223		} else {
224	15	$str .= $char;
225		}
226		}
227		}
228	15	}
229	15	return $str;
230		}
231
232		/**
233		* @return Token
234		*/
235	10	protected function readNumber() {
236	10	$hasDot = false;
237	10	$number = $this->readWhile(
238		function ($char) use (&$hasDot) {
239	10	if ($char == '.') {
240	1	if ($hasDot) {
241		return false;
242		}
243	1	$hasDot = true;
244	1	return true;
245		}
246	10	return $this->isDigit($char);
247		}
248	10	);
249	10	return new Token(Token::TYPE_NUMBER, $hasDot ? floatval($number) : intval($number));
250		}
251
252		/**
253		* @return Token
254		*/
255	17	protected function readIdentifier() {
256	17	$first = false;
257	17	$identifier = $this->readWhile(
258	17	function ($char) use (&$first) {
259	17	if ($first) {
260		$first = false;
261		return $this->isStartIdentifierCharacter($char);
262		}
263	17	return $this->isIdentifierCharacter($char);
264		}
265	17	);
266	17	if ($this->isKeyword($identifier)) {
267	6	return new Token(Token::TYPE_KEYWORD, $identifier);
268		}
269	13	return new Token(Token::TYPE_IDENTIFIER, $identifier);
270		}
271
272		/**
273		* @return Token
274		*/
275	15	protected function readPunctuation() {
276	15	return new Token(Token::TYPE_PUNCTUATION, $this->input->next());
277		}
278
279		/**
280		* @param callable $predicate
281		*
282		* @return string
283		*/
284	40	protected function readWhile(callable $predicate) {
285	40	$str = "";
286	40	while (!$this->input->eof() && call_user_func($predicate, $this->input->peek())) {
287	21	$str .= $this->input->next();
288	21	}
289	40	return $str;
290		}
291
292		/**
293		* @param string $char
294		*
295		* @return bool
296		*/
297	40	protected function isWhitespace($char) {
298	40	return strpos(" \t\n\r", $char) !== false;
299		}
300
301		/**
302		* @param string $char
303		*
304		* @return bool
305		*/
306	25	protected function isDigit($char) {
307	25	return is_numeric($char);
308		}
309
310		/**
311		* @return bool
312		*/
313	34	protected function isDoubleBracketString() {
314	34	return $this->input->peek() == '[' && !$this->input->eof(1) && ($this->input->peek(1) == '[' \|\| $this->input->peek(1) == '=');
315		}
316
317		/**
318		* @return bool
319		*/
320	40	protected function isComment() {
321	40	return $this->input->peek() == '-' && !$this->input->eof(1) && $this->input->peek(1) == '-';
322		}
323
324		/**
325		* @param string $char
326		*
327		* @return bool
328		*/
329	21	protected function isStartIdentifierCharacter($char) {
330	21	return preg_match('/[a-zA-Z_]/', $char) === 1;
331		}
332
333		/**
334		* @param string $char
335		*
336		* @return bool
337		*/
338	17	protected function isIdentifierCharacter($char) {
339	17	return preg_match('/[a-zA-Z0-9_]/', $char) === 1;
340		}
341
342		/**
343		* @param string $char
344		*
345		* @return bool
346		*/
347	16	protected function isPunctuation($char) {
348	16	return strpos(",{}=[]", $char) !== false;
349		}
350
351		/**
352		* @param string $text
353		*
354		* @return bool
355		*/
356	17	protected function isKeyword($text) {
357	17	return in_array($text, Lua::$luaKeywords);
358		}
359		}

koesie10 / LuaSerializer

TokenStream::readEscaped() D last analyzed 2016-10-20 19:27 UTC

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like

TokenStream::readEscaped() D
last analyzed 2016-10-20 19:27 UTC