|
1
|
|
|
<?php declare(strict_types=1); |
|
2
|
|
|
/** |
|
3
|
|
|
* pgn-splitter (https://github.com/chesszebra/pgn-splitter) |
|
4
|
|
|
* |
|
5
|
|
|
* @link https://github.com/chesszebra/pgn-splitter for the canonical source repository |
|
6
|
|
|
* @copyright Copyright (c) 2017 Chess Zebra (https://chesszebra.com) |
|
7
|
|
|
* @license https://github.com/chesszebra/pgn-splitter/blob/master/LICENSE MIT |
|
8
|
|
|
*/ |
|
9
|
|
|
|
|
10
|
|
|
namespace ChessZebra\Chess\Pgn; |
|
11
|
|
|
|
|
12
|
|
|
use ChessZebra\Chess\Pgn\Exception\InvalidStreamException; |
|
13
|
|
|
|
|
14
|
|
|
final class Splitter |
|
15
|
|
|
{ |
|
16
|
|
|
const SPLIT_GAMES = 0; |
|
17
|
|
|
const SPLIT_CHUNKS = 1; |
|
18
|
|
|
|
|
19
|
|
|
const STATE_LIMBO = 0; |
|
20
|
|
|
const STATE_TAGS = 1; |
|
21
|
|
|
const STATE_MOVES = 2; |
|
22
|
|
|
|
|
23
|
|
|
/** |
|
24
|
|
|
* The stream to split. |
|
25
|
|
|
* |
|
26
|
|
|
* @var resource |
|
27
|
|
|
*/ |
|
28
|
|
|
private $stream; |
|
29
|
|
|
|
|
30
|
|
|
/** |
|
31
|
|
|
* The mode to split the stream on. |
|
32
|
|
|
* |
|
33
|
|
|
* @var int |
|
34
|
|
|
*/ |
|
35
|
|
|
private $mode; |
|
36
|
|
|
|
|
37
|
|
|
/** |
|
38
|
|
|
* The current state of the splitter. |
|
39
|
|
|
* |
|
40
|
|
|
* @var int |
|
41
|
|
|
*/ |
|
42
|
|
|
private $state; |
|
43
|
|
|
|
|
44
|
|
|
/** |
|
45
|
|
|
* The parsed buffer. |
|
46
|
|
|
* |
|
47
|
|
|
* @var string |
|
48
|
|
|
*/ |
|
49
|
|
|
private $buffer; |
|
50
|
|
|
|
|
51
|
|
|
/** |
|
52
|
|
|
* Initializes a new instance of this class. |
|
53
|
|
|
* |
|
54
|
|
|
* @param resource $stream The stream to split. |
|
55
|
|
|
* @param int $mode The mode to split on. |
|
56
|
|
|
* @throws InvalidStreamException Thrown when an invalid stream is provided. |
|
57
|
|
|
*/ |
|
58
|
|
|
public function __construct($stream, int $mode = self::SPLIT_GAMES) |
|
59
|
|
|
{ |
|
60
|
|
|
if (!is_resource($stream)) { |
|
61
|
|
|
throw new InvalidStreamException('The provided stream is not a valid resource.'); |
|
62
|
|
|
} |
|
63
|
|
|
|
|
64
|
|
|
$this->stream = $stream; |
|
65
|
|
|
$this->mode = $mode; |
|
66
|
|
|
} |
|
67
|
|
|
|
|
68
|
|
|
/** |
|
69
|
|
|
* Splits the stream into loose chunks and provides them to the given callback. |
|
70
|
|
|
* |
|
71
|
|
|
* @param callable $callback The callback that is called for each found chunk. |
|
72
|
|
|
* @return int Returns the amount of chunks found in the stream. |
|
73
|
|
|
*/ |
|
74
|
|
|
public function split(callable $callback): int |
|
75
|
|
|
{ |
|
76
|
|
|
$result = 0; |
|
77
|
|
|
|
|
78
|
|
|
$this->buffer = ''; |
|
79
|
|
|
$this->state = self::STATE_LIMBO; |
|
80
|
|
|
|
|
81
|
|
|
while (!feof($this->stream)) { |
|
82
|
|
|
$line = fgets($this->stream); |
|
83
|
|
|
$line = str_replace("\r\n", "\n", $line); // Replace Windows line endings with unix |
|
84
|
|
|
$line = str_replace("\r", "\n", $line); // Replace mac line endings with unix |
|
85
|
|
|
|
|
86
|
|
|
if (!$line) { |
|
87
|
|
|
continue; |
|
88
|
|
|
} |
|
89
|
|
|
|
|
90
|
|
|
switch ($this->state) { |
|
91
|
|
|
case self::STATE_LIMBO: |
|
92
|
|
|
$this->parseLimbo($line); |
|
93
|
|
|
break; |
|
94
|
|
|
|
|
95
|
|
|
case self::STATE_TAGS: |
|
96
|
|
|
$this->parseTags($callback, $result, $line); |
|
97
|
|
|
break; |
|
98
|
|
|
|
|
99
|
|
|
case self::STATE_MOVES: |
|
100
|
|
|
$this->parseMoves($callback, $result, $line); |
|
101
|
|
|
break; |
|
102
|
|
|
|
|
103
|
|
|
default: |
|
104
|
|
|
throw new \RuntimeException(sprintf('The state "%d" is not implemented.', $this->state)); |
|
105
|
|
|
} |
|
106
|
|
|
} |
|
107
|
|
|
|
|
108
|
|
|
if (trim($this->buffer) !== '') { |
|
109
|
|
|
$result++; |
|
110
|
|
|
$callback($this->buffer); |
|
111
|
|
|
$this->buffer = ''; |
|
112
|
|
|
} |
|
113
|
|
|
|
|
114
|
|
|
return $result; |
|
115
|
|
|
} |
|
116
|
|
|
|
|
117
|
|
|
private function parseLimbo(string $line): void |
|
118
|
|
|
{ |
|
119
|
|
|
if (trim($line) === '') { |
|
120
|
|
|
return; |
|
121
|
|
|
} |
|
122
|
|
|
|
|
123
|
|
|
if (strlen($line) >= 3 && $line[0] === "\xEF" && $line[1] === "\xBB" && $line[2] === "\xBF") { |
|
124
|
|
|
$line = substr($line, 3); |
|
125
|
|
|
} |
|
126
|
|
|
|
|
127
|
|
|
if ($line[0] === '[') { |
|
128
|
|
|
$this->buffer .= $line; |
|
129
|
|
|
$this->state = self::STATE_TAGS; |
|
130
|
|
|
} else { |
|
131
|
|
|
$this->buffer .= $line; |
|
132
|
|
|
$this->state = self::STATE_MOVES; |
|
133
|
|
|
} |
|
134
|
|
|
} |
|
135
|
|
|
|
|
136
|
|
|
private function parseTags(callable $callback, int &$result, string $line): void |
|
137
|
|
|
{ |
|
138
|
|
|
if ($line[0] === '[') { |
|
139
|
|
|
$this->buffer .= $line; |
|
140
|
|
|
return; |
|
141
|
|
|
} |
|
142
|
|
|
|
|
143
|
|
|
if ($this->mode === self::SPLIT_CHUNKS && trim($this->buffer) !== '') { |
|
144
|
|
|
$result++; |
|
145
|
|
|
|
|
146
|
|
|
$callback($this->buffer); |
|
147
|
|
|
|
|
148
|
|
|
$this->buffer = ''; |
|
149
|
|
|
return; |
|
150
|
|
|
} |
|
151
|
|
|
|
|
152
|
|
|
$this->buffer .= $line; |
|
153
|
|
|
|
|
154
|
|
|
$this->state = self::STATE_MOVES; |
|
155
|
|
|
} |
|
156
|
|
|
|
|
157
|
|
|
private function parseMoves(callable $callback, int &$result, string $line): void |
|
158
|
|
|
{ |
|
159
|
|
|
if ($line === "\n") { |
|
160
|
|
|
$this->state = self::STATE_LIMBO; |
|
161
|
|
|
|
|
162
|
|
|
$result++; |
|
163
|
|
|
|
|
164
|
|
|
$callback($this->buffer); |
|
165
|
|
|
|
|
166
|
|
|
$this->buffer = ''; |
|
167
|
|
|
} |
|
168
|
|
|
|
|
169
|
|
|
$this->buffer .= $line; |
|
170
|
|
|
} |
|
171
|
|
|
} |
|
172
|
|
|
|
chesszebra /
pgn-splitter
