1
|
|
|
<?php |
2
|
|
|
/** |
3
|
|
|
* @author: Viskov Sergey |
4
|
|
|
* @date: 3/10/16 |
5
|
|
|
* @time: 1:33 PM |
6
|
|
|
*/ |
7
|
|
|
|
8
|
|
|
namespace LTDBeget\sphinx\configurator\parser; |
9
|
|
|
|
10
|
|
|
|
11
|
|
|
use BadMethodCallException; |
12
|
|
|
use LTDBeget\ascii\AsciiChar; |
13
|
|
|
use LTDBeget\sphinx\configurator\exceptions\SyntaxErrorException; |
14
|
|
|
use LTDBeget\sphinx\enums\eSection; |
15
|
|
|
use LTDBeget\stringstream\StringStream; |
16
|
|
|
|
17
|
|
|
/** |
18
|
|
|
* Class SphinxConfigurationParser |
19
|
|
|
* @package LTDBeget\sphinx\configurator\parser |
20
|
|
|
*/ |
21
|
|
|
final class SphinxConfigurationParser |
22
|
|
|
{ |
23
|
|
|
/** |
24
|
|
|
* parse and tokenize input string |
25
|
|
|
* @param string $plainData |
26
|
|
|
* @throws SyntaxErrorException |
27
|
|
|
* @throws BadMethodCallException |
28
|
|
|
* @return array |
29
|
|
|
*/ |
30
|
|
|
public static function parse(string $plainData) : array |
31
|
|
|
{ |
32
|
|
|
return (new self($plainData))->tokenize()->tokens; |
33
|
|
|
} |
34
|
|
|
|
35
|
|
|
/** |
36
|
|
|
* SphinxConfigurationParser constructor. |
37
|
|
|
* @internal |
38
|
|
|
* @param string $string |
39
|
|
|
* @throws BadMethodCallException |
40
|
|
|
*/ |
41
|
|
|
private function __construct(string $string) |
42
|
|
|
{ |
43
|
|
|
$this->originalString = $string; |
|
|
|
|
44
|
|
|
$string = $this->removeComments($string); |
45
|
|
|
$this->stream = new StringStream($string); |
|
|
|
|
46
|
|
|
} |
47
|
|
|
|
48
|
|
|
/** |
49
|
|
|
* @internal |
50
|
|
|
* @param string $string |
51
|
|
|
* @return string |
52
|
|
|
*/ |
53
|
|
|
private function removeComments(string $string) : string |
54
|
|
|
{ |
55
|
|
|
return preg_replace("/(^#| #| #).*\n/im", "\n", $string); |
56
|
|
|
} |
57
|
|
|
|
58
|
|
|
/** |
59
|
|
|
* @internal |
60
|
|
|
* @return SphinxConfigurationParser |
61
|
|
|
* @throws SyntaxErrorException |
62
|
|
|
*/ |
63
|
|
|
private function tokenize() : SphinxConfigurationParser |
64
|
|
|
{ |
65
|
|
|
do { |
66
|
|
|
$this->extractSection(); |
67
|
|
|
$this->saveCurrentSection(); |
68
|
|
|
|
69
|
|
|
} while (! $this->stream->isEnd()); |
70
|
|
|
|
71
|
|
|
return $this; |
72
|
|
|
} |
73
|
|
|
|
74
|
|
|
/** |
75
|
|
|
* @internal |
76
|
|
|
* @throws SyntaxErrorException |
77
|
|
|
*/ |
78
|
|
|
private function extractSection() |
79
|
|
|
{ |
80
|
|
|
$this->extractSectionType(); |
81
|
|
|
|
82
|
|
|
switch ($this->currentSection["type"]) { |
83
|
|
|
case eSection::SOURCE(): |
84
|
|
|
case eSection::INDEX(): |
85
|
|
|
$this->extractSectionName(); |
86
|
|
|
|
87
|
|
|
$this->extractInheritance(); |
88
|
|
|
break; |
89
|
|
|
case eSection::INDEXER(): |
90
|
|
|
case eSection::SEARCHD(): |
91
|
|
|
case eSection::COMMON(): |
92
|
|
|
break; |
93
|
|
|
default: |
94
|
|
|
throw new SyntaxErrorException($this->stream); |
95
|
|
|
} |
96
|
|
|
|
97
|
|
|
$this->extractOptions(); |
98
|
|
|
|
99
|
|
|
|
100
|
|
|
|
101
|
|
|
$this->stream->ignoreWhitespace(); |
102
|
|
|
} |
103
|
|
|
|
104
|
|
|
/** |
105
|
|
|
* @internal |
106
|
|
|
* @throws SyntaxErrorException |
107
|
|
|
*/ |
108
|
|
|
private function extractSectionType() |
109
|
|
|
{ |
110
|
|
|
$this->stream->ignoreWhitespace(); |
111
|
|
|
start: |
112
|
|
|
$char = $this->stream->currentAscii(); |
113
|
|
|
$this->stream->next(); |
114
|
|
|
if ($char->isLetter()) { |
115
|
|
|
$this->currentSection["type"] .= (string) $char; |
116
|
|
|
goto start; |
117
|
|
|
} elseif ($char->isWhiteSpace()) { |
118
|
|
|
return; |
119
|
|
|
} else { |
120
|
|
|
throw new SyntaxErrorException($this->stream); |
121
|
|
|
} |
122
|
|
|
} |
123
|
|
|
|
124
|
|
|
/** |
125
|
|
|
* @internal |
126
|
|
|
* @throws SyntaxErrorException |
127
|
|
|
*/ |
128
|
|
|
private function extractSectionName() |
129
|
|
|
{ |
130
|
|
|
$this->stream->ignoreHorizontalSpace(); |
131
|
|
|
|
132
|
|
|
start: |
133
|
|
|
$char = $this->stream->currentAscii(); |
134
|
|
|
$this->stream->next(); |
135
|
|
|
|
136
|
|
|
if ($char->isLetter() || $char->isDigit() || $char->is(AsciiChar::UNDERSCORE())) { |
137
|
|
|
$this->currentSection["name"] .= (string) $char; |
138
|
|
|
goto start; |
139
|
|
|
} elseif ($char->isWhiteSpace()) { |
140
|
|
|
return; |
141
|
|
|
} elseif ($char->is(AsciiChar::COLON())) { |
142
|
|
|
$this->stream->previous(); |
143
|
|
|
return; |
144
|
|
|
} else { |
145
|
|
|
throw new SyntaxErrorException($this->stream); |
146
|
|
|
} |
147
|
|
|
} |
148
|
|
|
|
149
|
|
|
/** |
150
|
|
|
* @internal |
151
|
|
|
* @throws SyntaxErrorException |
152
|
|
|
*/ |
153
|
|
|
private function extractInheritance() |
154
|
|
|
{ |
155
|
|
|
$this->stream->ignoreHorizontalSpace(); |
156
|
|
|
|
157
|
|
|
$char = $this->stream->currentAscii(); |
158
|
|
|
|
159
|
|
|
if($char->isVerticalSpace() || $char->is(AsciiChar::OPENING_BRACE())) { |
160
|
|
|
return; |
161
|
|
|
} |
162
|
|
|
|
163
|
|
|
if($char->is(AsciiChar::COLON())) { |
164
|
|
|
$this->stream->next(); |
165
|
|
|
$this->extractInheritanceName(); |
166
|
|
|
} else { |
167
|
|
|
throw new SyntaxErrorException($this->stream); |
168
|
|
|
} |
169
|
|
|
} |
170
|
|
|
|
171
|
|
|
/** |
172
|
|
|
* @internal |
173
|
|
|
* @throws SyntaxErrorException |
174
|
|
|
*/ |
175
|
|
View Code Duplication |
private function extractInheritanceName() |
|
|
|
|
176
|
|
|
{ |
177
|
|
|
$this->stream->ignoreHorizontalSpace(); |
178
|
|
|
start: |
179
|
|
|
$char = $this->stream->currentAscii(); |
180
|
|
|
$this->stream->next(); |
181
|
|
|
|
182
|
|
|
if ($char->isLetter() || $char->isDigit() || $char->is(AsciiChar::UNDERSCORE())) { |
183
|
|
|
$this->currentSection["inheritance"] .= (string) $char; |
184
|
|
|
goto start; |
185
|
|
|
} elseif ($char->isWhiteSpace()) { |
186
|
|
|
return; |
187
|
|
|
} else { |
188
|
|
|
throw new SyntaxErrorException($this->stream); |
189
|
|
|
} |
190
|
|
|
} |
191
|
|
|
|
192
|
|
|
/** |
193
|
|
|
* @internal |
194
|
|
|
* @throws SyntaxErrorException |
195
|
|
|
*/ |
196
|
|
|
private function extractOptions() |
197
|
|
|
{ |
198
|
|
|
$this->stream->ignoreWhitespace(); |
199
|
|
|
|
200
|
|
|
if ($this->stream->currentAscii()->is(AsciiChar::OPENING_BRACE())) { |
201
|
|
|
$this->stream->next(); |
202
|
|
|
|
203
|
|
|
start: |
204
|
|
|
$this->stream->ignoreWhitespace(); |
205
|
|
|
if ($this->stream->currentAscii()->is(AsciiChar::CLOSING_BRACE())) { |
206
|
|
|
$this->stream->next(); |
207
|
|
|
return; |
208
|
|
|
} |
209
|
|
|
$this->extractOption(); |
210
|
|
|
goto start; |
211
|
|
|
} else { |
212
|
|
|
throw new SyntaxErrorException($this->stream); |
213
|
|
|
} |
214
|
|
|
} |
215
|
|
|
|
216
|
|
|
/** |
217
|
|
|
* @internal |
218
|
|
|
* @throws SyntaxErrorException |
219
|
|
|
*/ |
220
|
|
|
private function extractOption() |
221
|
|
|
{ |
222
|
|
|
$this->extractOptionName(); |
223
|
|
|
$this->extractOptionValue(); |
224
|
|
|
$this->saveCurrentOption(); |
225
|
|
|
} |
226
|
|
|
|
227
|
|
|
/** |
228
|
|
|
* @internal |
229
|
|
|
* @throws SyntaxErrorException |
230
|
|
|
*/ |
231
|
|
View Code Duplication |
private function extractOptionName() |
|
|
|
|
232
|
|
|
{ |
233
|
|
|
$this->stream->ignoreWhitespace(); |
234
|
|
|
|
235
|
|
|
start: |
236
|
|
|
$char = $this->stream->currentAscii(); |
237
|
|
|
$this->stream->next(); |
238
|
|
|
|
239
|
|
|
if ($char->isLetter() || $char->isDigit() || $char->is(AsciiChar::UNDERSCORE())) { |
240
|
|
|
$this->currentOption["name"] .= (string) $char; |
241
|
|
|
goto start; |
242
|
|
|
} elseif ($char->isHorizontalSpace()) { |
243
|
|
|
return; |
244
|
|
|
} else { |
245
|
|
|
throw new SyntaxErrorException($this->stream); |
246
|
|
|
} |
247
|
|
|
} |
248
|
|
|
|
249
|
|
|
/** |
250
|
|
|
* @internal |
251
|
|
|
* @throws SyntaxErrorException |
252
|
|
|
*/ |
253
|
|
|
private function extractOptionValue() |
254
|
|
|
{ |
255
|
|
|
$this->stream->ignoreHorizontalSpace(); |
256
|
|
|
|
257
|
|
|
$char = $this->stream->currentAscii(); |
258
|
|
|
$this->stream->next(); |
259
|
|
|
|
260
|
|
|
if (! $char->is(AsciiChar::EQUALS())) { |
261
|
|
|
throw new SyntaxErrorException($this->stream); |
262
|
|
|
} |
263
|
|
|
|
264
|
|
|
$this->stream->ignoreHorizontalSpace(); |
265
|
|
|
|
266
|
|
|
start: |
267
|
|
|
$char = $this->stream->currentAscii(); |
268
|
|
|
$this->stream->next(); |
269
|
|
|
|
270
|
|
|
if ($char->isPrintableChar()) { |
271
|
|
|
if ($char->is(AsciiChar::BACKSLASH())) { // if possibility of multi-line |
272
|
|
|
$char = $this->stream->currentAscii(); |
273
|
|
|
|
274
|
|
|
if ($char->isVerticalSpace()) { // multi-line opened |
275
|
|
|
$this->currentOption["value"] .= (string) AsciiChar::SPACE(); |
276
|
|
|
$this->stream->next(); // ignore end line |
277
|
|
|
$this->stream->ignoreHorizontalSpace(); |
278
|
|
|
goto start; |
279
|
|
|
} else { // backslash as mean symbol |
280
|
|
|
$this->currentOption["value"] .= (string) AsciiChar::BACKSLASH(); |
281
|
|
|
goto start; |
282
|
|
|
} |
283
|
|
|
} else { |
284
|
|
|
$this->currentOption["value"] .= (string) $char; |
285
|
|
|
goto start; |
286
|
|
|
} |
287
|
|
|
} elseif ($char->isVerticalSpace()) { |
288
|
|
|
return; |
289
|
|
|
} else { |
290
|
|
|
throw new SyntaxErrorException($this->stream); |
291
|
|
|
} |
292
|
|
|
} |
293
|
|
|
|
294
|
|
|
/** |
295
|
|
|
* @internal |
296
|
|
|
*/ |
297
|
|
|
private function saveCurrentSection() |
298
|
|
|
{ |
299
|
|
|
$this->currentSection = array_filter($this->currentSection); |
300
|
|
|
$this->tokens[] = $this->currentSection; |
301
|
|
|
$this->currentSection = $this->getEmptySectionData(); |
302
|
|
|
} |
303
|
|
|
|
304
|
|
|
/** |
305
|
|
|
* @internal |
306
|
|
|
*/ |
307
|
|
|
private function saveCurrentOption() |
308
|
|
|
{ |
309
|
|
|
$this->currentSection["options"][] = $this->currentOption; |
310
|
|
|
$this->currentOption = $this->getEmptyOptionData(); |
311
|
|
|
} |
312
|
|
|
|
313
|
|
|
/** |
314
|
|
|
* @internal |
315
|
|
|
* @return array |
316
|
|
|
*/ |
317
|
|
|
private function getEmptySectionData() : array |
318
|
|
|
{ |
319
|
|
|
return [ |
320
|
|
|
"type" => "", |
321
|
|
|
"name" => "", |
322
|
|
|
"inheritance" => "", |
323
|
|
|
"options" => [] |
324
|
|
|
]; |
325
|
|
|
} |
326
|
|
|
|
327
|
|
|
/** |
328
|
|
|
* @internal |
329
|
|
|
* @return array |
330
|
|
|
*/ |
331
|
|
|
private function getEmptyOptionData() : array |
332
|
|
|
{ |
333
|
|
|
return [ |
334
|
|
|
"name" => "", |
335
|
|
|
"value" => "" |
336
|
|
|
]; |
337
|
|
|
} |
338
|
|
|
|
339
|
|
|
/** |
340
|
|
|
* Result of tokenize input string |
341
|
|
|
* @var array |
342
|
|
|
*/ |
343
|
|
|
private $tokens = []; |
344
|
|
|
|
345
|
|
|
/** |
346
|
|
|
* temporary storage of tokens for one section |
347
|
|
|
* @var array |
348
|
|
|
*/ |
349
|
|
|
private $currentSection = [ |
350
|
|
|
"type" => "", |
351
|
|
|
"name" => "", |
352
|
|
|
"inheritance" => "", |
353
|
|
|
"options" => [] |
354
|
|
|
]; |
355
|
|
|
/** |
356
|
|
|
* temporary storage of tokens for one option |
357
|
|
|
* @var array |
358
|
|
|
*/ |
359
|
|
|
private $currentOption = [ |
360
|
|
|
"name" => "", |
361
|
|
|
"value" => "" |
362
|
|
|
]; |
363
|
|
|
} |
In PHP it is possible to write to properties without declaring them. For example, the following is perfectly valid PHP code:
Generally, it is a good practice to explictly declare properties to avoid accidental typos and provide IDE auto-completion: