1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
/** |
4
|
|
|
* @package s9e\TextFormatter |
5
|
|
|
* @copyright Copyright (c) The s9e authors |
6
|
|
|
* @license http://www.opensource.org/licenses/mit-license.php The MIT License |
7
|
|
|
*/ |
8
|
|
|
namespace s9e\TextFormatter\Configurator\RendererGenerators\PHP; |
9
|
|
|
|
10
|
|
|
class BranchOutputOptimizer |
11
|
|
|
{ |
12
|
|
|
/** |
13
|
|
|
* @var integer Number of tokens |
14
|
|
|
*/ |
15
|
|
|
protected $cnt; |
16
|
|
|
|
17
|
|
|
/** |
18
|
|
|
* @var integer Current token index |
19
|
|
|
*/ |
20
|
|
|
protected $i; |
21
|
|
|
|
22
|
|
|
/** |
23
|
|
|
* @var array Tokens from current source |
24
|
|
|
*/ |
25
|
|
|
protected $tokens; |
26
|
|
|
|
27
|
|
|
/** |
28
|
|
|
* Optimize the code used to output content |
29
|
|
|
* |
30
|
|
|
* This method will go through the array of tokens, identify if/elseif/else blocks that contain |
31
|
|
|
* identical code at the beginning or the end and move the common code outside of the block |
32
|
|
|
* |
33
|
|
|
* @param array $tokens Array of tokens from token_get_all() |
34
|
|
|
* @return string Optimized code |
35
|
|
|
*/ |
36
|
12 |
|
public function optimize(array $tokens) |
37
|
|
|
{ |
38
|
12 |
|
$this->tokens = $tokens; |
39
|
12 |
|
$this->i = 0; |
40
|
12 |
|
$this->cnt = count($this->tokens); |
41
|
|
|
|
42
|
12 |
|
$php = ''; |
43
|
12 |
|
while (++$this->i < $this->cnt) |
44
|
|
|
{ |
45
|
12 |
|
if ($this->tokens[$this->i][0] === T_IF) |
46
|
|
|
{ |
47
|
12 |
|
$php .= $this->serializeIfBlock($this->parseIfBlock()); |
48
|
|
|
} |
49
|
|
|
else |
50
|
|
|
{ |
51
|
2 |
|
$php .= $this->serializeToken($this->tokens[$this->i]); |
52
|
|
|
} |
53
|
|
|
} |
54
|
|
|
|
55
|
|
|
// Free the memory taken up by the tokens |
56
|
12 |
|
unset($this->tokens); |
57
|
|
|
|
58
|
12 |
|
return $php; |
59
|
|
|
} |
60
|
|
|
|
61
|
|
|
/** |
62
|
|
|
* Capture the expressions used in any number of consecutive output statements |
63
|
|
|
* |
64
|
|
|
* Starts looking at current index. Ends at the first token that's not part of an output |
65
|
|
|
* statement |
66
|
|
|
* |
67
|
|
|
* @return string[] |
68
|
|
|
*/ |
69
|
12 |
|
protected function captureOutput() |
70
|
|
|
{ |
71
|
12 |
|
$expressions = []; |
72
|
12 |
|
while ($this->skipOutputAssignment()) |
73
|
|
|
{ |
74
|
|
|
do |
75
|
|
|
{ |
76
|
12 |
|
$expressions[] = $this->captureOutputExpression(); |
77
|
|
|
} |
78
|
12 |
|
while ($this->tokens[$this->i++] === '.'); |
79
|
|
|
} |
80
|
|
|
|
81
|
12 |
|
return $expressions; |
82
|
|
|
} |
83
|
|
|
|
84
|
|
|
/** |
85
|
|
|
* Capture an expression used in output at current index |
86
|
|
|
* |
87
|
|
|
* Ends on "." or ";" |
88
|
|
|
* |
89
|
|
|
* @return string |
90
|
|
|
*/ |
91
|
12 |
|
protected function captureOutputExpression() |
92
|
|
|
{ |
93
|
12 |
|
$parens = 0; |
94
|
12 |
|
$php = ''; |
95
|
|
|
do |
96
|
|
|
{ |
97
|
12 |
|
if ($this->tokens[$this->i] === ';') |
98
|
|
|
{ |
99
|
12 |
|
break; |
100
|
|
|
} |
101
|
12 |
|
elseif ($this->tokens[$this->i] === '.' && !$parens) |
102
|
|
|
{ |
103
|
7 |
|
break; |
104
|
|
|
} |
105
|
12 |
|
elseif ($this->tokens[$this->i] === '(') |
106
|
|
|
{ |
107
|
1 |
|
++$parens; |
108
|
|
|
} |
109
|
12 |
|
elseif ($this->tokens[$this->i] === ')') |
110
|
|
|
{ |
111
|
1 |
|
--$parens; |
112
|
|
|
} |
113
|
|
|
|
114
|
12 |
|
$php .= $this->serializeToken($this->tokens[$this->i]); |
115
|
|
|
} |
116
|
12 |
|
while (++$this->i < $this->cnt); |
117
|
|
|
|
118
|
12 |
|
return $php; |
119
|
|
|
} |
120
|
|
|
|
121
|
|
|
/** |
122
|
|
|
* Capture the source of a control structure from its keyword to its opening brace |
123
|
|
|
* |
124
|
|
|
* Ends after the brace, but the brace itself is not returned |
125
|
|
|
* |
126
|
|
|
* @return string |
127
|
|
|
*/ |
128
|
12 |
|
protected function captureStructure() |
129
|
|
|
{ |
130
|
12 |
|
$php = ''; |
131
|
|
|
do |
132
|
|
|
{ |
133
|
12 |
|
$php .= $this->serializeToken($this->tokens[$this->i]); |
134
|
|
|
} |
135
|
12 |
|
while ($this->tokens[++$this->i] !== '{'); |
136
|
|
|
|
137
|
|
|
// Move past the { |
138
|
12 |
|
++$this->i; |
139
|
|
|
|
140
|
12 |
|
return $php; |
141
|
|
|
} |
142
|
|
|
|
143
|
|
|
/** |
144
|
|
|
* Test whether the token at current index is an if/elseif/else token |
145
|
|
|
* |
146
|
|
|
* @return bool |
147
|
|
|
*/ |
148
|
12 |
|
protected function isBranchToken() |
149
|
|
|
{ |
150
|
12 |
|
return in_array($this->tokens[$this->i][0], [T_ELSE, T_ELSEIF, T_IF], true); |
151
|
|
|
} |
152
|
|
|
|
153
|
|
|
/** |
154
|
|
|
* Merge the branches of an if/elseif/else block |
155
|
|
|
* |
156
|
|
|
* Returns an array that contains the following: |
157
|
|
|
* |
158
|
|
|
* - before: array of PHP expressions to be output before the block |
159
|
|
|
* - source: PHP code for the if block |
160
|
|
|
* - after: array of PHP expressions to be output after the block |
161
|
|
|
* |
162
|
|
|
* @param array $branches |
163
|
|
|
* @return array |
164
|
|
|
*/ |
165
|
12 |
|
protected function mergeIfBranches(array $branches) |
166
|
|
|
{ |
167
|
|
|
// Test whether the branches cover all code paths. Without a "else" branch at the end, we |
168
|
|
|
// cannot optimize |
169
|
12 |
|
$lastBranch = end($branches); |
170
|
12 |
|
if ($lastBranch['structure'] === 'else') |
171
|
|
|
{ |
172
|
11 |
|
$before = $this->optimizeBranchesHead($branches); |
173
|
11 |
|
$after = $this->optimizeBranchesTail($branches); |
174
|
|
|
} |
175
|
|
|
else |
176
|
|
|
{ |
177
|
1 |
|
$before = $after = []; |
178
|
|
|
} |
179
|
|
|
|
180
|
12 |
|
$source = ''; |
181
|
12 |
|
foreach ($branches as $branch) |
182
|
|
|
{ |
183
|
12 |
|
$source .= $this->serializeBranch($branch); |
184
|
|
|
} |
185
|
|
|
|
186
|
|
|
return [ |
187
|
12 |
|
'before' => $before, |
188
|
12 |
|
'source' => $source, |
189
|
12 |
|
'after' => $after |
190
|
|
|
]; |
191
|
|
|
} |
192
|
|
|
|
193
|
|
|
/** |
194
|
|
|
* Merge two consecutive series of consecutive output expressions together |
195
|
|
|
* |
196
|
|
|
* @param array $left First series |
197
|
|
|
* @param array $right Second series |
198
|
|
|
* @return array Merged series |
199
|
|
|
*/ |
200
|
12 |
|
protected function mergeOutput(array $left, array $right) |
201
|
|
|
{ |
202
|
12 |
|
if (empty($left)) |
203
|
|
|
{ |
204
|
12 |
|
return $right; |
205
|
|
|
} |
206
|
|
|
|
207
|
2 |
|
if (empty($right)) |
208
|
|
|
{ |
209
|
1 |
|
return $left; |
210
|
|
|
} |
211
|
|
|
|
212
|
|
|
// Test whether we can merge the last expression on the left with the first expression on |
213
|
|
|
// the right |
214
|
1 |
|
$k = count($left) - 1; |
215
|
|
|
|
216
|
1 |
|
if (substr($left[$k], -1) === "'" && $right[0][0] === "'") |
217
|
|
|
{ |
218
|
1 |
|
$right[0] = substr($left[$k], 0, -1) . substr($right[0], 1); |
|
|
|
|
219
|
1 |
|
unset($left[$k]); |
220
|
|
|
} |
221
|
|
|
|
222
|
1 |
|
return array_merge($left, $right); |
223
|
|
|
} |
224
|
|
|
|
225
|
|
|
/** |
226
|
|
|
* Optimize the "head" part of a series of branches in-place |
227
|
|
|
* |
228
|
|
|
* @param array &$branches Array of branches, modified in-place |
229
|
|
|
* @return string[] PHP expressions removed from the "head" part of the branches |
230
|
|
|
*/ |
231
|
11 |
|
protected function optimizeBranchesHead(array &$branches) |
232
|
|
|
{ |
233
|
|
|
// Capture common output |
234
|
11 |
|
$before = $this->optimizeBranchesOutput($branches, 'head'); |
235
|
|
|
|
236
|
|
|
// Move the branch output to the tail for branches that have no body |
237
|
11 |
|
foreach ($branches as &$branch) |
238
|
|
|
{ |
239
|
11 |
|
if ($branch['body'] !== '' || !empty($branch['tail'])) |
240
|
|
|
{ |
241
|
7 |
|
continue; |
242
|
|
|
} |
243
|
|
|
|
244
|
8 |
|
$branch['tail'] = array_reverse($branch['head']); |
245
|
8 |
|
$branch['head'] = []; |
246
|
|
|
} |
247
|
11 |
|
unset($branch); |
248
|
|
|
|
249
|
11 |
|
return $before; |
250
|
|
|
} |
251
|
|
|
|
252
|
|
|
/** |
253
|
|
|
* Optimize the output of given branches |
254
|
|
|
* |
255
|
|
|
* @param array &$branches Array of branches |
256
|
|
|
* @param string $which Which end to optimize ("head" or "tail") |
257
|
|
|
* @return string[] PHP expressions removed from the given part of the branches |
258
|
|
|
*/ |
259
|
11 |
|
protected function optimizeBranchesOutput(array &$branches, $which) |
260
|
|
|
{ |
261
|
11 |
|
$expressions = []; |
262
|
11 |
|
while (isset($branches[0][$which][0])) |
263
|
|
|
{ |
264
|
11 |
|
$expr = $branches[0][$which][0]; |
265
|
11 |
|
foreach ($branches as $branch) |
266
|
|
|
{ |
267
|
11 |
|
if (!isset($branch[$which][0]) || $branch[$which][0] !== $expr) |
268
|
|
|
{ |
269
|
8 |
|
break 2; |
270
|
|
|
} |
271
|
|
|
} |
272
|
|
|
|
273
|
11 |
|
$expressions[] = $expr; |
274
|
11 |
|
foreach ($branches as &$branch) |
275
|
|
|
{ |
276
|
11 |
|
array_shift($branch[$which]); |
277
|
|
|
} |
278
|
11 |
|
unset($branch); |
279
|
|
|
} |
280
|
|
|
|
281
|
11 |
|
return $expressions; |
282
|
|
|
} |
283
|
|
|
|
284
|
|
|
/** |
285
|
|
|
* Optimize the "tail" part of a series of branches in-place |
286
|
|
|
* |
287
|
|
|
* @param array &$branches Array of branches, modified in-place |
288
|
|
|
* @return string[] PHP expressions removed from the "tail" part of the branches |
289
|
|
|
*/ |
290
|
11 |
|
protected function optimizeBranchesTail(array &$branches) |
291
|
|
|
{ |
292
|
11 |
|
return $this->optimizeBranchesOutput($branches, 'tail'); |
293
|
|
|
} |
294
|
|
|
|
295
|
|
|
/** |
296
|
|
|
* Parse the if, elseif or else branch starting at current index |
297
|
|
|
* |
298
|
|
|
* Ends at the last } |
299
|
|
|
* |
300
|
|
|
* @return array Branch's data ("structure", "head", "body", "tail") |
301
|
|
|
*/ |
302
|
12 |
|
protected function parseBranch() |
303
|
|
|
{ |
304
|
|
|
// Record the control structure |
305
|
12 |
|
$structure = $this->captureStructure(); |
306
|
|
|
|
307
|
|
|
// Record the output expressions at the start of this branch |
308
|
12 |
|
$head = $this->captureOutput(); |
309
|
12 |
|
$body = ''; |
310
|
12 |
|
$tail = []; |
311
|
|
|
|
312
|
12 |
|
$braces = 0; |
313
|
|
|
do |
314
|
|
|
{ |
315
|
12 |
|
$tail = $this->mergeOutput($tail, array_reverse($this->captureOutput())); |
316
|
12 |
|
if ($this->tokens[$this->i] === '}' && !$braces) |
317
|
|
|
{ |
318
|
12 |
|
break; |
319
|
|
|
} |
320
|
|
|
|
321
|
8 |
|
$body .= $this->serializeOutput(array_reverse($tail)); |
322
|
8 |
|
$tail = []; |
323
|
|
|
|
324
|
8 |
|
if ($this->tokens[$this->i][0] === T_IF) |
325
|
|
|
{ |
326
|
4 |
|
$child = $this->parseIfBlock(); |
327
|
|
|
|
328
|
|
|
// If this is the start of current branch, what's been optimized away and moved |
329
|
|
|
// outside, before the child branch is the head of this one. Otherwise it's just |
330
|
|
|
// part of its body |
331
|
4 |
|
if ($body === '') |
332
|
|
|
{ |
333
|
3 |
|
$head = $this->mergeOutput($head, $child['before']); |
334
|
|
|
} |
335
|
|
|
else |
336
|
|
|
{ |
337
|
1 |
|
$body .= $this->serializeOutput($child['before']); |
338
|
|
|
} |
339
|
|
|
|
340
|
4 |
|
$body .= $child['source']; |
341
|
4 |
|
$tail = $child['after']; |
342
|
|
|
} |
343
|
|
|
else |
344
|
|
|
{ |
345
|
5 |
|
$body .= $this->serializeToken($this->tokens[$this->i]); |
346
|
|
|
|
347
|
5 |
|
if ($this->tokens[$this->i] === '{') |
348
|
|
|
{ |
349
|
1 |
|
++$braces; |
350
|
|
|
} |
351
|
5 |
|
elseif ($this->tokens[$this->i] === '}') |
352
|
|
|
{ |
353
|
1 |
|
--$braces; |
354
|
|
|
} |
355
|
|
|
} |
356
|
|
|
} |
357
|
8 |
|
while (++$this->i < $this->cnt); |
358
|
|
|
|
359
|
|
|
return [ |
360
|
12 |
|
'structure' => $structure, |
361
|
12 |
|
'head' => $head, |
362
|
12 |
|
'body' => $body, |
363
|
12 |
|
'tail' => $tail |
364
|
|
|
]; |
365
|
|
|
} |
366
|
|
|
|
367
|
|
|
/** |
368
|
|
|
* Parse the if block (including elseif/else branches) starting at current index |
369
|
|
|
* |
370
|
|
|
* @return array |
371
|
|
|
*/ |
372
|
12 |
|
protected function parseIfBlock() |
373
|
|
|
{ |
374
|
12 |
|
$branches = []; |
375
|
|
|
do |
376
|
|
|
{ |
377
|
12 |
|
$branches[] = $this->parseBranch(); |
378
|
|
|
} |
379
|
12 |
|
while (++$this->i < $this->cnt && $this->isBranchToken()); |
380
|
|
|
|
381
|
|
|
// Move the index back to the last token used |
382
|
12 |
|
--$this->i; |
383
|
|
|
|
384
|
12 |
|
return $this->mergeIfBranches($branches); |
385
|
|
|
} |
386
|
|
|
|
387
|
|
|
/** |
388
|
|
|
* Serialize a recorded branch back to PHP |
389
|
|
|
* |
390
|
|
|
* @param array $branch |
391
|
|
|
* @return string |
392
|
|
|
*/ |
393
|
12 |
|
protected function serializeBranch(array $branch) |
394
|
|
|
{ |
395
|
|
|
// Optimize away "else{}" completely |
396
|
12 |
|
if ($branch['structure'] === 'else' |
397
|
12 |
|
&& $branch['body'] === '' |
398
|
12 |
|
&& empty($branch['head']) |
399
|
12 |
|
&& empty($branch['tail'])) |
400
|
|
|
{ |
401
|
3 |
|
return ''; |
402
|
|
|
} |
403
|
|
|
|
404
|
12 |
|
return $branch['structure'] . '{' . $this->serializeOutput($branch['head']) . $branch['body'] . $this->serializeOutput(array_reverse($branch['tail'])) . '}'; |
405
|
|
|
} |
406
|
|
|
|
407
|
|
|
/** |
408
|
|
|
* Serialize a series of recorded branch back to PHP |
409
|
|
|
* |
410
|
|
|
* @param array $block |
411
|
|
|
* @return string |
412
|
|
|
*/ |
413
|
12 |
|
protected function serializeIfBlock(array $block) |
414
|
|
|
{ |
415
|
12 |
|
return $this->serializeOutput($block['before']) . $block['source'] . $this->serializeOutput(array_reverse($block['after'])); |
416
|
|
|
} |
417
|
|
|
|
418
|
|
|
/** |
419
|
|
|
* Serialize a series of output expressions |
420
|
|
|
* |
421
|
|
|
* @param string[] $expressions Array of PHP expressions |
422
|
|
|
* @return string PHP code used to append given expressions to the output |
423
|
|
|
*/ |
424
|
12 |
|
protected function serializeOutput(array $expressions) |
425
|
|
|
{ |
426
|
12 |
|
if (empty($expressions)) |
427
|
|
|
{ |
428
|
12 |
|
return ''; |
429
|
|
|
} |
430
|
|
|
|
431
|
12 |
|
return '$this->out.=' . implode('.', $expressions) . ';'; |
432
|
|
|
} |
433
|
|
|
|
434
|
|
|
/** |
435
|
|
|
* Serialize a token back to PHP |
436
|
|
|
* |
437
|
|
|
* @param array|string $token Token from token_get_all() |
438
|
|
|
* @return string PHP code |
439
|
|
|
*/ |
440
|
12 |
|
protected function serializeToken($token) |
441
|
|
|
{ |
442
|
12 |
|
return (is_array($token)) ? $token[1] : $token; |
443
|
|
|
} |
444
|
|
|
|
445
|
|
|
/** |
446
|
|
|
* Attempt to move past output assignment at current index |
447
|
|
|
* |
448
|
|
|
* @return bool Whether if an output assignment was skipped |
449
|
|
|
*/ |
450
|
12 |
|
protected function skipOutputAssignment() |
451
|
|
|
{ |
452
|
12 |
|
if ($this->tokens[$this->i ][0] !== T_VARIABLE |
453
|
12 |
|
|| $this->tokens[$this->i ][1] !== '$this' |
454
|
12 |
|
|| $this->tokens[$this->i + 1][0] !== T_OBJECT_OPERATOR |
455
|
12 |
|
|| $this->tokens[$this->i + 2][0] !== T_STRING |
456
|
12 |
|
|| $this->tokens[$this->i + 2][1] !== 'out' |
457
|
12 |
|
|| $this->tokens[$this->i + 3][0] !== T_CONCAT_EQUAL) |
458
|
|
|
{ |
459
|
12 |
|
return false; |
460
|
|
|
} |
461
|
|
|
|
462
|
|
|
// Move past the concat assignment |
463
|
12 |
|
$this->i += 4; |
464
|
|
|
|
465
|
12 |
|
return true; |
466
|
|
|
} |
467
|
|
|
} |