1
|
|
|
<?php declare(strict_types = 1); |
2
|
|
|
|
3
|
|
|
namespace Suilven\RandomEnglish; |
4
|
|
|
|
5
|
|
|
use Faker\Factory; |
6
|
|
|
|
7
|
|
|
class RandomEnglishGenerator |
8
|
|
|
{ |
9
|
|
|
|
10
|
|
|
private const POSSIBLE_WORD_TYPES = [ |
11
|
|
|
'noun', |
12
|
|
|
'countable_noun', |
13
|
|
|
|
14
|
|
|
'adjective', |
15
|
|
|
'preposition', |
16
|
|
|
'conjunction', |
17
|
|
|
'contraction', |
18
|
|
|
'control_verb', |
19
|
|
|
|
20
|
|
|
'sequence_adverb', |
21
|
|
|
'frequency_adverb', |
22
|
|
|
|
23
|
|
|
// pluralize and ing these? |
24
|
|
|
'intransitive_verb', |
25
|
|
|
'irregular_verb', |
26
|
|
|
'transitive_verb', |
27
|
|
|
'verb', |
28
|
|
|
'adverb', |
29
|
|
|
|
30
|
|
|
'intransitive_verb_ing', |
31
|
|
|
'irregular_verb_ing', |
32
|
|
|
'transitive_verb_ing', |
33
|
|
|
'verb_ing', |
34
|
|
|
|
35
|
|
|
|
36
|
|
|
// positions |
37
|
|
|
'locative', |
38
|
|
|
|
39
|
|
|
'plural_noun', |
40
|
|
|
|
41
|
|
|
|
42
|
|
|
// names |
43
|
|
|
'mens_name', |
44
|
|
|
'womens_name', |
45
|
|
|
'country', |
46
|
|
|
'colour', |
47
|
|
|
]; |
48
|
|
|
|
49
|
|
|
private const POSSIBLE_FAKER_TYPES=[ |
50
|
|
|
'address', |
51
|
|
|
'name', |
52
|
|
|
'randomDigit', |
53
|
|
|
'randomLetter', |
54
|
|
|
'randomNumber', |
55
|
|
|
'title', |
56
|
|
|
'titleMale', |
57
|
|
|
'titleFemale', |
58
|
|
|
'firstNameMale', |
59
|
|
|
'firstNameFemale', |
60
|
|
|
'lastName', |
61
|
|
|
|
62
|
|
|
'catchPhrase', |
63
|
|
|
'bs', |
64
|
|
|
'company', |
65
|
|
|
'companySuffix', |
66
|
|
|
'jobTitle', |
67
|
|
|
|
68
|
|
|
'realText', |
69
|
|
|
|
70
|
|
|
'dayOfWeek', |
71
|
|
|
'monthName', |
72
|
|
|
|
73
|
|
|
]; |
74
|
|
|
|
75
|
|
|
/** @var string configuration file for sentences */ |
76
|
|
|
private $config = ''; |
77
|
|
|
|
78
|
|
|
public function __construct() |
79
|
|
|
{ |
80
|
|
|
$configFile = \dirname(__FILE__) . '/../config/sentence-structure.cfg'; |
81
|
|
|
$this->setConfig(\file_get_contents($configFile)); |
82
|
|
|
} |
83
|
|
|
|
84
|
|
|
|
85
|
|
|
/** @param string $newConfig configuration file of how sentence structures */ |
86
|
|
|
public function setConfig(string $newConfig): void |
87
|
|
|
{ |
88
|
|
|
$this->config = \trim($newConfig); |
89
|
|
|
} |
90
|
|
|
|
91
|
|
|
|
92
|
|
|
/** |
93
|
|
|
* Generate a random sentence |
94
|
|
|
* |
95
|
|
|
* @param bool $titleCase true to generate a title, false (default) to generate a sentence |
96
|
|
|
* @return string a random sentence |
97
|
|
|
*/ |
98
|
|
|
public function sentence(bool $titleCase = false): string |
99
|
|
|
{ |
100
|
|
|
// choose a random sentence structure |
101
|
|
|
$structures = \explode(\PHP_EOL, $this->config); |
102
|
|
|
\shuffle($structures); |
103
|
|
|
$structure = $structures[0]; |
104
|
|
|
|
105
|
|
|
$expression = '/[\s]+/'; |
106
|
|
|
$splits = \preg_split($expression, $structure, -1, \PREG_SPLIT_NO_EMPTY); |
107
|
|
|
|
108
|
|
|
$sentenceArray = []; |
109
|
|
|
|
110
|
|
|
/** @var string $possiblyRandomWord */ |
111
|
|
|
foreach ($splits as $possiblyRandomWord) { |
112
|
|
|
$word = $this->getWordOrRandomWord($possiblyRandomWord); |
113
|
|
|
$sentenceArray[] = $word; |
114
|
|
|
} |
115
|
|
|
|
116
|
|
|
$result = $this->makeArrayIntoSentence($sentenceArray); |
117
|
|
|
$this->augmentSentenceTitleCase($titleCase, $result); |
118
|
|
|
|
119
|
|
|
return $result; |
120
|
|
|
} |
121
|
|
|
|
122
|
|
|
|
123
|
|
|
/** |
124
|
|
|
* Generate a random title |
125
|
|
|
* |
126
|
|
|
* @return string a random sentence, all in caps, no trailing full stop |
127
|
|
|
*/ |
128
|
|
|
public function title(): string |
129
|
|
|
{ |
130
|
|
|
return $this->sentence(true); |
131
|
|
|
} |
132
|
|
|
|
133
|
|
|
|
134
|
|
|
/** |
135
|
|
|
* @param int $maxSentences The maximum number of sentences |
136
|
|
|
* @return string a paragraph of random sentences |
137
|
|
|
*/ |
138
|
|
|
public function paragraph(int $maxSentences = 10): string |
139
|
|
|
{ |
140
|
|
|
$nParagraph = \rand(1, $maxSentences); |
141
|
|
|
$sentences = []; |
142
|
|
|
|
143
|
|
|
for ($i = 0; $i < $nParagraph; $i++) { |
144
|
|
|
$sentences[] = $this->sentence(); |
145
|
|
|
} |
146
|
|
|
|
147
|
|
|
return \implode(" ", $sentences); |
148
|
|
|
} |
149
|
|
|
|
150
|
|
|
|
151
|
|
|
/** @return string a plural noun */ |
152
|
|
|
public function pluralNoun(): string |
153
|
|
|
{ |
154
|
|
|
$plurableNoun = $this->getRandomWord('countable_noun'); |
155
|
|
|
|
156
|
|
|
return $plurableNoun . 's'; |
157
|
|
|
} |
158
|
|
|
|
159
|
|
|
|
160
|
|
|
/** @return string the ing version of a verb, e.g. run -> running */ |
161
|
|
|
public function verbing(): string |
162
|
|
|
{ |
163
|
|
|
// @todo Choose a random verb source file |
164
|
|
|
$ingVerb = $this->getRandomWord('verb'); |
165
|
|
|
|
166
|
|
|
return $ingVerb . 'ing'; |
167
|
|
|
} |
168
|
|
|
|
169
|
|
|
|
170
|
|
|
/** |
171
|
|
|
* @param string $wordType The word type, e.g. 'noun' or 'verb' |
172
|
|
|
* @return bool true if this is a plural noun |
173
|
|
|
*/ |
174
|
|
|
public function pluralNounCheck(string &$wordType, string &$possiblyRandomWord): bool |
175
|
|
|
{ |
176
|
|
|
$pluralNoun = ($wordType === 'plural_noun'); |
177
|
|
|
if ($pluralNoun) { |
178
|
|
|
$wordType = 'noun'; |
179
|
|
|
$possiblyRandomWord = \str_replace('plural_', '', $possiblyRandomWord); |
180
|
|
|
} |
181
|
|
|
|
182
|
|
|
return $pluralNoun; |
183
|
|
|
} |
184
|
|
|
|
185
|
|
|
|
186
|
|
|
/** @return bool true if this is a verb with ing */ |
187
|
|
|
public function ingVerbCheck(string &$wordType, string &$possiblyRandomWord): bool |
188
|
|
|
{ |
189
|
|
|
$ing = \substr($wordType, -4) === '_ing'; |
190
|
|
|
if ($ing) { |
191
|
|
|
$wordType = \str_replace('_ing', '', $wordType); |
192
|
|
|
$possiblyRandomWord = \str_replace('_ing', '', $possiblyRandomWord); |
193
|
|
|
} |
194
|
|
|
|
195
|
|
|
return $ing; |
196
|
|
|
} |
197
|
|
|
|
198
|
|
|
|
199
|
|
|
public function augmentIfPluralNoun(bool $pluralNoun, string &$randomWord): void |
200
|
|
|
{ |
201
|
|
|
if (!$pluralNoun) { |
202
|
|
|
return; |
203
|
|
|
} |
204
|
|
|
|
205
|
|
|
$randomWord = $this->pluralNoun(); |
206
|
|
|
} |
207
|
|
|
|
208
|
|
|
|
209
|
|
|
public function augmentIfIngVerb(bool $ing, string &$randomWord): void |
210
|
|
|
{ |
211
|
|
|
if (!$ing) { |
212
|
|
|
return; |
213
|
|
|
} |
214
|
|
|
|
215
|
|
|
$randomWord = $this->verbing(); |
216
|
|
|
} |
217
|
|
|
|
218
|
|
|
|
219
|
|
|
public function augmentSentenceTitleCase(bool $titleCase, string &$result): void |
220
|
|
|
{ |
221
|
|
|
if (!$titleCase) { |
222
|
|
|
return; |
223
|
|
|
} |
224
|
|
|
|
225
|
|
|
$result = \ucwords($result); |
226
|
|
|
$result = \substr_replace($result, "", -1); |
227
|
|
|
} |
228
|
|
|
|
229
|
|
|
|
230
|
|
|
/** |
231
|
|
|
* Convert an array of strings into a sentence |
232
|
|
|
* 1) Capitalize first letter |
233
|
|
|
* 2) Glue words together with a space |
234
|
|
|
* 3) Add a full stop |
235
|
|
|
* 4) Deal with diferent sentence endings such as ! or ? |
236
|
|
|
* |
237
|
|
|
* @param array<string> $sentenceArray a sentence whose words are in an array of strings |
238
|
|
|
* @return string a sentence |
239
|
|
|
*/ |
240
|
|
|
public function makeArrayIntoSentence(array $sentenceArray): string |
241
|
|
|
{ |
242
|
|
|
// ensure sentence starts with a capital |
243
|
|
|
$sentenceArray[0] = \ucfirst($sentenceArray[0]); |
244
|
|
|
|
245
|
|
|
$sentence = \implode(" ", $sentenceArray) . '.'; |
246
|
|
|
$this->fixEndOfSentence($sentence); |
247
|
|
|
|
248
|
|
|
return $sentence; |
249
|
|
|
} |
250
|
|
|
|
251
|
|
|
|
252
|
|
|
/** |
253
|
|
|
* If a sentence ended with a question or exclamation mark prior to having a full stop appended, deal with it |
254
|
|
|
* |
255
|
|
|
* @param string $result The sentence, by reference, prior to being fixed |
256
|
|
|
*/ |
257
|
|
|
public function fixEndOfSentence(string &$result): void |
258
|
|
|
{ |
259
|
|
|
$result = \str_replace('?.', '?', $result); |
260
|
|
|
$result = \str_replace('!.', '?', $result); |
261
|
|
|
$result = \str_replace('..', '.', $result); |
262
|
|
|
} |
263
|
|
|
|
264
|
|
|
|
265
|
|
|
/** |
266
|
|
|
* @param string $wordType the type of word, e.g. noun, verb |
267
|
|
|
* @return string a random word for the given word type |
268
|
|
|
*/ |
269
|
|
|
private function getRandomWord(string $wordType): string |
270
|
|
|
{ |
271
|
|
|
$wordsFile = \dirname(__FILE__) . '/../words/english_' . $wordType . 's.txt'; |
272
|
|
|
$words = \explode(\PHP_EOL, \trim(\file_get_contents($wordsFile))); |
273
|
|
|
\shuffle($words); |
274
|
|
|
|
275
|
|
|
return $words[0]; |
276
|
|
|
} |
277
|
|
|
|
278
|
|
|
|
279
|
|
|
/** @return string the above string as is, or a possibly randomized version, e.g. a random noun */ |
280
|
|
|
private function getWordOrRandomWord(string $possiblyRandomWord): string |
281
|
|
|
{ |
282
|
|
|
$result = $possiblyRandomWord; |
283
|
|
|
|
284
|
|
|
if (\strstr($possiblyRandomWord, '|')) { |
285
|
|
|
$splits = \explode('|', $possiblyRandomWord); |
286
|
|
|
\shuffle($splits); |
287
|
|
|
$result = $splits[0]; |
288
|
|
|
} else { |
289
|
|
|
foreach (self::POSSIBLE_WORD_TYPES as $wordType) { |
290
|
|
|
$pluralNoun = $this->pluralNounCheck($wordType, $possiblyRandomWord); |
291
|
|
|
$ing = $this->ingVerbCheck($wordType, $possiblyRandomWord); |
292
|
|
|
|
293
|
|
|
$start = '[' . $wordType . ']'; |
294
|
|
|
|
295
|
|
|
// check the start of the word as it may be suffixed by likes of a question of exclamation mark. |
296
|
|
|
// If no match for the possible word type continue until the next one |
297
|
|
|
if (\substr($possiblyRandomWord, 0, \strlen($start)) !== $start) { |
298
|
|
|
continue; |
299
|
|
|
} |
300
|
|
|
|
301
|
|
|
// ---- we are now getting a random word from a file ---- |
302
|
|
|
$result = $this->chooseRandomWord($possiblyRandomWord, $wordType, $pluralNoun, $ing); |
303
|
|
|
|
304
|
|
|
break; |
305
|
|
|
} |
306
|
|
|
|
307
|
|
|
$faker = Factory::create(); |
308
|
|
|
foreach (self::POSSIBLE_FAKER_TYPES as $wordType) { |
309
|
|
|
$start = '[' . $wordType . ']'; |
310
|
|
|
|
311
|
|
|
// check the start of the word as it may be suffixed by likes of a question of exclamation mark. |
312
|
|
|
// If no match for the possible word type continue until the next one |
313
|
|
|
if (\substr($possiblyRandomWord, 0, \strlen($start)) !== $start) { |
314
|
|
|
continue; |
315
|
|
|
} |
316
|
|
|
|
317
|
|
|
$result = $faker->$wordType; |
318
|
|
|
} |
319
|
|
|
} |
320
|
|
|
|
321
|
|
|
// no randomized word has been found, aka no [verb] or [noun], thus append the possibly random word as is |
322
|
|
|
return $result; |
323
|
|
|
} |
324
|
|
|
|
325
|
|
|
|
326
|
|
|
/** |
327
|
|
|
* @param string $wordType noun, verb, adjective etc |
328
|
|
|
* @param bool $pluralizeNoun if true pluralize a noun |
329
|
|
|
* @param bool $makeVerbIng if true make a verb an ing verb, e.g. run -> running |
330
|
|
|
*/ |
331
|
|
|
private function chooseRandomWord( |
332
|
|
|
string $randomWordTypeWithSuffix, |
333
|
|
|
string $wordType, |
334
|
|
|
bool $pluralizeNoun, |
335
|
|
|
bool $makeVerbIng |
336
|
|
|
): string { |
337
|
|
|
$start = '[' . $wordType . ']'; |
338
|
|
|
|
339
|
|
|
// note the suffix of the word, and append this to the random word |
340
|
|
|
$restOfWord = \str_replace($start, '', $randomWordTypeWithSuffix); |
341
|
|
|
|
342
|
|
|
// This avoids having to use 'countrie' in the sentence structure file : country -> countries |
343
|
|
|
$wordType = \str_replace('country', 'countrie', $wordType); |
344
|
|
|
|
345
|
|
|
/** @var string $randomWord */ |
346
|
|
|
$randomWord = $this->getRandomWord($wordType); |
347
|
|
|
|
348
|
|
|
// augment the random word if a plural noun or an ing verb |
349
|
|
|
$this->augmentIfPluralNoun($pluralizeNoun, $randomWord); |
350
|
|
|
$this->augmentIfIngVerb($makeVerbIng, $randomWord); |
351
|
|
|
|
352
|
|
|
// add the random word, and then the suffix of the word such as ?! or !! |
353
|
|
|
return $randomWord . $restOfWord; |
354
|
|
|
} |
355
|
|
|
} |
356
|
|
|
|