1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
declare(strict_types=1); |
4
|
|
|
|
5
|
|
|
namespace ColinODell\Indentation; |
6
|
|
|
|
7
|
|
|
final class Indentation |
8
|
|
|
{ |
9
|
|
|
public const TYPE_SPACE = 'space'; |
10
|
|
|
public const TYPE_TAB = 'tab'; |
11
|
|
|
public const TYPE_UNKNOWN = 'unknown'; |
12
|
|
|
|
13
|
|
|
public int $amount; |
14
|
|
|
|
15
|
|
|
/** @var self::TYPE_* */ |
16
|
|
|
public string $type; |
17
|
|
|
|
18
|
|
|
/** |
19
|
|
|
* @param self::TYPE_* $type |
20
|
|
|
*/ |
21
|
|
|
public function __construct(int $amount, string $type) |
22
|
|
|
{ |
23
|
|
|
$this->amount = $amount; |
24
|
|
|
$this->type = $type; |
|
|
|
|
25
|
|
|
} |
26
|
|
|
|
27
|
|
|
public function __toString(): string |
28
|
|
|
{ |
29
|
|
|
if ($this->amount === 0 || $this->type === self::TYPE_UNKNOWN) { |
30
|
|
|
return ''; |
31
|
|
|
} |
32
|
|
|
|
33
|
|
|
$indentCharacter = $this->type === self::TYPE_SPACE ? ' ' : "\t"; |
34
|
|
|
|
35
|
|
|
return \str_repeat($indentCharacter, $this->amount); |
36
|
|
|
} |
37
|
|
|
|
38
|
|
|
/** |
39
|
|
|
* Detect the indentation of the given string. |
40
|
|
|
*/ |
41
|
|
|
public static function detect(string $string): Indentation |
42
|
|
|
{ |
43
|
|
|
// Identify indents while skipping single space indents to avoid common edge cases (e.g. code comments) |
44
|
|
|
$indents = self::makeIndentsMap($string, true); |
45
|
|
|
// If no indents are identified, run again and include all indents for comprehensive detection |
46
|
|
|
if (\count($indents) === 0) { |
47
|
|
|
$indents = self::makeIndentsMap($string, false); |
48
|
|
|
} |
49
|
|
|
|
50
|
|
|
$keyOfMostUsedIndent = self::getMostUsedKey($indents); |
51
|
|
|
if ($keyOfMostUsedIndent === null) { |
52
|
|
|
return new self(0, self::TYPE_UNKNOWN); |
53
|
|
|
} |
54
|
|
|
|
55
|
|
|
[$amount, $type] = self::decodeIndentsKey($keyOfMostUsedIndent); |
56
|
|
|
|
57
|
|
|
return new self($amount, $type); |
58
|
|
|
} |
59
|
|
|
|
60
|
|
|
public static function change(string $string, Indentation $newStyle): string |
61
|
|
|
{ |
62
|
|
|
$oldStyle = self::detect($string); |
63
|
|
|
|
64
|
|
|
if ($oldStyle->type === self::TYPE_UNKNOWN || $oldStyle->amount === 0) { |
65
|
|
|
return $string; |
66
|
|
|
} |
67
|
|
|
|
68
|
|
|
$lines = \preg_split('/(\R)/', $string, flags: \PREG_SPLIT_DELIM_CAPTURE); |
69
|
|
|
if ($lines === false) { |
70
|
|
|
throw new \InvalidArgumentException('Bad input string'); |
71
|
|
|
} |
72
|
|
|
|
73
|
|
|
$newContent = ''; |
74
|
|
|
foreach ($lines as $i => $line) { |
75
|
|
|
// Newline characters are in the odd-numbered positions |
76
|
|
|
if ($i % 2 === 1) { |
77
|
|
|
$newContent .= $line; |
78
|
|
|
continue; |
79
|
|
|
} |
80
|
|
|
|
81
|
|
|
if (\preg_match('/^(?:' . \preg_quote($oldStyle->__toString(), '/') . ')+/', $line, $matches) !== 1) { |
82
|
|
|
$newContent .= $line; |
83
|
|
|
continue; |
84
|
|
|
} |
85
|
|
|
|
86
|
|
|
$indentLevel = (int) (\strlen($matches[0]) / $oldStyle->amount); |
87
|
|
|
$newContent .= \str_repeat($newStyle->__toString(), $indentLevel) . \substr($line, $indentLevel * $oldStyle->amount); |
88
|
|
|
} |
89
|
|
|
|
90
|
|
|
return $newContent; |
91
|
|
|
} |
92
|
|
|
|
93
|
|
|
/** |
94
|
|
|
* @return array<string, array{0: int, 1: int}> |
95
|
|
|
*/ |
96
|
|
|
private static function makeIndentsMap(string $string, bool $ignoreSingleSpaces): array |
97
|
|
|
{ |
98
|
|
|
$indents = []; |
99
|
|
|
|
100
|
|
|
// Remember the size of previous line's indentation |
101
|
|
|
$previousSize = 0; |
102
|
|
|
$previousIndentType = null; |
103
|
|
|
|
104
|
|
|
// Indents key (ident type + size of the indents/unindents) |
105
|
|
|
$key = null; |
106
|
|
|
|
107
|
|
|
$lines = \preg_split('/\R/', $string); |
108
|
|
|
if ($lines === false) { |
109
|
|
|
throw new \InvalidArgumentException('Invalid string'); |
110
|
|
|
} |
111
|
|
|
|
112
|
|
|
foreach ($lines as $line) { |
113
|
|
|
if ($line === '') { |
114
|
|
|
// Ignore empty lines |
115
|
|
|
continue; |
116
|
|
|
} |
117
|
|
|
|
118
|
|
|
// Detect either spaces or tabs but not both to properly handle tabs for indentation and spaces for alignment |
119
|
|
|
if (\preg_match('/^(?:( )+|\t+)/', $line, $matches) !== 1) { |
120
|
|
|
$previousSize = 0; |
121
|
|
|
$previousIndentType = ''; |
122
|
|
|
continue; |
123
|
|
|
} |
124
|
|
|
|
125
|
|
|
$indent = \strlen($matches[0]); |
126
|
|
|
$indentType = isset($matches[1]) ? self::TYPE_SPACE : self::TYPE_TAB; |
127
|
|
|
// Ignore single space unless it's the only indent detected to prevent common false positives |
128
|
|
|
if ($ignoreSingleSpaces && $indentType === self::TYPE_SPACE && $indent === 1) { |
129
|
|
|
continue; |
130
|
|
|
} |
131
|
|
|
|
132
|
|
|
if ($indentType !== $previousIndentType) { |
133
|
|
|
$previousSize = 0; |
134
|
|
|
} |
135
|
|
|
|
136
|
|
|
$previousIndentType = $indentType; |
137
|
|
|
$weight = 0; |
138
|
|
|
$indentDifference = $indent - $previousSize; |
139
|
|
|
$previousSize = $indent; |
140
|
|
|
|
141
|
|
|
// Previous line have same indent? |
142
|
|
|
if ($indentDifference === 0) { |
143
|
|
|
$weight++; |
144
|
|
|
// We use the key from previous loop |
145
|
|
|
\assert(isset($key) && \is_string($key)); |
146
|
|
|
} else { |
147
|
|
|
$key = self::encodeIndentsKey($indentType, $indentDifference > 0 ? $indentDifference : -$indentDifference); |
148
|
|
|
} |
149
|
|
|
|
150
|
|
|
// Update the stats |
151
|
|
|
if (! isset($indents[$key])) { |
152
|
|
|
$indents[$key] = [1, 0]; |
153
|
|
|
} else { |
154
|
|
|
$indents[$key][0]++; |
155
|
|
|
$indents[$key][1] += $weight; |
156
|
|
|
} |
157
|
|
|
} |
158
|
|
|
|
159
|
|
|
return $indents; |
160
|
|
|
} |
161
|
|
|
|
162
|
|
|
/** |
163
|
|
|
* Encode the indent type and amount as a string (e.g. 's4') for use as a compound key in the indents map. |
164
|
|
|
*/ |
165
|
|
|
private static function encodeIndentsKey(string $indentType, int $indentAmount): string |
166
|
|
|
{ |
167
|
|
|
$typeCharacter = $indentType === self::TYPE_SPACE ? 's' : 't'; |
168
|
|
|
|
169
|
|
|
return $typeCharacter . $indentAmount; |
170
|
|
|
} |
171
|
|
|
|
172
|
|
|
/** |
173
|
|
|
* Extract the indent type and amount from a key of the indents map. |
174
|
|
|
* |
175
|
|
|
* @return array{0: int, 1: self::TYPE_*} |
|
|
|
|
176
|
|
|
*/ |
177
|
|
|
private static function decodeIndentsKey(string $indentsKey): array |
178
|
|
|
{ |
179
|
|
|
$keyHasTypeSpace = $indentsKey[0] === 's'; |
180
|
|
|
$type = $keyHasTypeSpace ? self::TYPE_SPACE : self::TYPE_TAB; |
181
|
|
|
|
182
|
|
|
$amount = \intval(\substr($indentsKey, 1)); |
183
|
|
|
|
184
|
|
|
return [$amount, $type]; |
185
|
|
|
} |
186
|
|
|
|
187
|
|
|
/** |
188
|
|
|
* Return the key (e.g. 's4') from the indents map that represents the most common indent, |
189
|
|
|
* or return undefined if there are no indents. |
190
|
|
|
* |
191
|
|
|
* @param array<string, array{int, int}> $indents |
|
|
|
|
192
|
|
|
*/ |
193
|
|
|
private static function getMostUsedKey(array $indents): string|null |
194
|
|
|
{ |
195
|
|
|
$result = null; |
196
|
|
|
$maxUsed = 0; |
197
|
|
|
$maxWeight = 0; |
198
|
|
|
|
199
|
|
|
foreach ($indents as $key => [$usedCount, $weight]) { |
200
|
|
|
if ($usedCount <= $maxUsed && ($usedCount !== $maxUsed || $weight <= $maxWeight)) { |
201
|
|
|
continue; |
202
|
|
|
} |
203
|
|
|
|
204
|
|
|
$maxUsed = $usedCount; |
205
|
|
|
$maxWeight = $weight; |
206
|
|
|
$result = $key; |
207
|
|
|
} |
208
|
|
|
|
209
|
|
|
return $result; |
210
|
|
|
} |
211
|
|
|
} |
212
|
|
|
|
Our type inference engine has found an assignment to a property that is incompatible with the declared type of that property.
Either this assignment is in error or the assigned type should be added to the documentation/type hint for that property..