Code Coverage
 
Lines
Functions and Methods
Classes and Traits
Total
95.74%
90 / 94
81.82%
9 / 11
 CRAP
0.00%
0 / 1
AnyMatch
95.74%
90 / 94
81.82%
9 / 11
 34
0.00%
0 / 1
 __construct
100.00%
1 / 1
100.00%
1 / 1
 1
 __toString
100.00%
1 / 1
100.00%
1 / 1
 1
 getRegexStringLength
100.00%
1 / 1
100.00%
1 / 1
 1
 getMinimalPossibleLength
100.00%
1 / 1
100.00%
1 / 1
 1
 getMaximumPossibleLength
100.00%
1 / 1
100.00%
1 / 1
 1
 makeAllIncludedArray
93.48%
43 / 46
0.00%
0 / 1
 15.06
 toCaseInsensitive
100.00%
9 / 9
100.00%
1 / 1
 2
 toDotAll
90.91%
10 / 11
0.00%
0 / 1
 3.01
 createRange
100.00%
16 / 16
100.00%
1 / 1
 5
 renderRange
100.00%
6 / 6
100.00%
1 / 1
 3
 removeStartAndEndMarkers
100.00%
1 / 1
100.00%
1 / 1
 1
1<?php
2namespace Apie\RegexTools\Parts;
3
4final class AnyMatch implements RegexPartInterface
5{
6    public function __construct(
7        public readonly string $part
8    ) {
9    }
10
11    public function __toString(): string
12    {
13        return '[' . $this->part . ']';
14    }
15
16    public function getRegexStringLength(): int
17    {
18        return 2 + strlen($this->part);
19    }
20
21    public function getMinimalPossibleLength(): int
22    {
23        return 1;
24    }
25
26    public function getMaximumPossibleLength(): int
27    {
28        return 1;
29    }
30
31    private function makeAllIncludedArray(?callable $callback): array
32    {
33        $split = mb_str_split($this->part);
34        $included = [];
35        $nextEscaped = false;
36        $skip = 0;
37        if ($split[0] === '^') {
38            $skip = 1;
39        }
40        foreach ($split as $key => $character) {
41            if ($skip > 0) {
42                $skip--;
43                continue;
44            }
45            // escaped character \
46            if ($character === '\\') {
47                if ($nextEscaped) {
48                    $included[] = '\\\\';
49                    $nextEscaped = false;
50                    continue;
51                }
52                $nextEscaped = true;
53                continue;
54            }
55            // previous character was \
56            if ($nextEscaped) {
57                $included[] = '\\' . $character;
58                $nextEscaped = false;
59                // todo \p and \P
60                continue;
61            }
62            // range
63            if (($split[$key + 1] ?? null) === '-' && !empty($split[$key + 2]) && $split[$key + 2] > $character) {
64                $minCode = mb_ord($character);
65                $maxCode = mb_ord($split[$key + 2]);
66
67                for ($code = $minCode; $code <= $maxCode; $code++) {
68                    $chr = mb_chr($code);
69                    foreach ($callback($chr) as $added) {
70                        $included[] = $added;
71                    }
72                }
73                $skip = 2;
74            } else {
75                foreach ($callback($character) as $added) {
76                    $included[] = $added;
77                }
78            }
79        }
80        $included = array_filter(
81            array_map(
82                function (string $word) {
83                    if (mb_substr($word, 0, 1) === '\\') {
84                        return mb_substr($word, 1);
85                    }
86                    return $word;
87                },
88                array_unique($included)
89            ),
90            function (string $input) {
91                // for some reason mb_strtoupper('ß') returns 'SS'
92                return $input !== '' && $input !== 'SS';
93            }
94        );
95        sort($included);
96        return $included;
97    }
98
99    public function toCaseInsensitive(): RegexPartInterface
100    {
101        $prefix = '';
102        if ($this->part[0] === '^') {
103            $prefix = '^';
104        }
105        $included = $this->makeAllIncludedArray(function (string $chr): array {
106            return [$chr, mb_strtoupper($chr), mb_strtolower($chr)];
107        });
108        return new AnyMatch(
109            $prefix . self::createRange($included)
110        );
111    }
112
113    public function toDotAll(): RegexPartInterface
114    {
115        $prefix = '';
116        if ($this->part[0] === '^') {
117            $prefix = '^';
118        }
119        $included = $this->makeAllIncludedArray(function (string $chr): array {
120            if ($chr === '.') {
121                return ['.', "\n", "\r"];
122            }
123            return [$chr];
124        });
125        return new AnyMatch(
126            $prefix . self::createRange($included)
127        );
128    }
129
130    /**
131     * @param list<string> $included
132     */
133    private static function createRange(array $included): string
134    {
135        $result = '';
136        $currentRange = null;
137        $startRange = null;
138
139        while (!empty($included)) {
140            $character = array_shift($included);
141            if ($startRange === null) {
142                $startRange = $character;
143                $currentRange = mb_ord($character) + 1;
144            } else {
145                if (mb_ord($character) === $currentRange) {
146                    $currentRange++;
147                } else {
148                    $result .= self::renderRange($startRange, $currentRange);
149                    
150                    $startRange = $character;
151                    $currentRange = mb_ord($character) + 1;
152                }
153            }
154        }
155        if ($startRange !== null) {
156            $result .= self::renderRange($startRange, $currentRange);
157        }
158
159        return $result;
160    }
161
162    private static function renderRange(?string $startRange, ?int $currentRange): string
163    {
164        $last = mb_chr($currentRange - 1);
165        if ($last === $startRange) {
166            return preg_quote($startRange);
167        }
168
169        if ($currentRange - 1 === mb_ord($startRange) + 1) {
170            return preg_quote($startRange) . preg_quote($last);
171        }
172        
173        return preg_quote($startRange) . '-' . preg_quote($last);
174    }
175
176    public function removeStartAndEndMarkers(): ?RegexPartInterface
177    {
178        return $this;
179    }
180}