Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
Total | |
100.00% |
45 / 45 |
|
100.00% |
7 / 7 |
CRAP | |
100.00% |
1 / 1 |
ConvertsMarkdownToPlainText | |
100.00% |
45 / 45 |
|
100.00% |
7 / 7 |
19 | |
100.00% |
1 / 1 |
__construct | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
execute | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
applyRegexTransformations | |
100.00% |
22 / 22 |
|
100.00% |
1 / 1 |
2 | |||
applyStringTransformations | |
100.00% |
7 / 7 |
|
100.00% |
1 / 1 |
2 | |||
removeTables | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
5 | |||
removeBlockquotes | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
3 | |||
trimWhitespace | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
5 |
1 | <?php |
2 | |
3 | declare(strict_types=1); |
4 | |
5 | namespace Hyde\Framework\Actions; |
6 | |
7 | use function trim; |
8 | use function rtrim; |
9 | use function is_numeric; |
10 | use function str_ends_with; |
11 | use function str_starts_with; |
12 | use function substr; |
13 | use function explode; |
14 | use function implode; |
15 | use function array_keys; |
16 | use function str_replace; |
17 | use function array_values; |
18 | use function preg_replace; |
19 | |
20 | /** |
21 | * Converts Markdown to plain text. |
22 | */ |
23 | class ConvertsMarkdownToPlainText |
24 | { |
25 | protected const ATX_HEADERS = ['/^(\n)?\s{0,}#{1,6}\s+| {0,}(\n)?\s{0,}#{0,} {0,}(\n)?\s{0,}$/m' => '$1$2$3']; |
26 | protected const SETEXT_HEADERS = ['/\n={2,}/' => "\n"]; |
27 | protected const HORIZONTAL_RULES = ['/^(-\s*?|\*\s*?|_\s*?){3,}\s*/m' => '']; |
28 | protected const HTML_TAGS = ['/<[^>]*>/' => '']; |
29 | protected const CODE_BLOCKS = ['/(`{3,})(.*?)\1/m' => '$2']; |
30 | protected const FENCED_CODEBLOCKS = ['/`{3}.*\n/' => '', '/`{3}/' => '']; |
31 | protected const TILDE_FENCED_CODEBLOCKS = ['/~{3}.*\n/' => '', '/~{3}/' => '']; |
32 | protected const INLINE_CODE = ['/`(.+?)`/' => '$1']; |
33 | protected const IMAGES = ['/\!\[(.*?)\][\[\(].*?[\]\)]/' => '$1']; |
34 | protected const INLINE_LINKS = ['/\[(.*?)\][\[\(].*?[\]\)]/' => '$1']; |
35 | protected const REFERENCE_LINKS = ['/^\s{1,2}\[(.*?)\]: (\S+)( ".*?")?\s*$/' => '']; |
36 | protected const STRIKETHROUGH = ['/~~/' => '']; |
37 | protected const BLOCKQUOTES = ['/^\s{0,3}>\s?/' => '']; |
38 | protected const FOOTNOTES = ['/\[\^.+?\](\: .*?$)?/' => '']; |
39 | protected const EMPHASIS = ['/([\*_]{1,3})(\S.*?\S{0,1})\1/' => '$2']; |
40 | |
41 | /** Emphasis (repeat the line to remove double emphasis) */ |
42 | protected const DOUBLE_EMPHASIS = self::EMPHASIS; |
43 | |
44 | /** Replace two or more newlines with exactly two */ |
45 | protected const REPEATED_NEWLINES = ['/\n{2,}/' => "\n\n"]; |
46 | |
47 | protected string $markdown; |
48 | |
49 | public function __construct(string $markdown) |
50 | { |
51 | $this->markdown = $markdown; |
52 | } |
53 | |
54 | /** |
55 | * Regex based on https://github.com/stiang/remove-markdown, licensed under MIT. |
56 | */ |
57 | public function execute(): string |
58 | { |
59 | return $this->applyStringTransformations($this->applyRegexTransformations($this->markdown)); |
60 | } |
61 | |
62 | protected function applyRegexTransformations(string $markdown): string |
63 | { |
64 | /** @var array<array-key, array<string, string>> $patterns */ |
65 | $patterns = [ |
66 | static::ATX_HEADERS, |
67 | static::SETEXT_HEADERS, |
68 | static::HORIZONTAL_RULES, |
69 | static::HTML_TAGS, |
70 | static::CODE_BLOCKS, |
71 | static::FENCED_CODEBLOCKS, |
72 | static::TILDE_FENCED_CODEBLOCKS, |
73 | static::INLINE_CODE, |
74 | static::IMAGES, |
75 | static::INLINE_LINKS, |
76 | static::REFERENCE_LINKS, |
77 | static::STRIKETHROUGH, |
78 | static::BLOCKQUOTES, |
79 | static::FOOTNOTES, |
80 | static::EMPHASIS, |
81 | static::DOUBLE_EMPHASIS, |
82 | static::REPEATED_NEWLINES, |
83 | ]; |
84 | |
85 | foreach ($patterns as $pattern) { |
86 | $markdown = preg_replace(array_keys($pattern), array_values($pattern), $markdown) ?? $markdown; |
87 | } |
88 | |
89 | return $markdown; |
90 | } |
91 | |
92 | protected function applyStringTransformations(string $markdown): string |
93 | { |
94 | $lines = explode("\n", $markdown); |
95 | foreach ($lines as $line => $contents) { |
96 | $contents = $this->removeTables($contents); |
97 | $contents = $this->removeBlockquotes($contents); |
98 | $contents = $this->trimWhitespace($contents); |
99 | |
100 | $lines[$line] = $contents; |
101 | } |
102 | |
103 | return implode("\n", $lines); |
104 | } |
105 | |
106 | protected function removeTables(string $contents): string |
107 | { |
108 | // Remove dividers |
109 | if (str_starts_with($contents, '|--') && str_ends_with($contents, '--|')) { |
110 | $contents = str_replace(['|', '-'], ['', ''], $contents); |
111 | } |
112 | // Remove cells |
113 | if (str_starts_with($contents, '| ') && str_ends_with($contents, '|')) { |
114 | $contents = rtrim(str_replace(['| ', ' | ', ' |'], ['', '', ''], $contents), ' '); |
115 | } |
116 | |
117 | return $contents; |
118 | } |
119 | |
120 | protected function removeBlockquotes(string $contents): string |
121 | { |
122 | // Remove blockquotes |
123 | if (str_starts_with($contents, '> ')) { |
124 | $contents = substr($contents, 2); |
125 | } |
126 | // Remove multiline blockquotes |
127 | if (str_starts_with($contents, '>')) { |
128 | $contents = substr($contents, 1); |
129 | } |
130 | |
131 | return $contents; |
132 | } |
133 | |
134 | protected function trimWhitespace(string $contents): string |
135 | { |
136 | // If it is a list, don't trim the whitespace |
137 | $firstCharacter = substr(trim($contents), 0, 1); |
138 | |
139 | if ($firstCharacter === '-' || $firstCharacter === '*' || $firstCharacter === '+' || is_numeric($firstCharacter)) { |
140 | return $contents; |
141 | } |
142 | |
143 | return trim($contents); |
144 | } |
145 | } |