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