1 | <?php |
||
2 | |||
3 | declare(strict_types=1); |
||
4 | |||
5 | namespace EdmondsCommerce\PHPQA\Markdown; |
||
6 | |||
7 | use EdmondsCommerce\PHPQA\Helper; |
||
8 | use Exception; |
||
9 | use RecursiveDirectoryIterator; |
||
10 | use RecursiveIteratorIterator; |
||
11 | use RecursiveRegexIterator; |
||
12 | use RegexIterator; |
||
13 | use RuntimeException; |
||
14 | use Throwable; |
||
15 | |||
16 | use function current; |
||
17 | use function dirname; |
||
18 | use function explode; |
||
19 | use function preg_match; |
||
20 | use function preg_replace; |
||
21 | use function realpath; |
||
22 | use function rtrim; |
||
23 | use function sprintf; |
||
24 | use function str_repeat; |
||
25 | use function str_replace; |
||
26 | use function strlen; |
||
27 | use function strpos; |
||
28 | use function trim; |
||
29 | |||
30 | final class LinksChecker |
||
31 | { |
||
32 | /** |
||
33 | * @throws Exception |
||
34 | * @SuppressWarnings(PHPMD.StaticAccess) |
||
35 | */ |
||
36 | 4 | public static function main(string $projectRootDirectory = null): int |
|
37 | { |
||
38 | 4 | $return = 0; |
|
39 | 4 | $projectRootDirectory = $projectRootDirectory ?? Helper::getProjectRootDirectory(); |
|
40 | 4 | $files = static::getFiles($projectRootDirectory); |
|
41 | 3 | foreach ($files as $file) { |
|
42 | 3 | $relativeFile = str_replace($projectRootDirectory, '', $file); |
|
43 | 3 | $title = "\n{$relativeFile}\n" . str_repeat('-', strlen($relativeFile)) . "\n"; |
|
44 | 3 | $errors = []; |
|
45 | 3 | $links = static::getLinks($file); |
|
46 | 3 | foreach ($links as $link) { |
|
47 | 3 | static::checkLink($projectRootDirectory, $link, $file, $errors, $return); |
|
48 | } |
||
49 | 3 | if ([] !== $errors) { |
|
50 | 2 | echo $title . implode('', $errors); |
|
51 | } |
||
52 | } |
||
53 | |||
54 | 3 | return $return; |
|
55 | } |
||
56 | |||
57 | /** |
||
58 | * @return string[] |
||
59 | * @SuppressWarnings(PHPMD.StaticAccess) |
||
60 | */ |
||
61 | 4 | private static function getFiles(string $projectRootDirectory): array |
|
62 | { |
||
63 | 4 | $files = self::getDocsFiles($projectRootDirectory); |
|
64 | 4 | $files[] = self::getMainReadme($projectRootDirectory); |
|
65 | |||
66 | 3 | return $files; |
|
67 | } |
||
68 | |||
69 | /** |
||
70 | * @SuppressWarnings(PHPMD.StaticAccess) |
||
71 | * |
||
72 | * @return string[] |
||
73 | */ |
||
74 | 4 | private static function getDocsFiles(string $projectRootDirectory): array |
|
75 | { |
||
76 | 4 | $files = []; |
|
77 | 4 | $dir = $projectRootDirectory . '/docs'; |
|
78 | 4 | if (!is_dir($dir)) { |
|
79 | 2 | return $files; |
|
80 | } |
||
81 | 2 | $directory = new RecursiveDirectoryIterator($dir); |
|
82 | 2 | $recursive = new RecursiveIteratorIterator($directory); |
|
83 | 2 | $regex = new RegexIterator( |
|
84 | 2 | $recursive, |
|
85 | 2 | '/^.+\.md/i', |
|
86 | 2 | RecursiveRegexIterator::GET_MATCH |
|
87 | ); |
||
88 | 2 | foreach ($regex as $file) { |
|
89 | 2 | if ($file[0] !== '') { |
|
90 | 2 | $files[] = $file[0]; |
|
91 | } |
||
92 | } |
||
93 | |||
94 | 2 | return $files; |
|
95 | } |
||
96 | |||
97 | /** |
||
98 | * @SuppressWarnings(PHPMD.StaticAccess) |
||
99 | */ |
||
100 | 4 | private static function getMainReadme(string $projectRootDirectory): string |
|
101 | { |
||
102 | 4 | $path = $projectRootDirectory . '/README.md'; |
|
103 | 4 | if (!is_file($path)) { |
|
104 | 1 | throw new RuntimeException( |
|
105 | "\n\nYou have no README.md file in your project" |
||
106 | 1 | . "\n\nAs the bear minimum you need to have this file to pass QA" |
|
107 | ); |
||
108 | } |
||
109 | |||
110 | 3 | return $path; |
|
111 | } |
||
112 | |||
113 | /** |
||
114 | * @SuppressWarnings(PHPMD.StaticAccess) |
||
115 | * |
||
116 | * @return array<array<string>> |
||
117 | */ |
||
118 | 3 | private static function getLinks(string $file): array |
|
119 | { |
||
120 | 3 | $links = []; |
|
121 | 3 | $contents = (string)file_get_contents($file); |
|
122 | 3 | $matches = null; |
|
123 | if ( |
||
124 | 3 | preg_match_all( |
|
125 | 3 | '/\[(.+?)\].*?\((.+?)\)/', |
|
126 | 3 | $contents, |
|
127 | 3 | $matches, |
|
128 | 3 | PREG_SET_ORDER |
|
129 | 3 | ) !== false |
|
130 | ) { |
||
131 | 3 | $links = array_merge($links, $matches); |
|
132 | } |
||
133 | |||
134 | 3 | return $links; |
|
135 | } |
||
136 | |||
137 | /** |
||
138 | * @SuppressWarnings(PHPMD.StaticAccess) |
||
139 | * |
||
140 | * @param string[] $link |
||
141 | * @param string[] $errors |
||
142 | */ |
||
143 | 3 | private static function checkLink( |
|
144 | string $projectRootDirectory, |
||
145 | array $link, |
||
146 | string $file, |
||
147 | array &$errors, |
||
148 | int &$return |
||
149 | ): void { |
||
150 | 3 | $path = trim($link[2]); |
|
151 | 3 | if (strpos($path, '#') === 0) { |
|
152 | 1 | return; |
|
153 | } |
||
154 | 3 | if (preg_match('%^(http|//)%', $path) === 1) { |
|
155 | 1 | self::validateHttpLink($link, $errors, $return); |
|
156 | |||
157 | 1 | return; |
|
158 | } |
||
159 | |||
160 | 2 | $path = current(explode('#', $path, 2)); |
|
161 | 2 | $start = rtrim($projectRootDirectory, '/'); |
|
162 | 2 | if ($path[0] !== '/' || strpos($path, './') === 0) { |
|
163 | 2 | $relativeSubdirs = preg_replace( |
|
164 | 2 | '%^' . $projectRootDirectory . '%', |
|
165 | 2 | '', |
|
166 | 2 | dirname($file) |
|
167 | ); |
||
168 | 2 | if ($relativeSubdirs !== null) { |
|
169 | 2 | $start .= '/' . rtrim($relativeSubdirs, '/'); |
|
170 | } |
||
171 | } |
||
172 | 2 | $realpath = realpath($start . '/' . $path); |
|
173 | 2 | if ($realpath === false) { |
|
174 | 1 | $errors[] = sprintf("\nBad link for \"%s\" to \"%s\"\n", $link[1], $link[2]); |
|
175 | 1 | $return = 1; |
|
176 | } |
||
177 | 2 | } |
|
178 | |||
179 | /** |
||
180 | * @param string[] $link |
||
181 | * @param string[] $errors |
||
182 | * @SuppressWarnings(PHPMD.UndefinedVariable) - seems to not understand the static variable |
||
183 | */ |
||
184 | 1 | private static function validateHttpLink(array $link, array &$errors, int &$return): void |
|
185 | { |
||
186 | 1 | static $checked = []; |
|
187 | 1 | list(, $anchor, $href) = $link; |
|
188 | 1 | $hashPos = (int)strpos($href, '#'); |
|
189 | 1 | if ($hashPos > 0) { |
|
190 | $href = substr($href, 0, $hashPos); |
||
191 | } |
||
192 | 1 | if (isset($checked[$href])) { |
|
193 | return; |
||
194 | } |
||
195 | 1 | $checked[$href] = true; |
|
196 | 1 | $context = stream_context_create( |
|
197 | [ |
||
198 | 1 | 'http' => [ |
|
199 | 'method' => 'HEAD', |
||
200 | 'protocol_version' => 1.1, |
||
201 | 'header' => [ |
||
202 | 'Connection: close', |
||
203 | ], |
||
204 | ], |
||
205 | ] |
||
206 | ); |
||
207 | 1 | $result = null; |
|
208 | try { |
||
209 | 1 | $headers = get_headers($href, 0, $context); |
|
0 ignored issues
–
show
Bug
introduced
by
Loading history...
|
|||
210 | 1 | if ($headers === false) { |
|
211 | throw new RuntimeException('Failed getting headers for href ' . $href); |
||
212 | } |
||
213 | 1 | foreach ($headers as $header) { |
|
214 | 1 | if (strpos($header, ' 200 ') !== false) { |
|
215 | //$time = round(microtime(true) - $start, 2); |
||
216 | //fwrite(STDERR, "\n".'OK ('.$time.' seconds): '.$href); |
||
217 | |||
218 | 1 | return; |
|
219 | } |
||
220 | } |
||
221 | } catch (Throwable $e) { |
||
222 | throw new RuntimeException('Unexpected error ' . $e->getMessage(), $e->getCode(), $e); |
||
223 | } |
||
224 | |||
225 | 1 | $errors[] = sprintf( |
|
226 | 1 | "\nBad link for \"%s\" to \"%s\"\nresult: %s\n", |
|
227 | 1 | $anchor, |
|
228 | 1 | $href, |
|
229 | 1 | var_export($result, true) |
|
230 | ); |
||
231 | 1 | $return = 1; |
|
232 | //$time = round(microtime(true) - $start, 2); |
||
233 | //fwrite(STDERR, "\n".'Failed ('.$time.' seconds): '.$href); |
||
234 | 1 | } |
|
235 | } |
||
236 |