santiago zrewidował ten Gist . Przejdź do rewizji
1 file changed, 154 insertions
mastobook_to_readeck.php(stworzono plik)
| @@ -0,0 +1,154 @@ | |||
| 1 | + | #!/usr/bin/php | |
| 2 | + | <?php | |
| 3 | + | ||
| 4 | + | //----------------------------- | |
| 5 | + | // CREDENTIALS (fill this up) | |
| 6 | + | // put this in a cron to regularly import all your mastodon / snac | |
| 7 | + | // bookmarks in your readeck instance | |
| 8 | + | //----------------------------- | |
| 9 | + | ||
| 10 | + | $MASTODON_TOKEN = 'xxxx'; | |
| 11 | + | $MASTODON_HOST = 'go.lema.org'; | |
| 12 | + | ||
| 13 | + | $READECK_TOKEN = 'xxxx'; | |
| 14 | + | $READECK_HOST = 'read.lema.org'; | |
| 15 | + | ||
| 16 | + | $MINIMUM_TEXT_SIZE = 500; // article with less characters of content will be ignored | |
| 17 | + | ||
| 18 | + | ||
| 19 | + | //----------------------------- | |
| 20 | + | // FETCH MASTODON BOOKMARKS | |
| 21 | + | //----------------------------- | |
| 22 | + | echo "# Fetching mastodon / snac bookmarks...\n"; | |
| 23 | + | $ch = curl_init("https://$MASTODON_HOST/api/v1/bookmarks"); | |
| 24 | + | curl_setopt_array($ch, [ | |
| 25 | + | CURLOPT_RETURNTRANSFER => true, | |
| 26 | + | CURLOPT_HTTPHEADER => [ | |
| 27 | + | "Authorization: Bearer $MASTODON_TOKEN", | |
| 28 | + | "Accept: application/json" | |
| 29 | + | ] | |
| 30 | + | ]); | |
| 31 | + | ||
| 32 | + | $bookmarksJson = curl_exec($ch); | |
| 33 | + | $bookmarks = json_decode($bookmarksJson, true); | |
| 34 | + | if (!is_array($bookmarks)) { | |
| 35 | + | die("❌ Failed to parse Mastodon bookmarks.\n"); | |
| 36 | + | } | |
| 37 | + | ||
| 38 | + | echo "Found bookmarks:".count($bookmarks)."\n"; | |
| 39 | + | ||
| 40 | + | //----------------------------- | |
| 41 | + | // FIND VALID URLs in posts | |
| 42 | + | //----------------------------- | |
| 43 | + | ||
| 44 | + | foreach ($bookmarks as $status) { | |
| 45 | + | if (!isset($status['content'])) { | |
| 46 | + | continue; | |
| 47 | + | } | |
| 48 | + | $content = strip_tags($status['content']); | |
| 49 | + | preg_match_all('/https?:\/\/[^\s"<]+/', $content, $matches); | |
| 50 | + | if (!empty($matches[0])) { | |
| 51 | + | ||
| 52 | + | $oneLink = $matches[0][0]; | |
| 53 | + | if (filter_var($oneLink, FILTER_VALIDATE_URL)) { | |
| 54 | + | $links[] = $oneLink; | |
| 55 | + | } else { | |
| 56 | + | // This happens for example if URL has an emoji at the end | |
| 57 | + | echo "INVALID URL: $oneLink\n"; | |
| 58 | + | } | |
| 59 | + | } | |
| 60 | + | } | |
| 61 | + | ||
| 62 | + | ||
| 63 | + | echo "Valid URLS:".count($links)."\n"; | |
| 64 | + | ||
| 65 | + | print_r($links); | |
| 66 | + | ||
| 67 | + | ||
| 68 | + | //----------------------------- | |
| 69 | + | // SEND LINKS TO READECK | |
| 70 | + | //----------------------------- | |
| 71 | + | ||
| 72 | + | $apiUrl = "https://$READECK_HOST/api/bookmarks"; | |
| 73 | + | ||
| 74 | + | $ch = curl_init(); | |
| 75 | + | curl_setopt($ch, CURLOPT_RETURNTRANSFER, true); | |
| 76 | + | curl_setopt($ch, CURLOPT_POST, true); | |
| 77 | + | ||
| 78 | + | $headers = [ | |
| 79 | + | "Authorization: Bearer $READECK_TOKEN", | |
| 80 | + | 'Accept: application/json', | |
| 81 | + | 'Content-Type: application/json' | |
| 82 | + | ]; | |
| 83 | + | ||
| 84 | + | $alreadySentDir = __DIR__ . "/already_sent"; | |
| 85 | + | ||
| 86 | + | if (!is_dir($alreadySentDir)) { | |
| 87 | + | mkdir($alreadySentDir, 0755, true); // recursive mkdir | |
| 88 | + | } | |
| 89 | + | ||
| 90 | + | ||
| 91 | + | ||
| 92 | + | foreach ($links as $link) { | |
| 93 | + | ||
| 94 | + | // READECK will accept several times the same URL ! | |
| 95 | + | // Make sure we don't send it several times by keeping an archive here | |
| 96 | + | $hash = md5($link); | |
| 97 | + | $filePath = __DIR__ . "/already_sent/{$hash}.txt"; | |
| 98 | + | ||
| 99 | + | if (file_exists($filePath)) { | |
| 100 | + | echo "ℹ️ Already sent: $link\n"; | |
| 101 | + | continue; | |
| 102 | + | } | |
| 103 | + | ||
| 104 | + | ||
| 105 | + | $content = file_get_contents($link); | |
| 106 | + | if ($content === false) { | |
| 107 | + | echo "❌ Failed to fetch $link\n"; | |
| 108 | + | continue; | |
| 109 | + | } | |
| 110 | + | $plainText = strip_tags($content); | |
| 111 | + | ||
| 112 | + | if (strlen($plainText) < $MINIMUM_TEXT_SIZE) { | |
| 113 | + | echo "⚠️ Skipping $link\ncontent too small (".strlen($plainText)." chars < $MINIMUM_TEXT_SIZE )\n"; | |
| 114 | + | continue; | |
| 115 | + | } | |
| 116 | + | ||
| 117 | + | echo "🟢 Will add to Readeck $link\nLength: " . strlen($plainText)."\n"; | |
| 118 | + | ||
| 119 | + | //not passing title here, since we don't have it | |
| 120 | + | $payload = json_encode([ | |
| 121 | + | "labels" => ["automasto"], | |
| 122 | + | "url" => $link | |
| 123 | + | ]); | |
| 124 | + | ||
| 125 | + | curl_setopt($ch, CURLOPT_URL, $apiUrl); | |
| 126 | + | curl_setopt($ch, CURLOPT_POSTFIELDS, $payload); | |
| 127 | + | curl_setopt($ch, CURLOPT_HTTPHEADER, $headers); | |
| 128 | + | ||
| 129 | + | $response = curl_exec($ch); | |
| 130 | + | $httpCode = curl_getinfo($ch, CURLINFO_HTTP_CODE); | |
| 131 | + | ||
| 132 | + | ||
| 133 | + | if (curl_errno($ch)) { | |
| 134 | + | echo "❌ Error adding $link: " . curl_error($ch) . "\n"; | |
| 135 | + | } else { | |
| 136 | + | ||
| 137 | + | // Store already sent file only if connection worked | |
| 138 | + | file_put_contents($filePath, $link); | |
| 139 | + | ||
| 140 | + | $json = json_decode($response, true); | |
| 141 | + | if (json_last_error() === JSON_ERROR_NONE) { | |
| 142 | + | if ($httpCode >= 200 && $httpCode < 300) { | |
| 143 | + | echo "✅ [$httpCode] Successfully added: $link\n"; | |
| 144 | + | } else { | |
| 145 | + | echo "⚠️ Server returned status $httpCode for $link\n"; | |
| 146 | + | } | |
| 147 | + | } else { | |
| 148 | + | echo "⚠️ Response is not valid JSON for $link: $response\n"; | |
| 149 | + | } | |
| 150 | + | } | |
| 151 | + | ||
| 152 | + | } | |
| 153 | + | ||
| 154 | + | curl_close($ch); | |
Nowsze
Starsze