diff --git a/inc/Data/Driver/HttpDriver.php b/inc/Data/Driver/HttpDriver.php deleted file mode 100644 index 2e379f27..00000000 --- a/inc/Data/Driver/HttpDriver.php +++ /dev/null @@ -1,135 +0,0 @@ -inner); - \curl_setopt_array($this->inner, [ - \CURLOPT_URL => $url, - \CURLOPT_TIMEOUT => $timeout, - \CURLOPT_USERAGENT => 'Tinyboard', - \CURLOPT_PROTOCOLS => \CURLPROTO_HTTP | \CURLPROTO_HTTPS, - ]); - } - - public function __construct(int $timeout, int $max_file_size) { - $this->inner = \curl_init(); - $this->timeout = $timeout; - $this->max_file_size = $max_file_size; - } - - public function __destruct() { - \curl_close($this->inner); - } - - /** - * Execute a GET request. - * - * @param string $endpoint Uri endpoint. - * @param ?array $data Optional GET parameters. - * @param ?array $data Optional HTTP headers. - * @param int $timeout Optional request timeout in seconds. Use the default timeout if 0. - * @return string Returns the body of the response. - * @throws RuntimeException Throws on IO error. - */ - public function requestGet(string $endpoint, ?array $data, ?array $headers = null, int $timeout = 0): string { - if (!empty($data)) { - $endpoint .= '?' . \http_build_query($data); - } - if ($timeout == 0) { - $timeout = $this->timeout; - } - - $this->resetTowards($endpoint, $timeout); - if (!empty($headers)) { - \curl_setopt($this->inner, \CURLOPT_HTTPHEADER, $headers); - } - \curl_setopt($this->inner, \CURLOPT_RETURNTRANSFER, true); - $ret = \curl_exec($this->inner); - - if ($ret === false) { - throw new \RuntimeException(\curl_error($this->inner)); - } - return $ret; - } - - /** - * Execute a POST request. - * - * @param string $endpoint Uri endpoint. - * @param ?array $data Optional POST parameters. - * @param int $timeout Optional request timeout in seconds. Use the default timeout if 0. - * @return string Returns the body of the response. - * @throws RuntimeException Throws on IO error. - */ - public function requestPost(string $endpoint, ?array $data, int $timeout = 0): string { - if ($timeout == 0) { - $timeout = $this->timeout; - } - - $this->resetTowards($endpoint, $timeout); - \curl_setopt($this->inner, \CURLOPT_POST, true); - if (!empty($data)) { - \curl_setopt($this->inner, \CURLOPT_POSTFIELDS, \http_build_query($data)); - } - \curl_setopt($this->inner, \CURLOPT_RETURNTRANSFER, true); - $ret = \curl_exec($this->inner); - - if ($ret === false) { - throw new \RuntimeException(\curl_error($this->inner)); - } - return $ret; - } - - /** - * Download the url's target with curl. - * - * @param string $url Url to the file to download. - * @param ?array $data Optional GET parameters. - * @param resource $fd File descriptor to save the content to. - * @param int $timeout Optional request timeout in seconds. Use the default timeout if 0. - * @return bool Returns true on success, false if the file was too large. - * @throws RuntimeException Throws on IO error. - */ - public function requestGetInto(string $endpoint, ?array $data, mixed $fd, int $timeout = 0): bool { - if (!empty($data)) { - $endpoint .= '?' . \http_build_query($data); - } - if ($timeout == 0) { - $timeout = $this->timeout; - } - - $this->resetTowards($endpoint, $timeout); - // Adapted from: https://stackoverflow.com/a/17642638 - $opt = (\PHP_MAJOR_VERSION >= 8 && \PHP_MINOR_VERSION >= 2) ? \CURLOPT_XFERINFOFUNCTION : \CURLOPT_PROGRESSFUNCTION; - \curl_setopt_array($this->inner, [ - \CURLOPT_NOPROGRESS => false, - $opt => fn($res, $next_dl, $dl, $next_up, $up) => (int)($dl <= $this->max_file_size), - \CURLOPT_FAILONERROR => true, - \CURLOPT_FOLLOWLOCATION => false, - \CURLOPT_FILE => $fd, - \CURLOPT_IPRESOLVE => CURL_IPRESOLVE_V4, - ]); - $ret = \curl_exec($this->inner); - - if ($ret === false) { - if (\curl_errno($this->inner) === CURLE_ABORTED_BY_CALLBACK) { - return false; - } - - throw new \RuntimeException(\curl_error($this->inner)); - } - return true; - } -} diff --git a/inc/Data/OembedResponse.php b/inc/Data/OembedResponse.php deleted file mode 100644 index 0e99f5ff..00000000 --- a/inc/Data/OembedResponse.php +++ /dev/null @@ -1,11 +0,0 @@ -log = $log; - $this->oembed_extractor = $oembed_extractor; - $this->embed_entries = $embed_entries; - $this->thumb_download_timeout = $thumb_download_timeout; - } - - private function make_tmp_file(): string { - $ret = \tempnam($this->tmp_dir, self::TMP_FILE_PREFIX); - if ($ret === false) { - throw new \RuntimeException("Could not create temporary file in {$this->tmp_dir}"); - } - \register_shutdown_function(fn() => @unlink($ret)); - return $ret; - } - - /** - * Downloads the thumbnail into a temporary file. - * - * @return ?string The path to the temporary file, null if the file was too large. - */ - private function fetchThumbnail(string $thumbnail_url): ?string { - $tmp_file = $this->make_tmp_file(); - $fd = \fopen($tmp_file, 'w+b'); - if ($fd === false) { - throw new \RuntimeException("Could not open temporary file $tmp_file for read/write"); - } - - $ret = $this->http->requestGetInto($thumbnail_url, null, $fd, $this->thumb_download_timeout); - return $ret ? $tmp_file : null; - } - - /** - * Returns the path to the thumbnail from a matched url, if any. - * - * @param string $url The url to embed. - * @param int $entry_index The index of the embedding entry. - * @return ?array Returns the url to the thumbnail and the path to the fallback. - */ - private function extractThumb(string $url, int $entry_index) { - $embed_entry = $this->embed_entries[$entry_index]; - $match_regex = $embed_entry['match_regex']; - $type = $embed_entry['type']; - - if ($type === 'oembed') { - $thumbnail_url_fallback = $embed_entry['thumbnail_url_fallback'] ?? null; - $provider = $embed_entry['provider_url']; - $oembed_resp = $this->oembed_extractor->fetch($provider, $url); - - return [ $oembed_resp->thumbnail_url, $thumbnail_url_fallback ]; - } elseif ($type === 'regex') { - $thumbnail_url_regex = $embed_entry['thumbnail_url']; - // Plz somebody review this. - return [ \preg_replace($match_regex, $thumbnail_url_regex, $url), null ]; - } else { - $this->log->log(LogDriver::ERROR, "Unknown embed type '$type' in embed entry $entry_index, ignoring the entry"); - return [ null, null ]; - } - } - - /** - * Find the embed entry matching with the url, if any. - * - * @param string $url Url to embed. MUST BE ALREADY VALIDATED. - * @return int The index of the matched embed entry or null. - */ - public function matchEmbed(string $url): ?int { - for ($i = 0; $i < \count($this->embed_entries); $i++) { - $match_regex = $this->embed_entries[$i]['match_regex']; - if (\preg_match($match_regex, $url, $matches)) { - return $i; - } - } - - return null; - } - - /** - * Get the embed's thumbnail if possible. May download it from the network into a temporary file, or use a static file. - * - * @param string $url Url to embed. MUST BE ALREADY VALIDATED. - * @param int The index of the matched embed entry. - * @return ?array Null if no thumbnail can be selected, otherwise an array with the local file path to the thumbnail - * and if the the file is a temporary or a static one. - */ - public function getEmbedThumb(string $url, int $entry_index): ?array { - $ret = $this->extractThumb($url, $entry_index); - list($thumbnail_url, $thumbnail_url_fallback) = $ret; - if (!isset($thumbnail_url, $thumbnail_url_fallback)) { - return null; - } - - if (\filter_var($thumbnail_url, \FILTER_VALIDATE_URL) === false) { - $this->log->log(LogDriver::ERROR, "Thumbnail URL '$thumbnail_url' from embed entry $entry_index is not a valid URL, trying fallback"); - } else { - $tmp_file = $this->fetchThumbnail($thumbnail_url); - if ($tmp_file !== null) { - return [ $tmp_file, true ]; - } - $this->log->log(LogDriver::NOTICE, "Thumbnail at '$thumbnail_url' was too large, trying fallback"); - } - - if ($thumbnail_url_fallback === null) { - return null; - } - return [ $thumbnail_url_fallback, false ]; - } - - public function renderEmbed(string $url, int $entry_index, string $thumbnail_path): string { - $embed_entry = $this->embed_entries[$entry_index]; - $match_regex = $embed_entry['match_regex']; - $html = $embed_entry['html']; - - $ret = \preg_replace($match_regex, $html, $url); - if (!\is_string($ret)) { - throw new \RuntimeException("Error while applying regex replacement for embed entry $entry_index"); - } - - \str_replace('%%embed_url%%', $url, $ret); - \str_replace('%%thumbnail_path%%', $thumbnail_path, $ret); - return $ret; - } -} diff --git a/inc/Service/Embed/OembedExtractor.php b/inc/Service/Embed/OembedExtractor.php deleted file mode 100644 index f86b2013..00000000 --- a/inc/Service/Embed/OembedExtractor.php +++ /dev/null @@ -1,66 +0,0 @@ -cache = $cache; - $this->http = $http; - $this->provider_timeout = $provider_timeout; - } - - /** - * Fetch the oembed data from the given provider with the given url. - * - * @param string $identifier Opaque identifier for caching, must be unique for each $url-$provider combination. - * @return OembedResponse The serialized remove response. May be cached. - */ - public function fetch(string $provider_url, string $url): OembedResponse { - $ret = $this->cache->get("oembed_embedder_$provider_url$url"); - if ($ret === null) { - $body = $this->http->requestGet( - $provider_url, - [ - 'url' => $url, - 'format' => 'json' - ], - [ - 'Content-Type: application/json' - ], - $this->provider_timeout - ); - $json = \json_decode($body, true, 512, \JSON_THROW_ON_ERROR); - - $ret = [ - 'title' => $json['title'] ?? null, - 'thumbnail_url' => $json['thumbnail_url'] ?? null, - ]; - - $cache_timeout = self::DEFAULT_CACHE_TIMEOUT; - if (isset($json['cache_age'])) { - $cache_age = \intval($json['cache_age']); - if ($cache_age > 0) { - $cache_age = \max($cache_age, self::MIN_CACHE_TIMEOUT); - } - } - - $this->cache->set("oembed_embedder_$provider_url$url", $ret, $cache_timeout); - } - - $resp = new OembedResponse(); - $resp->title = $ret['title']; - $resp->thumbnail_url = $ret['thumbnail_url']; - return $resp; - } -} diff --git a/inc/config.php b/inc/config.php index 8afb699f..71b0fbf4 100644 --- a/inc/config.php +++ b/inc/config.php @@ -943,10 +943,6 @@ // Location of thumbnail to use for deleted images. $config['image_deleted'] = 'static/deleted.png'; - // When a thumbnailed image is going to be the same (in dimension), just copy the entire file and use - // that as a thumbnail instead of resizing/redrawing. - $config['minimum_copy_resize'] = false; - // Maximum image upload size in bytes. $config['max_filesize'] = 10 * 1024 * 1024; // 10MB // Maximum image dimensions. @@ -985,15 +981,6 @@ // Set this to true if you're using Linux and you can execute `md5sum` binary. $config['gnu_md5'] = false; - // Use Tesseract OCR to retrieve text from images, so you can use it as a spamfilter. - $config['tesseract_ocr'] = false; - - // Tesseract parameters - $config['tesseract_params'] = ''; - - // Tesseract preprocess command - $config['tesseract_preprocess_command'] = 'convert -monochrome %s -'; - // Number of posts in a "View Last X Posts" page $config['noko50_count'] = 50; // Number of posts a thread needs before it gets a "View Last X Posts" page. @@ -1265,37 +1252,6 @@ $config['embed_width'] = 300; $config['embed_height'] = 246; - // Download timeout for the remove embed thumbnails in seconds. - $config['embed_thumb_timeout'] = 2; - - /** - * Replacement parameters: - * - $1-$N: matched arguments from 'match_regex'. - * - %%thumbnail_path%%: Path to the downloaded thumbnail. - */ - $config['embedding_2'] = [ - [ - 'match_regex' => '/^(?:(?:https?:)?\/\/)?((?:www|m)\.)?(?:(?:youtube(?:-nocookie)?\.com|youtu\.be))(?:\/(?:[\w\-]+\?v=|embed\/|live\/|v\/)?)([\w\-]{11})((?:\?|\&)\S+)?$/i', - 'type' => 'regex', - 'thumbnail_url' => 'https://img.youtube.com/vi/$2/0.jpg', - 'html' => '