← All changes | includes/AI/Providers/BaseProvider.php +174 -0 4.8.2 → 4.9.4 View file →

		@@ -123,8 +123,182 @@
123	123	return $data;
124	124	}
125	125
126	126	/**
	127	+ * POST a multipart/form-data body carrying one file, and decode the response.
	128	+ *
	129	+ * The JSON helper above cannot express a file upload, and WordPress ships no
	130	+ * multipart builder — `wp_remote_post()` sends `body` as an array only as
	131	+ * urlencoded form fields. So the body is assembled by hand here. The one
	132	+ * caller is OpenAI's transcription endpoint, which takes the audio as a real
	133	+ * file part; everything else in this layer is JSON.
	134	+ *
	135	+ * The file is streamed in from its path rather than passed around as a
	136	+ * string by callers, so the bytes live in exactly one variable and are
	137	+ * released when this method returns.
	138	+ *
	139	+ * @param string $url
	140	+ * @param array $headers Auth headers. Content-Type is set here — a
	141	+ * caller-supplied one would lack the boundary.
	142	+ * @param array $fields Scalar form fields, e.g. `model`.
	143	+ * @param array $file `[ 'name' => field name, 'filename' => …, 'type' => mime, 'path' => … ]`
	144	+ * @param int $timeout
	145	+ * @param int|null $status_code Out-param, as in post_json().
	146	+ * @return array|\WP_Error
	147	+ */
	148	+ protected function post_multipart( $url, $headers, $fields, $file, $timeout = 120, &$status_code = null ) {
	149	+ if ( ! is_readable( $file['path'] ) ) {
	150	+ $status_code = 0;
	151	+ return new \WP_Error( 'api_error', __( 'Could not read the uploaded file.', 'betterdocs' ) );
	152	+ }
	153	+
	154	+ $response = $this->post_multipart_streamed( $url, $headers, $fields, $file, $timeout );
	155	+
	156	+ if ( null === $response ) {
	157	+ $response = $this->post_multipart_buffered( $url, $headers, $fields, $file, $timeout );
	158	+ }
	159	+
	160	+ if ( is_wp_error( $response ) ) {
	161	+ $status_code = 0;
	162	+ return new \WP_Error( 'api_error', sprintf(
	163	+ /* translators: 1: provider label, 2: error message */
	164	+ __( 'Failed to connect to %1$s: %2$s', 'betterdocs' ),
	165	+ $this->label(),
	166	+ $response->get_error_message()
	167	+ ) );
	168	+ }
	169	+
	170	+ $status_code = (int) wp_remote_retrieve_response_code( $response );
	171	+
	172	+ $data = json_decode( wp_remote_retrieve_body( $response ), true );
	173	+ if ( ! is_array( $data ) ) {
	174	+ return new \WP_Error( 'no_content', sprintf(
	175	+ /* translators: %s: provider label */
	176	+ __( 'Empty or invalid response from %s.', 'betterdocs' ),
	177	+ $this->label()
	178	+ ) );
	179	+ }
	180	+
	181	+ return $data;
	182	+ }
	183	+
	184	+ /**
	185	+ * Send the multipart body with cURL reading the file straight from disk.
	186	+ *
	187	+ * Building the body as a PHP string holds the whole recording in memory two
	188	+ * or three times over (the bytes, the body, cURL's own copy) — a 24 MB file
	189	+ * pushed a 128M host over its limit. Handing cURL a CURLFile instead lets it
	190	+ * stream the part from the temp upload, so the file is never in PHP memory.
	191	+ *
	192	+ * The request still goes through wp_remote_post(), so proxies, SSL settings
	193	+ * and every `http_*` filter apply as normal; only the body and the headers
	194	+ * are replaced on the handle, right before it is sent.
	195	+ *
	196	+ * @return array|\WP_Error|null Null when cURL is not the transport, so the
	197	+ * caller can fall back to the buffered body.
	198	+ */
	199	+ protected function post_multipart_streamed( $url, $headers, $fields, $file, $timeout ) {
	200	+ if ( ! function_exists( 'curl_init' ) || ! class_exists( '\CURLFile' ) ) {
	201	+ return null;
	202	+ }
	203	+
	204	+ $streamed = false;
	205	+ $attach = static function ( $handle, $r, $request_url ) use ( $url, $headers, $fields, $file, &$streamed ) {
	206	+ if ( $request_url !== $url || $streamed ) {
	207	+ return;
	208	+ }
	209	+
	210	+ $post = $fields;
	211	+ $post[ $file['name'] ] = new \CURLFile( $file['path'], $file['type'], $file['filename'] );
	212	+
	213	+ // cURL writes its own multipart Content-Type, boundary included, so
	214	+ // the header list is set without one. `Expect:` stops a 100-continue
	215	+ // round trip on large bodies.
	216	+ $lines = array( 'Expect:' );
	217	+ foreach ( $headers as $name => $value ) {
	218	+ if ( 'content-type' !== strtolower( $name ) ) {
	219	+ $lines[] = $name . ': ' . $value;
	220	+ }
	221	+ }
	222	+
	223	+ curl_setopt( $handle, CURLOPT_POSTFIELDS, $post ); // phpcs:ignore WordPress.WP.AlternativeFunctions.curl_curl_setopt -- the handle WordPress is about to send.
	224	+ curl_setopt( $handle, CURLOPT_HTTPHEADER, $lines ); // phpcs:ignore WordPress.WP.AlternativeFunctions.curl_curl_setopt
	225	+ $streamed = true;
	226	+ };
	227	+
	228	+ add_action( 'http_api_curl', $attach, PHP_INT_MAX, 3 );
	229	+
	230	+ $response = wp_remote_post( $url, array(
	231	+ 'headers' => $headers,
	232	+ // Replaced on the handle above. Non-empty so the request is sent
	233	+ // as a POST with a body on every transport.
	234	+ 'body' => ' ',
	235	+ 'timeout' => (int) $timeout,
	236	+ ) );
	237	+
	238	+ remove_action( 'http_api_curl', $attach, PHP_INT_MAX );
	239	+
	240	+ // The hook never ran: something other than cURL sent that request,
	241	+ // carrying the one-byte placeholder. Nothing was transcribed (the
	242	+ // provider rejects an empty upload), so the buffered path runs it again.
	243	+ return $streamed ? $response : null;
	244	+ }
	245	+
	246	+ /**
	247	+ * The same request with the body built in PHP, for a host without cURL.
	248	+ *
	249	+ * @return array|\WP_Error
	250	+ */
	251	+ protected function post_multipart_buffered( $url, $headers, $fields, $file, $timeout ) {
	252	+ $boundary = wp_generate_password( 24, false );
	253	+ $eol = "\r\n";
	254	+ $body = '';
	255	+
	256	+ foreach ( $fields as $key => $value ) {
	257	+ $body .= '--' . $boundary . $eol;
	258	+ $body .= 'Content-Disposition: form-data; name="' . $key . '"' . $eol . $eol;
	259	+ $body .= $value . $eol;
	260	+ }
	261	+
	262	+ $body .= '--' . $boundary . $eol;
	263	+ $body .= 'Content-Disposition: form-data; name="' . $file['name'] . '"; filename="' . $file['filename'] . '"' . $eol;
	264	+ $body .= 'Content-Type: ' . $file['type'] . $eol . $eol;
	265	+ // Appended straight onto the body, never as `$bytes . $eol`: that
	266	+ // expression builds a third full-size copy of the file before the append.
	267	+ $body .= (string) @file_get_contents( $file['path'] ); // phpcs:ignore WordPress.WP.AlternativeFunctions.file_get_contents_file_get_contents -- local temp upload, not a remote fetch.
	268	+ $body .= $eol;
	269	+ $body .= '--' . $boundary . '--' . $eol;
	270	+
	271	+ $headers['Content-Type'] = 'multipart/form-data; boundary=' . $boundary;
	272	+
	273	+ return wp_remote_post( $url, array(
	274	+ 'headers' => $headers,
	275	+ 'body' => $body,
	276	+ 'timeout' => (int) $timeout,
	277	+ ) );
	278	+ }
	279	+
	280	+ /**
	281	+ * Transcribe an audio or video file to plain text.
	282	+ *
	283	+ * Default: not supported. A provider that can do it overrides this; callers
	284	+ * are expected to check Core\WriteWithAI::platform_supports( 'transcription' )
	285	+ * first and show the switch-platform message, so reaching this is a bug
	286	+ * rather than a user-facing path.
	287	+ *
	288	+ * @param array $file `[ 'path', 'filename', 'mime' ]`
	289	+ * @param array $options
	290	+ * @return string|\WP_Error Transcript text.
	291	+ */
	292	+ public function transcribe( $file, $options = array() ) {
	293	+ return new \WP_Error( 'no_transcription', sprintf(
	294	+ /* translators: %s: provider label */
	295	+ __( '%s cannot transcribe audio or video.', 'betterdocs' ),
	296	+ $this->label()
	297	+ ) );
	298	+ }
	299	+
	300	+ /**
127	301	* Turn an HTTP status + the provider's raw error text into a clear, actionable
128	302	* message. Chiefly distinguishes a retired/unknown model (404) from a genuine
129	303	* quota / rate-limit rejection (429) — without this, a dead model id and a real
130	304	* quota error both surface the provider's raw text and look identical (a retired