| 1 |
<?php |
| 2 |
/** |
| 3 |
* Import Media — server-side ability used by the website builder in place of |
| 4 |
* the three wp-cli calls (`wp media import …`, `wp post get …`, |
| 5 |
* `wp post meta update _wp_attachment_image_alt …`) that are denylisted or |
| 6 |
* unsupported by RunWpCli's `verify_command_security`. |
| 7 |
* |
| 8 |
* Sideloads a remote image into the WordPress media library via WP core's |
| 9 |
* `media_sideload_image()` and `media_handle_sideload()` helpers. SSRF guards: |
| 10 |
* https:// only, public IP only, must content-type as image, capped at 25 MB. |
| 11 |
* |
| 12 |
* @since 0.0.5 |
| 13 |
* @package zip-ai |
| 14 |
*/ |
| 15 |
|
| 16 |
namespace ZipAI\MCP\Classes\Abilities\Zipai\Builder; |
| 17 |
|
| 18 |
use ZipAI\MCP\Classes\Abilities\Abstract_Ability; |
| 19 |
use ZipAI\MCP\Classes\Core\Tool_Types; |
| 20 |
use ZipAI\MCP\Classes\Core\Response; |
| 21 |
|
| 22 |
defined( 'ABSPATH' ) || exit; |
| 23 |
|
| 24 |
/** |
| 25 |
* Ability: sideload a remote image URL into the media library. |
| 26 |
*/ |
| 27 |
class ImportMedia extends Abstract_Ability { |
| 28 |
|
| 29 |
const MAX_BYTES = 26214400; // 25 MB. |
| 30 |
|
| 31 |
/** |
| 32 |
* Configures the ability's id, label, description and metadata. |
| 33 |
* |
| 34 |
* @return void |
| 35 |
*/ |
| 36 |
public function configure() { |
| 37 |
$this->id = 'zipai/import-media'; |
| 38 |
$this->label = 'Import Media'; |
| 39 |
$this->description = 'Sideload a remote image URL (e.g. a stock photo from Unsplash or Pexels) into the WordPress media library and return its attachment { id, url }. Use this to bring an external image URL into the site before applying it to a block. For AI-generated images use core ai_generate_image instead.'; |
| 40 |
$this->capability = 'upload_files'; |
| 41 |
} |
| 42 |
|
| 43 |
/** |
| 44 |
* Returns the tool-type classification for this ability. |
| 45 |
* |
| 46 |
* @return string One of the Tool_Types constants. |
| 47 |
*/ |
| 48 |
public function get_tool_type() { |
| 49 |
return Tool_Types::ACTION; |
| 50 |
} |
| 51 |
|
| 52 |
/** |
| 53 |
* Returns the JSON Schema for this ability's input arguments. |
| 54 |
* |
| 55 |
* @return array<string,mixed> JSON Schema describing accepted arguments. |
| 56 |
*/ |
| 57 |
public function get_input_schema() { |
| 58 |
return array( |
| 59 |
'type' => 'object', |
| 60 |
'required' => array( 'url' ), |
| 61 |
'additionalProperties' => false, |
| 62 |
'properties' => array( |
| 63 |
'url' => array( |
| 64 |
'type' => 'string', |
| 65 |
'format' => 'uri', |
| 66 |
'description' => 'Public https URL of the image. Internal/loopback IPs are refused.', |
| 67 |
), |
| 68 |
'title' => array( |
| 69 |
'type' => 'string', |
| 70 |
'description' => 'Optional attachment title.', |
| 71 |
'maxLength' => 250, |
| 72 |
), |
| 73 |
'alt_text' => array( |
| 74 |
'type' => 'string', |
| 75 |
'description' => 'Optional alt text — stored in the `_wp_attachment_image_alt` meta.', |
| 76 |
'maxLength' => 500, |
| 77 |
), |
| 78 |
), |
| 79 |
); |
| 80 |
} |
| 81 |
|
| 82 |
/** |
| 83 |
* Sideloads a remote image URL into the WordPress media library. |
| 84 |
* |
| 85 |
* @param array<string,mixed> $args Validated input arguments. |
| 86 |
* @return array<string,mixed> Standardized success or error response. |
| 87 |
*/ |
| 88 |
public function execute( $args ) { |
| 89 |
// Validate the RAW URL first. `esc_url_raw` silently rewrites things |
| 90 |
// like `file://…` and `javascript:…` to '' which would otherwise |
| 91 |
// land in the "url is required" branch with a misleading error. |
| 92 |
$raw_url = isset( $args['url'] ) && is_string( $args['url'] ) ? $args['url'] : ''; |
| 93 |
$title = isset( $args['title'] ) && is_string( $args['title'] ) ? sanitize_text_field( $args['title'] ) : ''; |
| 94 |
$alt_text = isset( $args['alt_text'] ) && is_string( $args['alt_text'] ) ? sanitize_text_field( $args['alt_text'] ) : ''; |
| 95 |
|
| 96 |
if ( '' === $raw_url ) { |
| 97 |
return Response::error( 'url is required.' ); |
| 98 |
} |
| 99 |
|
| 100 |
$validation = $this->validate_remote_url( $raw_url ); |
| 101 |
if ( is_wp_error( $validation ) ) { |
| 102 |
return Response::error( $validation->get_error_message() ); |
| 103 |
} |
| 104 |
// Safe to esc_url_raw now — we've confirmed scheme/host/IP pass. |
| 105 |
$url = esc_url_raw( $raw_url ); |
| 106 |
if ( '' === $url ) { |
| 107 |
return Response::error( 'url is not a valid URL.' ); |
| 108 |
} |
| 109 |
|
| 110 |
// WP core helpers used below. |
| 111 |
if ( ! function_exists( 'media_sideload_image' ) ) { |
| 112 |
require_once ABSPATH . 'wp-admin/includes/media.php'; |
| 113 |
} |
| 114 |
if ( ! function_exists( 'wp_tempnam' ) ) { |
| 115 |
require_once ABSPATH . 'wp-admin/includes/file.php'; |
| 116 |
} |
| 117 |
if ( ! function_exists( 'wp_read_image_metadata' ) ) { |
| 118 |
require_once ABSPATH . 'wp-admin/includes/image.php'; |
| 119 |
} |
| 120 |
|
| 121 |
// 1) Stream to a temp file with the size cap enforced mid-download |
| 122 |
// (download_url() has no streaming limit — it would write the full |
| 123 |
// body to disk before any size check could run). |
| 124 |
$tmp = wp_tempnam( $url ); |
| 125 |
if ( ! $tmp ) { |
| 126 |
return Response::error( 'Could not create a temporary file.' ); |
| 127 |
} |
| 128 |
// Cap + 1 so a transport that truncates at the limit (instead of |
| 129 |
// erroring) still leaves the file measurably over MAX_BYTES. |
| 130 |
$response = wp_safe_remote_get( |
| 131 |
$url, |
| 132 |
array( |
| 133 |
'timeout' => 30, |
| 134 |
'stream' => true, |
| 135 |
'filename' => $tmp, |
| 136 |
'limit_response_size' => self::MAX_BYTES + 1, |
| 137 |
) |
| 138 |
); |
| 139 |
if ( is_wp_error( $response ) ) { |
| 140 |
wp_delete_file( $tmp ); |
| 141 |
return Response::error( 'Download failed: ' . $response->get_error_message() ); |
| 142 |
} |
| 143 |
$status = (int) wp_remote_retrieve_response_code( $response ); |
| 144 |
if ( 200 !== $status ) { |
| 145 |
wp_delete_file( $tmp ); |
| 146 |
$reason = trim( wp_remote_retrieve_response_message( $response ) ); |
| 147 |
return Response::error( 'Download failed: ' . ( '' !== $reason ? $reason : "HTTP {$status}" ) ); |
| 148 |
} |
| 149 |
if ( filesize( $tmp ) > self::MAX_BYTES ) { |
| 150 |
wp_delete_file( $tmp ); |
| 151 |
return Response::error( sprintf( 'Remote file is larger than %d bytes.', self::MAX_BYTES ) ); |
| 152 |
} |
| 153 |
|
| 154 |
// 2) Verify it's actually an image and resolve a sideload-safe filename. |
| 155 |
$resolved = $this->resolve_image_file( $tmp, $url ); |
| 156 |
if ( is_wp_error( $resolved ) ) { |
| 157 |
wp_delete_file( $tmp ); |
| 158 |
return Response::error( $resolved->get_error_message() ); |
| 159 |
} |
| 160 |
|
| 161 |
// 3) Hand off to media_handle_sideload — creates the attachment, returns ID. |
| 162 |
$file_array = array( |
| 163 |
'name' => $resolved['name'], |
| 164 |
'tmp_name' => $tmp, |
| 165 |
); |
| 166 |
$post_data = array(); |
| 167 |
if ( '' !== $title ) { |
| 168 |
$post_data['post_title'] = $title; |
| 169 |
} |
| 170 |
$attachment_id = media_handle_sideload( $file_array, 0, '' !== $title ? $title : null, $post_data ); |
| 171 |
if ( is_wp_error( $attachment_id ) ) { |
| 172 |
wp_delete_file( $tmp ); |
| 173 |
return Response::error( 'media_handle_sideload failed: ' . $attachment_id->get_error_message() ); |
| 174 |
} |
| 175 |
|
| 176 |
// 4) Alt text. |
| 177 |
if ( '' !== $alt_text ) { |
| 178 |
update_post_meta( (int) $attachment_id, '_wp_attachment_image_alt', $alt_text ); |
| 179 |
} |
| 180 |
|
| 181 |
$attachment_url = wp_get_attachment_url( (int) $attachment_id ); |
| 182 |
if ( ! $attachment_url ) { |
| 183 |
// Roll back the just-created attachment so a URL-resolution failure |
| 184 |
// doesn't leave an orphan in the media library. |
| 185 |
wp_delete_attachment( (int) $attachment_id, true ); |
| 186 |
return Response::error( 'Image was imported but its URL could not be resolved.' ); |
| 187 |
} |
| 188 |
|
| 189 |
return Response::success( |
| 190 |
sprintf( 'Imported image into the media library (id %d).', (int) $attachment_id ), |
| 191 |
array( |
| 192 |
'id' => (int) $attachment_id, |
| 193 |
'url' => $attachment_url, |
| 194 |
'title' => $title, |
| 195 |
'alt' => $alt_text, |
| 196 |
'mime' => $resolved['type'], |
| 197 |
) |
| 198 |
); |
| 199 |
} |
| 200 |
|
| 201 |
/** |
| 202 |
* Resolve the mime type and a sideload-safe filename for a downloaded file. |
| 203 |
* |
| 204 |
* The downloaded bytes are the authority: stock/CDN image URLs (Unsplash, |
| 205 |
* dynamic image services) often have no extension in their path, and a URL |
| 206 |
* extension can lie about the content. The URL filename is only a naming |
| 207 |
* hint. Formats core cannot byte-sniff (SVG) fall back to the extension |
| 208 |
* check, which also enforces the site's allowed mime types. |
| 209 |
* |
| 210 |
* @param string $tmp Path to the downloaded temp file. |
| 211 |
* @param string $url Source URL — filename hint only, never trusted for type. |
| 212 |
* @return array{type: string, name: string}|\WP_Error |
| 213 |
*/ |
| 214 |
public function resolve_image_file( string $tmp, string $url ) { |
| 215 |
$url_path = wp_parse_url( $url, PHP_URL_PATH ); |
| 216 |
$path_name = is_string( $url_path ) ? basename( $url_path ) : ''; |
| 217 |
|
| 218 |
$type = (string) wp_get_image_mime( $tmp ); |
| 219 |
if ( '' !== $type ) { |
| 220 |
// Parity with the extension path: enforce the site's allowed mime |
| 221 |
// list here too, so a disallowed type fails with a precise error |
| 222 |
// instead of media_handle_sideload's generic one. |
| 223 |
if ( ! in_array( $type, get_allowed_mime_types(), true ) ) { |
| 224 |
return new \WP_Error( |
| 225 |
'disallowed_mime', |
| 226 |
sprintf( 'Image type "%s" is not allowed on this site.', $type ) |
| 227 |
); |
| 228 |
} |
| 229 |
$ext = wp_get_default_extension_for_mime_type( $type ); |
| 230 |
if ( ! $ext ) { |
| 231 |
return new \WP_Error( |
| 232 |
'unsupported_image', |
| 233 |
sprintf( 'Image type "%s" is not supported.', $type ) |
| 234 |
); |
| 235 |
} |
| 236 |
$base = pathinfo( $path_name, PATHINFO_FILENAME ); |
| 237 |
return array( |
| 238 |
'type' => $type, |
| 239 |
'name' => sanitize_file_name( ( '' !== $base ? $base : 'image' ) . '.' . $ext ), |
| 240 |
); |
| 241 |
} |
| 242 |
|
| 243 |
$mime_check = wp_check_filetype_and_ext( $tmp, $path_name ); |
| 244 |
$ext_type = (string) $mime_check['type']; |
| 245 |
if ( '' === $ext_type || 0 !== strpos( $ext_type, 'image/' ) ) { |
| 246 |
return new \WP_Error( |
| 247 |
'not_an_image', |
| 248 |
sprintf( 'URL does not resolve to an image (detected: "%s").', $ext_type ) |
| 249 |
); |
| 250 |
} |
| 251 |
$proper_filename = (string) $mime_check['proper_filename']; |
| 252 |
return array( |
| 253 |
'type' => $ext_type, |
| 254 |
'name' => '' !== $proper_filename ? $proper_filename : $path_name, |
| 255 |
); |
| 256 |
} |
| 257 |
|
| 258 |
/** |
| 259 |
* Reject URLs that wouldn't survive WP.org review. Returns true on success |
| 260 |
* or a WP_Error on the first failed check. |
| 261 |
* |
| 262 |
* @param string $url Raw URL. |
| 263 |
* @return true|\WP_Error |
| 264 |
*/ |
| 265 |
private function validate_remote_url( string $url ) { |
| 266 |
if ( ! filter_var( $url, FILTER_VALIDATE_URL ) ) { |
| 267 |
return new \WP_Error( 'invalid_url', 'url is not a valid URL.' ); |
| 268 |
} |
| 269 |
$parts = wp_parse_url( $url ); |
| 270 |
if ( ! is_array( $parts ) || empty( $parts['scheme'] ) || empty( $parts['host'] ) ) { |
| 271 |
return new \WP_Error( 'invalid_url', 'url is missing scheme or host.' ); |
| 272 |
} |
| 273 |
// HTTPS only — http allows MITM substitution; data:/file:/javascript: are obvious SSRF vectors. |
| 274 |
if ( 'https' !== strtolower( $parts['scheme'] ) ) { |
| 275 |
return new \WP_Error( 'insecure_scheme', 'Only https:// URLs are accepted.' ); |
| 276 |
} |
| 277 |
|
| 278 |
$host = strtolower( $parts['host'] ); |
| 279 |
// Strip bracketed IPv6 form (parse_url returns "[::1]") so it lines up |
| 280 |
// with the literal IP we compare against. |
| 281 |
if ( '[' === $host[0] && str_ends_with( $host, ']' ) ) { |
| 282 |
$host = substr( $host, 1, -1 ); |
| 283 |
} |
| 284 |
// Reject loopback hostnames outright (covers most local-dev SSRF). |
| 285 |
$blocked_hosts = array( 'localhost', '0.0.0.0', '127.0.0.1', '::1' ); |
| 286 |
if ( in_array( $host, $blocked_hosts, true ) ) { |
| 287 |
return new \WP_Error( 'blocked_host', 'Internal hostnames are not allowed.' ); |
| 288 |
} |
| 289 |
// If host is an IP literal, refuse private + link-local + loopback ranges. |
| 290 |
if ( filter_var( $host, FILTER_VALIDATE_IP ) ) { |
| 291 |
if ( ! filter_var( $host, FILTER_VALIDATE_IP, FILTER_FLAG_NO_PRIV_RANGE | FILTER_FLAG_NO_RES_RANGE ) ) { |
| 292 |
return new \WP_Error( 'private_ip', 'Private/reserved IPs are not allowed.' ); |
| 293 |
} |
| 294 |
} |
| 295 |
return true; |
| 296 |
} |
| 297 |
|
| 298 |
/** |
| 299 |
* Returns the JSON Schema for this ability's response. |
| 300 |
* |
| 301 |
* @return array<string,mixed> JSON Schema describing the response shape. |
| 302 |
*/ |
| 303 |
public function get_output_schema() { |
| 304 |
return array( |
| 305 |
'type' => 'object', |
| 306 |
'required' => array( 'success' ), |
| 307 |
'properties' => array( |
| 308 |
'success' => array( 'type' => 'boolean' ), |
| 309 |
'message' => array( 'type' => 'string' ), |
| 310 |
'data' => array( |
| 311 |
'type' => 'object', |
| 312 |
'properties' => array( |
| 313 |
'id' => array( 'type' => 'integer' ), |
| 314 |
'url' => array( 'type' => 'string' ), |
| 315 |
'title' => array( 'type' => 'string' ), |
| 316 |
'alt' => array( 'type' => 'string' ), |
| 317 |
'mime' => array( 'type' => 'string' ), |
| 318 |
), |
| 319 |
), |
| 320 |
), |
| 321 |
); |
| 322 |
} |
| 323 |
} |
| 324 |
|
