feat(media): multimodal attachments — image/video inlining per model capabilities

Adds media.rs for per-turn attachment routing, message_builder
partitioning by resolved model capabilities, controller endpoints
for uploads, and server routing for /data/* behind session auth.

See CLAUDE.md §Multimodal attachments for full design.
This commit is contained in:
2026-07-18 18:02:45 +01:00
parent 4fea04c57f
commit 2b35312abd
15 changed files with 631 additions and 35 deletions
+4
View File
@@ -210,6 +210,10 @@ impl ApiError {
pub fn forbidden(msg: impl Into<String>) -> Self {
Self { status: StatusCode::FORBIDDEN, message: msg.into() }
}
pub fn payload_too_large(msg: impl Into<String>) -> Self {
Self { status: StatusCode::PAYLOAD_TOO_LARGE, message: msg.into() }
}
}
/// Resolves the authenticated caller's per-user runtime context, or `401` when the
+38 -5
View File
@@ -9,17 +9,25 @@ use tokio::io::AsyncWriteExt;
use core_api::message_meta::Attachment;
use skald_core::session::handler::media::sniff_mime;
use skald_core::skald::Skald;
use skald_core::tools::fs as fs_tools;
use super::{ApiError, guard::AuthUser, require_context};
use super::sessions::SourcePath;
/// Max bytes accepted for a single uploaded file; anything larger is cut off
/// mid-stream, the partial file removed, and the request answered 413.
const MAX_UPLOAD_BYTES: u64 = 256 * 1024 * 1024;
/// `POST /api/{source}/uploads`
///
/// Accepts a `multipart/form-data` body with one or more file fields and saves
/// each under `data/uploads/{session_id}/`. Bytes are streamed straight to disk
/// (`field.chunk()` → file), never buffered whole in RAM, so arbitrarily large
/// files are fine — the route disables the default body-size limit (see router).
/// each under `data/uploads/{user_id}/{session_id}/` (per-user namespaced, so
/// colliding session ids across users never share a directory). Bytes are
/// streamed straight to disk (`field.chunk()` → file), never buffered whole in
/// RAM — the route disables the default body-size limit (see router) and
/// enforces [`MAX_UPLOAD_BYTES`] itself. When the magic bytes are recognized,
/// the sniffed MIME wins over the client-supplied `Content-Type`.
///
/// Returns the saved [`Attachment`]s (project-root-relative path, name, MIME,
/// size) so the client can show chips and echo them back when sending the message.
@@ -34,7 +42,7 @@ pub async fn upload(
// directory the message will reference.
let session_id = ctx.chat_hub.session_handler(&p.source).await?.session_id;
let dir_rel = format!("data/uploads/{session_id}");
let dir_rel = format!("data/uploads/{}/{session_id}", auth.user_id);
let dir_abs = fs_tools::resolve(&dir_rel)?;
tokio::fs::create_dir_all(&dir_abs).await?;
@@ -54,13 +62,30 @@ pub async fn upload(
.map_err(|e| ApiError::from(anyhow::anyhow!("cannot create {}: {e}", abs_path.display())))?;
let mut size: u64 = 0;
let mut too_large = false;
while let Some(chunk) = field.chunk().await
.map_err(|e| ApiError::bad_request(format!("upload read error: {e}")))?
{
file.write_all(&chunk).await?;
size += chunk.len() as u64;
if size > MAX_UPLOAD_BYTES {
too_large = true;
break;
}
file.write_all(&chunk).await?;
}
file.flush().await?;
drop(file);
if too_large {
let _ = tokio::fs::remove_file(&abs_path).await;
return Err(ApiError::payload_too_large(format!(
"'{final_name}' exceeds the {} MiB upload limit",
MAX_UPLOAD_BYTES / 1024 / 1024
)));
}
// The sniffed type wins over the client claim when we recognize the bytes.
let mimetype = sniff_head(&abs_path).await.map(String::from).or(mimetype);
saved.push(Attachment {
path: format!("{dir_rel}/{final_name}"),
@@ -73,6 +98,14 @@ pub async fn upload(
Ok(Json(saved))
}
/// Reads the first bytes of a saved upload and sniffs its real media type.
async fn sniff_head(path: &StdPath) -> Option<&'static str> {
let mut file = tokio::fs::File::open(path).await.ok()?;
let mut head = [0u8; 16];
let n = tokio::io::AsyncReadExt::read(&mut file, &mut head).await.ok()?;
sniff_mime(&head[..n])
}
/// Reduces an arbitrary client filename to a safe basename: directory components
/// are dropped and an empty/`.`/`..` result falls back to `"file"`.
fn sanitize_filename(raw: &str) -> String {
+14 -2
View File
@@ -78,6 +78,7 @@ impl WebServer {
Arc::clone(&skald),
api::guard::require_auth,
));
let skald_for_data = Arc::clone(&skald);
// Resolve the app state first so the resulting `Router<()>` can host the
// stateless plugin routers via `nest`.
@@ -87,7 +88,8 @@ impl WebServer {
for (id, plugin_router) in plugin_routers {
router = router.nest(&format!("/api/plugin/{id}"), plugin_router);
}
// Serve the data/ directory under /data/ (accessible via URL).
// Serve the data/ directory under /data/ (accessible via URL), behind the
// same session-cookie gate as /api — uploads are private user content.
let data_dir = Path::new(static_dir).parent().unwrap_or(Path::new(".")).join("data");
// Static responses (SPA assets + /data) get `Cache-Control: no-cache`:
// the browser may store them but MUST revalidate before use, so after a
@@ -98,7 +100,17 @@ impl WebServer {
header::CACHE_CONTROL,
HeaderValue::from_static("no-cache"),
));
router = router.nest_service("/data", static_assets().service(ServeDir::new(&data_dir)));
let data_service = ServiceBuilder::new()
.layer(axum::middleware::from_fn_with_state(
skald_for_data,
api::guard::require_auth,
))
.layer(SetResponseHeaderLayer::overriding(
header::CACHE_CONTROL,
HeaderValue::from_static("no-cache"),
))
.service(ServeDir::new(&data_dir));
router = router.nest_service("/data", data_service);
router = router.fallback_service(static_assets().service(ServeDir::new(static_dir)));
// Negotiated gzip/brotli compression (Accept-Encoding). Matters most for
// the mobile WebView, whose HTTP traffic is reverse-proxied byte-for-byte