| 1 | //! Attachment endpoints: upload (write-gated) and serve (read-gated). |
| 2 | //! |
| 3 | //! Images pasted/dropped into the file editor are POSTed here as a raw body, |
| 4 | //! stored content-addressed outside git (see [`anvil_core::attachments`]), and |
| 5 | //! served back so the Markdown only carries a URL. The serve route is gated by |
| 6 | //! the repo's read access, so private repos stay private. |
| 7 | |
| 8 | use anvil_core::{ |
| 9 | App, |
| 10 | access, |
| 11 | attachments, |
| 12 | storage, |
| 13 | }; |
| 14 | use axum::{ |
| 15 | Router, |
| 16 | body::Bytes, |
| 17 | extract::{ |
| 18 | DefaultBodyLimit, |
| 19 | Path, |
| 20 | State, |
| 21 | }, |
| 22 | http::{ |
| 23 | HeaderMap, |
| 24 | StatusCode, |
| 25 | header, |
| 26 | }, |
| 27 | response::{ |
| 28 | IntoResponse, |
| 29 | Response, |
| 30 | }, |
| 31 | routing::{ |
| 32 | get, |
| 33 | post, |
| 34 | }, |
| 35 | }; |
| 36 | |
| 37 | use crate::{ |
| 38 | auth::{ |
| 39 | CSRF_FIELD, |
| 40 | Csrf, |
| 41 | CurrentUser, |
| 42 | verify_csrf, |
| 43 | }, |
| 44 | ui::{ |
| 45 | forbidden, |
| 46 | not_found, |
| 47 | resolve_repo, |
| 48 | server_error, |
| 49 | }, |
| 50 | }; |
| 51 | |
| 52 | /// Register the attachment routes. `max_upload_bytes` bounds a single upload |
| 53 | /// (from `http.attachment_max_mb`); the body-limit layer rejects anything |
| 54 | /// larger before it is buffered. |
| 55 | pub fn routes(router: Router<App>, max_upload_bytes: usize) -> Router<App> { |
| 56 | router |
| 57 | .route( |
| 58 | "/{owner}/{repo}/-/attachments/{hash}", |
| 59 | get(serve_attachment), |
| 60 | ) |
| 61 | .route( |
| 62 | "/{owner}/{repo}/-/attachments", |
| 63 | post(upload_attachment).layer(DefaultBodyLimit::max(max_upload_bytes)), |
| 64 | ) |
| 65 | } |
| 66 | |
| 67 | /// A 64-char lowercase hex SHA-256 — the only shape a stored attachment name |
| 68 | /// can take. Guards the path component before it touches the filesystem. |
| 69 | fn is_valid_hash(hash: &str) -> bool { |
| 70 | hash.len() == 64 |
| 71 | && hash |
| 72 | .bytes() |
| 73 | .all(|b| b.is_ascii_hexdigit() && !b.is_ascii_uppercase()) |
| 74 | } |
| 75 | |
| 76 | /// Detect a supported raster image type from leading magic bytes, returning its |
| 77 | /// MIME type. SVG is deliberately excluded — it can carry script, and we serve |
| 78 | /// attachments from our own origin. The client's `Content-Type` is ignored. |
| 79 | fn sniff_image(bytes: &[u8]) -> Option<&'static str> { |
| 80 | if bytes.starts_with(&[0x89, b'P', b'N', b'G', 0x0D, 0x0A, 0x1A, 0x0A]) { |
| 81 | Some("image/png") |
| 82 | } else if bytes.starts_with(&[0xFF, 0xD8, 0xFF]) { |
| 83 | Some("image/jpeg") |
| 84 | } else if bytes.starts_with(b"GIF87a") || bytes.starts_with(b"GIF89a") { |
| 85 | Some("image/gif") |
| 86 | } else if bytes.len() >= 12 && &bytes[0..4] == b"RIFF" && &bytes[8..12] == b"WEBP" { |
| 87 | Some("image/webp") |
| 88 | } else { |
| 89 | None |
| 90 | } |
| 91 | } |
| 92 | |
| 93 | /// `GET /{owner}/{repo}/-/attachments/{hash}` — serve an attachment's bytes, |
| 94 | /// gated by the repo's read access. Content is immutable (addressed by hash), |
| 95 | /// so it is cached aggressively and served with active content neutralized. |
| 96 | async fn serve_attachment( |
| 97 | State(app): State<App>, |
| 98 | CurrentUser(user): CurrentUser, |
| 99 | Path((owner, repo, hash)): Path<(String, String, String)>, |
| 100 | ) -> Response { |
| 101 | let (_, meta) = match resolve_repo(&app, user.as_ref(), &owner, &repo).await { |
| 102 | Ok(v) => v, |
| 103 | Err(resp) => return resp, |
| 104 | }; |
| 105 | if !is_valid_hash(&hash) { |
| 106 | return not_found("no such attachment"); |
| 107 | } |
| 108 | let row = match attachments::find(&app.db, meta.id, &hash).await { |
| 109 | Ok(Some(r)) => r, |
| 110 | Ok(None) => return not_found("no such attachment"), |
| 111 | Err(e) => return server_error(e), |
| 112 | }; |
| 113 | let path = storage::attachment_path(&app.config.attachments_dir(), meta.id, &hash); |
| 114 | let bytes = match std::fs::read(&path) { |
| 115 | Ok(b) => b, |
| 116 | Err(_) => return not_found("attachment data missing on disk"), |
| 117 | }; |
| 118 | ( |
| 119 | [ |
| 120 | (header::CONTENT_TYPE, row.content_type), |
| 121 | (header::X_CONTENT_TYPE_OPTIONS, "nosniff".to_string()), |
| 122 | // Content-addressed ⇒ immutable: cache for a year. |
| 123 | ( |
| 124 | header::CACHE_CONTROL, |
| 125 | "public, max-age=31536000, immutable".to_string(), |
| 126 | ), |
| 127 | (header::CONTENT_DISPOSITION, "inline".to_string()), |
| 128 | // Neutralize any active content even if a type slips through. |
| 129 | ( |
| 130 | header::CONTENT_SECURITY_POLICY, |
| 131 | "default-src 'none'; sandbox".to_string(), |
| 132 | ), |
| 133 | ], |
| 134 | bytes, |
| 135 | ) |
| 136 | .into_response() |
| 137 | } |
| 138 | |
| 139 | /// `POST /{owner}/{repo}/-/attachments` — store a raw image body and return its |
| 140 | /// serve URL. Requires write access and a valid CSRF token (sent as the |
| 141 | /// `X-CSRF-Token` header, since the body is the raw file, not a form). |
| 142 | async fn upload_attachment( |
| 143 | State(app): State<App>, |
| 144 | CurrentUser(user): CurrentUser, |
| 145 | csrf: Csrf, |
| 146 | Path((owner, repo)): Path<(String, String)>, |
| 147 | headers: HeaderMap, |
| 148 | body: Bytes, |
| 149 | ) -> Response { |
| 150 | let (_, meta) = match resolve_repo(&app, user.as_ref(), &owner, &repo).await { |
| 151 | Ok(v) => v, |
| 152 | Err(resp) => return resp, |
| 153 | }; |
| 154 | let Some(user) = user else { |
| 155 | return (StatusCode::UNAUTHORIZED, "sign in to upload").into_response(); |
| 156 | }; |
| 157 | if !access::can_write(&meta, Some(&user)) { |
| 158 | return forbidden(); |
| 159 | } |
| 160 | let submitted = headers |
| 161 | .get("x-csrf-token") |
| 162 | .and_then(|v| v.to_str().ok()) |
| 163 | .or_else(|| headers.get(CSRF_FIELD).and_then(|v| v.to_str().ok())) |
| 164 | .unwrap_or_default(); |
| 165 | if let Err(resp) = verify_csrf(&csrf, submitted) { |
| 166 | return resp; |
| 167 | } |
| 168 | |
| 169 | let cap = app.config.http.attachment_max_mb.saturating_mul(1 << 20); |
| 170 | if cap != 0 && body.len() > cap { |
| 171 | return (StatusCode::PAYLOAD_TOO_LARGE, "attachment too large").into_response(); |
| 172 | } |
| 173 | let Some(content_type) = sniff_image(&body) else { |
| 174 | return ( |
| 175 | StatusCode::BAD_REQUEST, |
| 176 | "unsupported file type (png, jpeg, gif, webp only)", |
| 177 | ) |
| 178 | .into_response(); |
| 179 | }; |
| 180 | |
| 181 | let hash = attachments::content_hash(&body); |
| 182 | let dir = app.config.attachments_dir().join(meta.id.to_string()); |
| 183 | if let Err(e) = std::fs::create_dir_all(&dir) { |
| 184 | return server_error(e); |
| 185 | } |
| 186 | // Content-addressed: the file is immutable, so only write if it's new. |
| 187 | let path = dir.join(&hash); |
| 188 | let is_new = !path.exists(); |
| 189 | // Per-repo quota: only new content adds bytes, so deduped re-uploads are |
| 190 | // always allowed even at the cap. |
| 191 | let quota = app.config.http.attachment_quota_mb.saturating_mul(1 << 20); |
| 192 | if quota != 0 && is_new && storage::dir_size(&dir) + body.len() as u64 > quota as u64 { |
| 193 | return ( |
| 194 | StatusCode::PAYLOAD_TOO_LARGE, |
| 195 | "repository attachment quota exceeded", |
| 196 | ) |
| 197 | .into_response(); |
| 198 | } |
| 199 | if is_new && let Err(e) = std::fs::write(&path, &body) { |
| 200 | return server_error(e); |
| 201 | } |
| 202 | if let Err(e) = attachments::add( |
| 203 | &app.db, |
| 204 | meta.id, |
| 205 | &hash, |
| 206 | content_type, |
| 207 | body.len() as i64, |
| 208 | user.id, |
| 209 | ) |
| 210 | .await |
| 211 | { |
| 212 | return server_error(e); |
| 213 | } |
| 214 | |
| 215 | let url = format!("/{owner}/{repo}/-/attachments/{hash}"); |
| 216 | let markdown = format!(""); |
| 217 | axum::Json(serde_json::json!({ "url": url, "markdown": markdown })).into_response() |
| 218 | } |
| 219 | |
| 220 | #[cfg(test)] |
| 221 | mod tests { |
| 222 | use super::*; |
| 223 | |
| 224 | #[test] |
| 225 | fn hash_shape_is_strict() { |
| 226 | assert!(is_valid_hash(&"a".repeat(64))); |
| 227 | assert!(!is_valid_hash(&"a".repeat(63))); |
| 228 | assert!(!is_valid_hash(&"A".repeat(64))); // uppercase rejected |
| 229 | assert!(!is_valid_hash("../etc/passwd")); |
| 230 | assert!(!is_valid_hash(&"g".repeat(64))); // non-hex |
| 231 | } |
| 232 | |
| 233 | #[test] |
| 234 | fn sniffs_supported_images_only() { |
| 235 | assert_eq!( |
| 236 | sniff_image(&[0x89, b'P', b'N', b'G', 0x0D, 0x0A, 0x1A, 0x0A, 0, 0]), |
| 237 | Some("image/png") |
| 238 | ); |
| 239 | assert_eq!(sniff_image(&[0xFF, 0xD8, 0xFF, 0xE0]), Some("image/jpeg")); |
| 240 | assert_eq!(sniff_image(b"GIF89a..."), Some("image/gif")); |
| 241 | assert_eq!(sniff_image(b"RIFF\0\0\0\0WEBPVP8 "), Some("image/webp")); |
| 242 | assert_eq!(sniff_image(b"<svg></svg>"), None); |
| 243 | assert_eq!(sniff_image(b"<!doctype html>"), None); |
| 244 | assert_eq!(sniff_image(b""), None); |
| 245 | } |
| 246 | } |