| 1 | /** |
| 2 | * Off-main-thread image encoder. |
| 3 | * |
| 4 | * Accepts a ``File``, validates it via magic bytes (ignoring the extension to |
| 5 | * defeat rename-based spoofs), and either passes through or *normalizes* the |
| 6 | * bytes so the resulting base64 data URL stays ≤ ``TARGET_MAX_BYTES``. The |
| 7 | * normalization path uses ``createImageBitmap`` + ``OffscreenCanvas`` so the |
| 8 | * full decode/resize/re-encode cycle never blocks the UI thread. |
| 9 | * |
| 10 | * Output contract: |
| 11 | * ``{ok: true, dataUrl, mime, bytes, origBytes, normalized}`` on success, or |
| 12 | * ``{ok: false, reason}`` for every recoverable failure — magic-bytes |
| 13 | * mismatch, unsupported MIME, decode error, or a post-normalization payload |
| 14 | * that *still* exceeds the budget (extreme aspect ratios). |
| 15 | */ |
| 16 | |
| 17 | /// <reference lib="webworker" /> |
| 18 | |
| 19 | // --- Types ------------------------------------------------------------------- |
| 20 | |
| 21 | export type EncodeInput = { |
| 22 | id: string; |
| 23 | file: File; |
| 24 | }; |
| 25 | |
| 26 | export type EncodeSuccess = { |
| 27 | id: string; |
| 28 | ok: true; |
| 29 | dataUrl: string; |
| 30 | mime: string; |
| 31 | bytes: number; |
| 32 | origBytes: number; |
| 33 | /** True iff the Worker re-encoded the image to hit the size budget. */ |
| 34 | normalized: boolean; |
| 35 | }; |
| 36 | |
| 37 | export type EncodeFailure = { |
| 38 | id: string; |
| 39 | ok: false; |
| 40 | reason: |
| 41 | | "invalid_mime" |
| 42 | | "magic_mismatch" |
| 43 | | "too_large_after_normalize" |
| 44 | | "decode_failed" |
| 45 | | "io"; |
| 46 | }; |
| 47 | |
| 48 | export type EncodeResponse = EncodeSuccess | EncodeFailure; |
| 49 | |
| 50 | // --- Budgets ----------------------------------------------------------------- |
| 51 | |
| 52 | /** Upper bound for the final base64-decoded payload. Matches the server-side |
| 53 | * safeguard (8 MB) minus safety margin; anything this function yields should |
| 54 | * safely pass ``_MAX_IMAGE_BYTES`` on the server. */ |
| 55 | export const TARGET_MAX_BYTES = 6 * 1024 * 1024; |
| 56 | |
| 57 | /** Long-edge pixel cap when we resize a large image. 2048 keeps retina UIs |
| 58 | * crisp while bounding decode cost and matching most LLM vision tiers' |
| 59 | * internal downscale target. */ |
| 60 | const NORMALIZE_MAX_EDGE = 2048; |
| 61 | |
| 62 | /** JPEG/WebP quality during normalization. 0.85 is the sweet spot — visually |
| 63 | * lossless for content photography, ~30% smaller than libjpeg default. */ |
| 64 | const WEBP_QUALITY = 0.85; |
| 65 | |
| 66 | /** PNG / GIF kept as PNG after normalization so crisp UI screenshots stay |
| 67 | * lossless. JPEG / WebP re-encode as WebP for better compression. */ |
| 68 | const NORMALIZE_LOSSY_MIMES = new Set(["image/jpeg", "image/webp"]); |
| 69 | |
| 70 | const SUPPORTED_MIMES = new Set([ |
| 71 | "image/png", |
| 72 | "image/jpeg", |
| 73 | "image/webp", |
| 74 | "image/gif", |
| 75 | ]); |
| 76 | |
| 77 | // --- Magic bytes ------------------------------------------------------------- |
| 78 | |
| 79 | /** Sniff the first 12 bytes; returns the canonical MIME or ``null``. |
| 80 | * |
| 81 | * Covers PNG, JPEG, WebP, GIF — the same whitelist honoured by the server. |
| 82 | */ |
| 83 | export function sniffImageMime(bytes: Uint8Array): string | null { |
| 84 | if (bytes.length >= 8) { |
| 85 | if ( |
| 86 | bytes[0] === 0x89 && |
| 87 | bytes[1] === 0x50 && |
| 88 | bytes[2] === 0x4e && |
| 89 | bytes[3] === 0x47 && |
| 90 | bytes[4] === 0x0d && |
| 91 | bytes[5] === 0x0a && |
| 92 | bytes[6] === 0x1a && |
| 93 | bytes[7] === 0x0a |
| 94 | ) { |
| 95 | return "image/png"; |
| 96 | } |
| 97 | } |
| 98 | if (bytes.length >= 3) { |
| 99 | if (bytes[0] === 0xff && bytes[1] === 0xd8 && bytes[2] === 0xff) { |
| 100 | return "image/jpeg"; |
| 101 | } |
| 102 | } |
| 103 | if (bytes.length >= 6) { |
| 104 | const g1 = |
| 105 | bytes[0] === 0x47 && bytes[1] === 0x49 && bytes[2] === 0x46 && |
| 106 | bytes[3] === 0x38 && bytes[5] === 0x61; |
| 107 | if (g1 && (bytes[4] === 0x37 || bytes[4] === 0x39)) { |
| 108 | return "image/gif"; |
| 109 | } |
| 110 | } |
| 111 | if (bytes.length >= 12) { |
| 112 | const riff = |
| 113 | bytes[0] === 0x52 && bytes[1] === 0x49 && bytes[2] === 0x46 && bytes[3] === 0x46; |
| 114 | const webp = |
| 115 | bytes[8] === 0x57 && bytes[9] === 0x45 && bytes[10] === 0x42 && bytes[11] === 0x50; |
| 116 | if (riff && webp) return "image/webp"; |
| 117 | } |
| 118 | return null; |
| 119 | } |
| 120 | |
| 121 | // --- Encoder ----------------------------------------------------------------- |
| 122 | |
| 123 | function bufferToBase64(buf: ArrayBuffer): string { |
| 124 | // ``btoa`` can't take large strings — chunk through 32 KB windows. |
| 125 | const bytes = new Uint8Array(buf); |
| 126 | let binary = ""; |
| 127 | const CHUNK = 0x8000; |
| 128 | for (let i = 0; i < bytes.length; i += CHUNK) { |
| 129 | binary += String.fromCharCode.apply( |
| 130 | null, |
| 131 | bytes.subarray(i, i + CHUNK) as unknown as number[], |
| 132 | ); |
| 133 | } |
| 134 | return self.btoa(binary); |
| 135 | } |
| 136 | |
| 137 | function computeScaledDims( |
| 138 | srcW: number, |
| 139 | srcH: number, |
| 140 | maxEdge: number, |
| 141 | ): { w: number; h: number } { |
| 142 | const longest = Math.max(srcW, srcH); |
| 143 | if (longest <= maxEdge) return { w: srcW, h: srcH }; |
| 144 | const scale = maxEdge / longest; |
| 145 | return { |
| 146 | w: Math.max(1, Math.round(srcW * scale)), |
| 147 | h: Math.max(1, Math.round(srcH * scale)), |
| 148 | }; |
| 149 | } |
| 150 | |
| 151 | async function normalize( |
| 152 | file: File, |
| 153 | sourceMime: string, |
| 154 | ): Promise<{ dataUrl: string; mime: string; bytes: number } | { error: EncodeFailure["reason"] }> { |
| 155 | // Re-encode paths: JPEG/WebP → WebP q=0.85; PNG/GIF → PNG (keep crisp). |
| 156 | const targetMime = NORMALIZE_LOSSY_MIMES.has(sourceMime) |
| 157 | ? "image/webp" |
| 158 | : "image/png"; |
| 159 | let bitmap: ImageBitmap; |
| 160 | try { |
| 161 | bitmap = await createImageBitmap(file); |
| 162 | } catch { |
| 163 | return { error: "decode_failed" }; |
| 164 | } |
| 165 | const { w, h } = computeScaledDims(bitmap.width, bitmap.height, NORMALIZE_MAX_EDGE); |
| 166 | try { |
| 167 | const canvas = new OffscreenCanvas(w, h); |
| 168 | const ctx = canvas.getContext("2d", { alpha: true }); |
| 169 | if (!ctx) { |
| 170 | bitmap.close(); |
| 171 | return { error: "decode_failed" }; |
| 172 | } |
| 173 | ctx.imageSmoothingQuality = "high"; |
| 174 | ctx.drawImage(bitmap, 0, 0, w, h); |
| 175 | bitmap.close(); |
| 176 | const options: ImageEncodeOptions = { type: targetMime }; |
| 177 | if (targetMime === "image/webp") options.quality = WEBP_QUALITY; |
| 178 | const blob = await canvas.convertToBlob(options); |
| 179 | if (blob.size > TARGET_MAX_BYTES) { |
| 180 | return { error: "too_large_after_normalize" }; |
| 181 | } |
| 182 | const buf = await blob.arrayBuffer(); |
| 183 | const dataUrl = `data:${targetMime};base64,${bufferToBase64(buf)}`; |
| 184 | return { dataUrl, mime: targetMime, bytes: blob.size }; |
| 185 | } catch { |
| 186 | try { |
| 187 | bitmap.close(); |
| 188 | } catch { |
| 189 | // bitmap already closed |
| 190 | } |
| 191 | return { error: "decode_failed" }; |
| 192 | } |
| 193 | } |
| 194 | |
| 195 | export async function encodeImageInWorker( |
| 196 | input: EncodeInput, |
| 197 | ): Promise<EncodeResponse> { |
| 198 | const { id, file } = input; |
| 199 | const origBytes = file.size; |
| 200 | |
| 201 | let buffer: ArrayBuffer; |
| 202 | try { |
| 203 | buffer = await file.arrayBuffer(); |
| 204 | } catch { |
| 205 | return { id, ok: false, reason: "io" }; |
| 206 | } |
| 207 | |
| 208 | const head = new Uint8Array(buffer.slice(0, 12)); |
| 209 | const sniffed = sniffImageMime(head); |
| 210 | if (!sniffed) return { id, ok: false, reason: "magic_mismatch" }; |
| 211 | if (!SUPPORTED_MIMES.has(sniffed)) { |
| 212 | return { id, ok: false, reason: "invalid_mime" }; |
| 213 | } |
| 214 | // Defend against MIME spoofing: the declared ``file.type`` can lie. |
| 215 | if (file.type && SUPPORTED_MIMES.has(file.type) && file.type !== sniffed) { |
| 216 | // Trust the magic bytes; proceed with the sniffed MIME. |
| 217 | } |
| 218 | |
| 219 | if (origBytes <= TARGET_MAX_BYTES) { |
| 220 | const dataUrl = `data:${sniffed};base64,${bufferToBase64(buffer)}`; |
| 221 | return { |
| 222 | id, |
| 223 | ok: true, |
| 224 | dataUrl, |
| 225 | mime: sniffed, |
| 226 | bytes: origBytes, |
| 227 | origBytes, |
| 228 | normalized: false, |
| 229 | }; |
| 230 | } |
| 231 | |
| 232 | const result = await normalize(file, sniffed); |
| 233 | if ("error" in result) { |
| 234 | return { id, ok: false, reason: result.error }; |
| 235 | } |
| 236 | return { |
| 237 | id, |
| 238 | ok: true, |
| 239 | dataUrl: result.dataUrl, |
| 240 | mime: result.mime, |
| 241 | bytes: result.bytes, |
| 242 | origBytes, |
| 243 | normalized: true, |
| 244 | }; |
| 245 | } |
| 246 | |
| 247 | // --- Worker boot ------------------------------------------------------------- |
| 248 | // Only attach the message listener when running *inside* a Worker so the same |
| 249 | // module can be imported by tests (and by the thin ``imageEncode.ts`` wrapper |
| 250 | // in the main thread, which also calls ``encodeImageInWorker`` as a |
| 251 | // fall-through path when the Worker isn't available). |
| 252 | |
| 253 | declare const self: DedicatedWorkerGlobalScope; |
| 254 | |
| 255 | if ( |
| 256 | typeof self !== "undefined" && |
| 257 | typeof (self as unknown as { importScripts?: unknown }).importScripts === |
| 258 | "function" |
| 259 | ) { |
| 260 | self.addEventListener("message", async (event: MessageEvent<EncodeInput>) => { |
| 261 | const response = await encodeImageInWorker(event.data); |
| 262 | self.postMessage(response); |
| 263 | }); |
| 264 | } |
| 265 |