initial commit

app/lib/bundle.ts

@@ -0,0 +1,101 @@
+import type { DumpColumns } from './dump-parse'
+
+/**
+ * A parsed dump serialised as one flat file.
+ *
+ *   [0..4)   magic
+ *   [4..8)   manifest length in bytes
+ *   [8..12)  payload base offset
+ *   [12..16) reserved
+ *   [16..)   JSON manifest, then zero padding, then the column buffers
+ *
+ * The payload offset lives in the fixed header rather than the manifest so
+ * that writing it cannot change the manifest's own length.
+ *
+ * This exists for the case where the dump ships with the site instead of
+ * being uploaded. Parse once at build time and the browser does one fetch
+ * plus a few typed-array views over the response - no parsing, no IndexedDB.
+ * It also gzips far better than JSON, because each column is homogeneous.
+ */
+
+// Bumped to \x02 when `kind` and `value` were added; a \x01 bundle has no way
+// to tell enum constants from objects, so it is rejected rather than upgraded.
+const MAGIC = 0x50445802 // "PDX\x02"
+const MAGIC_V1 = 0x50445801
+const HEADER = 16
+
+const COLUMNS = [
+  ['addr', Float64Array],
+  ['typeId', Uint16Array],
+  ['pathStart', Uint32Array],
+  ['pathBlob', Uint8Array],
+  ['nameOff', Uint16Array],
+  ['rootOff', Uint16Array],
+  ['outer', Float64Array],
+  ['offset', Int32Array],
+  ['kind', Uint8Array],
+  ['value', BigInt64Array],
+] as const
+
+const align8 = (n: number) => (8 - (n % 8)) % 8
+
+export function encodeBundle(cols: DumpColumns): Uint8Array {
+  const layout: Record<string, { offset: number; length: number }> = {}
+  const parts: { at: number; bytes: Uint8Array }[] = []
+  let size = 0
+
+  for (const [key] of COLUMNS) {
+    const view = cols[key] as ArrayBufferView & { length: number }
+    size += align8(size)
+    layout[key] = { offset: size, length: view.length }
+    parts.push({
+      at: size,
+      bytes: new Uint8Array(view.buffer, view.byteOffset, view.byteLength),
+    })
+    size += view.byteLength
+  }
+
+  const json = new TextEncoder().encode(
+    JSON.stringify({
+      count: cols.count,
+      types: cols.types,
+      skipped: cols.skipped,
+      damaged: cols.damaged,
+      columns: layout,
+    }),
+  )
+  const base = HEADER + json.length + align8(HEADER + json.length)
+
+  const out = new Uint8Array(base + size)
+  const dv = new DataView(out.buffer)
+  dv.setUint32(0, MAGIC, true)
+  dv.setUint32(4, json.length, true)
+  dv.setUint32(8, base, true)
+  out.set(json, HEADER)
+  for (const { at, bytes } of parts) out.set(bytes, base + at)
+  return out
+}
+
+export function decodeBundle(buf: ArrayBuffer): DumpColumns {
+  const dv = new DataView(buf)
+  const magic = dv.getUint32(0, true)
+  if (magic === MAGIC_V1) {
+    throw new Error('This .pdx was built by an older parser. Re-run `npm run prebake`.')
+  }
+  if (magic !== MAGIC) throw new Error('Not a .pdx bundle.')
+  const jsonLen = dv.getUint32(4, true)
+  const base = dv.getUint32(8, true)
+  const manifest = JSON.parse(new TextDecoder().decode(new Uint8Array(buf, HEADER, jsonLen)))
+
+  const cols: Record<string, unknown> = {
+    count: manifest.count,
+    types: manifest.types,
+    skipped: manifest.skipped,
+    damaged: manifest.damaged ?? 0,
+  }
+  for (const [key, Ctor] of COLUMNS) {
+    const { offset, length } = manifest.columns[key]
+    cols[key] = new Ctor(buf, base + offset, length)
+  }
+  return cols as unknown as DumpColumns
+}