1 files changed, 21 insertions, 0 deletions
diff --git a/packages/util/src/encode.ts b/packages/util/src/encode.ts
index 138cf1608..e4c6e70ac 100644
--- a/packages/util/src/encode.ts
+++ b/packages/util/src/encode.ts
@@ -28,3 +28,24 @@ export function checksum(content: string): string | undefined {
   }
   return (hash >>> 0).toString(36)
 }
+
+export function sampledChecksum(content: string, limit = 500_000): string | undefined {
+  if (!content) return undefined
+  if (content.length <= limit) return checksum(content)
+
+  const size = 4096
+  const points = [
+    0,
+    Math.floor(content.length * 0.25),
+    Math.floor(content.length * 0.5),
+    Math.floor(content.length * 0.75),
+    content.length - size,
+  ]
+  const hashes = points
+    .map((point) => {
+      const start = Math.max(0, Math.min(content.length - size, point - Math.floor(size / 2)))
+      return checksum(content.slice(start, start + size)) ?? ""
+    })
+    .join(":")
+  return `${content.length}:${hashes}`
+}