@shpitdev/convex-cliproxy-observability 0.3.6 → 0.3.7

package/src/protocols/checkpoint.ts

@@ -1,4 +1,4 @@
-import { expandStockHookBlock } from "./stock-block.js";
+import { stockHookPieces } from "./body-parts.js";
 import {
   providerIdentity,
   TOKEN_FIELDS,
@@ -7,7 +7,7 @@ import {
   type PrivateContentReference,
 } from "../model-call/index.js";
 import type { CaptureObservationV1 } from "../capture/index.js";
-import { decodeBody } from "../capture/index.js";
+import { continuesBodyPart, decodeBody } from "../capture/index.js";
 import { assembleResponsesStream } from "./responses.js";
 import { SSEReader, type SSECheckpoint } from "./framing.js";
 import { count, list, nanoTime, object, parseObject } from "./values.js";
@@ -28,6 +28,10 @@ export type ProjectionCheckpoint = {
   streamed?: boolean;
   inputSeen?: boolean;
   chatFinished?: boolean;
+  /** First segment of an unfinished split body; projection reassembles from there. */
+  bodyPartSegment?: number;
+  /** Last part of an unfinished split body; the next observation must continue it. */
+  openBodyPart?: Pick<CaptureObservationV1, "kind" | "sequence" | "bodyPart">;
 };
 const usageFields = [
   "input_tokens",
@@ -186,10 +190,22 @@ function observeProtocol(
     }
   }
 }
+/**
+ * Split bodies up to this size are reassembled for scalar projection. Larger ones
+ * stay complete in raw segments; projection does not allocate them in a 64 MiB
+ * Convex action.
+ */
+export const MAX_PROJECTED_SPLIT_BYTES = 8 * 1024 * 1024;
+/**
+ * Earlier parts of a split body only advance the sequence. The final part needs
+ * the exact reassembled body (`BodyPartAssembler`) as `assembled`, or `null` when
+ * the body exceeds `MAX_PROJECTED_SPLIT_BYTES` and its semantics are not projected.
+ */
 export function applyObservation(
   call: ModelCallV1,
   state: ProjectionCheckpoint,
   o: CaptureObservationV1,
+  assembled?: Uint8Array | null,
 ): void {
   if (o.sequence !== call.capture.projectedThroughSequence + 1)
     throw new Error("projection sequence gap");
@@ -204,7 +220,25 @@ export function applyObservation(
     if (!call.capture.gaps.includes(o.gap) && call.capture.gaps.length < 8)
       call.capture.gaps.push(o.gap.slice(0, 128));
   }
-  const body = decodeBody(o);
+  const part = o.bodyPart;
+  // Segments validate only their own neighbors; continuity across them is checked here.
+  if (
+    (state.openBodyPart || (part && part.index > 0)) &&
+    !(state.openBodyPart && continuesBodyPart(state.openBodyPart, o))
+  )
+    throw new Error("body part interrupted");
+  const deferred = part !== undefined && part.index < part.count - 1;
+  state.openBodyPart = deferred
+    ? { kind: o.kind, sequence: o.sequence, bodyPart: part }
+    : undefined;
+  if (part && !deferred && assembled === undefined)
+    throw new Error("split body requires reassembly");
+  const unprojected = part !== undefined && !deferred && assembled === null;
+  if (unprojected && part.totalBytes <= MAX_PROJECTED_SPLIT_BYTES)
+    throw new Error("split body within projection bound");
+  // Like an over-limit SSE frame: content is retained, its semantics are not decoded.
+  if (unprojected && o.kind !== "request") state.invalid = state.frameLimit = true;
+  const body = deferred || unprojected ? new Uint8Array() : (assembled ?? decodeBody(o));
   const decoder = new TextDecoder("utf-8", { fatal: true });
   // A new after-auth event replaces the selection, including unknown fields.
   // Stream initialization may omit auth; an explicit selection replaces the pair.
@@ -227,14 +261,21 @@ export function applyObservation(
     call.correlationConflicts = o.correlationConflicts ?? [];
     call.traceId = o.correlation?.traceId;
     call.sourceTraceId = o.sourceTraceId;
-    try {
-      const input = parseObject(decoder.decode(body));
-      state.streamed = input.stream === true;
-      call.streamed = state.streamed;
-    } catch {
-      state.invalid = true;
+    if (!deferred && !unprojected) {
+      try {
+        const input = parseObject(decoder.decode(body));
+        state.streamed = input.stream === true;
+        call.streamed = state.streamed;
+      } catch {
+        state.invalid = true;
+      }
     }
   }
+  // An unparsed request (e.g. a split body over the projection bound) takes
+  // streaming from the protocol evidence that follows it.
+  if (call.streamed === undefined && (o.kind === "stream_init" || o.kind === "stream_chunk"))
+    call.streamed = true;
+  else if (call.streamed === undefined && o.kind === "response") call.streamed = false;
   // Derived read fields: reads fill them only for summaries projected before 0.3.0.
   Object.assign(call, providerIdentity(call), { costProvenance: call.cost.kind });
   if (o.kind === "response" || o.kind === "stream_chunk" || o.kind === "completion") {
@@ -251,10 +292,7 @@ export function applyObservation(
       }
       if ((o.kind === "stream_chunk" || o.kind === "completion") && body.length) {
         const reader = new SSEReader(state.sse);
-        const pieces =
-          o.bodyFraming === "stock_hook_block"
-            ? expandStockHookBlock(body, o.stockHookChunkLengths ?? [])
-            : [body];
+        const pieces = o.bodyFraming === "stock_hook_block" ? stockHookPieces(o, body) : [body];
         const frames = pieces.flatMap((piece) =>
           o.bodyFraming === "stock_hook_chunk" || o.bodyFraming === "stock_hook_block"
             ? reader.feedStock(piece, call.clientProtocol, o.observedAt)

package/src/protocols/index.ts

@@ -144,3 +144,4 @@ export function projectCapturedPayloads(args: {
 }
 
 export { expandStockHookBlock } from "./stock-block.js";
+export { BodyPartAssembler, stockHookPieces } from "./body-parts.js";