@gmickel/gno 1.18.0 → 1.19.0

package/src/core/retrieval-trace-management.ts

@@ -0,0 +1,422 @@
+/** Shared local management service for private retrieval trace receipts. */
+
+import type {
+  RetrievalTraceBundle,
+  RetrievalTraceExportFormat,
+  StorePort,
+  StoreResult,
+} from "../store/types";
+import type {
+  ReplayRetrievalTraceInput,
+  ReplayRetrievalTraceResult,
+} from "./retrieval-replay-types";
+import type {
+  DeleteRetrievalTraceResult,
+  ExportRetrievalTracesInput,
+  ExportRetrievalTracesResult,
+  LabelRetrievalTraceInput,
+  LabelRetrievalTraceResult,
+  PurgeRetrievalTracesResult,
+  RetrievalTraceDetail,
+  RetrievalTraceDetailOptions,
+  RetrievalTraceListOptions,
+  RetrievalTraceListResult,
+  RetrievalTraceSummary,
+} from "./retrieval-trace-management-types";
+
+import {
+  canonicalTraceJson,
+  hashTraceCanonical,
+} from "../store/retrieval-trace-codec";
+import { err, ok } from "../store/types";
+import {
+  replayRetrievalTraces,
+  type RetrievalReplayDeps,
+} from "./retrieval-replay";
+import { exportRetrievalTraces } from "./retrieval-trace-export";
+import {
+  allEvidence,
+  applyTargetKind,
+  compactTarget,
+  decodeCursor,
+  encodeCursor,
+  type EvidenceTarget,
+  latestForTarget,
+  refMatches,
+  summaryOf,
+  targetKey,
+} from "./retrieval-trace-management-helpers";
+
+const DEFAULT_LIST_LIMIT = 50;
+const MAX_LIST_LIMIT = 1000;
+const DEFAULT_DETAIL_LIMIT = 1000;
+const MAX_DETAIL_LIMIT = 10_000;
+const SHA256_PATTERN = /^[a-f0-9]{64}$/;
+const DOCID_PATTERN = /^#[a-f0-9]{6,}$/;
+
+interface ManagementDeps {
+  clock?: () => number;
+}
+
+export class RetrievalTraceManagementService {
+  private readonly clock: () => number;
+
+  constructor(
+    private readonly store: StorePort,
+    deps: ManagementDeps = {}
+  ) {
+    this.clock = deps.clock ?? Date.now;
+  }
+
+  async list(
+    options: RetrievalTraceListOptions = {}
+  ): Promise<StoreResult<RetrievalTraceListResult>> {
+    const limit = options.limit ?? DEFAULT_LIST_LIMIT;
+    if (!Number.isSafeInteger(limit) || limit < 1 || limit > MAX_LIST_LIMIT) {
+      return err(
+        "INVALID_INPUT",
+        `Trace limit must be from 1 to ${MAX_LIST_LIMIT}`
+      );
+    }
+    const cursor = decodeCursor(options.cursor);
+    if (!cursor.ok) return cursor;
+    const rows = await this.store.listRetrievalTraces(limit + 1, cursor.value);
+    if (!rows.ok) return rows;
+    const page = rows.value.slice(0, limit);
+    const last = page.at(-1);
+    return ok({
+      schemaVersion: "1.0",
+      traces: page.map(summaryOf),
+      nextCursor:
+        rows.value.length > limit && last
+          ? encodeCursor({
+              createdAtMs: last.createdAtMs,
+              traceId: last.traceId,
+            })
+          : null,
+    });
+  }
+
+  async show(
+    traceId: string,
+    options: RetrievalTraceDetailOptions = {}
+  ): Promise<StoreResult<RetrievalTraceDetail>> {
+    const limit = options.detailLimit ?? DEFAULT_DETAIL_LIMIT;
+    if (!Number.isSafeInteger(limit) || limit < 1 || limit > MAX_DETAIL_LIMIT) {
+      return err(
+        "INVALID_INPUT",
+        `Trace detail limit must be from 1 to ${MAX_DETAIL_LIMIT}`
+      );
+    }
+    const stored = await this.store.getBoundedRetrievalTrace(traceId, limit);
+    if (!stored.ok) return stored;
+    if (!stored.value) return err("NOT_FOUND", "Retrieval trace not found");
+    const { trace, runs, events, judgments, exports } = stored.value.bundle;
+    const totals = stored.value.totals;
+    return ok({
+      schemaVersion: "1.0",
+      trace,
+      runs: runs.slice(0, limit),
+      events: events.slice(0, limit),
+      judgments: judgments.slice(0, limit),
+      exports: exports.slice(0, limit),
+      totals,
+      truncated: {
+        runs: totals.runs > runs.length,
+        events: totals.events > events.length,
+        judgments: totals.judgments > judgments.length,
+        exports: totals.exports > exports.length,
+      },
+    });
+  }
+
+  async label(
+    input: LabelRetrievalTraceInput
+  ): Promise<StoreResult<LabelRetrievalTraceResult>> {
+    if (
+      typeof input.traceId !== "string" ||
+      typeof input.targetRef !== "string"
+    ) {
+      return err("INVALID_INPUT", "Invalid retrieval trace label");
+    }
+    const stored = await this.store.getRetrievalTrace(input.traceId);
+    if (!stored.ok) return stored;
+    if (!stored.value) return err("NOT_FOUND", "Retrieval trace not found");
+    const resolved = await this.resolveLabelTarget(stored.value, input);
+    if (!resolved.ok) return resolved;
+    const { target, targetKind, runId } = resolved.value;
+    const key = targetKey(targetKind, target);
+    const latest = latestForTarget(stored.value.judgments, key);
+    if (latest?.label === input.label) {
+      return ok({
+        schemaVersion: "1.0",
+        result: "duplicate",
+        judgment: latest,
+      });
+    }
+    const now = Math.max(this.clock(), (latest?.createdAtMs ?? -1) + 1);
+    if (!Number.isSafeInteger(now) || now < 0) {
+      return err("INVALID_INPUT", "Trace clock must return epoch milliseconds");
+    }
+    const idempotencyKey =
+      input.idempotencyKey ??
+      `label:${hashTraceCanonical({
+        traceId: input.traceId,
+        key,
+        label: input.label,
+        supersedes: latest?.judgmentId ?? null,
+      })}`;
+    const judgmentId = `judgment-${hashTraceCanonical({
+      traceId: input.traceId,
+      idempotencyKey,
+    }).slice(0, 40)}`;
+    const targetRef = await this.redactTargetRef(
+      stored.value.trace.redactionMode,
+      input.targetRef
+    );
+    if (!targetRef.ok) return targetRef;
+    const judgment = {
+      judgmentId,
+      traceId: input.traceId,
+      runId,
+      idempotencyKey,
+      label: input.label,
+      targetKind,
+      targetRef: targetRef.value,
+      target,
+      createdAtMs: now,
+    };
+    const appended = await this.store.appendRetrievalTraceJudgment(judgment);
+    if (!appended.ok) {
+      if (appended.error.code !== "CONSTRAINT_VIOLATION") return appended;
+      const concurrent = await this.store.getRetrievalTrace(input.traceId);
+      if (!concurrent.ok) return concurrent;
+      const settled = concurrent.value
+        ? latestForTarget(concurrent.value.judgments, key)
+        : undefined;
+      if (
+        settled?.label === input.label &&
+        settled.idempotencyKey === idempotencyKey
+      ) {
+        return ok({
+          schemaVersion: "1.0",
+          result: "duplicate",
+          judgment: settled,
+        });
+      }
+      return appended;
+    }
+    const refreshed = await this.store.getRetrievalTrace(input.traceId);
+    if (!refreshed.ok) return refreshed;
+    const row = refreshed.value?.judgments.find(
+      ({ judgmentId: id }) => id === judgmentId
+    );
+    return row
+      ? ok({ schemaVersion: "1.0", result: appended.value, judgment: row })
+      : err("QUERY_FAILED", "Stored retrieval trace judgment is unavailable");
+  }
+
+  async export<Format extends RetrievalTraceExportFormat = "agentic-receipt">(
+    input: ExportRetrievalTracesInput<Format>
+  ): Promise<StoreResult<ExportRetrievalTracesResult<Format>>> {
+    return exportRetrievalTraces(this.store, this.clock, input);
+  }
+
+  async replay(
+    input: ReplayRetrievalTraceInput,
+    deps: Omit<RetrievalReplayDeps, "store">
+  ): Promise<StoreResult<ReplayRetrievalTraceResult>> {
+    return replayRetrievalTraces({ ...deps, store: this.store }, input);
+  }
+
+  async delete(
+    traceId: string
+  ): Promise<StoreResult<DeleteRetrievalTraceResult>> {
+    const deleted = await this.store.deleteRetrievalTrace(traceId);
+    if (!deleted.ok) return deleted;
+    if (deleted.value.traces === 0) {
+      return err("NOT_FOUND", "Retrieval trace not found");
+    }
+    return ok({
+      schemaVersion: "1.0",
+      traceId,
+      deleted: true,
+      counts: deleted.value,
+    });
+  }
+
+  async purge(): Promise<StoreResult<PurgeRetrievalTracesResult>> {
+    const purged = await this.store.purgeRetrievalTraces();
+    return purged.ok ? ok({ schemaVersion: "1.0", ...purged.value }) : purged;
+  }
+
+  private async resolveLabelTarget(
+    bundle: RetrievalTraceBundle,
+    input: LabelRetrievalTraceInput
+  ): Promise<
+    StoreResult<{
+      target: EvidenceTarget;
+      targetKind: NonNullable<LabelRetrievalTraceInput["targetKind"]>;
+      runId: string | null;
+    }>
+  > {
+    if (
+      typeof input.traceId !== "string" ||
+      typeof input.targetRef !== "string" ||
+      !["relevant", "irrelevant", "missing_expected"].includes(input.label) ||
+      input.targetRef.length < 1 ||
+      input.targetRef.length > 4096 ||
+      (input.targetKind !== undefined &&
+        !["document", "chunk", "span"].includes(input.targetKind)) ||
+      (input.sourceHash !== undefined &&
+        !SHA256_PATTERN.test(input.sourceHash)) ||
+      (input.docid !== undefined && !DOCID_PATTERN.test(input.docid))
+    ) {
+      return err("INVALID_INPUT", "Invalid retrieval trace label");
+    }
+    if (input.label === "missing_expected") {
+      return await this.resolveMissingExpected(bundle, input);
+    }
+    const match = allEvidence(bundle).find(({ target }) =>
+      refMatches(target, input.targetRef)
+    );
+    if (!match) {
+      return err(
+        "INVALID_INPUT",
+        "Relevant and irrelevant labels must reference recorded evidence"
+      );
+    }
+    const target = applyTargetKind(match.target, input);
+    if (!target.ok) return target;
+    const targetKind =
+      input.targetKind ??
+      (input.startLine !== undefined
+        ? "span"
+        : target.value.seq !== undefined
+          ? "chunk"
+          : "document");
+    return ok({ target: target.value, targetKind, runId: match.runId });
+  }
+
+  private async resolveMissingExpected(
+    bundle: RetrievalTraceBundle,
+    input: LabelRetrievalTraceInput
+  ): Promise<
+    StoreResult<{
+      target: EvidenceTarget;
+      targetKind: "document";
+      runId: null;
+    }>
+  > {
+    if (input.targetKind !== undefined && input.targetKind !== "document") {
+      return err(
+        "INVALID_INPUT",
+        "missing_expected judgments must target a document"
+      );
+    }
+    let document: Awaited<ReturnType<StorePort["getDocumentByUri"]>> | null =
+      null;
+    if (input.targetRef.startsWith("gno://")) {
+      document = await this.store.getDocumentByUri(input.targetRef);
+    } else if (DOCID_PATTERN.test(input.targetRef)) {
+      document = await this.store.getDocumentByDocid(input.targetRef);
+    } else if (!SHA256_PATTERN.test(input.targetRef)) {
+      return err(
+        "INVALID_INPUT",
+        "Expected documents require a gno:// URI, docid, or source hash"
+      );
+    }
+    if (document && !document.ok) return document;
+    const row = document?.ok ? document.value : null;
+    const target: EvidenceTarget = compactTarget(
+      row
+        ? {
+            docid: row.docid,
+            sourceHash: row.sourceHash,
+            mirrorHash: row.mirrorHash ?? undefined,
+            uri: row.uri,
+          }
+        : {
+            docid:
+              input.docid ??
+              (DOCID_PATTERN.test(input.targetRef)
+                ? input.targetRef
+                : undefined),
+            sourceHash:
+              input.sourceHash ??
+              (SHA256_PATTERN.test(input.targetRef)
+                ? input.targetRef
+                : undefined),
+            uri: input.targetRef.startsWith("gno://")
+              ? input.targetRef
+              : undefined,
+          }
+    );
+    if (!(target.uri || target.docid || target.sourceHash)) {
+      return err("INVALID_INPUT", "Expected document identity is incomplete");
+    }
+    if (input.sourceHash && row && row.sourceHash !== input.sourceHash) {
+      return err("INVALID_INPUT", "Expected document source hash mismatch");
+    }
+    if (input.docid && row && row.docid !== input.docid) {
+      return err("INVALID_INPUT", "Expected document docid mismatch");
+    }
+    if (
+      allEvidence(bundle).some(({ target: evidence }) =>
+        [target.uri, target.docid, target.sourceHash]
+          .filter(Boolean)
+          .some((identity) => refMatches(evidence, identity!))
+      )
+    ) {
+      return err(
+        "CONSTRAINT_VIOLATION",
+        "Expected document is already present in recorded evidence"
+      );
+    }
+    return ok({ target, targetKind: "document", runId: null });
+  }
+
+  private async redactTargetRef(
+    mode: RetrievalTraceBundle["trace"]["redactionMode"],
+    targetRef: string
+  ): Promise<StoreResult<string>> {
+    if (mode === "replay") return ok(targetRef.normalize("NFC"));
+    const secret = await this.store.getOrCreateRetrievalTraceRedactionSecret();
+    if (!secret.ok) return secret;
+    return ok(
+      `redacted:${new Bun.CryptoHasher("sha256")
+        .update(`${secret.value}\0${targetRef.normalize("NFC")}`)
+        .digest("hex")}`
+    );
+  }
+}
+
+export type {
+  DeleteRetrievalTraceResult,
+  ExportRetrievalTracesInput,
+  ExportRetrievalTracesResult,
+  LabelRetrievalTraceInput,
+  LabelRetrievalTraceResult,
+  PurgeRetrievalTracesResult,
+  RetrievalTraceDetail,
+  RetrievalTraceDetailOptions,
+  RetrievalTraceListOptions,
+  RetrievalTraceListRequest,
+  RetrievalTraceListResult,
+  RetrievalTraceLabelRequest,
+  RetrievalTraceLabelResult,
+  RetrievalTraceExportRequest,
+  RetrievalTraceExportResult,
+  RetrievalTraceDeleteResult,
+  RetrievalTracePurgeResult,
+  RetrievalTraceSummary,
+} from "./retrieval-trace-management-types";
+export type {
+  ReplayRetrievalTraceInput,
+  ReplayRetrievalTraceResult,
+  RetrievalReplayCandidate,
+} from "./retrieval-replay-types";
+
+export const serializeRetrievalTraceArtifact = (
+  result: ExportRetrievalTracesResult
+): string => canonicalTraceJson(result.artifact);

package/src/core/retrieval-trace-request.ts

@@ -0,0 +1,141 @@
+/** Surface-boundary request setup for private retrieval traces. */
+
+import type { Config } from "../config/types";
+import type { HybridSearchOptions, SearchOptions } from "../pipeline/types";
+import type { StorePort, StoreResult } from "../store/types";
+import type { RetrievalTraceFingerprints } from "../store/types";
+import type { RetrievalTraceTerminalStatus } from "../store/types";
+
+import { canonicalTraceJson } from "../store/retrieval-trace-codec";
+import { err, ok } from "../store/types";
+import { RetrievalTraceSession } from "./retrieval-trace-session";
+
+export const retrievalTraceFailureStatus = (
+  cause: unknown
+): RetrievalTraceTerminalStatus => {
+  if (
+    cause instanceof Error &&
+    (cause.name === "AbortError" ||
+      (cause as Error & { code?: string }).code === "ABORT_ERR")
+  ) {
+    return "cancelled";
+  }
+  return "failed";
+};
+
+export const finishRetrievalTraceAfterError = async (
+  session: RetrievalTraceSession | null | undefined,
+  cause: unknown
+): Promise<void> => {
+  await session?.finish(retrievalTraceFailureStatus(cause));
+};
+
+const fingerprint = (value: unknown): string =>
+  new Bun.CryptoHasher("sha256")
+    .update(canonicalTraceJson(JSON.parse(JSON.stringify(value))))
+    .digest("hex");
+
+export const buildRetrievalTraceFingerprints = async (input: {
+  store: StorePort;
+  config: Config;
+  pipeline: string;
+  pipelineOptions?: Record<string, unknown>;
+  indexName?: string;
+  modelUris?: string[];
+}): Promise<RetrievalTraceFingerprints> => {
+  const collections = await input.store.getCollections();
+  if (!collections.ok) throw new Error(collections.error.message);
+  const snapshots = [];
+  for (const collection of [...collections.value].sort((left, right) =>
+    left.name.localeCompare(right.name)
+  )) {
+    const snapshot = await input.store.getActivationIndexSnapshot(
+      collection.name
+    );
+    if (!snapshot.ok) throw new Error(snapshot.error.message);
+    snapshots.push({ collection: collection.name, value: snapshot.value });
+  }
+  return {
+    pipeline: fingerprint({
+      contract: "retrieval-trace-v1",
+      pipeline: input.pipeline,
+      ...(input.pipelineOptions ? { options: input.pipelineOptions } : {}),
+    }),
+    model: fingerprint(
+      [...(input.modelUris ?? [])].sort((left, right) =>
+        left.localeCompare(right)
+      )
+    ),
+    config: fingerprint(input.config),
+    index: fingerprint({
+      indexName: input.indexName ?? "default",
+      snapshots,
+    }),
+  };
+};
+
+/** Start a real surface session; all expensive fingerprint work stays lazy. */
+export const startRetrievalTraceRequest = async (input: {
+  store: StorePort;
+  config: Config;
+  query: string;
+  goal?: string;
+  filters?: Record<string, unknown>;
+  pipeline: string;
+  indexName?: string;
+  modelUris?: string[];
+}): Promise<StoreResult<RetrievalTraceSession | null>> => {
+  if (input.config.retrievalTraces?.enabled !== true) return ok(null);
+  try {
+    return await RetrievalTraceSession.start({
+      store: input.store,
+      config: input.config.retrievalTraces,
+      query: input.query,
+      goal: input.goal,
+      filters: JSON.parse(JSON.stringify(input.filters ?? {})),
+      fingerprints: () => buildRetrievalTraceFingerprints(input),
+    });
+  } catch (cause) {
+    return err(
+      "QUERY_FAILED",
+      cause instanceof Error
+        ? `Trace fingerprinting failed: ${cause.message}`
+        : "Trace fingerprinting failed",
+      cause
+    );
+  }
+};
+
+export const retrievalTraceFilters = (
+  options: SearchOptions | HybridSearchOptions
+): Record<string, unknown> =>
+  JSON.parse(
+    JSON.stringify({
+      limit: options.limit,
+      minScore: options.minScore,
+      collection: options.collection,
+      lang: options.lang,
+      full: options.full,
+      lineNumbers: options.lineNumbers,
+      tagsAll: options.tagsAll,
+      tagsAny: options.tagsAny,
+      since: options.since,
+      until: options.until,
+      categories: options.categories,
+      author: options.author,
+      intent: options.intent,
+      exclude: options.exclude,
+      ...("noExpand" in options
+        ? {
+            noExpand: options.noExpand,
+            noRerank: options.noRerank,
+            candidateLimit: options.candidateLimit,
+            explain: options.explain,
+            graph: options.graph,
+            noGraph: options.noGraph,
+            queryLanguageHint: options.queryLanguageHint,
+            queryModes: options.queryModes,
+          }
+        : {}),
+    })
+  );