@happyvertical/smrt-ingestion 0.55.6 → 0.55.8

package/dist/server.js

@@ -1,6 +1,6 @@
 import "./models.js";
 import { n as generationSnapshotRead } from "./chunks/execution-internal-DWteDHM2.js";
-import { n as extractAnalysis, r as proposeDocumentSplits, t as createSDKExtractionAdapter } from "./chunks/extraction-bseWZuWN.js";
+import { i as proposeDocumentSplits, n as createSDKExtractionAdapter, r as extractAnalysis, t as GenerationSnapshotStaleError } from "./chunks/proposal-errors-z24VKWnQ.js";
 import { detectEngine, isEmbeddedDatabase, isTransientDatabaseError, withEmbeddedWriteTransaction } from "@happyvertical/smrt-core";
 import { withTenant } from "@happyvertical/smrt-tenancy";
 import { createHash, randomUUID } from "node:crypto";
@@ -645,6 +645,9 @@ var IngestionService = class {
 	async reconcileAction(actionId) {
 		return (await this.execution()).reconcileAction(actionId);
 	}
+	async listReviews(itemId, input = {}) {
+		return (await this.execution()).listReviews(itemId, input);
+	}
 	async getAction(actionId) {
 		return (await this.execution()).getAction(actionId);
 	}
@@ -677,6 +680,28 @@ var IngestionService = class {
 	async previewGeneratedProposals(input) {
 		return (await this.proposals()).preview(input);
 	}
+	async feedback() {
+		const { IntakeFeedbackService } = await import("./chunks/feedback-C3OHD4sA.js").then((n) => n.n);
+		return new IntakeFeedbackService(this.options);
+	}
+	async recordFeedback(input) {
+		return (await this.feedback()).record(input);
+	}
+	async observeFeedback(input) {
+		return (await this.feedback()).observe(input);
+	}
+	async retrieveFeedback(input) {
+		return (await this.proposals()).retrieveFeedback(input);
+	}
+	async suggestRoutingRule(input) {
+		return (await this.feedback()).suggest(input);
+	}
+	async getRoutingRule(input) {
+		return (await this.feedback()).readRule(input);
+	}
+	async adoptRoutingRule(input) {
+		return (await this.feedback()).adopt(input);
+	}
 	now() {
 		return this.options.now?.() ?? /* @__PURE__ */ new Date();
 	}
@@ -1063,62 +1088,129 @@ var IngestionService = class {
 	}
 	/** Append an immutable input revision. Reprocessing never creates actions. */
 	async analyze(itemId, configuration, requestKey) {
+		const frozen = structuredClone(configuration);
+		if (Object.hasOwn(frozen, "humanCorrection") && frozen.stage !== "interpret") throw new Error("Human corrections require authenticated split revision");
+		return this.tx((db) => this.appendAnalysis(db, itemId, frozen, requestKey));
+	}
+	async appendAnalysis(db, itemId, configuration, requestKey) {
 		nonempty(requestKey);
 		const configDigest = digest(configuration);
-		return this.tx(async (db) => {
-			await this.lock(db, itemId);
-			const item = await this.item(itemId, "process", db);
-			if (item.receipt_state !== "ready" || item.cancelled) throw new Error("Intake not processable");
-			const replay = (await this.rows(db, "intake_analyses", "item_id=?", [itemId])).find((a) => object(a.data).requestKey === requestKey);
-			if (replay) {
-				if (object(replay.data).configDigest !== configDigest) throw new Error("Analysis request conflict");
-				return {
-					id: String(replay.id),
-					revision: Number(replay.revision)
-				};
-			}
-			const evidence = await this.rows(db, "intake_evidence", "item_id=? AND state='durable'", [itemId]);
-			await db.query("UPDATE intake_analysis_attempts SET state='superseded' WHERE tenant_id=? AND confidential_scope_id=? AND item_id=? AND state IN ('queued','running')", this.scope.tenantId, this.scope.confidentialScopeId, itemId);
-			await db.query("UPDATE intake_analyses SET state='superseded' WHERE tenant_id=? AND confidential_scope_id=? AND item_id=? AND state IN ('queued','running')", this.scope.tenantId, this.scope.confidentialScopeId, itemId);
-			const revision = Number(item.analysis_revision) + 1;
-			await db.query("UPDATE intake_dispatches SET state='completed' WHERE tenant_id=? AND confidential_scope_id=? AND item_id=? AND revision<? AND state<>'completed'", this.scope.tenantId, this.scope.confidentialScopeId, itemId, revision);
-			const inputs = evidence.map((e) => ({
-				id: e.id,
-				hash: e.content_hash
-			})).sort((a, b) => String(a.id).localeCompare(String(b.id)));
-			const [analysis] = await this.insert(db, "intake_analyses", {
-				item_id: itemId,
-				revision,
-				fence: 0,
-				attempt_number: 0,
-				current_attempt_id: null,
-				state: "queued",
-				input_digest: digest({
-					inputs,
-					configuration
-				}),
-				data: canonical({
-					inputs,
-					configuration,
-					requestKey,
-					configDigest
-				})
-			});
-			await db.query("UPDATE intake_items SET analysis_revision=?,processing_state='queued' WHERE tenant_id=? AND confidential_scope_id=? AND id=?", revision, this.scope.tenantId, this.scope.confidentialScopeId, itemId);
-			await this.insert(db, "intake_dispatches", {
-				item_id: itemId,
-				stage: "analyze",
-				revision,
-				state: "pending",
-				job_id: "",
-				deliveries: 0,
-				data: "{}"
-			}, "tenant_id,item_id,stage,revision");
+		await this.lock(db, itemId);
+		const item = await this.item(itemId, "process", db);
+		if (item.receipt_state !== "ready" || item.cancelled) throw new Error("Intake not processable");
+		const replay = (await this.rows(db, "intake_analyses", "item_id=?", [itemId])).find((a) => object(a.data).requestKey === requestKey);
+		if (replay) {
+			if (object(replay.data).configDigest !== configDigest) throw new Error("Analysis request conflict");
 			return {
-				id: String(analysis.id),
-				revision
+				id: String(replay.id),
+				revision: Number(replay.revision)
 			};
+		}
+		const evidence = await this.rows(db, "intake_evidence", "item_id=? AND state='durable'", [itemId]);
+		await db.query("UPDATE intake_analysis_attempts SET state='superseded' WHERE tenant_id=? AND confidential_scope_id=? AND item_id=? AND state IN ('queued','running')", this.scope.tenantId, this.scope.confidentialScopeId, itemId);
+		await db.query("UPDATE intake_analyses SET state='superseded' WHERE tenant_id=? AND confidential_scope_id=? AND item_id=? AND state IN ('queued','running')", this.scope.tenantId, this.scope.confidentialScopeId, itemId);
+		const revision = Number(item.analysis_revision) + 1;
+		await db.query("UPDATE intake_dispatches SET state='completed' WHERE tenant_id=? AND confidential_scope_id=? AND item_id=? AND revision<? AND state<>'completed'", this.scope.tenantId, this.scope.confidentialScopeId, itemId, revision);
+		const inputs = evidence.map((e) => ({
+			id: e.id,
+			hash: e.content_hash
+		})).sort((a, b) => String(a.id).localeCompare(String(b.id)));
+		const [analysis] = await this.insert(db, "intake_analyses", {
+			item_id: itemId,
+			revision,
+			fence: 0,
+			attempt_number: 0,
+			current_attempt_id: null,
+			state: "queued",
+			input_digest: digest({
+				inputs,
+				configuration
+			}),
+			data: canonical({
+				inputs,
+				configuration,
+				requestKey,
+				configDigest
+			})
 		});
+		await db.query("UPDATE intake_items SET analysis_revision=?,processing_state='queued' WHERE tenant_id=? AND confidential_scope_id=? AND id=?", revision, this.scope.tenantId, this.scope.confidentialScopeId, itemId);
+		await this.insert(db, "intake_dispatches", {
+			item_id: itemId,
+			stage: "analyze",
+			revision,
+			state: "pending",
+			job_id: "",
+			deliveries: 0,
+			data: "{}"
+		}, "tenant_id,item_id,stage,revision");
+		return {
+			id: String(analysis.id),
+			revision
+		};
+	}
+	/** Record an authenticated split correction and queue fresh extraction, never copy model output. */
+	async reviseLogicalSplit(input) {
+		const frozen = structuredClone(input);
+		nonempty(frozen.requestId);
+		integer(frozen.expectedRevision);
+		if (!Array.isArray(frozen.groups) || !frozen.groups.length || frozen.groups.length > 100 || frozen.groups.some((group) => !Array.isArray(group) || !group.length || group.length > 1e3 || group.some((page) => !Number.isSafeInteger(page) || page < 1))) throw new Error("Invalid logical split");
+		const pages = frozen.groups.flat();
+		if (pages.length > 1e3 || new Set(pages).size !== pages.length) throw new Error("Invalid logical split");
+		const requestKey = `split:${frozen.requestId}`;
+		const correction = {
+			...frozen,
+			actorId: this.scope.actorId,
+			kind: "logical_split"
+		};
+		const replay = await this.tx(async (db) => {
+			await this.lock(db, frozen.itemId);
+			await this.item(frozen.itemId, "process", db);
+			const found = (await this.rows(db, "intake_analyses", "item_id=?", [frozen.itemId])).find((row) => object(row.data).requestKey === requestKey);
+			if (!found) return null;
+			if (digest(object(object(found.data).configuration).humanCorrection) !== digest(correction)) throw new Error("Split request conflict");
+			return {
+				id: String(found.id),
+				revision: Number(found.revision)
+			};
+		});
+		if (replay) {
+			await (await this.proposals())[generationSnapshotRead](frozen.itemId, (work) => this.tx(work), (db) => this.completedAnalysisSnapshot(db, frozen.itemId, frozen.attemptId, false), void 0, true);
+			return replay;
+		}
+		let result;
+		await (await this.proposals())[generationSnapshotRead](frozen.itemId, (work) => this.tx(work), (db) => this.completedAnalysisSnapshot(db, frozen.itemId, frozen.attemptId, true), async (db, current) => {
+			if (current.revision !== frozen.expectedRevision) throw new Error("Stale logical split");
+			const pin = object(object(current.configuration.proposals).source);
+			const source = await this.completedAnalysisSnapshot(db, frozen.itemId, String(pin.attemptId), false);
+			for (const key of [
+				"attemptId",
+				"revision",
+				"inputDigest",
+				"outputDigest",
+				"evidenceDigest"
+			]) if (pin[key] !== source[key]) throw new Error("Split source changed");
+			const extraction = source.result.output.results;
+			if (!Array.isArray(extraction)) throw new Error("Split source unavailable");
+			const evidence = extraction.find((value) => object(object(value).evidence).id === frozen.evidenceId);
+			if (!evidence) throw new Error("Split evidence unavailable");
+			const observed = /* @__PURE__ */ new Set();
+			for (const value of object(evidence).segments) {
+				const location = object(object(value).location);
+				if (location.kind === "page") observed.add(Number(location.page));
+				if (location.kind === "pages") {
+					const from = Number(location.startPage), to = Number(location.endPage);
+					if (!Number.isSafeInteger(from) || !Number.isSafeInteger(to) || to - from > 1e3) throw new Error("Invalid source range");
+					for (let page = from; page <= to; page++) observed.add(page);
+				}
+			}
+			if (!observed.size || pages.length !== observed.size || pages.some((page) => !observed.has(page))) throw new Error("Split must cover observed pages exactly");
+			result = await this.appendAnalysis(db, frozen.itemId, {
+				...source.configuration,
+				humanCorrection: correction
+			}, requestKey);
+		}, true);
+		if (!result) throw new Error("Split unavailable");
+		return result;
 	}
 	/** Allocate a fenced attempt; active leases cannot be stolen before expiry. */
 	async claimAnalysis(itemId, revision, workerId) {
@@ -1259,7 +1351,7 @@ var IngestionService = class {
 			"outputDigest",
 			"evidenceDigest"
 		]) if (pin[key] !== source[key]) throw new Error("Generation source changed");
-		if (source.result.provider !== "smrt-ingestion-extraction" || source.configuration.stage === "interpret" || source.revision !== generation.revision - 1 || digest(generation.evidence.map((entry) => ({
+		if (source.result.provider !== "smrt-ingestion-extraction" || source.configuration.stage === "interpret" || source.revision !== generation.revision - 1 || digest(configuration.humanCorrection ?? null) !== digest(source.configuration.humanCorrection ?? null) || digest(generation.evidence.map((entry) => ({
 			id: entry.id,
 			hash: entry.contentHash
 		}))) !== source.evidenceDigest) throw new Error("Generation evidence changed");
@@ -1583,6 +1675,6 @@ var IngestionService = class {
 	}
 };
 //#endregion
-export { EmailSourceAdapter, IngestionService, WatchFolderSourceAdapter, createSDKExtractionAdapter, createSourceDeliveryHandler, createVendorWebhookHandler, extractAnalysis, proposeDocumentSplits };
+export { EmailSourceAdapter, GenerationSnapshotStaleError, IngestionService, WatchFolderSourceAdapter, createSDKExtractionAdapter, createSourceDeliveryHandler, createVendorWebhookHandler, extractAnalysis, proposeDocumentSplits };
 
 //# sourceMappingURL=server.js.map