@remit/storage-service 0.0.10 → 0.0.12

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@remit/storage-service",
-  "version": "0.0.10",
+  "version": "0.0.12",
   "type": "module",
   "main": "src/index.ts",
   "types": "src/index.ts",

package/src/backends/filesystem.ts

@@ -29,9 +29,6 @@ import {
 	buildBodyPartKey,
 	buildDeduplicatedKey,
 	buildExportArchiveKey,
-	buildExtractedPrefix,
-	buildExtractedSkippedKey,
-	buildExtractedTextKey,
 	buildMessageBodyKey,
 	buildOutboxAttachmentKey,
 	buildOutboxAttachmentPrefix,
@@ -174,105 +171,6 @@ export const createFilesystemStorageService = (
 			: buffer;
 	};
 
-	const storeExtractedText: StorageService["storeExtractedText"] = (params) => {
-		const { accountConfigId, accountId, messageId, partPath, text } = params;
-		return storeInternal({
-			key: buildExtractedTextKey(
-				accountConfigId,
-				accountId,
-				messageId,
-				partPath,
-			),
-			content: Buffer.from(text, "utf8"),
-		});
-	};
-
-	const storeExtractedSkipped: StorageService["storeExtractedSkipped"] = (
-		params,
-	) => {
-		const { accountConfigId, accountId, messageId, partPath, marker } = params;
-		return storeInternal({
-			key: buildExtractedSkippedKey(
-				accountConfigId,
-				accountId,
-				messageId,
-				partPath,
-			),
-			content: Buffer.from(JSON.stringify(marker), "utf8"),
-			compress: false,
-		});
-	};
-
-	const extractedResultExists: StorageService["extractedResultExists"] = async (
-		accountConfigId,
-		accountId,
-		messageId,
-		partPath,
-	) => {
-		const textPath = join(
-			basePath,
-			buildExtractedTextKey(accountConfigId, accountId, messageId, partPath),
-		);
-		const skippedPath = join(
-			basePath,
-			buildExtractedSkippedKey(accountConfigId, accountId, messageId, partPath),
-		);
-		const exists = (fullPath: string): Promise<boolean> =>
-			stat(fullPath)
-				.then(() => true)
-				.catch(() => false);
-
-		const [textExists, skippedExists] = await Promise.all([
-			exists(textPath),
-			exists(skippedPath),
-		]);
-		return textExists || skippedExists;
-	};
-
-	const retrieveExtractedText: StorageService["retrieveExtractedText"] = async (
-		accountConfigId,
-		accountId,
-		messageId,
-		partPath,
-	) => {
-		const fullPath = join(
-			basePath,
-			buildExtractedTextKey(accountConfigId, accountId, messageId, partPath),
-		);
-		const buffer = await readFile(fullPath).catch((error: unknown) => {
-			if (isStorageNotFoundError(error)) return null;
-			throw error;
-		});
-		if (!buffer) return null;
-		const decoded =
-			buffer[0] === 0x1f && buffer[1] === 0x8b ? gunzipSync(buffer) : buffer;
-		return decoded.toString("utf8");
-	};
-
-	const listExtractedTexts: StorageService["listExtractedTexts"] = async (
-		accountConfigId,
-		accountId,
-		messageId,
-	) => {
-		const prefix = buildExtractedPrefix(accountConfigId, accountId, messageId);
-		const suffix = ".txt.gz";
-		const dirPath = join(basePath, prefix);
-
-		const entries = await readdir(dirPath, { recursive: true }).catch(
-			(error: unknown) => {
-				if (isStorageNotFoundError(error)) return [];
-				throw error;
-			},
-		);
-
-		return entries
-			.filter((entry) => entry.endsWith(suffix))
-			.map((entry) => ({
-				partPath: entry.slice(0, -suffix.length),
-				key: `${prefix}${entry}`,
-			}));
-	};
-
 	const storeOutboxAttachment: StorageService["storeOutboxAttachment"] = (
 		params,
 	) => {
@@ -400,6 +298,27 @@ export const createFilesystemStorageService = (
 		return stats ? { sizeBytes: stats.size } : null;
 	};
 
+	const retrieveOutboxAttachment: StorageService["retrieveOutboxAttachment"] = (
+		accountConfigId,
+		accountId,
+		outboxMessageId,
+		outboxAttachmentId,
+	) =>
+		readFile(
+			join(
+				basePath,
+				buildOutboxAttachmentKey(
+					accountConfigId,
+					accountId,
+					outboxMessageId,
+					outboxAttachmentId,
+				),
+			),
+		).catch((error: unknown) => {
+			if (isStorageNotFoundError(error)) return null;
+			throw error;
+		});
+
 	// This backend has no presigned anything, so the URL addresses the
 	// deployment's own upload route. Its authority is the same HMAC the read side
 	// uses on /content, under the write label, covering the storage key, the
@@ -540,15 +459,11 @@ export const createFilesystemStorageService = (
 		deleteOutboxAttachments,
 		deleteOutboxAttachment,
 		statOutboxAttachment,
+		retrieveOutboxAttachment,
 		createOutboxAttachmentUploadUrl,
 		storeDeduplicated,
 		storeParsedBody,
 		retrieveParsedBody,
-		storeExtractedText,
-		storeExtractedSkipped,
-		extractedResultExists,
-		retrieveExtractedText,
-		listExtractedTexts,
 		retrieveMessageBody: async (accountConfigId, accountId, messageId) => {
 			const key = buildMessageBodyKey(accountConfigId, accountId, messageId);
 			const fullPath = join(basePath, key);

package/src/backends/s3.test.ts

@@ -367,160 +367,6 @@ describe("createS3StorageService", () => {
 		);
 		assert.strictEqual(miss, null);
 	});
-
-	test("storeExtractedText writes gzipped text keyed under .../extracted/{partPath}.txt.gz and round-trips", async () => {
-		const { client, commands, stored } = createFakeS3Client();
-		const storage = createS3StorageService(client, "bucket");
-
-		const ref = await storage.storeExtractedText({
-			accountConfigId: "cfg-x",
-			accountId: "acc-x",
-			messageId: "msg-x",
-			partPath: "1.2",
-			text: "extracted attachment text",
-		});
-
-		assert.strictEqual(
-			ref.storageKey,
-			"accounts/cfg-x/acc-x/messages/msg-x/extracted/1.2.txt.gz",
-		);
-		const put = commands.find((c) => c.name === "PutObjectCommand");
-		assert.strictEqual(put?.input.ContentEncoding, "gzip");
-		assert.strictEqual(put?.input.ContentType, "text/plain; charset=utf-8");
-
-		const entry = stored.get(
-			"accounts/cfg-x/acc-x/messages/msg-x/extracted/1.2.txt.gz",
-		);
-		assert.ok(entry);
-		assert.strictEqual(
-			gunzipSync(entry.body).toString("utf8"),
-			"extracted attachment text",
-		);
-
-		const retrieved = await storage.retrieveExtractedText(
-			"cfg-x",
-			"acc-x",
-			"msg-x",
-			"1.2",
-		);
-		assert.strictEqual(retrieved, "extracted attachment text");
-	});
-
-	test("retrieveExtractedText returns null on miss", async () => {
-		const { client } = createFakeS3Client();
-		const storage = createS3StorageService(client, "bucket");
-		const result = await storage.retrieveExtractedText(
-			"cfg-x",
-			"acc-x",
-			"msg-x",
-			"nope",
-		);
-		assert.strictEqual(result, null);
-	});
-
-	test("storeExtractedSkipped writes an uncompressed JSON marker keyed under .../extracted/{partPath}.skipped.json", async () => {
-		const { client, commands, stored } = createFakeS3Client();
-		const storage = createS3StorageService(client, "bucket");
-
-		const ref = await storage.storeExtractedSkipped({
-			accountConfigId: "cfg-s",
-			accountId: "acc-s",
-			messageId: "msg-s",
-			partPath: "3",
-			marker: { status: "failed", reason: "pdf: corrupt" },
-		});
-
-		assert.strictEqual(
-			ref.storageKey,
-			"accounts/cfg-s/acc-s/messages/msg-s/extracted/3.skipped.json",
-		);
-		const put = commands.find((c) => c.name === "PutObjectCommand");
-		assert.strictEqual(put?.input.ContentEncoding, undefined);
-		assert.strictEqual(put?.input.ContentType, "application/json");
-
-		const entry = stored.get(
-			"accounts/cfg-s/acc-s/messages/msg-s/extracted/3.skipped.json",
-		);
-		assert.ok(entry);
-		assert.deepStrictEqual(JSON.parse(entry.body.toString("utf8")), {
-			status: "failed",
-			reason: "pdf: corrupt",
-		});
-	});
-
-	test("extractedResultExists is true once either the text artifact or the skip marker is stored", async () => {
-		const { client } = createFakeS3Client();
-		const storage = createS3StorageService(client, "bucket");
-
-		assert.strictEqual(
-			await storage.extractedResultExists("cfg-e", "acc-e", "msg-e", "1"),
-			false,
-		);
-
-		await storage.storeExtractedText({
-			accountConfigId: "cfg-e",
-			accountId: "acc-e",
-			messageId: "msg-e",
-			partPath: "1",
-			text: "hi",
-		});
-		assert.strictEqual(
-			await storage.extractedResultExists("cfg-e", "acc-e", "msg-e", "1"),
-			true,
-		);
-
-		await storage.storeExtractedSkipped({
-			accountConfigId: "cfg-e",
-			accountId: "acc-e",
-			messageId: "msg-e",
-			partPath: "2",
-			marker: { status: "skipped", reason: "too-large" },
-		});
-		assert.strictEqual(
-			await storage.extractedResultExists("cfg-e", "acc-e", "msg-e", "2"),
-			true,
-		);
-	});
-
-	test("listExtractedTexts returns only .txt.gz artifacts for the given message, excluding skip markers and other messages", async () => {
-		const { client } = createFakeS3Client();
-		const storage = createS3StorageService(client, "bucket");
-
-		await storage.storeExtractedText({
-			accountConfigId: "cfg-l",
-			accountId: "acc-l",
-			messageId: "msg-l",
-			partPath: "1",
-			text: "one",
-		});
-		await storage.storeExtractedText({
-			accountConfigId: "cfg-l",
-			accountId: "acc-l",
-			messageId: "msg-l",
-			partPath: "2.1",
-			text: "two",
-		});
-		await storage.storeExtractedSkipped({
-			accountConfigId: "cfg-l",
-			accountId: "acc-l",
-			messageId: "msg-l",
-			partPath: "3",
-			marker: { status: "failed", reason: "doc: corrupt" },
-		});
-		await storage.storeExtractedText({
-			accountConfigId: "cfg-l",
-			accountId: "acc-l",
-			messageId: "other-msg",
-			partPath: "1",
-			text: "unrelated",
-		});
-
-		const items = await storage.listExtractedTexts("cfg-l", "acc-l", "msg-l");
-		assert.deepStrictEqual(items.map((i) => i.partPath).sort(), ["1", "2.1"]);
-		for (const item of items) {
-			assert.ok(item.key.endsWith(".txt.gz"));
-		}
-	});
 });
 
 describe("createS3StorageService: outbox attachments", () => {

package/src/backends/s3.ts

@@ -15,7 +15,6 @@ import { Upload } from "@aws-sdk/lib-storage";
 import { getSignedUrl } from "@aws-sdk/s3-request-presigner";
 import { ContentEncoding, StorageType } from "@remit/domain-enums";
 import type {
-	ExtractedTextListItem,
 	OutboxAttachmentListItem,
 	ParsedBody,
 	StorageReference,
@@ -25,9 +24,6 @@ import {
 	buildBodyPartKey,
 	buildDeduplicatedKey,
 	buildExportArchiveKey,
-	buildExtractedPrefix,
-	buildExtractedSkippedKey,
-	buildExtractedTextKey,
 	buildMessageBodyKey,
 	buildOutboxAttachmentKey,
 	buildOutboxAttachmentPrefix,
@@ -201,134 +197,6 @@ export const createS3StorageService = (
 		return response.ContentEncoding === "gzip" ? gunzipSync(buffer) : buffer;
 	};
 
-	const storeExtractedText: StorageService["storeExtractedText"] = (params) => {
-		const { accountConfigId, accountId, messageId, partPath, text } = params;
-		return storeInternal({
-			key: buildExtractedTextKey(
-				accountConfigId,
-				accountId,
-				messageId,
-				partPath,
-			),
-			content: Buffer.from(text, "utf8"),
-			contentType: "text/plain; charset=utf-8",
-		});
-	};
-
-	const storeExtractedSkipped: StorageService["storeExtractedSkipped"] = (
-		params,
-	) => {
-		const { accountConfigId, accountId, messageId, partPath, marker } = params;
-		return storeInternal({
-			key: buildExtractedSkippedKey(
-				accountConfigId,
-				accountId,
-				messageId,
-				partPath,
-			),
-			content: Buffer.from(JSON.stringify(marker), "utf8"),
-			contentType: "application/json",
-			compress: false,
-		});
-	};
-
-	const extractedResultExists: StorageService["extractedResultExists"] = async (
-		accountConfigId,
-		accountId,
-		messageId,
-		partPath,
-	) => {
-		const textKey = buildExtractedTextKey(
-			accountConfigId,
-			accountId,
-			messageId,
-			partPath,
-		);
-		const skippedKey = buildExtractedSkippedKey(
-			accountConfigId,
-			accountId,
-			messageId,
-			partPath,
-		);
-
-		const headExists = (key: string): Promise<boolean> =>
-			client
-				.send(new HeadObjectCommand({ Bucket: bucketName, Key: key }))
-				.then(() => true)
-				.catch(() => false);
-
-		const [textExists, skippedExists] = await Promise.all([
-			headExists(textKey),
-			headExists(skippedKey),
-		]);
-		return textExists || skippedExists;
-	};
-
-	const retrieveExtractedText: StorageService["retrieveExtractedText"] = async (
-		accountConfigId,
-		accountId,
-		messageId,
-		partPath,
-	) => {
-		const key = buildExtractedTextKey(
-			accountConfigId,
-			accountId,
-			messageId,
-			partPath,
-		);
-
-		const response = await client
-			.send(new GetObjectCommand({ Bucket: bucketName, Key: key }))
-			.catch((error: unknown) => {
-				if (isStorageNotFoundError(error)) return null;
-				throw error;
-			});
-
-		if (!response) return null;
-		if (!response.Body) {
-			throw new Error(`Empty response body for extracted text: ${key}`);
-		}
-
-		const body = await response.Body.transformToByteArray();
-		const buffer = Buffer.from(body);
-		const decoded =
-			response.ContentEncoding === "gzip" ? gunzipSync(buffer) : buffer;
-		return decoded.toString("utf8");
-	};
-
-	const listExtractedTexts: StorageService["listExtractedTexts"] = async (
-		accountConfigId,
-		accountId,
-		messageId,
-	) => {
-		const prefix = buildExtractedPrefix(accountConfigId, accountId, messageId);
-		const suffix = ".txt.gz";
-		const items: ExtractedTextListItem[] = [];
-		let continuationToken: string | undefined;
-
-		do {
-			const response = await client.send(
-				new ListObjectsV2Command({
-					Bucket: bucketName,
-					Prefix: prefix,
-					ContinuationToken: continuationToken,
-				}),
-			);
-			for (const object of response.Contents ?? []) {
-				if (!object.Key || !object.Key.endsWith(suffix)) continue;
-				items.push({
-					partPath: object.Key.slice(prefix.length, -suffix.length),
-					key: object.Key,
-				});
-			}
-			continuationToken = response.IsTruncated
-				? response.NextContinuationToken
-				: undefined;
-		} while (continuationToken);
-
-		return items;
-	};
-
 	const storeOutboxAttachment: StorageService["storeOutboxAttachment"] = (
 		params,
 	) => {
@@ -467,6 +335,28 @@ export const createS3StorageService = (
 		return { sizeBytes: head.ContentLength ?? 0 };
 	};
 
+	const retrieveOutboxAttachment: StorageService["retrieveOutboxAttachment"] =
+		async (accountConfigId, accountId, outboxMessageId, outboxAttachmentId) => {
+			const key = buildOutboxAttachmentKey(
+				accountConfigId,
+				accountId,
+				outboxMessageId,
+				outboxAttachmentId,
+			);
+			const response = await client
+				.send(new GetObjectCommand({ Bucket: bucketName, Key: key }))
+				.catch((error: unknown) => {
+					if (isStorageNotFoundError(error)) return null;
+					throw error;
+				});
+			if (!response) return null;
+			if (!response.Body) {
+				throw new Error(`Empty response body for outbox attachment: ${key}`);
+			}
+			const bytes = await response.Body.transformToByteArray();
+			return Buffer.from(bytes.buffer, bytes.byteOffset, bytes.byteLength);
+		};
+
 	// The browser PUTs straight to the bucket and these bytes never touch the
 	// API. `ContentLength` on the command hoists `content-length` into
 	// X-Amz-SignedHeaders, so S3 recomputes the signature over the length it was
@@ -718,15 +608,11 @@ export const createS3StorageService = (
 		deleteOutboxAttachments,
 		deleteOutboxAttachment,
 		statOutboxAttachment,
+		retrieveOutboxAttachment,
 		createOutboxAttachmentUploadUrl,
 		storeDeduplicated,
 		storeParsedBody,
 		retrieveParsedBody,
-		storeExtractedText,
-		storeExtractedSkipped,
-		extractedResultExists,
-		retrieveExtractedText,
-		listExtractedTexts,
 		retrieveMessageBody,
 		retrieveMessageBodyStream,
 		retrieve,

package/src/index.ts

@@ -20,9 +20,6 @@ export {
 	buildBodyPartKey,
 	buildDeduplicatedKey,
 	buildExportArchiveKey,
-	buildExtractedPrefix,
-	buildExtractedSkippedKey,
-	buildExtractedTextKey,
 	buildMessageBodyKey,
 	buildOutboxAttachmentKey,
 	buildOutboxAttachmentPrefix,
@@ -31,8 +28,6 @@ export {
 	type CreateOutboxAttachmentUploadUrlParams,
 	computeChecksum,
 	createMockStorageService,
-	type ExtractedTextListItem,
-	type ExtractedTextMarker,
 	isStorageNotFoundError,
 	type OutboxAttachmentListItem,
 	type OutboxAttachmentUploadTarget,
@@ -47,8 +42,6 @@ export {
 	type StorageTypeValue,
 	type StoreBodyPartParams,
 	type StoreDeduplicatedParams,
-	type StoreExtractedSkippedParams,
-	type StoreExtractedTextParams,
 	type StoreMessageBodyParams,
 	type StoreOutboxAttachmentParams,
 	type StoreParsedBodyParams,

package/src/outbox-attachment-retrieve.test.ts

@@ -0,0 +1,56 @@
+import assert from "node:assert/strict";
+import { mkdtemp, rm } from "node:fs/promises";
+import { tmpdir } from "node:os";
+import { join } from "node:path";
+import { describe, it } from "node:test";
+import { createFilesystemStorageService } from "./backends/filesystem.js";
+import { createMockStorageService, type StorageService } from "./storage.js";
+
+const OWNER = {
+	accountConfigId: "cfg-1",
+	accountId: "acc-1",
+	outboxMessageId: "ob-1",
+};
+
+const readsBackWhatWasStored = async (storage: StorageService) => {
+	const content = Buffer.from([0x25, 0x50, 0x44, 0x46, 0x00, 0xff]);
+	await storage.storeOutboxAttachment({
+		...OWNER,
+		outboxAttachmentId: "att-1",
+		content,
+	});
+
+	assert.deepEqual(
+		await storage.retrieveOutboxAttachment(
+			OWNER.accountConfigId,
+			OWNER.accountId,
+			OWNER.outboxMessageId,
+			"att-1",
+		),
+		content,
+	);
+	assert.equal(
+		await storage.retrieveOutboxAttachment(
+			OWNER.accountConfigId,
+			OWNER.accountId,
+			OWNER.outboxMessageId,
+			"att-missing",
+		),
+		null,
+	);
+};
+
+describe("retrieveOutboxAttachment", () => {
+	it("reads back the exact bytes on the filesystem backend, and null for none", async () => {
+		const basePath = await mkdtemp(join(tmpdir(), "remit-outbox-retrieve-"));
+		try {
+			await readsBackWhatWasStored(createFilesystemStorageService(basePath));
+		} finally {
+			await rm(basePath, { recursive: true, force: true });
+		}
+	});
+
+	it("reads back the exact bytes on the mock backend, and null for none", async () => {
+		await readsBackWhatWasStored(createMockStorageService());
+	});
+});

package/src/storage.test.ts

@@ -10,9 +10,6 @@ import { createFilesystemStorageService } from "./backends/filesystem.js";
 import {
 	buildBodyPartKey,
 	buildDeduplicatedKey,
-	buildExtractedPrefix,
-	buildExtractedSkippedKey,
-	buildExtractedTextKey,
 	buildMessageBodyKey,
 	buildOutboxAttachmentKey,
 	buildOutboxAttachmentPrefix,
@@ -120,38 +117,6 @@ describe("path builders", () => {
 		assert.strictEqual(segments[1], "CFG");
 		assert.strictEqual(segments[2], "ACC");
 	});
-
-	test("buildExtractedTextKey nests accountId under accountConfigId and suffixes .txt.gz", () => {
-		const key = buildExtractedTextKey("cfg1", "acc123", "msg456", "1.2");
-		assert.strictEqual(
-			key,
-			"accounts/cfg1/acc123/messages/msg456/extracted/1.2.txt.gz",
-		);
-	});
-
-	test("buildExtractedSkippedKey nests accountId under accountConfigId and suffixes .skipped.json", () => {
-		const key = buildExtractedSkippedKey("cfg1", "acc123", "msg456", "1.2");
-		assert.strictEqual(
-			key,
-			"accounts/cfg1/acc123/messages/msg456/extracted/1.2.skipped.json",
-		);
-	});
-
-	test("buildExtractedTextKey and buildExtractedSkippedKey share the same prefix", () => {
-		const prefix = buildExtractedPrefix("cfg1", "acc123", "msg456");
-		assert.strictEqual(
-			prefix,
-			"accounts/cfg1/acc123/messages/msg456/extracted/",
-		);
-		assert.ok(
-			buildExtractedTextKey("cfg1", "acc123", "msg456", "1").startsWith(prefix),
-		);
-		assert.ok(
-			buildExtractedSkippedKey("cfg1", "acc123", "msg456", "1").startsWith(
-				prefix,
-			),
-		);
-	});
 });
 
 describe("parseContentStorageKey", () => {
@@ -374,115 +339,6 @@ describe("createMockStorageService", () => {
 		);
 		assert.strictEqual(miss, null);
 	});
-
-	test("storeExtractedText / retrieveExtractedText round-trip", async () => {
-		const storage = createMockStorageService();
-
-		const ref = await storage.storeExtractedText({
-			accountConfigId: "cfg1",
-			accountId: "acc123",
-			messageId: "msg456",
-			partPath: "1.2",
-			text: "extracted attachment text",
-		});
-
-		assert.strictEqual(
-			ref.storageKey,
-			"accounts/cfg1/acc123/messages/msg456/extracted/1.2.txt.gz",
-		);
-
-		const retrieved = await storage.retrieveExtractedText(
-			"cfg1",
-			"acc123",
-			"msg456",
-			"1.2",
-		);
-		assert.strictEqual(retrieved, "extracted attachment text");
-	});
-
-	test("retrieveExtractedText returns null on miss", async () => {
-		const storage = createMockStorageService();
-		const result = await storage.retrieveExtractedText(
-			"cfg1",
-			"acc123",
-			"msg456",
-			"nope",
-		);
-		assert.strictEqual(result, null);
-	});
-
-	test("extractedResultExists is true after either the text artifact or the skip marker is stored", async () => {
-		const storage = createMockStorageService();
-
-		assert.strictEqual(
-			await storage.extractedResultExists("cfg1", "acc123", "msg456", "1"),
-			false,
-		);
-
-		await storage.storeExtractedText({
-			accountConfigId: "cfg1",
-			accountId: "acc123",
-			messageId: "msg456",
-			partPath: "1",
-			text: "hi",
-		});
-		assert.strictEqual(
-			await storage.extractedResultExists("cfg1", "acc123", "msg456", "1"),
-			true,
-		);
-
-		await storage.storeExtractedSkipped({
-			accountConfigId: "cfg1",
-			accountId: "acc123",
-			messageId: "msg456",
-			partPath: "2",
-			marker: { status: "skipped", reason: "type-not-allowed" },
-		});
-		assert.strictEqual(
-			await storage.extractedResultExists("cfg1", "acc123", "msg456", "2"),
-			true,
-		);
-	});
-
-	test("listExtractedTexts returns only .txt.gz artifacts, not skip markers", async () => {
-		const storage = createMockStorageService();
-
-		await storage.storeExtractedText({
-			accountConfigId: "cfg1",
-			accountId: "acc123",
-			messageId: "msg456",
-			partPath: "1",
-			text: "first",
-		});
-		await storage.storeExtractedText({
-			accountConfigId: "cfg1",
-			accountId: "acc123",
-			messageId: "msg456",
-			partPath: "2.1",
-			text: "second",
-		});
-		await storage.storeExtractedSkipped({
-			accountConfigId: "cfg1",
-			accountId: "acc123",
-			messageId: "msg456",
-			partPath: "3",
-			marker: { status: "failed", reason: "pdf: corrupt" },
-		});
-		// A different message must not leak into the list.
-		await storage.storeExtractedText({
-			accountConfigId: "cfg1",
-			accountId: "acc123",
-			messageId: "other-msg",
-			partPath: "1",
-			text: "unrelated",
-		});
-
-		const items = await storage.listExtractedTexts("cfg1", "acc123", "msg456");
-		assert.deepStrictEqual(items.map((i) => i.partPath).sort(), ["1", "2.1"]);
-		for (const item of items) {
-			assert.ok(item.key.endsWith(".txt.gz"));
-		}
-	});
 });
 
 describe("createFilesystemStorageService", () => {
@@ -670,102 +526,6 @@ describe("createFilesystemStorageService", () => {
 		assert.strictEqual(miss, null);
 	});
 
-	test("storeExtractedText / retrieveExtractedText round-trip with gzip", async () => {
-		const storage = createFilesystemStorageService(testBasePath);
-
-		const ref = await storage.storeExtractedText({
-			accountConfigId: "cfg1",
-			accountId: "acc123",
-			messageId: "msg-extracted-fs",
-			partPath: "1",
-			text: "fs extracted text",
-		});
-
-		assert.strictEqual(
-			ref.storageKey,
-			"accounts/cfg1/acc123/messages/msg-extracted-fs/extracted/1.txt.gz",
-		);
-		assert.strictEqual(ref.contentEncoding, ContentEncoding.Gzip);
-
-		const retrieved = await storage.retrieveExtractedText(
-			"cfg1",
-			"acc123",
-			"msg-extracted-fs",
-			"1",
-		);
-		assert.strictEqual(retrieved, "fs extracted text");
-	});
-
-	test("storeExtractedSkipped writes an uncompressed JSON marker that satisfies extractedResultExists", async () => {
-		const storage = createFilesystemStorageService(testBasePath);
-
-		assert.strictEqual(
-			await storage.extractedResultExists("cfg1", "acc123", "msg-skip-fs", "1"),
-			false,
-		);
-
-		const ref = await storage.storeExtractedSkipped({
-			accountConfigId: "cfg1",
-			accountId: "acc123",
-			messageId: "msg-skip-fs",
-			partPath: "1",
-			marker: { status: "skipped", reason: "too-large" },
-		});
-
-		assert.strictEqual(
-			ref.storageKey,
-			"accounts/cfg1/acc123/messages/msg-skip-fs/extracted/1.skipped.json",
-		);
-		assert.strictEqual(ref.contentEncoding, ContentEncoding.None);
-		assert.strictEqual(
-			await storage.extractedResultExists("cfg1", "acc123", "msg-skip-fs", "1"),
-			true,
-		);
-	});
-
-	test("listExtractedTexts lists only .txt.gz artifacts for the given message", async () => {
-		const storage = createFilesystemStorageService(testBasePath);
-
-		await storage.storeExtractedText({
-			accountConfigId: "cfg1",
-			accountId: "acc123",
-			messageId: "msg-list-fs",
-			partPath: "1",
-			text: "one",
-		});
-		await storage.storeExtractedText({
-			accountConfigId: "cfg1",
-			accountId: "acc123",
-			messageId: "msg-list-fs",
-			partPath: "2.1",
-			text: "two",
-		});
-		await storage.storeExtractedSkipped({
-			accountConfigId: "cfg1",
-			accountId: "acc123",
-			messageId: "msg-list-fs",
-			partPath: "3",
-			marker: { status: "failed", reason: "doc: corrupt" },
-		});
-
-		const items = await storage.listExtractedTexts(
-			"cfg1",
-			"acc123",
-			"msg-list-fs",
-		);
-		assert.deepStrictEqual(items.map((i) => i.partPath).sort(), ["1", "2.1"]);
-	});
-
-	test("listExtractedTexts returns an empty array when no extraction has happened yet", async () => {
-		const storage = createFilesystemStorageService(testBasePath);
-		const items = await storage.listExtractedTexts(
-			"cfg1",
-			"acc123",
-			"msg-never-scanned",
-		);
-		assert.deepStrictEqual(items, []);
-	});
-
 	test("storeOutboxAttachment writes the bytes uncompressed so the size on disk is the size the cap counts", async () => {
 		const storage = createFilesystemStorageService(testBasePath);
 		// Highly compressible: gzip would make the stored size disagree with the

package/src/storage.ts

@@ -128,41 +128,6 @@ export interface StoreParsedBodyParams {
 	parsed: ParsedBody;
 }
 
-/** Parameters for storing extracted attachment text (attachment-scan worker, #450) */
-export interface StoreExtractedTextParams {
-	accountConfigId: string;
-	accountId: string;
-	messageId: string;
-	partPath: string;
-	text: string;
-}
-
-/**
- * Marker written in place of extracted text when extraction is skipped or
- * fails, so the idempotency check terminates retries on a poison or
- * unsupported attachment instead of re-running extraction on every
- * redelivery.
- */
-export interface ExtractedTextMarker {
-	status: "skipped" | "failed";
-	reason: string;
-}
-
-/** Parameters for storing an extraction skip/failure marker */
-export interface StoreExtractedSkippedParams {
-	accountConfigId: string;
-	accountId: string;
-	messageId: string;
-	partPath: string;
-	marker: ExtractedTextMarker;
-}
-
-/** One extracted-text artifact found under a message's `extracted/` prefix */
-export interface ExtractedTextListItem {
-	partPath: string;
-	key: string;
-}
-
 export interface StorageService {
 	/** Store a message body (raw RFC822 content) */
 	storeMessageBody(params: StoreMessageBodyParams): Promise<StorageReference>;
@@ -266,6 +231,17 @@ export interface StorageService {
 		outboxAttachmentId: string,
 	): Promise<{ sizeBytes: number } | null>;
 
+	/**
+	 * The bytes of one attachment, for building the MIME part it goes out as.
+	 * Null when nothing is stored under the key.
+	 */
+	retrieveOutboxAttachment(
+		accountConfigId: string,
+		accountId: string,
+		outboxMessageId: string,
+		outboxAttachmentId: string,
+	): Promise<Buffer | null>;
+
 	/**
 	 * Mint somewhere to PUT one file. The backend decides what that is — block
 	 * storage direct, or this deployment's upload route — and the caller never
@@ -290,48 +266,6 @@ export interface StorageService {
 		messageId: string,
 	): Promise<ParsedBody | null>;
 
-	/** Store extracted attachment text (attachment-scan worker, #450) */
-	storeExtractedText(
-		params: StoreExtractedTextParams,
-	): Promise<StorageReference>;
-
-	/** Store a skip/failure marker in place of extracted text */
-	storeExtractedSkipped(
-		params: StoreExtractedSkippedParams,
-	): Promise<StorageReference>;
-
-	/**
-	 * Whether an extraction result — either the text artifact or the
-	 * skip/failure marker — already exists for this part. Makes extraction
-	 * idempotent: a redelivered SQS message short-circuits instead of
-	 * re-running extraction (#450).
-	 */
-	extractedResultExists(
-		accountConfigId: string,
-		accountId: string,
-		messageId: string,
-		partPath: string,
-	): Promise<boolean>;
-
-	/** Retrieve previously extracted text; returns null on NoSuchKey */
-	retrieveExtractedText(
-		accountConfigId: string,
-		accountId: string,
-		messageId: string,
-		partPath: string,
-	): Promise<string | null>;
-
-	/**
-	 * List every extracted-text artifact stored for a message — the
-	 * `.skipped.json` markers are excluded. Used by the search-index worker
-	 * (#452) to fold attachment text into a message's vectors.
-	 */
-	listExtractedTexts(
-		accountConfigId: string,
-		accountId: string,
-		messageId: string,
-	): Promise<ExtractedTextListItem[]>;
-
 	/** Retrieve a raw message body (RFC822 .eml) by account/message id; returns null on NoSuchKey */
 	retrieveMessageBody(
 		accountConfigId: string,
@@ -409,29 +343,6 @@ export const buildBodyPartKey = (
 ): string =>
 	`accounts/${accountConfigId}/${accountId}/messages/${messageId}/parts/${partPath}`;
 
-export const buildExtractedTextKey = (
-	accountConfigId: string,
-	accountId: string,
-	messageId: string,
-	partPath: string,
-): string =>
-	`${buildExtractedPrefix(accountConfigId, accountId, messageId)}${partPath}.txt.gz`;
-
-export const buildExtractedSkippedKey = (
-	accountConfigId: string,
-	accountId: string,
-	messageId: string,
-	partPath: string,
-): string =>
-	`${buildExtractedPrefix(accountConfigId, accountId, messageId)}${partPath}.skipped.json`;
-
-export const buildExtractedPrefix = (
-	accountConfigId: string,
-	accountId: string,
-	messageId: string,
-): string =>
-	`accounts/${accountConfigId}/${accountId}/messages/${messageId}/extracted/`;
-
 export const buildOutboxAttachmentPrefix = (
 	accountConfigId: string,
 	accountId: string,
@@ -650,6 +561,12 @@ export const createMockStorageService = (): StorageService => {
 		return content ? { sizeBytes: content.length } : null;
 	};
 
+	const retrieveOutboxAttachment: StorageService["retrieveOutboxAttachment"] =
+		async (accountConfigId, accountId, outboxMessageId, outboxAttachmentId) =>
+			storage.get(
+				`mock://${buildOutboxAttachmentKey(accountConfigId, accountId, outboxMessageId, outboxAttachmentId)}`,
+			) ?? null;
+
 	const createOutboxAttachmentUploadUrl: StorageService["createOutboxAttachmentUploadUrl"] =
 		async (params) => ({
 			uploadUrl: `mock://upload/${buildOutboxAttachmentKey(params.accountConfigId, params.accountId, params.outboxMessageId, params.outboxAttachmentId)}?max=${params.sizeBytes}&exp=${params.expiresAt}`,
@@ -704,68 +621,6 @@ export const createMockStorageService = (): StorageService => {
 		return storage.get(uri) ?? null;
 	};
 
-	const storeExtractedText: StorageService["storeExtractedText"] = async (
-		params,
-	) => {
-		const { accountConfigId, accountId, messageId, partPath, text } = params;
-		return storeInternal(
-			buildExtractedTextKey(accountConfigId, accountId, messageId, partPath),
-			Buffer.from(text, "utf8"),
-		);
-	};
-
-	const storeExtractedSkipped: StorageService["storeExtractedSkipped"] = async (
-		params,
-	) => {
-		const { accountConfigId, accountId, messageId, partPath, marker } = params;
-		return storeInternal(
-			buildExtractedSkippedKey(accountConfigId, accountId, messageId, partPath),
-			Buffer.from(JSON.stringify(marker), "utf8"),
-		);
-	};
-
-	const extractedResultExists: StorageService["extractedResultExists"] = async (
-		accountConfigId,
-		accountId,
-		messageId,
-		partPath,
-	) => {
-		const textUri = `mock://${buildExtractedTextKey(accountConfigId, accountId, messageId, partPath)}`;
-		const skippedUri = `mock://${buildExtractedSkippedKey(accountConfigId, accountId, messageId, partPath)}`;
-		return storage.has(textUri) || storage.has(skippedUri);
-	};
-
-	const retrieveExtractedText: StorageService["retrieveExtractedText"] = async (
-		accountConfigId,
-		accountId,
-		messageId,
-		partPath,
-	) => {
-		const uri = `mock://${buildExtractedTextKey(accountConfigId, accountId, messageId, partPath)}`;
-		const content = storage.get(uri);
-		if (!content) return null;
-		return content.toString("utf8");
-	};
-
-	const listExtractedTexts: StorageService["listExtractedTexts"] = async (
-		accountConfigId,
-		accountId,
-		messageId,
-	) => {
-		const prefix = `mock://${buildExtractedPrefix(accountConfigId, accountId, messageId)}`;
-		const suffix = ".txt.gz";
-		const items: ExtractedTextListItem[] = [];
-		for (const uri of storage.keys()) {
-			if (!uri.startsWith(prefix) || !uri.endsWith(suffix)) continue;
-			const key = uri.slice("mock://".length);
-			items.push({
-				partPath: uri.slice(prefix.length, -suffix.length),
-				key,
-			});
-		}
-		return items;
-	};
-
 	const retrieveMessageBody: StorageService["retrieveMessageBody"] = async (
 		accountConfigId,
 		accountId,
@@ -805,15 +660,11 @@ export const createMockStorageService = (): StorageService => {
 		deleteOutboxAttachments,
 		deleteOutboxAttachment,
 		statOutboxAttachment,
+		retrieveOutboxAttachment,
 		createOutboxAttachmentUploadUrl,
 		storeDeduplicated,
 		storeParsedBody,
 		retrieveParsedBody,
-		storeExtractedText,
-		storeExtractedSkipped,
-		extractedResultExists,
-		retrieveExtractedText,
-		listExtractedTexts,
 		retrieveMessageBody,
 		retrieveMessageBodyStream,
 		retrieve: async (uri) => {