@remit/mailbox-service 0.0.98 → 0.0.99

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@remit/mailbox-service",
3
- "version": "0.0.98",
3
+ "version": "0.0.99",
4
4
  "type": "module",
5
5
  "main": "src/index.ts",
6
6
  "types": "src/index.ts",
@@ -381,10 +381,8 @@ describe("Address.flags.category overrides classification at sync time (issue #2
381
381
  );
382
382
 
383
383
  assert.equal(harness.messageUpdates.length, 1);
384
- assert.equal(
385
- harness.messageUpdates[0].input.category,
386
- MessageCategory.marketing,
387
- );
384
+ // Re-store writes only bodyStorageKey — category untouched on row + thread.
385
+ assert.equal(harness.messageUpdates[0].input.category, undefined);
388
386
  assert.equal(harness.message.category, MessageCategory.marketing);
389
387
  assert.equal(harness.rows[0].category, MessageCategory.marketing);
390
388
  });
@@ -422,10 +420,8 @@ describe("Address.flags.category overrides classification at sync time (issue #2
422
420
 
423
421
  assert.deepEqual(result.syncedMessageIds, ["m-1"]);
424
422
  assert.equal(harness.messageUpdates.length, 1);
425
- assert.equal(
426
- harness.messageUpdates[0].input.category,
427
- MessageCategory.marketing,
428
- );
423
+ // Re-store writes only bodyStorageKey — category untouched on row + thread.
424
+ assert.equal(harness.messageUpdates[0].input.category, undefined);
429
425
  assert.equal(harness.message.category, MessageCategory.marketing);
430
426
  assert.equal(harness.rows[0].category, MessageCategory.marketing);
431
427
  });
@@ -2,15 +2,13 @@
2
2
  * RFC 034 Decision 3.1: `Message.category` is written once and never mutated
3
3
  * after — RFC 030's message-list GSI sort key depends on it never churning.
4
4
  *
5
- * `applyPostStoreSteps` computes a fresh classification on every pass and
6
- * folds it into the single Message update, with no check for whether the
7
- * message already carries a decided category. Two shipped paths re-enter it
8
- * on an already-classified message — the `NoSuchKey` fallback in
9
- * `fetchAndGetBody`, and `syncBodies(..., force: true)` — and both must leave
10
- * `category` untouched. Each test below feeds a re-entrant pass a body that
11
- * would classify differently from the message's existing category, so a
12
- * regression that drops the guard shows up even though header classification
13
- * is otherwise deterministic (issue #355).
5
+ * The re-entrant paths — `fetchAndGetBody`'s `NoSuchKey` fallback and
6
+ * `syncBodies(..., force: true)` — are now body-only re-stores (issue #1011):
7
+ * `applyPostStoreSteps(isReStore: true)` skips the entire decision pass and
8
+ * writes only `bodyStorageKey`, so the existing category on the row is never
9
+ * touched. Each test below feeds a re-store a body that would classify
10
+ * differently, so a regression that drops the isReStore skip shows up as the
11
+ * category field appearing in the update input (or changing on the row).
14
12
  */
15
13
 
16
14
  import assert from "node:assert/strict";
@@ -157,10 +155,11 @@ describe("Message.category survives a re-entrant classification pass", () => {
157
155
  );
158
156
 
159
157
  assert.equal(harness.messageUpdates.length, 1);
160
- assert.equal(
161
- harness.messageUpdates[0].input.category,
162
- MessageCategory.marketing,
163
- );
158
+ // Re-store writes only bodyStorageKey — category is untouched on the row.
159
+ assert.equal(harness.messageUpdates[0].input.category, undefined);
160
+ assert.deepEqual(Object.keys(harness.messageUpdates[0].input).sort(), [
161
+ "bodyStorageKey",
162
+ ]);
164
163
  assert.equal(harness.message.category, MessageCategory.marketing);
165
164
  });
166
165
 
@@ -196,10 +195,11 @@ describe("Message.category survives a re-entrant classification pass", () => {
196
195
 
197
196
  assert.deepEqual(result.syncedMessageIds, ["m-1"]);
198
197
  assert.equal(harness.messageUpdates.length, 1);
199
- assert.equal(
200
- harness.messageUpdates[0].input.category,
201
- MessageCategory.marketing,
202
- );
198
+ // Re-store writes only bodyStorageKey — category is untouched on the row.
199
+ assert.equal(harness.messageUpdates[0].input.category, undefined);
200
+ assert.deepEqual(Object.keys(harness.messageUpdates[0].input).sort(), [
201
+ "bodyStorageKey",
202
+ ]);
203
203
  assert.equal(harness.message.category, MessageCategory.marketing);
204
204
  });
205
205
 
@@ -1,20 +1,18 @@
1
1
  /**
2
- * RFC 039 Non-goals / issue #383: placement is meant to run once per message.
3
- * #378 (issue #355) guarded `Message.category` against the same two re-entrant
4
- * paths re-running `applyPostStoreSteps` on an already-processed message —
5
- * `fetchAndGetBody`'s `NoSuchKey` fallback and `syncBodies(..., force: true)`
6
- * — but did not guard `resolvePlacement`/`computePlacement`. Without a guard,
7
- * a message the provider originally junked and a user later rescued by hand
8
- * (never touched by Remit, so `movedByRemit` records nothing) gets
9
- * `classifyPlacement` re-evaluated against the same demote signals that
10
- * junked it in the first place, and can be silently moved right back.
2
+ * RFC 039 Non-goals / issues #383, #1011: placement is meant to run once per
3
+ * message. The two re-entrant paths — `fetchAndGetBody`'s `NoSuchKey` fallback
4
+ * and `syncBodies(..., force: true)` — are now marked as re-stores by
5
+ * `applyPostStoreSteps(isReStore: true)`, which skips the entire decision pass
6
+ * (placement, filters, classification) and writes only `bodyStorageKey`. This
7
+ * replaces the old per-field guards (`hasDecidedPlacement`,
8
+ * `hasDecidedCategory`, `hasClassifiedBody`) that each individual derivation
9
+ * carried.
11
10
  *
12
- * Each "already decided" test below presets `placementDecidedAt` on the
13
- * fixture (as a genuine first pass would have left it) and feeds the
14
- * re-entrant pass a body whose headers WOULD trigger a confident demote if
15
- * `classifyPlacement` ran fresh — so a regression that drops the guard shows
16
- * up as an unwanted move, not as a passing test relying on deterministic
17
- * heuristics happening to agree.
11
+ * The last test below previously asserted that a legacy row (movedByRemit: true,
12
+ * no placementDecidedAt) gets backfilled with placementDecidedAt on re-enter —
13
+ * that behavior is gone by design: re-stores never re-decide placement, so a
14
+ * pre-#1011 row that never got placementDecidedAt simply keeps not having it
15
+ * (and never gets moved by Remit either).
18
16
  */
19
17
 
20
18
  import assert from "node:assert/strict";
@@ -201,6 +199,11 @@ describe("placement survives a re-entrant computePlacement pass (issue #383)", (
201
199
  assert.equal(harness.message.mailboxId, MAILBOXES.inbox.mailboxId);
202
200
  assert.equal(harness.messageUpdates[0]?.input.movedByRemit, undefined);
203
201
  assert.equal(harness.messageUpdates[0]?.input.placementVerdict, undefined);
202
+ // The re-store wrote only bodyStorageKey — no placement fields.
203
+ assert.equal(
204
+ harness.messageUpdates[0]?.input.placementDecidedAt,
205
+ undefined,
206
+ );
204
207
  });
205
208
 
206
209
  it("keeps a user-rescued message in Inbox when syncBodies re-fetches with force", async () => {
@@ -279,16 +282,18 @@ describe("placement survives a re-entrant computePlacement pass (issue #383)", (
279
282
  );
280
283
  });
281
284
 
282
- it("continues to protect a message Remit itself already moved (movedByRemit: true), unchanged behavior", async () => {
285
+ it("a legacy row with movedByRemit: true and no placementDecidedAt is not re-decided on re-store", async () => {
283
286
  const harness = buildHarness(
284
287
  {
285
288
  messageId: "m-1",
286
289
  mailboxId: MAILBOXES.inbox.mailboxId,
287
290
  bodyStorageKey: "s3://bodies/m-1",
288
291
  movedByRemit: true,
289
- // No placementDecidedAt — a legacy row synced before issue #383's
290
- // guard existed. `classifyPlacement`'s own `movedByRemit` check
291
- // must still hold on its own.
292
+ // No placementDecidedAt and no category — a legacy row synced before
293
+ // the write-once guards existed. Under #1011 a re-store skips the
294
+ // decision pass entirely, so placement is never re-evaluated and the
295
+ // row's missing fields are never backfilled — the re-store writes only
296
+ // bodyStorageKey.
292
297
  },
293
298
  async () => {
294
299
  throw noSuchKeyError();
@@ -313,10 +318,20 @@ describe("placement survives a re-entrant computePlacement pass (issue #383)", (
313
318
  [],
314
319
  "movedByRemit must keep protecting a legacy row with no placementDecidedAt of its own",
315
320
  );
321
+ // The re-store wrote only bodyStorageKey — no placement fields were
322
+ // re-decided or backfilled.
316
323
  assert.equal(harness.message.mailboxId, MAILBOXES.inbox.mailboxId);
317
- assert.ok(
318
- typeof harness.messageUpdates[0]?.input.placementDecidedAt === "number",
319
- "the legacy row is backfilled with placementDecidedAt going forward, self-healing for future passes",
324
+ assert.equal(
325
+ harness.messageUpdates[0]?.input.placementDecidedAt,
326
+ undefined,
327
+ );
328
+ assert.equal(harness.messageUpdates[0]?.input.movedByRemit, undefined);
329
+ assert.equal(harness.messageUpdates[0]?.input.placementVerdict, undefined);
330
+ // The only field written is bodyStorageKey.
331
+ assert.equal(
332
+ Object.keys(harness.messageUpdates[0]?.input ?? {}).length,
333
+ 1,
334
+ "re-store must write only bodyStorageKey",
320
335
  );
321
336
  });
322
337
  });
@@ -0,0 +1,605 @@
1
+ /**
2
+ * Issue #1011: a forced body re-sync must not re-fire standing rules on an
3
+ * already-seen message.
4
+ *
5
+ * The two re-entrant paths — `fetchAndGetBody`'s `NoSuchKey` IMAP fallback, and
6
+ * `syncBodies(..., force: true)` (the read-miss re-arm cue) — both re-fetch a
7
+ * body whose row already carries `bodyStorageKey`. That is a body-only fetch:
8
+ * the decision pass (placement, filters, classification) already ran when the
9
+ * body first landed. Re-running it now would move mail the user has since filed
10
+ * by hand, re-fire the engagement counter, re-apply labels, and rewrite the
11
+ * write-once category.
12
+ *
13
+ * Each test below presets a message whose body was already stored and whose
14
+ * headers WOULD trigger a confident placement demote or a standing filter move
15
+ * if the decision pass ran fresh. A regression that drops the `isReStore` skip
16
+ * shows up as an unwanted move, not as a passing test relying on the heuristics
17
+ * happening to agree.
18
+ */
19
+
20
+ import assert from "node:assert/strict";
21
+ import { Readable } from "node:stream";
22
+ import { describe, it } from "node:test";
23
+ import type {
24
+ FilterItem,
25
+ IAddressRepository,
26
+ IEnvelopeRepository,
27
+ IFilterAnchorRepository,
28
+ IFilterRepository,
29
+ IMailboxSpecialUseRepository,
30
+ IMessageLabelRepository,
31
+ IMessageRepository,
32
+ IThreadMessageRepository,
33
+ MessageItem,
34
+ UpdateMessageInput,
35
+ } from "@remit/data-ports";
36
+ import { NotFoundError } from "@remit/data-ports/errors";
37
+ import {
38
+ FilterClauseField,
39
+ FilterMatchOperator,
40
+ FilterState,
41
+ MessageCategory,
42
+ } from "@remit/domain-enums";
43
+ import type { StorageService } from "@remit/storage-service";
44
+ import type { PlacementConfig } from "./body-sync.js";
45
+ import { BodySyncService } from "./body-sync.js";
46
+ import { NO_ACTION } from "./filters/match.js";
47
+ import type { FilterConfig } from "./filters/pipeline.js";
48
+ import type { PlacementMoveService } from "./placement-move.js";
49
+ import type { IImapConnection } from "./types.js";
50
+
51
+ const MAILBOXES = {
52
+ inbox: { mailboxId: "mb-inbox", fullPath: "INBOX" },
53
+ junk: { mailboxId: "mb-junk", fullPath: "Junk" },
54
+ travel: { mailboxId: "mb-travel", fullPath: "Travel" },
55
+ };
56
+
57
+ /**
58
+ * Demote headers: DKIM domain mismatches From, dmarc=fail, untrusted sender.
59
+ * `classifyPlacement`'s demote branch (inbox → junk, HIGH bar) confidently acts
60
+ * on this set when the message currently sits in Inbox.
61
+ */
62
+ const DEMOTE_EML = Buffer.from(
63
+ [
64
+ "From: Support <support@evil-mimic.example>",
65
+ "To: me@example.com",
66
+ "Subject: Verify your account",
67
+ "Authentication-Results: mx.example.com; dmarc=fail",
68
+ "DKIM-Signature: v=1; a=rsa-sha256; d=relay.example.net; s=sel; b=xxx",
69
+ "X-Spam-Status: No, score=0.1",
70
+ "Content-Type: text/plain",
71
+ "",
72
+ "body",
73
+ ].join("\r\n"),
74
+ );
75
+
76
+ /**
77
+ * A standing literal filter that matches subjects containing "invitation" and
78
+ * moves them to `destinationMailboxId`. Used to assert that a re-store does not
79
+ * re-fire the filter on an already-seen message.
80
+ */
81
+ const buildInvitationFilter = (destinationMailboxId: string): FilterItem =>
82
+ ({
83
+ filterId: "flt-invite",
84
+ accountConfigId: "cfg-1",
85
+ name: "Invitations",
86
+ scope: "Standing",
87
+ state: FilterState.Active,
88
+ hasAnchor: false,
89
+ ruleChangedAt: 1,
90
+ matchOperator: FilterMatchOperator.And,
91
+ literalClauses: [{ field: FilterClauseField.Subject, value: "invitation" }],
92
+ actionLabelId: NO_ACTION,
93
+ actionMailboxId: destinationMailboxId,
94
+ createdAt: 1,
95
+ updatedAt: 1,
96
+ }) as unknown as FilterItem;
97
+
98
+ const INVITATION_EML = Buffer.from(
99
+ [
100
+ "From: Someone <someone@example.com>",
101
+ "To: me@example.com",
102
+ "Subject: You have a new invitation to Travel",
103
+ "Content-Type: text/plain",
104
+ "",
105
+ "details",
106
+ ].join("\r\n"),
107
+ );
108
+
109
+ interface Harness {
110
+ service: BodySyncService;
111
+ message: MessageItem;
112
+ messageUpdates: Array<{ messageId: string; input: UpdateMessageInput }>;
113
+ moves: Array<{ messageId: string; destinationMailboxId: string }>;
114
+ }
115
+
116
+ const noSuchKeyError = () =>
117
+ Object.assign(new Error("missing"), { name: "NoSuchKey" });
118
+
119
+ /**
120
+ * Build a BodySyncService harness with both placement and filter configs wired,
121
+ * so we can assert that neither fires during a re-store.
122
+ */
123
+ const buildHarness = (
124
+ message: Partial<MessageItem> & Pick<MessageItem, "messageId">,
125
+ retrieve: () => Promise<Buffer>,
126
+ filter: FilterItem | null,
127
+ ): Harness => {
128
+ const messageUpdates: Array<{
129
+ messageId: string;
130
+ input: UpdateMessageInput;
131
+ }> = [];
132
+ const moves: Array<{ messageId: string; destinationMailboxId: string }> = [];
133
+
134
+ const messageRow = {
135
+ uid: 1,
136
+ mailboxId: MAILBOXES.inbox.mailboxId,
137
+ ...message,
138
+ } as unknown as MessageItem;
139
+
140
+ const messageService = {
141
+ get: async (messageId: string) =>
142
+ messageId === message.messageId ? messageRow : undefined,
143
+ update: async (messageId: string, input: UpdateMessageInput) => {
144
+ messageUpdates.push({ messageId, input });
145
+ Object.assign(messageRow, input);
146
+ },
147
+ } as unknown as IMessageRepository;
148
+
149
+ const threadMessageService = {
150
+ findAllByMessageId: async () => [
151
+ {
152
+ threadMessageId: "tm-1",
153
+ messageId: message.messageId,
154
+ mailboxId: messageRow.mailboxId,
155
+ sentDate: 1,
156
+ isRead: false,
157
+ isDeleted: false,
158
+ hasStars: false,
159
+ hasAttachment: false,
160
+ },
161
+ ],
162
+ update: async () => {},
163
+ } as unknown as IThreadMessageRepository;
164
+
165
+ const storageService = {
166
+ retrieve,
167
+ storeMessageBody: async () => ({ uri: `s3://bodies/${message.messageId}` }),
168
+ storeMessageBodyStream: async () => ({
169
+ uri: `s3://bodies/${message.messageId}`,
170
+ }),
171
+ storeParsedBody: async () => {},
172
+ listBodyParts: async () => [],
173
+ } as unknown as StorageService;
174
+
175
+ const addressService = {
176
+ getAddress: async () => {
177
+ throw new NotFoundError("Address not found");
178
+ },
179
+ incrementInboundCount: async () => {},
180
+ } as unknown as IAddressRepository;
181
+
182
+ const envelopeService = {
183
+ listBodyParts: async () => [],
184
+ } as unknown as IEnvelopeRepository;
185
+
186
+ const mailboxSpecialUseService = {
187
+ findJunkMailbox: async () => MAILBOXES.junk,
188
+ findArchiveMailbox: async () => null,
189
+ findInboxMailbox: async () => MAILBOXES.inbox,
190
+ } as unknown as IMailboxSpecialUseRepository;
191
+
192
+ const placementMoveService = {
193
+ moveMessage: async (
194
+ _accountConfigId: string,
195
+ messageId: string,
196
+ destinationMailboxId: string,
197
+ ) => {
198
+ moves.push({ messageId, destinationMailboxId });
199
+ messageRow.mailboxId = destinationMailboxId;
200
+ },
201
+ } as unknown as PlacementMoveService;
202
+
203
+ const placementConfig: PlacementConfig = {
204
+ mailboxSpecialUseService,
205
+ placementMoveService,
206
+ };
207
+
208
+ const filterConfig: FilterConfig | undefined = filter
209
+ ? {
210
+ filterService: {
211
+ listByAccountAndState: async () => [filter],
212
+ refreshExpiry: async (f: FilterItem) => f,
213
+ } as unknown as IFilterRepository,
214
+ filterAnchorService: {
215
+ get: async () => undefined,
216
+ } as unknown as IFilterAnchorRepository,
217
+ messageLabelService: {
218
+ apply: async () => {},
219
+ } as unknown as IMessageLabelRepository,
220
+ placementMoveService,
221
+ }
222
+ : undefined;
223
+
224
+ const service = new BodySyncService(
225
+ messageService,
226
+ storageService,
227
+ threadMessageService,
228
+ addressService,
229
+ envelopeService,
230
+ { info: () => {}, error: () => {}, debug: () => {}, warn: () => {} },
231
+ placementConfig,
232
+ filterConfig,
233
+ );
234
+
235
+ return { service, message: messageRow, messageUpdates, moves };
236
+ };
237
+
238
+ const noSuchKeyConn = (body: Buffer): IImapConnection =>
239
+ ({
240
+ openBox: async () => {},
241
+ fetchMessageBody: async () => body,
242
+ }) as unknown as IImapConnection;
243
+
244
+ const streamBody = (body: Buffer): Readable =>
245
+ Readable.from([body]) as unknown as Readable;
246
+
247
+ describe("issue #1011: forced re-store does not re-fire standing rules", () => {
248
+ it("syncBodies(force: true) does not re-evaluate placement on a stored message", async () => {
249
+ const harness = buildHarness(
250
+ {
251
+ messageId: "m-1",
252
+ mailboxId: MAILBOXES.inbox.mailboxId,
253
+ bodyStorageKey: "s3://bodies/m-1",
254
+ // Already classified and placed — category and placementDecidedAt
255
+ // set by a prior, genuine first-store pass.
256
+ category: MessageCategory.personal,
257
+ placementDecidedAt: 500,
258
+ },
259
+ async () => {
260
+ throw new Error("force path must not retrieve from storage");
261
+ },
262
+ null, // no filter config
263
+ );
264
+
265
+ const connection = {
266
+ openBox: async () => {},
267
+ async *fetchMessageBodies(uids: number[]) {
268
+ for (const uid of uids) {
269
+ yield { uid, source: streamBody(DEMOTE_EML) };
270
+ }
271
+ },
272
+ } as unknown as IImapConnection;
273
+
274
+ const result = await harness.service.syncBodies(
275
+ ["m-1"],
276
+ "acc-1",
277
+ "cfg-1",
278
+ "INBOX",
279
+ async () => connection,
280
+ true,
281
+ );
282
+
283
+ assert.deepEqual(result.syncedMessageIds, ["m-1"], "message is synced");
284
+
285
+ // No placement move — a hand-rescued message stays where the user put it.
286
+ assert.deepEqual(
287
+ harness.moves,
288
+ [],
289
+ "a forced re-store must not re-decide placement, even though headers would demote",
290
+ );
291
+ assert.equal(harness.message.mailboxId, MAILBOXES.inbox.mailboxId);
292
+
293
+ // The update writes only bodyStorageKey — no placement or category churn.
294
+ const update = harness.messageUpdates[0];
295
+ assert.equal(update?.input.movedByRemit, undefined);
296
+ assert.equal(update?.input.placementVerdict, undefined);
297
+ assert.equal(update?.input.placementDecidedAt, undefined);
298
+ assert.equal(update?.input.category, undefined);
299
+ assert.deepEqual(
300
+ Object.keys(update?.input ?? {}).sort(),
301
+ ["bodyStorageKey"],
302
+ "re-store must write only bodyStorageKey",
303
+ );
304
+ });
305
+
306
+ it("syncBodies(force: true) does not re-fire a standing filter on a stored message", async () => {
307
+ const harness = buildHarness(
308
+ {
309
+ messageId: "m-1",
310
+ mailboxId: MAILBOXES.travel.mailboxId,
311
+ bodyStorageKey: "s3://bodies/m-1",
312
+ category: MessageCategory.personal,
313
+ },
314
+ async () => {
315
+ throw new Error("force path must not retrieve from storage");
316
+ },
317
+ buildInvitationFilter(MAILBOXES.junk.mailboxId),
318
+ );
319
+
320
+ const connection = {
321
+ openBox: async () => {},
322
+ async *fetchMessageBodies(uids: number[]) {
323
+ for (const uid of uids) {
324
+ yield { uid, source: streamBody(INVITATION_EML) };
325
+ }
326
+ },
327
+ } as unknown as IImapConnection;
328
+
329
+ const result = await harness.service.syncBodies(
330
+ ["m-1"],
331
+ "acc-1",
332
+ "cfg-1",
333
+ "INBOX",
334
+ async () => connection,
335
+ true,
336
+ );
337
+
338
+ assert.deepEqual(result.syncedMessageIds, ["m-1"]);
339
+
340
+ // No filter move — the message already sits in its filter destination
341
+ // (Travel) and a re-store must not re-fire the filter.
342
+ assert.deepEqual(
343
+ harness.moves,
344
+ [],
345
+ "a forced re-store must not re-fire standing filters",
346
+ );
347
+ // The update writes only bodyStorageKey — no filterMove marker.
348
+ const update = harness.messageUpdates[0];
349
+ assert.equal(update?.input.filterMove, undefined);
350
+ assert.equal(update?.input.movedByRemit, undefined);
351
+ assert.deepEqual(
352
+ Object.keys(update?.input ?? {}).sort(),
353
+ ["bodyStorageKey"],
354
+ "re-store must write only bodyStorageKey",
355
+ );
356
+ });
357
+
358
+ it("fetchAndGetBody NoSuchKey fallback does not re-fire placement or filters", async () => {
359
+ const harness = buildHarness(
360
+ {
361
+ messageId: "m-1",
362
+ mailboxId: MAILBOXES.inbox.mailboxId,
363
+ bodyStorageKey: "s3://bodies/m-1",
364
+ category: MessageCategory.personal,
365
+ placementDecidedAt: 500,
366
+ },
367
+ async () => {
368
+ throw noSuchKeyError();
369
+ },
370
+ buildInvitationFilter(MAILBOXES.junk.mailboxId),
371
+ );
372
+
373
+ await harness.service.fetchAndGetBody(
374
+ "m-1",
375
+ "acc-1",
376
+ "cfg-1",
377
+ "INBOX",
378
+ async () => noSuchKeyConn(DEMOTE_EML),
379
+ );
380
+
381
+ assert.deepEqual(
382
+ harness.moves,
383
+ [],
384
+ "the NoSuchKey fallback must not re-decide placement or re-fire filters",
385
+ );
386
+ assert.equal(harness.message.mailboxId, MAILBOXES.inbox.mailboxId);
387
+
388
+ const update = harness.messageUpdates[0];
389
+ assert.equal(update?.input.movedByRemit, undefined);
390
+ assert.equal(update?.input.placementVerdict, undefined);
391
+ assert.equal(update?.input.filterMove, undefined);
392
+ assert.deepEqual(
393
+ Object.keys(update?.input ?? {}).sort(),
394
+ ["bodyStorageKey"],
395
+ "re-store must write only bodyStorageKey",
396
+ );
397
+ });
398
+
399
+ it("a first-store via fetchAndGetBody (no bodyStorageKey) still runs placement and filters", async () => {
400
+ // Regression guard: the re-store skip must not disable decisions on a
401
+ // genuine first store that happens to arrive via the read path.
402
+ const harness = buildHarness(
403
+ {
404
+ messageId: "m-1",
405
+ mailboxId: MAILBOXES.inbox.mailboxId,
406
+ // No bodyStorageKey — a genuine first store.
407
+ },
408
+ async () => {
409
+ throw new Error("must not retrieve from storage on a first store");
410
+ },
411
+ buildInvitationFilter(MAILBOXES.travel.mailboxId),
412
+ );
413
+
414
+ await harness.service.fetchAndGetBody(
415
+ "m-1",
416
+ "acc-1",
417
+ "cfg-1",
418
+ "INBOX",
419
+ async () => noSuchKeyConn(INVITATION_EML),
420
+ );
421
+
422
+ // The filter fires — it is a first store, not a re-store.
423
+ assert.deepEqual(
424
+ harness.moves,
425
+ [{ messageId: "m-1", destinationMailboxId: MAILBOXES.travel.mailboxId }],
426
+ "a genuine first store must run the standing filter",
427
+ );
428
+ const update = harness.messageUpdates[0];
429
+ assert.equal(update?.input.filterMove?.filterId, "flt-invite");
430
+ assert.equal(update?.input.movedByRemit, true);
431
+ });
432
+
433
+ it("a forced batch re-stores a stored message (bytes only) but first-stores a never-stored one", async () => {
434
+ // A forced batch may contain both never-stored messages (first store —
435
+ // full decisions) and already-stored ones (re-store — bytes only).
436
+ const storedMessage = {
437
+ messageId: "m-stored",
438
+ mailboxId: MAILBOXES.inbox.mailboxId,
439
+ uid: 1,
440
+ bodyStorageKey: "s3://bodies/m-stored",
441
+ category: MessageCategory.personal,
442
+ } as unknown as MessageItem;
443
+ const newMessage = {
444
+ messageId: "m-new",
445
+ mailboxId: MAILBOXES.inbox.mailboxId,
446
+ uid: 2,
447
+ } as unknown as MessageItem;
448
+
449
+ const messageUpdates: Array<{
450
+ messageId: string;
451
+ input: UpdateMessageInput;
452
+ }> = [];
453
+ const moves: Array<{ messageId: string; destinationMailboxId: string }> =
454
+ [];
455
+
456
+ const messageService = {
457
+ get: async (messageId: string) =>
458
+ messageId === "m-stored"
459
+ ? storedMessage
460
+ : messageId === "m-new"
461
+ ? newMessage
462
+ : undefined,
463
+ update: async (id: string, input: UpdateMessageInput) => {
464
+ messageUpdates.push({ messageId: id, input });
465
+ const row = id === "m-stored" ? storedMessage : newMessage;
466
+ Object.assign(row, input);
467
+ },
468
+ } as unknown as IMessageRepository;
469
+
470
+ const threadMessageService = {
471
+ findAllByMessageId: async (
472
+ _accountConfigId: string,
473
+ messageId: string,
474
+ ) => [
475
+ {
476
+ threadMessageId: `tm-${messageId}`,
477
+ messageId,
478
+ mailboxId:
479
+ messageId === "m-stored"
480
+ ? storedMessage.mailboxId
481
+ : newMessage.mailboxId,
482
+ sentDate: 1,
483
+ isRead: false,
484
+ isDeleted: false,
485
+ hasStars: false,
486
+ hasAttachment: false,
487
+ },
488
+ ],
489
+ update: async () => {},
490
+ } as unknown as IThreadMessageRepository;
491
+
492
+ const storageService = {
493
+ retrieve: async () => {
494
+ throw new Error("must not retrieve on force path");
495
+ },
496
+ storeMessageBodyStream: async () => ({
497
+ uri: "s3://bodies/m",
498
+ }),
499
+ storeParsedBody: async () => {},
500
+ listBodyParts: async () => [],
501
+ } as unknown as StorageService;
502
+
503
+ const addressService = {
504
+ getAddress: async () => {
505
+ throw new NotFoundError("not found");
506
+ },
507
+ incrementInboundCount: async () => {},
508
+ } as unknown as IAddressRepository;
509
+
510
+ const envelopeService = {
511
+ listBodyParts: async () => [],
512
+ } as unknown as IEnvelopeRepository;
513
+
514
+ const mailboxSpecialUseService = {
515
+ findJunkMailbox: async () => MAILBOXES.junk,
516
+ findArchiveMailbox: async () => null,
517
+ findInboxMailbox: async () => MAILBOXES.inbox,
518
+ } as unknown as IMailboxSpecialUseRepository;
519
+
520
+ const placementMoveService = {
521
+ moveMessage: async (
522
+ _accountConfigId: string,
523
+ messageId: string,
524
+ destinationMailboxId: string,
525
+ ) => {
526
+ moves.push({ messageId, destinationMailboxId });
527
+ },
528
+ } as unknown as PlacementMoveService;
529
+
530
+ const filterConfig: FilterConfig = {
531
+ filterService: {
532
+ listByAccountAndState: async () => [
533
+ buildInvitationFilter(MAILBOXES.junk.mailboxId),
534
+ ],
535
+ refreshExpiry: async (f: FilterItem) => f,
536
+ } as unknown as IFilterRepository,
537
+ filterAnchorService: {
538
+ get: async () => undefined,
539
+ } as unknown as IFilterAnchorRepository,
540
+ messageLabelService: {
541
+ apply: async () => {},
542
+ } as unknown as IMessageLabelRepository,
543
+ placementMoveService,
544
+ };
545
+
546
+ const service = new BodySyncService(
547
+ messageService,
548
+ storageService,
549
+ threadMessageService,
550
+ addressService,
551
+ envelopeService,
552
+ { info: () => {}, error: () => {}, debug: () => {}, warn: () => {} },
553
+ { mailboxSpecialUseService, placementMoveService },
554
+ filterConfig,
555
+ );
556
+
557
+ // UID 1 is the stored message, UID 2 is new. fetchMessageBodies yields
558
+ // in order; the stored one (DEMOTE_EML) would demote if placement ran,
559
+ // the new one (INVITATION_EML) would trigger the filter.
560
+ const connection = {
561
+ openBox: async () => {},
562
+ async *fetchMessageBodies() {
563
+ yield { uid: 1, source: streamBody(DEMOTE_EML) };
564
+ yield { uid: 2, source: streamBody(INVITATION_EML) };
565
+ },
566
+ } as unknown as IImapConnection;
567
+
568
+ const result = await service.syncBodies(
569
+ ["m-stored", "m-new"],
570
+ "acc-1",
571
+ "cfg-1",
572
+ "INBOX",
573
+ async () => connection,
574
+ true,
575
+ );
576
+
577
+ // Both messages were handled.
578
+ assert.deepEqual(result.syncedMessageIds.sort(), ["m-new", "m-stored"]);
579
+
580
+ // The stored message is a re-store: no placement move, no filter run.
581
+ const storedUpdate = messageUpdates.find((u) => u.messageId === "m-stored");
582
+ assert.equal(storedUpdate?.input.movedByRemit, undefined);
583
+ assert.equal(storedUpdate?.input.filterMove, undefined);
584
+ assert.deepEqual(
585
+ Object.keys(storedUpdate?.input ?? {}).sort(),
586
+ ["bodyStorageKey"],
587
+ "stored message in a forced batch must re-store (bytes only)",
588
+ );
589
+
590
+ // The new message is a first-store: the filter fires.
591
+ assert.deepEqual(
592
+ moves,
593
+ [
594
+ {
595
+ messageId: "m-new",
596
+ destinationMailboxId: MAILBOXES.junk.mailboxId,
597
+ },
598
+ ],
599
+ "only the never-stored message in the forced batch takes the decision pass",
600
+ );
601
+ const newUpdate = messageUpdates.find((u) => u.messageId === "m-new");
602
+ assert.equal(newUpdate?.input.movedByRemit, true);
603
+ assert.equal(newUpdate?.input.filterMove?.filterId, "flt-invite");
604
+ });
605
+ });
package/src/body-sync.ts CHANGED
@@ -166,6 +166,7 @@ const extractSnippet = (parsed: ParsedMail): string =>
166
166
  SNIPPET_LENGTH,
167
167
  );
168
168
 
169
+ /**
169
170
  /**
170
171
  * RFC 034 Decision 3.1: `Message.category` is written once and never mutated
171
172
  * after — RFC 030's message-list GSI sort key depends on it never churning.
@@ -183,30 +184,20 @@ const hasDecidedCategory = (
183
184
  ): boolean =>
184
185
  category !== undefined && category !== MessageCategory.uncategorized;
185
186
 
186
- /**
187
- * Issue #383 (RFC 039 Non-goals): whether {@link BodySyncService.computePlacement}
188
- * has already produced a verdict for this message — moved, left in place, or
189
- * archived, confident or unsure alike. Absence means placement has genuinely
190
- * never been evaluated. The same two re-entrant paths `hasDecidedCategory`
191
- * guards (`fetchAndGetBody`'s `NoSuchKey` fallback, `syncBodies(..., force:
192
- * true)`) also re-enter `computePlacement`; without this guard a message a
193
- * user manually rescued (never touched by Remit, so `movedByRemit` never
194
- * recorded anything) can be silently re-evaluated and moved right back.
195
- */
196
- const hasDecidedPlacement = (placementDecidedAt: number | undefined): boolean =>
197
- placementDecidedAt !== undefined;
198
-
199
187
  /**
200
188
  * Issue #499: whether a completed body-sync pass has already stored this
201
189
  * message's body. `bodyStorageKey` is written last, once every derivation of
202
- * that pass has run, so its presence means that pass finished. The same two
203
- * re-entrant paths `hasDecidedCategory` and `hasDecidedPlacement` guard
204
- * (`fetchAndGetBody`'s `NoSuchKey` fallback, `syncBodies(..., force: true)`)
205
- * are the ones that reach the derivations again with it already set.
190
+ * that pass has run, so its presence means that pass finished.
206
191
  *
207
192
  * Key presence, and nothing more. What the classifier decided is
208
193
  * `Message.category`; whether it ran at all is `Message.classificationState`.
209
194
  * Neither is inferred from here — the field this reads is the storage fact.
195
+ *
196
+ * The re-fetch paths that re-enter with the key already set (`syncBodies(...,
197
+ * force: true)` and `fetchAndGetBody`'s `NoSuchKey` fallback) are re-stores
198
+ * that skip the derivations entirely (issue #1011), so this guard is what
199
+ * keeps a first store that lands on an already-stored row from re-deciding
200
+ * anything.
210
201
  */
211
202
  const hasStoredBody = (bodyStorageKey: string | undefined): boolean =>
212
203
  Boolean(bodyStorageKey);
@@ -217,9 +208,9 @@ const hasStoredBody = (bodyStorageKey: string | undefined): boolean =>
217
208
  * means "nothing confident to say":
218
209
  *
219
210
  * - `already-moved-by-remit` is {@link classifyPlacement}'s guard against
220
- * re-deciding a settled placement, and auto-archive is a re-decision. The
221
- * re-entrant paths `hasDecidedPlacement` names reach it on messages Remit
222
- * itself placed, or that a user moved back afterwards.
211
+ * re-deciding a settled placement, and auto-archive is a re-decision. It
212
+ * fires on messages Remit itself placed, or that a user moved back
213
+ * afterwards.
223
214
  * - A message already in Junk gets an unsure `leave` whatever the sender's
224
215
  * `blocked` flag says, since the demote branch only fires outside Junk.
225
216
  * Moving mail out of Junk is a rescue, which this file reserves for a
@@ -397,6 +388,12 @@ export class BodySyncService {
397
388
  // messageId so we can match FETCH rows back and re-enqueue any UID the
398
389
  // server never returns.
399
390
  const pending = new Map<number, string>();
391
+ // Message IDs in this batch whose row already carried `bodyStorageKey`
392
+ // when `force` re-added them to `pending`: a forced re-fetch of an
393
+ // already-stored body is a re-store (issue #1011) — bytes only, no
394
+ // decision re-run. A message with no stored body inside a forced batch
395
+ // is still a first store and takes the full decision pass.
396
+ const reStoredMessageIds = new Set<string>();
400
397
 
401
398
  // One read per round, not per message (issue #72). The list is small by
402
399
  // design and almost always empty, so a lookup per message would put a
@@ -438,6 +435,7 @@ export class BodySyncService {
438
435
  skippedCount++;
439
436
  continue;
440
437
  }
438
+ if (message.bodyStorageKey) reStoredMessageIds.add(messageId);
441
439
  pending.set(message.uid, messageId);
442
440
  }
443
441
 
@@ -468,6 +466,7 @@ export class BodySyncService {
468
466
  accountId,
469
467
  accountConfigId,
470
468
  source,
469
+ reStoredMessageIds.has(messageId),
471
470
  );
472
471
  } catch (error) {
473
472
  // A per-message store failure (e.g. the parsed-body S3 write) is a
@@ -644,12 +643,18 @@ export class BodySyncService {
644
643
  * cache, per-part objects). The S3 upload never sees a whole-body concat —
645
644
  * the storage service streams it — but mailparser still needs the full bytes,
646
645
  * so we collect them in parallel. Later issues move parsing off the hot path.
646
+ *
647
+ * `isReStore` marks a re-fetch of a message whose row already carried
648
+ * `bodyStorageKey` (the `force` re-arm in {@link syncBodies}): the
649
+ * post-store steps then re-materialize the body artifacts without
650
+ * re-running any decision (issue #1011).
647
651
  */
648
652
  private async storeStreamedBody(
649
653
  messageId: string,
650
654
  accountId: string,
651
655
  accountConfigId: string,
652
656
  source: Readable,
657
+ isReStore: boolean,
653
658
  ): Promise<void> {
654
659
  const toStorage = new PassThrough();
655
660
  const chunks: Buffer[] = [];
@@ -693,6 +698,7 @@ export class BodySyncService {
693
698
  {
694
699
  uri: ref.uri,
695
700
  },
701
+ isReStore,
696
702
  );
697
703
  }
698
704
 
@@ -724,6 +730,16 @@ export class BodySyncService {
724
730
  accountConfigId: string,
725
731
  body: Buffer,
726
732
  bodyRef: { uri: string },
733
+ /**
734
+ * Whether this call is a re-store of an already-classified message
735
+ * rather than a first store. The two re-entrant paths (`force: true`
736
+ * in `syncBodies`, and `fetchAndGetBody`'s `NoSuchKey` IMAP fallback) are
737
+ * both body-only re-fetches: a body re-sync fetches and stores bytes and
738
+ * runs no decision logic (issue #1011). When `isReStore` is true the
739
+ * placement/filter/category machinery is skipped entirely and the
740
+ * message's existing category is carried forward unchanged.
741
+ */
742
+ isReStore: boolean,
727
743
  ): Promise<ParsedMail> {
728
744
  // The one operation on this path that can fail because of how the message
729
745
  // is built. Its own try block lives in `parseMessageBody`, so this frame —
@@ -733,6 +749,43 @@ export class BodySyncService {
733
749
  // feed happens later, alongside the Message write (see below), once the
734
750
  // write-once category is decided.
735
751
  const parsed = await parseMessageBody(body);
752
+
753
+ // Body-only re-fetch (issue #1011): the decision pass already ran when
754
+ // this message's body first landed, so a re-store re-materializes only
755
+ // the body-derived artifacts — parsed cache, body parts, the storage
756
+ // key — and re-decides nothing: no classification, placement, filters,
757
+ // labels, read state, and no second engagement-counter increment.
758
+ // Skipping the decisions here is what lets the write-once fields stay
759
+ // safe without per-derivation re-entrancy guards.
760
+ if (isReStore) {
761
+ // Same ordering rule as the first-store path below: the parsed-body
762
+ // cache must be durable before `bodyStorageKey` — the skip-guard
763
+ // signal — is written, so a failure here requeues the message and
764
+ // the retry genuinely re-attempts the artifact writes.
765
+ await this.storeParsedBodyCache(
766
+ accountConfigId,
767
+ accountId,
768
+ messageId,
769
+ parsed,
770
+ );
771
+ if (!isBodyPartDeferralEnabled()) {
772
+ await this.storeBodyPartContents(
773
+ accountConfigId,
774
+ accountId,
775
+ messageId,
776
+ parsed,
777
+ );
778
+ }
779
+ await this.messageService.update(messageId, {
780
+ bodyStorageKey: bodyRef.uri,
781
+ });
782
+ this.log.info(
783
+ { messageId, storageKey: bodyRef.uri },
784
+ "Body re-stored without re-running decisions",
785
+ );
786
+ return parsed;
787
+ }
788
+
736
789
  const snippet = extractSnippet(parsed);
737
790
  const listId = extractListId(parsed);
738
791
 
@@ -892,20 +945,19 @@ export class BodySyncService {
892
945
  // once the parsed cache AND the move (when any) are.
893
946
  //
894
947
  // `category` is RFC 034 D3.1's write-once field (RFC 030's message-list
895
- // GSI sort key depends on it never churning). This step re-enters on an
896
- // already-classified message through two shipped paths — the `NoSuchKey`
897
- // fallback in `fetchAndGetBody` and `syncBodies(..., force: true)` — so a
898
- // real, previously-decided category is carried forward unchanged instead
899
- // of the just-recomputed one.
900
- // This also protects a `flags.category` override (issue #299): without
901
- // this guard a re-entrant pass would let a *later* override silently
902
- // rewrite a category already decided on an earlier message, which is
903
- // exactly the churn RFC 030's GSI-safety argument forbids.
948
+ // GSI sort key depends on it never churning). The re-entrant re-store
949
+ // paths (issue #1011) return before this write; the carry-forward keeps
950
+ // the field write-once even when a first store lands on a row that
951
+ // already carries a real category — a previously-decided category wins
952
+ // over the just-recomputed one. It likewise protects a `flags.category`
953
+ // override (issue #299): recomputing would let a *later* override silently
954
+ // rewrite a category already decided, which is exactly the churn RFC
955
+ // 030's GSI-safety argument forbids.
904
956
  //
905
- // `resolved.placementDecidedAt` (issue #383) guards the same two
906
- // re-entrant paths for placement: `computePlacement` already declined to
907
- // recompute a verdict once this field is set, so it is only ever present
908
- // here on the pass that first decided it.
957
+ // `resolved.placementDecidedAt` (issue #383) is only ever present on
958
+ // the pass that first decided it: the re-store paths skip placement
959
+ // entirely (issue #1011), so `computePlacement` is reached only on a
960
+ // first store.
909
961
  const finalCategory = hasDecidedCategory(existingMessage.category)
910
962
  ? existingMessage.category
911
963
  : classification.category;
@@ -1026,13 +1078,17 @@ export class BodySyncService {
1026
1078
 
1027
1079
  // Same shared step the sync path (storeStreamedBody) runs — issue
1028
1080
  // #1271 unified the two body paths so classification/placement no
1029
- // longer depends on which one fetched the body first.
1081
+ // longer depends on which one fetched the body first. A store from
1082
+ // here is a re-store exactly when the row already claimed a stored
1083
+ // body — the NoSuchKey fallback above; a row with no bodyStorageKey is
1084
+ // a genuine first store and takes the full decision pass.
1030
1085
  parsed = await this.applyPostStoreSteps(
1031
1086
  messageId,
1032
1087
  accountId,
1033
1088
  accountConfigId,
1034
1089
  body,
1035
1090
  { uri: ref.uri },
1091
+ message.bodyStorageKey != null,
1036
1092
  );
1037
1093
  } else {
1038
1094
  parsed = await simpleParser(body);
@@ -1492,14 +1548,12 @@ export class BodySyncService {
1492
1548
 
1493
1549
  const message = await this.messageService.get(messageId);
1494
1550
 
1495
- // Issue #383: placement is meant to run once per message (RFC 039
1496
- // Non-goals). Once a verdict has EVER been decided for this message —
1497
- // moved, left in place, or archived — a re-entrant pass (the `NoSuchKey`
1498
- // fallback in `fetchAndGetBody`, `syncBodies(..., force: true)`) must not
1499
- // recompute it: a message a user has since moved by hand (which never
1500
- // touches `movedByRemit`) would otherwise be silently re-evaluated
1501
- // against the same signals that placed it in the first place.
1502
- if (hasDecidedPlacement(message.placementDecidedAt)) return {};
1551
+ // Issue #383 / #1011: placement still runs once per message (RFC 039
1552
+ // Non-goals), but the once-per-message guarantee now comes from the
1553
+ // caller rather than a guard here — the re-store paths
1554
+ // (`fetchAndGetBody`'s `NoSuchKey` fallback, `syncBodies(..., force:
1555
+ // true)`) skip the decision steps entirely, so only the pass that first
1556
+ // stores a body ever reaches this method.
1503
1557
 
1504
1558
  const junkMailbox =
1505
1559
  await mailboxSpecialUseService.findJunkMailbox(accountId);