@anthusai/papyrus 5.19.1 → 5.20.0

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.
Files changed (43) hide show
  1. package/amplify/data/resource.js +1 -0
  2. package/amplify/data/resource.ts +1 -0
  3. package/amplify/data/schema.js +1 -0
  4. package/amplify/data/schema.ts +1 -0
  5. package/amplify/functions/assignment-action/data-access.js +2 -0
  6. package/amplify/functions/assignment-action/data-access.ts +2 -0
  7. package/amplify/functions/assignment-action/handler.ts +1 -1
  8. package/amplify/functions/content-actions/handler.py +3 -0
  9. package/amplify/functions/shared/editorial-context.ts +63 -4
  10. package/amplify/plugins/reader-steering/config.js +111 -0
  11. package/amplify/plugins/reader-steering/config.ts +158 -0
  12. package/amplify/plugins/reader-steering/function/handler.py +544 -0
  13. package/amplify/plugins/reader-steering/function/resource.js +27 -0
  14. package/amplify/plugins/reader-steering/function/resource.ts +35 -0
  15. package/amplify/plugins/reader-steering/papyrus.plugin.js +139 -0
  16. package/amplify/plugins/reader-steering/papyrus.plugin.ts +162 -0
  17. package/amplify/plugins/reader-steering/schema.js +40 -0
  18. package/amplify/plugins/reader-steering/schema.ts +42 -0
  19. package/amplify/plugins/registry.js +2 -0
  20. package/amplify/plugins/registry.ts +2 -0
  21. package/amplify/plugins/types.ts +3 -0
  22. package/amplify/plugins/wire.js +1 -0
  23. package/amplify/plugins/wire.ts +1 -0
  24. package/app/api/reader/handle/route.ts +9 -0
  25. package/app/api/reader/me/route.ts +9 -0
  26. package/app/api/reader/rating/route.ts +9 -0
  27. package/app/api/reader/retired/route.ts +9 -0
  28. package/app/api/reader/vote/route.ts +9 -0
  29. package/components/reference-cyclotron-priming.tsx +54 -15
  30. package/infra/amplify-app-shell.js +15 -1
  31. package/infra/amplify-app-shell.ts +17 -1
  32. package/infra/site-config.js +12 -2
  33. package/infra/site-config.ts +19 -2
  34. package/lib/editorial-checks.ts +6 -2
  35. package/lib/prospectus-session.ts +1 -0
  36. package/lib/prospectus-sign-in-routes.ts +2 -2
  37. package/lib/prospectus-site-identity.ts +24 -1
  38. package/lib/reader-api.ts +143 -0
  39. package/lib/reader-retirement-route.ts +89 -0
  40. package/lib/reader-subject.md +29 -0
  41. package/lib/reader-subject.ts +53 -0
  42. package/package.json +9 -1
  43. package/routes.manifest.json +56 -1
@@ -0,0 +1,544 @@
1
+ """reader-actions: the one function behind the reader API of the reader-steering plugin.
2
+
3
+ The CMS app's reader routes invoke it with ``{publication, siteSubject, isStaff, action, payload}`` after they
4
+ have checked the session, the Origin and the body size. It validates every field, rate limits per reader,
5
+ and writes the ReaderProfile, ReaderHandle and ReaderVote tables directly with its own role. It answers
6
+ ``{statusCode, body}``; the body never contains the siteSubject.
7
+
8
+ The siteSubject is the per-site pairwise id that Prospectus returns. The global person id never reaches this
9
+ function. Only ``ss1_`` subjects are accepted, plus ``ssdev_`` ones when the site was synthesized for
10
+ development or test (docs/reader-steering.md).
11
+ """
12
+
13
+ from __future__ import annotations
14
+
15
+ import json
16
+ import os
17
+ import re
18
+ import time
19
+ import uuid
20
+ from dataclasses import dataclass
21
+ from datetime import datetime, timezone
22
+ from decimal import Decimal
23
+ from typing import Any, Callable, Protocol
24
+
25
+ SITE_SUBJECT = re.compile(r"ss1_[A-Za-z0-9_-]{43}")
26
+ INTERIM_SUBJECT = re.compile(r"ssdev_[A-Za-z0-9_-]{43}")
27
+ SUBJECT_KEY = re.compile(r"[A-Za-z0-9._:-]{1,200}")
28
+ HANDLE = re.compile(r"[a-z0-9_-]{3,24}")
29
+ TOGGLE_DIMENSIONS = ("up", "used-it")
30
+ PUBLISHED_ID_PREFIX = "published-"
31
+ MAXIMUM_SUBJECTS = 50
32
+ MINIMUM_RATING = 1
33
+ MAXIMUM_RATING = 5
34
+ HOUR_SECONDS = 3600
35
+ HANDLE_CHANGES_PER_HOUR = 10
36
+ STATE_ACTIVE = "active"
37
+ STATE_LIMITED = "limited"
38
+ STATE_BANNED = "banned"
39
+ ROLE_READER = "reader"
40
+ ROLE_EDITOR = "editor"
41
+ ACTION_ME = "me"
42
+ ACTION_HANDLE = "handle"
43
+ ACTION_VOTE = "vote"
44
+ ACTION_RATING = "rating"
45
+ RATE_BUCKET_REACTIONS = "react"
46
+ RATE_BUCKET_HANDLE = "handle"
47
+ SNAPSHOT_TASK = {"projectionTask": {"task": "reader-snapshot"}}
48
+ SNAPSHOT_SCHEDULE_PREFIX = "papyrus-reader-snapshot-"
49
+ RESERVED_HANDLES = frozenset(
50
+ "admin administrator root system sysadmin moderator mod mods staff editor editors support help security "
51
+ "official anonymous anon guest null undefined none me you everyone nobody deleted reader readers "
52
+ "papyrus anthus prospectus api www mail abuse postmaster webmaster".split()
53
+ )
54
+
55
+
56
+ @dataclass(frozen=True)
57
+ class ReaderSettings:
58
+ dimensions: tuple[str, ...] = ()
59
+ actions_per_hour: int = 120
60
+ new_account_hours: int = 24
61
+ new_account_divisor: int = 3
62
+ reserved_handles: frozenset[str] = frozenset()
63
+ allow_interim_subjects: bool = False
64
+ snapshot_debounce_seconds: int = 60
65
+
66
+
67
+ def settings_from_environment(environment: dict[str, str] | None = None) -> ReaderSettings:
68
+ environment = os.environ if environment is None else environment
69
+ raw = json.loads(environment.get("PAPYRUS_READER_SETTINGS") or "{}")
70
+ limits = raw.get("limits") or {}
71
+ defaults = ReaderSettings()
72
+ return ReaderSettings(
73
+ dimensions=tuple(raw.get("dimensions") or ()),
74
+ actions_per_hour=int(limits.get("actionsPerHour", defaults.actions_per_hour)),
75
+ new_account_hours=int(limits.get("newAccountHours", defaults.new_account_hours)),
76
+ new_account_divisor=int(limits.get("newAccountDivisor", defaults.new_account_divisor)),
77
+ reserved_handles=frozenset(raw.get("reservedHandles") or ()),
78
+ allow_interim_subjects=bool(raw.get("allowInterimSubjects", False)),
79
+ snapshot_debounce_seconds=int(raw.get("snapshotDebounceSeconds", defaults.snapshot_debounce_seconds)),
80
+ )
81
+
82
+
83
+ class ReaderStore(Protocol):
84
+ def get_profile(self, site_subject: str) -> dict[str, Any] | None: ...
85
+
86
+ def create_profile(self, profile: dict[str, Any]) -> bool: ...
87
+
88
+ def claim_handle(self, site_subject: str, reader_id: str, handle: str, now_iso: str) -> str: ...
89
+
90
+ def get_vote_values(self, vote_ids: list[str]) -> dict[str, int]: ...
91
+
92
+ def put_vote(self, vote: dict[str, Any], now_iso: str) -> None: ...
93
+
94
+ def delete_vote(self, vote_id: str) -> None: ...
95
+
96
+ def is_published(self, subject_key: str) -> bool: ...
97
+
98
+ def count_action(self, reader_id: str, bucket: str, window_start: int, limit: int, expires_at: int) -> bool: ...
99
+
100
+ def claim_snapshot_slot(self, now_seconds: float, debounce_seconds: int) -> int | None: ...
101
+
102
+ def release_snapshot_slot(self, locked_until: int) -> None: ...
103
+
104
+
105
+ def reply(status: int, body: dict[str, Any]) -> dict[str, Any]:
106
+ return {"statusCode": status, "body": body}
107
+
108
+
109
+ def refusal(status: int, error: str, message: str, **extra: Any) -> dict[str, Any]:
110
+ return reply(status, {"error": error, "message": message, **extra})
111
+
112
+
113
+ def iso(now_seconds: float) -> str:
114
+ return datetime.fromtimestamp(now_seconds, tz=timezone.utc).strftime("%Y-%m-%dT%H:%M:%S.000Z")
115
+
116
+
117
+ def seconds_of(iso_text: str) -> float:
118
+ return datetime.strptime(iso_text, "%Y-%m-%dT%H:%M:%S.000Z").replace(tzinfo=timezone.utc).timestamp()
119
+
120
+
121
+ def vote_id(subject_key: str, reader_id: str, dimension: str) -> str:
122
+ return f"{subject_key}#{reader_id}#{dimension}"
123
+
124
+
125
+ def valid_site_subject(value: Any, settings: ReaderSettings) -> bool:
126
+ if not isinstance(value, str):
127
+ return False
128
+ if SITE_SUBJECT.fullmatch(value):
129
+ return True
130
+ return settings.allow_interim_subjects and INTERIM_SUBJECT.fullmatch(value) is not None
131
+
132
+
133
+ def action_limit(profile: dict[str, Any], settings: ReaderSettings, now_seconds: float) -> int:
134
+ age_hours = (now_seconds - seconds_of(profile["createdAt"])) / HOUR_SECONDS
135
+ if age_hours < settings.new_account_hours:
136
+ return max(1, settings.actions_per_hour // settings.new_account_divisor)
137
+ return settings.actions_per_hour
138
+
139
+
140
+ def within_rate_limit(
141
+ store: ReaderStore, profile: dict[str, Any], bucket: str, limit: int, now_seconds: float
142
+ ) -> dict[str, Any] | None:
143
+ window_start = int(now_seconds // HOUR_SECONDS) * HOUR_SECONDS
144
+ window_end = window_start + HOUR_SECONDS
145
+ if store.count_action(profile["readerId"], bucket, window_start, limit, window_end + HOUR_SECONDS):
146
+ return None
147
+ return refusal(429, "rate_limited", "Too many actions. Try again later.", retryAfter=max(1, int(window_end - now_seconds)))
148
+
149
+
150
+ def ensure_profile(store: ReaderStore, site_subject: str, now_seconds: float) -> dict[str, Any]:
151
+ profile = store.get_profile(site_subject)
152
+ if profile is not None:
153
+ return profile
154
+ created = {
155
+ "id": site_subject,
156
+ "siteSubject": site_subject,
157
+ "readerId": str(uuid.uuid4()),
158
+ "state": STATE_ACTIVE,
159
+ "karma": 0,
160
+ "createdAt": iso(now_seconds),
161
+ }
162
+ if store.create_profile(created):
163
+ return created
164
+ existing = store.get_profile(site_subject)
165
+ if existing is None:
166
+ raise RuntimeError("The reader profile could not be created.")
167
+ return existing
168
+
169
+
170
+ def public_profile(profile: dict[str, Any]) -> dict[str, Any]:
171
+ return {
172
+ "readerId": profile["readerId"],
173
+ "handle": profile.get("handle"),
174
+ "karma": int(profile.get("karma") or 0),
175
+ "state": profile.get("state", STATE_ACTIVE),
176
+ }
177
+
178
+
179
+ def my_state(store: ReaderStore, profile: dict[str, Any], payload: dict[str, Any], settings: ReaderSettings) -> dict[str, Any]:
180
+ subjects = payload.get("subjects", [])
181
+ if (
182
+ not isinstance(subjects, list)
183
+ or len(subjects) > MAXIMUM_SUBJECTS
184
+ or not all(isinstance(subject, str) and SUBJECT_KEY.fullmatch(subject) for subject in subjects)
185
+ ):
186
+ return refusal(422, "invalid_subjects", f"Send at most {MAXIMUM_SUBJECTS} item keys.")
187
+ dimensions = [*TOGGLE_DIMENSIONS, *settings.dimensions]
188
+ wanted = {vote_id(subject, profile["readerId"], dimension): (subject, dimension) for subject in subjects for dimension in dimensions}
189
+ found = store.get_vote_values(list(wanted))
190
+ votes: dict[str, dict[str, int]] = {}
191
+ for stored_id, value in found.items():
192
+ subject, dimension = wanted[stored_id]
193
+ votes.setdefault(subject, {})[dimension] = value
194
+ return reply(200, {**public_profile(profile), "votes": votes})
195
+
196
+
197
+ def set_handle(
198
+ store: ReaderStore, profile: dict[str, Any], payload: dict[str, Any], settings: ReaderSettings, now_seconds: float
199
+ ) -> dict[str, Any]:
200
+ requested = payload.get("handle")
201
+ if not isinstance(requested, str):
202
+ return refusal(422, "invalid_handle", "Send a handle.")
203
+ handle = requested.lower()
204
+ if not HANDLE.fullmatch(handle):
205
+ return refusal(422, "invalid_handle", "A handle has 3 to 24 letters, digits, hyphens or underscores.")
206
+ if handle in RESERVED_HANDLES or handle in settings.reserved_handles:
207
+ return refusal(422, "reserved_handle", "That handle is reserved.")
208
+ if profile.get("handle"):
209
+ return refusal(409, "handle_already_set", "The handle is set once and cannot be changed.")
210
+ limited = within_rate_limit(store, profile, RATE_BUCKET_HANDLE, HANDLE_CHANGES_PER_HOUR, now_seconds)
211
+ if limited:
212
+ return limited
213
+ outcome = store.claim_handle(profile["siteSubject"], profile["readerId"], handle, iso(now_seconds))
214
+ if outcome == "taken":
215
+ return refusal(409, "handle_taken", "That handle is taken.")
216
+ if outcome == "already_set":
217
+ return refusal(409, "handle_already_set", "The handle is set once and cannot be changed.")
218
+ return reply(200, {**public_profile({**profile, "handle": handle})})
219
+
220
+
221
+ def react(
222
+ store: ReaderStore,
223
+ profile: dict[str, Any],
224
+ *,
225
+ subject: Any,
226
+ dimension: Any,
227
+ allowed_dimensions: tuple[str, ...],
228
+ value: int | None,
229
+ is_staff: bool,
230
+ settings: ReaderSettings,
231
+ now_seconds: float,
232
+ schedule_snapshot: Callable[[], None],
233
+ ) -> dict[str, Any]:
234
+ if not isinstance(subject, str) or not SUBJECT_KEY.fullmatch(subject):
235
+ return refusal(422, "invalid_subject", "Send the key of a published item.")
236
+ if not isinstance(dimension, str) or dimension not in allowed_dimensions:
237
+ return refusal(422, "invalid_dimension", "That dimension is not available.")
238
+ if profile.get("state") == STATE_BANNED:
239
+ return refusal(403, "reader_banned", "This reader cannot take part.")
240
+ if profile.get("state") == STATE_LIMITED:
241
+ return refusal(403, "reader_limited", "This reader cannot vote or rate right now.")
242
+ limited = within_rate_limit(store, profile, RATE_BUCKET_REACTIONS, action_limit(profile, settings, now_seconds), now_seconds)
243
+ if limited:
244
+ return limited
245
+ if not store.is_published(subject):
246
+ return refusal(404, "unknown_subject", "That item is not published.")
247
+ stored_id = vote_id(subject, profile["readerId"], dimension)
248
+ if value is None:
249
+ store.delete_vote(stored_id)
250
+ else:
251
+ store.put_vote(
252
+ {
253
+ "id": stored_id,
254
+ "subjectKey": subject,
255
+ "readerId": profile["readerId"],
256
+ "dimension": dimension,
257
+ "value": value,
258
+ "voterRole": ROLE_EDITOR if is_staff else ROLE_READER,
259
+ },
260
+ iso(now_seconds),
261
+ )
262
+ schedule_snapshot()
263
+ return reply(200, {"subject": subject, "dimension": dimension, "value": value})
264
+
265
+
266
+ def vote_action(store, profile, payload, settings, is_staff, now_seconds, schedule_snapshot):
267
+ on = payload.get("on")
268
+ if not isinstance(on, bool):
269
+ return refusal(422, "invalid_vote", "Send on as true or false.")
270
+ return react(
271
+ store, profile, subject=payload.get("subject"), dimension=payload.get("dimension"), allowed_dimensions=TOGGLE_DIMENSIONS,
272
+ value=1 if on else None, is_staff=is_staff, settings=settings, now_seconds=now_seconds, schedule_snapshot=schedule_snapshot,
273
+ )
274
+
275
+
276
+ def rating_action(store, profile, payload, settings, is_staff, now_seconds, schedule_snapshot):
277
+ value = payload.get("value")
278
+ if value is not None and (isinstance(value, bool) or not isinstance(value, int) or not MINIMUM_RATING <= value <= MAXIMUM_RATING):
279
+ return refusal(422, "invalid_rating", f"Send a whole number from {MINIMUM_RATING} to {MAXIMUM_RATING}, or null.")
280
+ return react(
281
+ store, profile, subject=payload.get("subject"), dimension=payload.get("dimension"), allowed_dimensions=settings.dimensions,
282
+ value=value, is_staff=is_staff, settings=settings, now_seconds=now_seconds, schedule_snapshot=schedule_snapshot,
283
+ )
284
+
285
+
286
+ def handle_event(
287
+ event: dict[str, Any],
288
+ *,
289
+ store: ReaderStore,
290
+ settings: ReaderSettings,
291
+ now_seconds: float,
292
+ schedule_snapshot: Callable[[], None],
293
+ ) -> dict[str, Any]:
294
+ site_subject, action, payload = event.get("siteSubject"), event.get("action"), event.get("payload")
295
+ if not valid_site_subject(site_subject, settings):
296
+ return refusal(403, "reader_identity_invalid", "The reader identity is not valid for this site.")
297
+ if not isinstance(payload, dict) or not isinstance(event.get("isStaff"), bool):
298
+ return refusal(400, "malformed_request", "The request is malformed.")
299
+ profile = ensure_profile(store, site_subject, now_seconds)
300
+ if action == ACTION_ME:
301
+ return my_state(store, profile, payload, settings)
302
+ if profile.get("state") == STATE_BANNED:
303
+ return refusal(403, "reader_banned", "This reader cannot take part.")
304
+ if action == ACTION_HANDLE:
305
+ return set_handle(store, profile, payload, settings, now_seconds)
306
+ if action == ACTION_VOTE:
307
+ return vote_action(store, profile, payload, settings, event["isStaff"], now_seconds, schedule_snapshot)
308
+ if action == ACTION_RATING:
309
+ return rating_action(store, profile, payload, settings, event["isStaff"], now_seconds, schedule_snapshot)
310
+ return refusal(400, "unknown_action", "Unknown action.")
311
+
312
+
313
+ def plain(value: Any) -> Any:
314
+ if isinstance(value, Decimal):
315
+ return int(value) if value == value.to_integral_value() else float(value)
316
+ if isinstance(value, dict):
317
+ return {key: plain(item) for key, item in value.items()}
318
+ if isinstance(value, list):
319
+ return [plain(item) for item in value]
320
+ return value
321
+
322
+
323
+ def is_conditional_failure(error: Exception) -> bool:
324
+ return str(getattr(error, "response", {}).get("Error", {}).get("Code", "")) == "ConditionalCheckFailedException"
325
+
326
+
327
+ class DynamoReaderStore:
328
+ """The Amplify model tables written directly: items carry ``__typename``, ``createdAt`` and ``updatedAt`` so staff can list them."""
329
+
330
+ def __init__(self, *, resource: Any, profile_table: str, handle_table: str, vote_table: str, published_item_table: str, rate_table: str) -> None:
331
+ self._resource = resource
332
+ self._names = {"profile": profile_table, "handle": handle_table, "vote": vote_table, "published": published_item_table, "rate": rate_table}
333
+
334
+ def _table(self, key: str) -> Any:
335
+ return self._resource.Table(self._names[key])
336
+
337
+ def get_profile(self, site_subject: str) -> dict[str, Any] | None:
338
+ item = self._table("profile").get_item(Key={"id": site_subject}, ConsistentRead=True).get("Item")
339
+ return plain(item) if item else None
340
+
341
+ def create_profile(self, profile: dict[str, Any]) -> bool:
342
+ item = {**profile, "__typename": "ReaderProfile", "updatedAt": profile["createdAt"]}
343
+ try:
344
+ self._table("profile").put_item(Item=item, ConditionExpression="attribute_not_exists(id)")
345
+ except Exception as error:
346
+ if is_conditional_failure(error):
347
+ return False
348
+ raise
349
+ return True
350
+
351
+ def claim_handle(self, site_subject: str, reader_id: str, handle: str, now_iso: str) -> str:
352
+ client = self._resource.meta.client
353
+ try:
354
+ client.transact_write_items(
355
+ TransactItems=[
356
+ {
357
+ "Put": {
358
+ "TableName": self._names["handle"],
359
+ "Item": {"id": handle, "__typename": "ReaderHandle", "readerId": reader_id, "createdAt": now_iso, "updatedAt": now_iso},
360
+ "ConditionExpression": "attribute_not_exists(id)",
361
+ }
362
+ },
363
+ {
364
+ "Update": {
365
+ "TableName": self._names["profile"],
366
+ "Key": {"id": site_subject},
367
+ "UpdateExpression": "SET #h = :h, updatedAt = :now",
368
+ "ConditionExpression": "attribute_exists(id) AND attribute_not_exists(#h)",
369
+ "ExpressionAttributeNames": {"#h": "handle"},
370
+ "ExpressionAttributeValues": {":h": handle, ":now": now_iso},
371
+ }
372
+ },
373
+ ]
374
+ )
375
+ except Exception as error:
376
+ reasons = getattr(error, "response", {}).get("CancellationReasons") or []
377
+ codes = [str(reason.get("Code", "")) for reason in reasons]
378
+ if codes[:1] == ["ConditionalCheckFailed"]:
379
+ return "taken"
380
+ if codes[1:2] == ["ConditionalCheckFailed"]:
381
+ return "already_set"
382
+ raise
383
+ return "claimed"
384
+
385
+ def get_vote_values(self, vote_ids: list[str]) -> dict[str, int]:
386
+ found: dict[str, int] = {}
387
+ table_name = self._names["vote"]
388
+ for start in range(0, len(vote_ids), 100):
389
+ pending = {table_name: {"Keys": [{"id": stored_id} for stored_id in vote_ids[start : start + 100]], "ProjectionExpression": "id, #v", "ExpressionAttributeNames": {"#v": "value"}}}
390
+ while pending:
391
+ response = self._resource.batch_get_item(RequestItems=pending)
392
+ for item in response.get("Responses", {}).get(table_name, []):
393
+ found[item["id"]] = int(item["value"])
394
+ pending = response.get("UnprocessedKeys") or {}
395
+ return found
396
+
397
+ def put_vote(self, vote: dict[str, Any], now_iso: str) -> None:
398
+ self._table("vote").update_item(
399
+ Key={"id": vote["id"]},
400
+ UpdateExpression=(
401
+ "SET #t = :t, subjectKey = :s, readerId = :r, #d = :d, #v = :v, voterRole = :role, "
402
+ "createdAt = if_not_exists(createdAt, :now), updatedAt = :now"
403
+ ),
404
+ ExpressionAttributeNames={"#t": "__typename", "#d": "dimension", "#v": "value"},
405
+ ExpressionAttributeValues={
406
+ ":t": "ReaderVote",
407
+ ":s": vote["subjectKey"],
408
+ ":r": vote["readerId"],
409
+ ":d": vote["dimension"],
410
+ ":v": vote["value"],
411
+ ":role": vote["voterRole"],
412
+ ":now": now_iso,
413
+ },
414
+ )
415
+
416
+ def delete_vote(self, vote_id_: str) -> None:
417
+ self._table("vote").delete_item(Key={"id": vote_id_})
418
+
419
+ def is_published(self, subject_key: str) -> bool:
420
+ item = self._table("published").get_item(Key={"id": PUBLISHED_ID_PREFIX + subject_key}, ProjectionExpression="#s", ExpressionAttributeNames={"#s": "status"}).get("Item")
421
+ return bool(item) and item.get("status") == "published"
422
+
423
+ def count_action(self, reader_id: str, bucket: str, window_start: int, limit: int, expires_at: int) -> bool:
424
+ try:
425
+ self._table("rate").update_item(
426
+ Key={"pk": f"{reader_id}#{bucket}#{window_start}"},
427
+ UpdateExpression="ADD hits :one SET expiresAt = :expires",
428
+ ConditionExpression="attribute_not_exists(hits) OR hits < :limit",
429
+ ExpressionAttributeValues={":one": 1, ":expires": expires_at, ":limit": limit},
430
+ )
431
+ except Exception as error:
432
+ if is_conditional_failure(error):
433
+ return False
434
+ raise
435
+ return True
436
+
437
+ def claim_snapshot_slot(self, now_seconds: float, debounce_seconds: int) -> int | None:
438
+ """Opens a debounce window and returns when it closes, or ``None`` when a window is already open."""
439
+ locked_until = int(now_seconds) + debounce_seconds
440
+ try:
441
+ self._table("rate").put_item(
442
+ Item={"pk": "snapshot-debounce", "lockedUntil": locked_until, "expiresAt": locked_until + debounce_seconds + HOUR_SECONDS},
443
+ ConditionExpression="attribute_not_exists(pk) OR lockedUntil <= :now",
444
+ ExpressionAttributeValues={":now": int(now_seconds)},
445
+ )
446
+ except Exception as error:
447
+ if is_conditional_failure(error):
448
+ return None
449
+ raise
450
+ return locked_until
451
+
452
+ def release_snapshot_slot(self, locked_until: int) -> None:
453
+ """Reopens the window this call opened, so the next write can schedule the run that failed to schedule."""
454
+ try:
455
+ self._table("rate").delete_item(
456
+ Key={"pk": "snapshot-debounce"},
457
+ ConditionExpression="lockedUntil = :until",
458
+ ExpressionAttributeValues={":until": locked_until},
459
+ )
460
+ except Exception as error:
461
+ if not is_conditional_failure(error):
462
+ raise
463
+
464
+
465
+ def snapshot_schedule_name(locked_until: int) -> str:
466
+ return f"{SNAPSHOT_SCHEDULE_PREFIX}{locked_until}"
467
+
468
+
469
+ def snapshot_scheduler(
470
+ store: ReaderStore, settings: ReaderSettings, now_seconds: float, target: "SnapshotTarget | None", scheduler_client: Any
471
+ ) -> Callable[[], None]:
472
+ """Trailing-edge debounce: the first write in a window schedules one run at the end of the window.
473
+
474
+ Later writes in the window schedule nothing, because the run at the end of the window reads them all. A failure never
475
+ fails the vote; it reopens the window so the next write tries again.
476
+ """
477
+
478
+ def schedule() -> None:
479
+ if target is None:
480
+ return
481
+ locked_until = None
482
+ try:
483
+ locked_until = store.claim_snapshot_slot(now_seconds, settings.snapshot_debounce_seconds)
484
+ if locked_until is None:
485
+ return
486
+ scheduler_client.create_schedule(
487
+ Name=snapshot_schedule_name(locked_until),
488
+ ScheduleExpression="at(" + time.strftime("%Y-%m-%dT%H:%M:%S", time.gmtime(locked_until)) + ")",
489
+ ScheduleExpressionTimezone="UTC",
490
+ FlexibleTimeWindow={"Mode": "OFF"},
491
+ ActionAfterCompletion="DELETE",
492
+ Target={
493
+ "Arn": target.function_arn,
494
+ "RoleArn": target.role_arn,
495
+ "Input": json.dumps(SNAPSHOT_TASK),
496
+ "RetryPolicy": {"MaximumRetryAttempts": 0},
497
+ },
498
+ )
499
+ except Exception as error:
500
+ print(json.dumps({"readerSnapshotScheduleFailed": type(error).__name__}))
501
+ if locked_until is not None and type(error).__name__ != "ConflictException":
502
+ try:
503
+ store.release_snapshot_slot(locked_until)
504
+ except Exception as release_error:
505
+ print(json.dumps({"readerSnapshotReleaseFailed": type(release_error).__name__}))
506
+
507
+ return schedule
508
+
509
+
510
+ @dataclass(frozen=True)
511
+ class SnapshotTarget:
512
+ function_arn: str
513
+ role_arn: str
514
+
515
+
516
+ def snapshot_target_from_environment(environment: dict[str, str] | None = None) -> SnapshotTarget | None:
517
+ environment = os.environ if environment is None else environment
518
+ function_arn = environment.get("PAPYRUS_READER_SNAPSHOT_FUNCTION_ARN", "")
519
+ role_arn = environment.get("PAPYRUS_READER_SCHEDULER_ROLE_ARN", "")
520
+ return SnapshotTarget(function_arn, role_arn) if function_arn and role_arn else None
521
+
522
+
523
+ def store_from_environment(environment: dict[str, str] | None = None) -> DynamoReaderStore:
524
+ import boto3
525
+
526
+ environment = os.environ if environment is None else environment
527
+ return DynamoReaderStore(
528
+ resource=boto3.resource("dynamodb"),
529
+ profile_table=environment["PAPYRUS_READER_PROFILE_TABLE"],
530
+ handle_table=environment["PAPYRUS_READER_HANDLE_TABLE"],
531
+ vote_table=environment["PAPYRUS_READER_VOTE_TABLE"],
532
+ published_item_table=environment["PAPYRUS_READER_PUBLISHED_ITEM_TABLE"],
533
+ rate_table=environment["PAPYRUS_READER_RATE_TABLE"],
534
+ )
535
+
536
+
537
+ def handler(event: dict[str, Any], _context: Any = None) -> dict[str, Any]:
538
+ import boto3
539
+
540
+ settings = settings_from_environment()
541
+ store = store_from_environment()
542
+ now_seconds = time.time()
543
+ scheduler = snapshot_scheduler(store, settings, now_seconds, snapshot_target_from_environment(), boto3.client("scheduler"))
544
+ return handle_event(event, store=store, settings=settings, now_seconds=now_seconds, schedule_snapshot=scheduler)
@@ -0,0 +1,27 @@
1
+ import { defineFunction } from "@aws-amplify/backend";
2
+ import { Duration, RemovalPolicy } from "aws-cdk-lib";
3
+ import { AttributeType, BillingMode, Table } from "aws-cdk-lib/aws-dynamodb";
4
+ import { Effect, PolicyStatement } from "aws-cdk-lib/aws-iam";
5
+ import { Architecture, Code, Function, Runtime } from "aws-cdk-lib/aws-lambda";
6
+ import path from "node:path";
7
+ import { fileURLToPath } from "node:url";
8
+ const functionDirectory = path.dirname(fileURLToPath(import.meta.url));
9
+ export const readerActions = defineFunction((scope) => {
10
+ const rateTable = new Table(scope, "PapyrusReaderRateWindows", {
11
+ partitionKey: { name: "pk", type: AttributeType.STRING },
12
+ billingMode: BillingMode.PAY_PER_REQUEST,
13
+ timeToLiveAttribute: "expiresAt",
14
+ removalPolicy: RemovalPolicy.DESTROY,
15
+ });
16
+ const lambda = new Function(scope, "papyrus-reader-actions", {
17
+ runtime: Runtime.PYTHON_3_12,
18
+ architecture: Architecture.ARM_64,
19
+ handler: "handler.handler",
20
+ timeout: Duration.seconds(10),
21
+ memorySize: 256,
22
+ code: Code.fromAsset(functionDirectory, { exclude: ["*.ts", "__pycache__", "*.pyc"] }),
23
+ environment: { PAPYRUS_READER_RATE_TABLE: rateTable.tableName },
24
+ });
25
+ lambda.addToRolePolicy(new PolicyStatement({ effect: Effect.ALLOW, actions: ["dynamodb:UpdateItem", "dynamodb:PutItem"], resources: [rateTable.tableArn] }));
26
+ return lambda;
27
+ }, { resourceGroupName: "data" });
@@ -0,0 +1,35 @@
1
+ import { defineFunction } from "@aws-amplify/backend";
2
+ import { Duration, RemovalPolicy } from "aws-cdk-lib";
3
+ import { AttributeType, BillingMode, Table } from "aws-cdk-lib/aws-dynamodb";
4
+ import { Effect, PolicyStatement } from "aws-cdk-lib/aws-iam";
5
+ import { Architecture, Code, Function, Runtime } from "aws-cdk-lib/aws-lambda";
6
+ import type { Construct } from "constructs";
7
+ import path from "node:path";
8
+ import { fileURLToPath } from "node:url";
9
+
10
+ const functionDirectory = path.dirname(fileURLToPath(import.meta.url));
11
+
12
+ export const readerActions = defineFunction(
13
+ (scope: Construct) => {
14
+ const rateTable = new Table(scope, "PapyrusReaderRateWindows", {
15
+ partitionKey: { name: "pk", type: AttributeType.STRING },
16
+ billingMode: BillingMode.PAY_PER_REQUEST,
17
+ timeToLiveAttribute: "expiresAt",
18
+ removalPolicy: RemovalPolicy.DESTROY,
19
+ });
20
+ const lambda = new Function(scope, "papyrus-reader-actions", {
21
+ runtime: Runtime.PYTHON_3_12,
22
+ architecture: Architecture.ARM_64,
23
+ handler: "handler.handler",
24
+ timeout: Duration.seconds(10),
25
+ memorySize: 256,
26
+ code: Code.fromAsset(functionDirectory, { exclude: ["*.ts", "__pycache__", "*.pyc"] }),
27
+ environment: { PAPYRUS_READER_RATE_TABLE: rateTable.tableName },
28
+ });
29
+ lambda.addToRolePolicy(
30
+ new PolicyStatement({ effect: Effect.ALLOW, actions: ["dynamodb:UpdateItem", "dynamodb:PutItem"], resources: [rateTable.tableArn] }),
31
+ );
32
+ return lambda;
33
+ },
34
+ { resourceGroupName: "data" },
35
+ );