feat(generator): добавлено состояние версии 2 для рестартов

- Зачем:
  - генератор должен переживать рестарт без потери популяции пользователей и коротко прерванных активных визитов.
- Что:
  - добавлен компактный state v2 для популяции, активных визитов и остатка бюджета рождений.
  - сервис генератора переведён на единый тиковый поток с сохранением и восстановлением состояния.
  - добавлена безопасная деградация для старого state v1 и битого state v2.
  - покрыты короткий и долгий простой, reset состояния и валидация вложенного state.
- Проверка:
  - uv run --with-requirements generator/requirements.txt pytest generator/tests -q.
  - git diff --check.
This commit is contained in:
Dmitry Dementiev
2026-06-11 18:01:57 +03:00
parent 8f1e997cc9
commit 640050e409
7 changed files with 964 additions and 35 deletions
+127 -3
View File
@@ -2,13 +2,16 @@
import logging
import random
from dataclasses import dataclass
from dataclasses import dataclass, field
from datetime import datetime
logger = logging.getLogger("generator")
STATE_VERSION = "2.0"
def _nested_list_to_tuple(obj):
"""Рекурсивно преобразует list в tuple для восстановления RNG state."""
if isinstance(obj, list):
@@ -16,6 +19,107 @@ def _nested_list_to_tuple(obj):
return obj
def _require_keys(item: dict, keys: tuple[str, ...], label: str) -> None:
missing = [key for key in keys if key not in item]
if missing:
raise ValueError(f"{label} missing fields: {', '.join(missing)}")
def _validate_v2_payload(data: dict) -> None:
population = data.get("population")
active_visits = data.get("active_visits")
pending_visit_births = data.get("pending_visit_births", 0.0)
if not isinstance(population, list):
raise ValueError("population must be a list")
if not population:
raise ValueError("population must be non-empty")
if not isinstance(active_visits, list):
raise ValueError("active_visits must be a list")
if not isinstance(pending_visit_births, int | float):
raise ValueError("pending_visit_births must be a number")
if not 0 <= pending_visit_births < 1_000_000:
raise ValueError("pending_visit_births is out of range")
users_by_id = {}
for index, user in enumerate(population):
if not isinstance(user, dict):
raise ValueError(f"population[{index}] must be an object")
_require_keys(user, ("user_domain_id", "seed_click_id"), f"population[{index}]")
user_domain_id = user["user_domain_id"]
seed_click_id = user["seed_click_id"]
active_click_id = user.get("active_click_id")
last_finished_at = user.get("last_finished_at")
if not isinstance(user_domain_id, str) or not user_domain_id:
raise ValueError(f"population[{index}].user_domain_id must be a string")
if not isinstance(seed_click_id, str) or not seed_click_id:
raise ValueError(f"population[{index}].seed_click_id must be a string")
if active_click_id is not None and not isinstance(active_click_id, str):
raise ValueError(f"population[{index}].active_click_id must be a string or null")
if last_finished_at is not None:
if not isinstance(last_finished_at, str):
raise ValueError(f"population[{index}].last_finished_at must be a string or null")
datetime.fromisoformat(last_finished_at)
if user_domain_id in users_by_id:
raise ValueError(f"duplicate population user_domain_id: {user_domain_id}")
users_by_id[user_domain_id] = user
active_visit_pairs = set()
for index, visit in enumerate(active_visits):
if not isinstance(visit, dict):
raise ValueError(f"active_visits[{index}] must be an object")
_require_keys(
visit,
(
"user_domain_id",
"click_id",
"next_index",
"started_at",
"offsets_us",
"page_url_paths",
),
f"active_visits[{index}]",
)
user_domain_id = visit["user_domain_id"]
click_id = visit["click_id"]
offsets = visit["offsets_us"]
page_url_paths = visit["page_url_paths"]
next_index = visit["next_index"]
if not isinstance(user_domain_id, str) or user_domain_id not in users_by_id:
raise ValueError(f"active_visits[{index}].user_domain_id is unknown")
if not isinstance(click_id, str) or not click_id:
raise ValueError(f"active_visits[{index}].click_id must be a string")
if not isinstance(offsets, list) or not offsets:
raise ValueError(f"active_visits[{index}].offsets_us must be a non-empty list")
if not all(isinstance(offset, int) and offset >= 0 for offset in offsets):
raise ValueError(f"active_visits[{index}].offsets_us must contain non-negative integers")
if offsets != sorted(offsets):
raise ValueError(f"active_visits[{index}].offsets_us must be sorted")
if not isinstance(page_url_paths, list) or len(page_url_paths) != len(offsets):
raise ValueError(
f"active_visits[{index}].page_url_paths must match offsets_us length"
)
if not all(isinstance(path, str) and path.startswith("/") for path in page_url_paths):
raise ValueError(f"active_visits[{index}].page_url_paths must contain paths")
if not isinstance(next_index, int) or not 0 <= next_index <= len(offsets):
raise ValueError(f"active_visits[{index}].next_index is out of range")
datetime.fromisoformat(visit["started_at"])
user_active_click_id = users_by_id[user_domain_id].get("active_click_id")
if user_active_click_id != click_id:
raise ValueError(
f"population active_click_id conflicts with active_visits[{index}]"
)
active_visit_pairs.add((user_domain_id, click_id))
for user_domain_id, user in users_by_id.items():
active_click_id = user.get("active_click_id")
if active_click_id and (user_domain_id, active_click_id) not in active_visit_pairs:
raise ValueError(
f"population user {user_domain_id} has active_click_id without active visit"
)
@dataclass
class GeneratorState:
"""Состояние генератора для восстановления после рестарта."""
@@ -24,7 +128,10 @@ class GeneratorState:
rng_state: tuple
last_batch_id: str
last_timestamp: datetime
version: str = "1.0"
version: str = STATE_VERSION
population: list[dict] = field(default_factory=list)
active_visits: list[dict] = field(default_factory=list)
pending_visit_births: float = 0.0
def to_dict(self) -> dict:
"""Конвертирует в словарь для JSON-сериализации."""
@@ -34,12 +141,26 @@ class GeneratorState:
"last_batch_id": self.last_batch_id,
"last_timestamp": self.last_timestamp.isoformat(),
"version": self.version,
"population": self.population,
"active_visits": self.active_visits,
"pending_visit_births": self.pending_visit_births,
}
@classmethod
def from_dict(cls, data: dict) -> "GeneratorState":
"""Создаёт состояние из словаря."""
try:
if not isinstance(data, dict):
raise ValueError("state must be an object")
version = data.get("version", "1.0")
if version != STATE_VERSION:
logger.warning(
"Unsupported generator state version %s, will start fresh",
version,
)
raise ValueError(f"unsupported state version: {version}")
_validate_v2_payload(data)
rng_state_raw = data.get("rng_state")
if not rng_state_raw:
logger.warning("State missing rng_state field")
@@ -65,7 +186,10 @@ class GeneratorState:
last_timestamp=datetime.fromisoformat(
data.get("last_timestamp", "1970-01-01T00:00:00+00:00")
),
version=data.get("version", "1.0"),
version=version,
population=data.get("population", []),
active_visits=data.get("active_visits", []),
pending_visit_births=data.get("pending_visit_births", 0.0),
)
except Exception as e:
logger.warning(f"Invalid state format, will start fresh: {e}")