@bymax-one/nest-ai-tokens 1.0.0

package/CHANGELOG.md

@@ -0,0 +1,113 @@
+# Changelog
+
+All notable changes to `@bymax-one/nest-ai-tokens` will be documented in this file.
+
+The format follows [Keep a Changelog](https://keepachangelog.com/en/1.0.0/) and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
+
+---
+
+## [Unreleased]
+
+## [1.0.0] - 2026-08-03
+
+First published release. Everything below ships in it.
+
+The `Fixed` and `Security` entries record defects found and corrected before
+publication, not regressions any consumer saw — there is no earlier release to
+have regressed from. They are kept because the reasoning is worth having.
+
+### Added
+
+- **Nine provider normalizers** — OpenAI Chat Completions, OpenAI Responses API, OpenAI-compatible (DeepSeek, xAI, Groq, Azure OpenAI), Anthropic Messages, Google Gemini / Vertex AI, AWS Bedrock Converse, Mistral, OpenRouter, Vercel AI SDK v5/v6. All are pure functions over plain objects; no provider SDK peer dependency.
+- **Versioned effective-dated pricing** — `PricingService` with a six-step model-resolution chain, in-memory TTL cache, `upsertPrice()`, and idempotent `MODEL_PRICES_SEED` snapshot seed. Rates are bigint nano-USD per million tokens; past records are never re-rated.
+- **Immutable append-only ledger** — `LedgerService` with content-derived idempotency (`deriveIdempotencyKey`), exactly-once semantics (payload-hash replay detection), lifecycle transitions (pending → posted → reversed/released), filtered queries, and optional per-tenant tamper-evident hash chain (`verifyChain()`).
+- **First-class markup / margin** — `MarkupResolver` with a fixed multiplier or a per-call `IMarkupPolicy`. Every record stores both `rawCostNanoUsd` (provider cost) and `billedCostNanoUsd` (after markup). The resale lever is a first-class configuration option, not application code.
+- **Prepaid wallets** — `WalletService` with grant/debit/refund/adjust entries, configurable burn order (expiry/priority/FIFO), overdraft support, and debit-allocation tracking.
+- **Multi-dimension budgets** — `BudgetService` with per-scope, per-feature, per-window (daily/weekly/monthly/total) caps on spend, token count, and operation count. Hard block (`policy: 'block'`) or soft alert with threshold events. Renewal-anchored windows via `anchorAt`.
+- **Optional Redis budget counter** — `RedisBudgetCounterStore` with a single atomic Lua `incrIfBelow` script for sub-ms cross-replica enforcement.
+- **Full hold → capture lifecycle** — `MeteringService.hold()` / `capture()` / `release()` / `meter()` for pre-flight spend reservation. `capture()` is idempotent.
+- **Streaming capture** — `StreamUsageCollector` accumulates SSE chunks, prefers provider-final usage, falls back to tokenizer on abort.
+- **NestJS enforcement** — `BudgetGuard` (CanActivate), `MeteringInterceptor` (NestInterceptor), and the `@Meter`, `@RequireBudget`, `@AiFeature` decorators for declarative controller-level metering.
+- **Usage reporting** — `UsageReportService` with SQL-aggregated `summarize()`, CSV/JSON `export()`, paginated history, per-model analytics, and currency conversion.
+- **Events** — Typed domain events (`ai_tokens.*`) delivered to `@nestjs/event-emitter` (optional peer) and/or an `IEventSink` port.
+- **OpenTelemetry** — Optional `ITelemetrySink` / `OtelTelemetrySink` via `@opentelemetry/api` (optional peer).
+- **PostgreSQL adapter** — `PrismaAiTokensStore` (`./prisma` subpath) implementing all four storage ports via parameterized raw SQL. Seven-table schema with migrations.
+- **Five subpaths** — `.` (server), `./shared` (zero-dep), `./prices` (data-only), `./prisma` (adapter), `./redis` (counter).
+- **Zero runtime dependencies** — `"dependencies": {}`. All runtime functionality via peer dependencies.
+- **100% unit test coverage** — 804 tests; 10-scenario Testcontainers e2e suite; Stryker mutation gate at 100.00% (0 surviving mutants, break 95).
+
+
+- **Prisma 7 support, alongside Prisma 6.** The declared peer range has always
+  been `>=6.0.0`, but nothing verified the claim; the end-to-end suite now runs
+  against Prisma 7 and a real PostgreSQL, and it is what surfaced the conflict
+  defect above. The library's own code needed no change — it reaches PostgreSQL
+  exclusively through parameterized raw SQL, and every `Prisma.*` helper it uses
+  is unchanged across the two majors.
+
+  What changed is the development schema and the test harness, mirroring what a
+  host application has to do: `url` left the `datasource` block for a new
+  `prisma.config.ts`, and clients are opened through `@prisma/adapter-pg` rather
+  than the removed `datasourceUrl`. Both setups are documented in the README.
+
+- **`pnpm check:exports`** runs `attw --pack . --profile strict` against the packed
+  tarball, which is what surfaced both resolution defects above.
+- **`pnpm check:runtime`** packs the tarball, lays it out the way npm would, and
+  asserts in ESM *and* CommonJS that an error thrown by `PrismaAiTokensStore`
+  satisfies `instanceof AiTokensException` against the class the root exports. No
+  source-based suite can observe this: the unit tests map the subpath specifiers to
+  `src` and therefore see a single copy. Both gates run in CI.
+
+### Fixed
+
+- **Idempotency conflicts were reported as generic store errors under Prisma 7.**
+  A raw-query unique violation carries the native SQLSTATE, and where it carries
+  it depends on how the client reaches the database. Prisma 6 puts it flat on
+  `meta.code`; a Prisma 7 driver adapter nests the driver's own error under
+  `meta.driverAdapterError.cause`. `isUniqueViolation` read only the first, so on
+  Prisma 7 the exactly-once replay-or-conflict path (§15.2) degraded a 409
+  `AI_TOKENS_IDEMPOTENCY_CONFLICT` into a 500-class `AI_TOKENS_STORE_ERROR` —
+  silently, since nothing threw. Both shapes are now read.
+
+- **Errors thrown by `PrismaAiTokensStore` were not recognised as
+  `AiTokensException`.** The store lives in the `./prisma` entry point and reached
+  the class through a relative path into `../server`, so the separate bundle got
+  its own copy. The copy carries the same name and the same shape, so nothing
+  crashed — but the four `instanceof AiTokensException` guards in
+  `bymax-ai-tokens.module.ts`, `hold-support.ts`, `wallet.service.ts` and
+  `metering.service.ts` all stopped matching, silently reclassifying store errors
+  such as `AI_TOKENS_INSUFFICIENT_CREDITS` and `AI_TOKENS_IDEMPOTENCY_CONFLICT` as
+  unexpected failures.
+
+  `./prisma` now imports the class by package specifier, which the bundler already
+  kept external for the lazily-loaded `./prices` subpath, so one identity is shared
+  in CommonJS as well as ESM. `isLedgerIdempotencyConflict` was unaffected: it is
+  deliberately duck-typed rather than `instanceof`-based.
+
+- **CommonJS consumers resolved ESM type declarations** on all five subpaths. The
+  `exports` map declared a single `types` condition, so `require()` landed on
+  `.d.ts` instead of `.d.cts`. Types are now declared per condition.
+
+- **`node10` type resolution failed outright**: the manifest carried no `main`,
+  `module` or `types`, and no `typesVersions`. All four are now present.
+
+### Security
+
+- **Peer floors raised to exclude known-vulnerable NestJS versions.** The declared
+  ranges were `@nestjs/common ^11.0.0` and `@nestjs/core ^11.0.0`, and both
+  admitted versions carrying published advisories:
+
+  | Peer             | Advisory                                                                                                                                    | Vulnerable                    | New floor  |
+  | ---------------- | ------------------------------------------------------------------------------------------------------------------------------------------- | ----------------------------- | ---------- |
+  | `@nestjs/common` | [GHSA-cj7v-w2c7-cp7c](https://github.com/advisories/GHSA-cj7v-w2c7-cp7c) — remote code execution via the `Content-Type` header              | `>= 11.0.0-next.1, < 11.0.16` | `^11.0.16` |
+  | `@nestjs/core`   | [GHSA-36xv-jgw5-4q75](https://github.com/advisories/GHSA-36xv-jgw5-4q75) — improper neutralization of special elements in downstream output | `<= 11.1.17`                  | `^11.1.18` |
+
+  A peer range is a statement about which versions this library supports. A floor
+  below a published advisory tells a consumer that a vulnerable install is a
+  supported one, and nothing in their tooling contradicts it — the install resolves
+  cleanly and silently. Corrected before the first publish, so no released version
+  ever carried the permissive range. No runtime behaviour changed.
+
+---
+
+[1.0.0]: https://github.com/bymaxone/nest-ai-tokens/releases/tag/v1.0.0
+[Unreleased]: https://github.com/bymaxone/nest-ai-tokens/compare/v1.0.0...HEAD

package/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 Bymax One
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.