diff --git a/docs/api/endpoints.md b/docs/api/endpoints.md index 8e524c72..ccd7a749 100644 --- a/docs/api/endpoints.md +++ b/docs/api/endpoints.md @@ -200,6 +200,84 @@ POST /v1/embeddings } ``` +## System One (TypeSafe Decisions) + +### Evaluate State + +Evaluate a state against typed questions (noul / choice / score) on a TypeSafe +System One decision model (e.g. `jev-latest`). Requires a `typesafe` upstream +provider on the node. + +```http +POST /v1/systemone +``` + +**Request Body:** + +```json +{ + "model": "jev-latest", + "state": "Help! My payouts have been failing for 3 days.", + "questions": { + "is_urgent": { + "type": "noul", + "instructions": "Does this convey urgency?" + } + } +} +``` + +**Parameters:** + +| Parameter | Type | Required | Default | Description | +|-----------|------|----------|---------|-------------| +| `model` | string | Yes | - | TypeSafe alias (`jev-latest`, `jev-preview`) or versioned id (`jev-1.13.0`) | +| `state` | string/object/array | Yes | - | Content to evaluate | +| `questions` | map | Yes | - | Typed questions; answers keyed identically | + +**Response:** + +```json +{ + "model": "jev-latest", + "answers": { + "is_urgent": { + "type": "noul", + "noul": 0.95 + } + }, + "usage": { + "input_tokens": 312, + "output_tokens": 48 + } +} +``` + +Billing is input-token based (output tokens are free on Jev); the response's +`usage` is the settlement seam, exactly like embeddings. + +**Notes:** + +- The response `model` echoes the id you requested (e.g. `jev-latest`), not the + resolved build (`jev-1.13.0`) TypeSafe returns. Routstr also adds its standard + `id`, `cost`, `metadata.routstr` and `usage.*_msats` fields. +- TypeSafe's `GET /v1/models` lists aliases only; the node additionally seeds + the known versioned ids so they can be requested directly. +- TypeSafe answers `429 Too Many Requests` and `529 Overloaded` when throttled. + Both are forwarded as upstream errors; retry with exponential backoff. + +**Enabling the provider:** + +1. Admin UI → Providers → *TypeSafe* (base URL is fixed to + `https://api.typesafe.ai/v1`), paste your `api.typesafe.ai` key. Or + `POST /admin/api/upstream-providers` with + `{"provider_type": "typesafe", "api_key": ""}`. +2. On a node with an empty provider table, setting `TYPESAFE_API_KEY` seeds + the provider automatically. +3. `jev-latest`, `jev-preview` and `jev-1.13.0` are catalogued with the + published rate ($0.042 per million input tokens, output free). Override the + model row if TypeSafe changes pricing. + ## Images (Coming Soon) ### Create Image diff --git a/docs/api/overview.md b/docs/api/overview.md index c82e4e22..c8069953 100644 --- a/docs/api/overview.md +++ b/docs/api/overview.md @@ -104,6 +104,7 @@ Standard OpenAI-compatible endpoints: - **Responses**: `/v1/responses` - **Chat Completions**: `/v1/chat/completions` - **Embeddings**: `/v1/embeddings` +- **System One**: `/v1/systemone` (TypeSafe decision models) - **Completions**: `/v1/completions` *(planned)* - **Images**: `/v1/images/generations` *(planned)* - **Audio**: `/v1/audio/transcriptions` *(planned)* diff --git a/routstr/proxy.py b/routstr/proxy.py index 1f517779..d3ba5212 100644 --- a/routstr/proxy.py +++ b/routstr/proxy.py @@ -262,6 +262,10 @@ _ALLOWED_ENDPOINTS: dict[str, frozenset[str]] = { "responses": frozenset({"POST"}), "messages": frozenset({"POST"}), "embeddings": frozenset({"POST"}), + # TypeSafe System One decision endpoint: POST {state, model, questions} + # -> {answers, usage}. Non-streaming, JSON in/out; billed from the + # response's usage exactly like embeddings. + "systemone": frozenset({"POST"}), "models": frozenset({"GET"}), "attestation": frozenset({"GET"}), "tee/attestation": frozenset({"GET"}), diff --git a/routstr/upstream/__init__.py b/routstr/upstream/__init__.py index 35e49bcf..edac0020 100644 --- a/routstr/upstream/__init__.py +++ b/routstr/upstream/__init__.py @@ -12,6 +12,7 @@ from .perplexity import PerplexityUpstreamProvider from .ppqai import PPQAIUpstreamProvider from .routstr import RoutstrUpstreamProvider from .tinfoil import TinfoilUpstreamProvider +from .typesafe import TypeSafeUpstreamProvider from .xai import XAIUpstreamProvider upstream_provider_classes: list[type[BaseUpstreamProvider]] = [ @@ -28,6 +29,7 @@ upstream_provider_classes: list[type[BaseUpstreamProvider]] = [ PPQAIUpstreamProvider, RoutstrUpstreamProvider, TinfoilUpstreamProvider, + TypeSafeUpstreamProvider, XAIUpstreamProvider, ] """List of all upstream classes""" diff --git a/routstr/upstream/base.py b/routstr/upstream/base.py index 217cf9fc..1a080eaf 100644 --- a/routstr/upstream/base.py +++ b/routstr/upstream/base.py @@ -303,7 +303,7 @@ def _openai_completion_path(path: str) -> str | None: def _x_cashu_path_has_settlement_handler(path: str) -> bool: canonical = path.rstrip("/") return _openai_completion_path(canonical) is not None or canonical.endswith( - ("embeddings", "messages", "messages/count_tokens") + ("embeddings", "messages", "messages/count_tokens", "systemone") ) @@ -3155,6 +3155,7 @@ class BaseUpstreamProvider: or path.endswith("embeddings") or path.endswith("messages") or path.endswith("messages/count_tokens") + or path.endswith("systemone") ): if path.endswith("messages"): client_wants_streaming = False diff --git a/routstr/upstream/helpers.py b/routstr/upstream/helpers.py index 7065800f..dc544b3a 100644 --- a/routstr/upstream/helpers.py +++ b/routstr/upstream/helpers.py @@ -273,6 +273,7 @@ async def _seed_providers_from_settings( ("FIREWORKS_API_KEY", "fireworks", None, None), ("XAI_API_KEY", "xai", None, None), ("TINFOIL_API_KEY", "tinfoil", None, None), + ("TYPESAFE_API_KEY", "typesafe", None, None), ] for env_key, provider_type, _, _ in env_mappings: diff --git a/routstr/upstream/typesafe.py b/routstr/upstream/typesafe.py new file mode 100644 index 00000000..5fd06a6d --- /dev/null +++ b/routstr/upstream/typesafe.py @@ -0,0 +1,143 @@ +"""Upstream provider for the TypeSafe System One API. + +``POST /v1/systemone`` takes ``{state, model, questions}`` and returns +``{model, answers, usage}``. The ``usage`` shape (``input_tokens`` / +``output_tokens``) is what :func:`routstr.payment.usage.normalize_usage` +already parses, so billing needs no dialect handling. This provider only +assembles the catalog: ``GET /v1/models`` lists names without prices. +""" + +from __future__ import annotations + +from datetime import datetime +from typing import TYPE_CHECKING, Any + +import httpx + +from ..core.logging import get_logger +from ..payment.models import Architecture, Model, Pricing, TopProvider +from .base import BaseUpstreamProvider + +if TYPE_CHECKING: + from ..core.db import UpstreamProviderRow + +logger = get_logger(__name__) + +# USD per token (https://docs.typesafe.ai/models). Output is free. +_INPUT_RATE_USD = 0.042 / 1_000_000 +_OUTPUT_RATE_USD = 0.0 + +# The listing returns aliases only; versioned ids are accepted but unlisted. +_VERSIONED_MODEL_IDS = ("jev-1.13.0",) + +_CONTEXT_LENGTH = 64_000 +_MODELS_TIMEOUT_SECONDS = 30.0 + + +def _parse_release_date(value: object) -> int: + if not isinstance(value, str) or not value: + return 0 + try: + return int(datetime.fromisoformat(value.replace("Z", "+00:00")).timestamp()) + except ValueError: + return 0 + + +def _build_model(name: str, entry: dict[str, Any] | None = None) -> Model: + entry = entry or {} + description = entry.get("description") + if not isinstance(description, str) or not description: + description = f"TypeSafe System One model {name}" + + return Model( + id=name, + name=name, + created=_parse_release_date(entry.get("release_date")), + description=description, + context_length=_CONTEXT_LENGTH, + architecture=Architecture( + modality="text->decisions", + input_modalities=["text"], + output_modalities=["decisions"], + tokenizer="Other", + instruct_type=None, + ), + pricing=Pricing(prompt=_INPUT_RATE_USD, completion=_OUTPUT_RATE_USD), + top_provider=TopProvider(context_length=_CONTEXT_LENGTH), + ) + + +def _models_from_listing(data: object) -> list[Model]: + entries = data.get("models", []) if isinstance(data, dict) else data + if not isinstance(entries, list): + entries = [] + + models: dict[str, Model] = {} + for entry in entries: + name = entry.get("name") if isinstance(entry, dict) else None + if isinstance(name, str) and name and name not in models: + models[name] = _build_model(name, entry) + + for name in _VERSIONED_MODEL_IDS: + if name not in models: + models[name] = _build_model(name) + return list(models.values()) + + +class TypeSafeUpstreamProvider(BaseUpstreamProvider): + """Upstream provider for the TypeSafe System One decision API.""" + + provider_type = "typesafe" + default_base_url = "https://api.typesafe.ai/v1" + platform_url = "https://docs.typesafe.ai" + + def __init__(self, api_key: str, provider_fee: float = 1.0): + super().__init__( + base_url=self.default_base_url, + api_key=api_key, + provider_fee=provider_fee, + ) + + @classmethod + def _build_from_row( + cls, provider_row: "UpstreamProviderRow" + ) -> "TypeSafeUpstreamProvider": + return cls( + api_key=provider_row.api_key, + provider_fee=provider_row.provider_fee, + ) + + @classmethod + def get_provider_metadata(cls) -> dict[str, object]: + return { + "id": cls.provider_type, + "name": "TypeSafe", + "default_base_url": cls.default_base_url, + "fixed_base_url": True, + "platform_url": cls.platform_url, + "can_create_account": False, + "can_topup": False, + "can_show_balance": False, + } + + def transform_model_name(self, model_id: str) -> str: + return model_id.removeprefix("typesafe/") + + async def fetch_models(self) -> list[Model]: + """Fetch the catalog; return an empty list on failure so init never breaks.""" + url = f"{self.base_url}/models" + headers = {"Authorization": f"Bearer {self.api_key}"} + try: + async with httpx.AsyncClient(timeout=_MODELS_TIMEOUT_SECONDS) as client: + response = await client.get(url, headers=headers) + response.raise_for_status() + data = response.json() + except Exception as exc: + logger.warning( + "Failed to fetch TypeSafe model catalog", + extra={"url": url, "error": str(exc)}, + exc_info=True, + ) + return [] + + return _models_from_listing(data) diff --git a/tests/integration/test_systemone.py b/tests/integration/test_systemone.py new file mode 100644 index 00000000..e079f0be --- /dev/null +++ b/tests/integration/test_systemone.py @@ -0,0 +1,221 @@ +"""Integration test: POST /v1/systemone proxied and billed end-to-end. + +The TypeSafe decision endpoint shares the request plumbing with embeddings: +model id in the top-level ``model`` field, flat ``usage`` in the response. +This test pins the whole path — allowlist, auth, forwarding, and settlement +billed from the response's usage — with only the network hop mocked. +""" + +import json +from typing import Any, AsyncGenerator +from unittest.mock import AsyncMock, patch + +import httpx +import pytest +from httpx import AsyncClient +from sqlmodel.ext.asyncio.session import AsyncSession + +from routstr.payment.models import Architecture, Model, Pricing +from routstr.proxy import refresh_model_maps +from routstr.upstream.base import BaseUpstreamProvider + +TYPESAFE_BASE_URL = "https://api.typesafe.ai/v1" + +SYSTEMONE_REQUEST = { + "model": "jev-latest", + "state": "Help! My payouts have been failing for 3 days.", + "questions": { + "is_urgent": {"type": "noul", "instructions": "Does this convey urgency?"} + }, +} + +SYSTEMONE_RESPONSE = { + "model": "jev-latest", + "answers": { + "is_urgent": {"type": "noul", "noul": 0.95}, + }, + "usage": {"input_tokens": 1000, "output_tokens": 200}, +} + + +class _StaticTypeSafeProvider(BaseUpstreamProvider): + """Upstream provider with a fixed TypeSafe model catalog.""" + + def __init__(self, base_url: str, api_key: str, fee: float, model: Model) -> None: + super().__init__(base_url, api_key, fee) + self.provider_type = "typesafe" + self._static_model = model + + def get_cached_models(self) -> list[Model]: + return [self._static_model] + + async def refresh_models_cache(self) -> None: + pass + + +def _jev_model(prompt_sats: float = 0.001, completion_sats: float = 0.0) -> Model: + return Model( + id="jev-latest", + name="jev-latest", + created=1, + description="TypeSafe System One decision model", + context_length=64_000, + architecture=Architecture( + modality="text->decisions", + input_modalities=["text"], + output_modalities=["decisions"], + tokenizer="Other", + instruct_type=None, + ), + pricing=Pricing( + prompt=prompt_sats, completion=completion_sats, max_cost=50.0 + ), + sats_pricing=Pricing( + prompt=prompt_sats, completion=completion_sats, max_cost=50.0 + ), + ) + + +@pytest.fixture +async def typesafe_provider_maps( + patched_db_engine: None, +) -> AsyncGenerator[_StaticTypeSafeProvider, None]: + """Install a TypeSafe provider with a priced jev-latest model.""" + provider = _StaticTypeSafeProvider( + TYPESAFE_BASE_URL, + "key-typesafe", + 1.0, + _jev_model(), + ) + + from routstr import proxy + + original_upstreams = proxy.get_upstreams() + with patch("routstr.proxy._upstreams", [provider]): + await refresh_model_maps() + yield provider + with patch("routstr.proxy._upstreams", original_upstreams): + await refresh_model_maps() + + +@pytest.mark.integration +@pytest.mark.asyncio +async def test_systemone_forwarded_and_billed_from_usage( + authenticated_client: AsyncClient, + typesafe_provider_maps: _StaticTypeSafeProvider, + integration_session: AsyncSession, +) -> None: + """A systemone request reaches api.typesafe.ai and bills input tokens.""" + sent_requests: list[httpx.Request] = [] + + async def fake_transport( + request: httpx.Request, *args: Any, **kwargs: Any + ) -> httpx.Response: + sent_requests.append(request) + return httpx.Response( + 200, + content=json.dumps(SYSTEMONE_RESPONSE).encode(), + headers={"content-type": "application/json"}, + ) + + with ( + patch( + "httpx.AsyncHTTPTransport.handle_async_request", + side_effect=fake_transport, + ), + patch( + "routstr.payment.cost_calculation.sats_usd_price", + return_value=0.0005, + ), + ): + response = await authenticated_client.post( + "/v1/systemone", + json=SYSTEMONE_REQUEST, + ) + + assert response.status_code == 200, response.text + payload = response.json() + + # The answers pass through untouched. + assert payload["answers"]["is_urgent"]["noul"] == pytest.approx(0.95) + + # Exactly one upstream hop, aimed at TypeSafe's endpoint with the + # provider's model spelling. + assert len(sent_requests) == 1 + sent = sent_requests[0] + assert str(sent.url) == "https://api.typesafe.ai/v1/systemone" + assert json.loads(sent.content)["model"] == "jev-latest" + + # Billed from usage: 1000 input tokens at 0.001 sats/token = 1000 msats; + # 200 output tokens at 0 sats = 0. Settled cost is exposed on the body. + assert payload["cost"]["input_msats"] == 1000 + assert payload["cost"]["output_msats"] == 0 + assert payload["cost"]["total_msats"] == 1000 + + +@pytest.mark.integration +@pytest.mark.asyncio +async def test_systemone_with_x_cashu_settles( + authenticated_client: AsyncClient, + typesafe_provider_maps: _StaticTypeSafeProvider, + testmint_wallet: Any, +) -> None: + """The X-Cashu settlement gate admits systemone and refunds the delta.""" + token = await testmint_wallet.mint_tokens(10_000) + + async def fake_transport( + request: httpx.Request, *args: Any, **kwargs: Any + ) -> httpx.Response: + return httpx.Response( + 200, + content=json.dumps(SYSTEMONE_RESPONSE).encode(), + headers={"content-type": "application/json"}, + ) + + # The redemption and refund helpers are bound into base.py at import time, + # so the app fixture's wallet patches do not reach the proxy's x-cashu path. + with ( + patch( + "httpx.AsyncHTTPTransport.handle_async_request", + side_effect=fake_transport, + ), + patch( + "routstr.upstream.base.recieve_token", + AsyncMock(return_value=(10_000, "sat", "https://mint.test")), + ), + patch( + "routstr.upstream.base.send_token", + AsyncMock(return_value="cashuBrefundtoken"), + ), + # send_refund derives the mint from the token it just created; the + # fake token has no mint to parse. + patch( + "routstr.upstream.base.token_mint_url", + return_value="https://mint.test", + ), + # cost_calculation binds sats_usd_price at import time, so the price + # patch in the app fixture does not reach it. + patch( + "routstr.payment.cost_calculation.sats_usd_price", + return_value=0.0005, + ), + # Mint trust is node state that other suites mutate; this test is + # about the endpoint gate, not mint policy. + patch( + "routstr.payment.helpers.is_trusted_source_mint", + return_value=True, + ), + ): + response = await authenticated_client.post( + "/v1/systemone", + json=SYSTEMONE_REQUEST, + headers={"X-Cashu": token}, + ) + + # Not rejected as an unsupported endpoint (that returns 400 with + # x_cashu_unsupported_endpoint), and settled rather than streamed raw. + assert response.status_code == 200, response.text + payload = response.json() + assert payload["answers"]["is_urgent"]["type"] == "noul" + # A refund header exists when the token exceeded the settled cost. + assert response.headers.get("X-Cashu") is not None diff --git a/tests/unit/test_typesafe_integration.py b/tests/unit/test_typesafe_integration.py new file mode 100644 index 00000000..2c85adfa --- /dev/null +++ b/tests/unit/test_typesafe_integration.py @@ -0,0 +1,178 @@ +"""Unit tests for the TypeSafe System One provider integration. + +Covers the three integration seams the systemone endpoint touches: + +* the proxy path allowlist admits ``systemone`` (POST only) and nothing that + merely looks like it; +* the X-Cashu settlement gate treats ``systemone`` as a settleable endpoint; +* the provider maps TypeSafe's pricing-less model listing onto priced + ``Model`` objects with usable rates. +""" + +from __future__ import annotations + +import os +from typing import Any + +os.environ.setdefault("UPSTREAM_BASE_URL", "http://test") +os.environ.setdefault("UPSTREAM_API_KEY", "test") + +from unittest.mock import AsyncMock, MagicMock, patch # noqa: E402 + +import pytest # noqa: E402 + +from routstr.proxy import _forwarding_allowed # noqa: E402 +from routstr.upstream.base import _x_cashu_path_has_settlement_handler # noqa: E402 +from routstr.upstream.typesafe import TypeSafeUpstreamProvider # noqa: E402 + + +@pytest.mark.parametrize( + ("path", "method"), + [ + ("v1/systemone", "POST"), + ("systemone", "POST"), + ("v1/systemone/", "POST"), + ], +) +def test_systemone_is_forwarded(path: str, method: str) -> None: + assert _forwarding_allowed(path, method) is True + + +@pytest.mark.parametrize( + ("path", "method"), + [ + ("v1/systemone", "GET"), # billed endpoint is POST-only + ("v1/systemone", "DELETE"), + ("systemonedump", "POST"), # longer segment must not match + ("v1/systemone/secret", "POST"), # trailing id segment widens nothing + ("v1/systemone/../admin", "POST"), # traversal spelling is screened + ], +) +def test_systemone_lookalikes_are_refused(path: str, method: str) -> None: + assert _forwarding_allowed(path, method) is False + + +def test_systemone_has_x_cashu_settlement_handler() -> None: + assert _x_cashu_path_has_settlement_handler("v1/systemone") is True + assert _x_cashu_path_has_settlement_handler("systemone/") is True + + +def test_provider_metadata_is_complete() -> None: + meta = TypeSafeUpstreamProvider.get_provider_metadata() + assert meta["id"] == "typesafe" + assert meta["default_base_url"] == "https://api.typesafe.ai/v1" + assert meta["fixed_base_url"] is True + + +def test_provider_transform_model_name() -> None: + provider = TypeSafeUpstreamProvider(api_key="test") + assert provider.transform_model_name("typesafe/jev-latest") == "jev-latest" + assert provider.transform_model_name("jev-latest") == "jev-latest" + + +def _mock_models_response(payload: dict) -> MagicMock: + mock_response = MagicMock() + mock_response.status_code = 200 + mock_response.raise_for_status = MagicMock() + mock_response.json.return_value = payload + return mock_response + + +def _patch_client(mock_response: MagicMock) -> Any: + mock_client = MagicMock() + mock_client.__aenter__ = AsyncMock(return_value=mock_client) + mock_client.__aexit__ = AsyncMock(return_value=None) + mock_client.get = AsyncMock(return_value=mock_response) + return patch( + "routstr.upstream.typesafe.httpx.AsyncClient", + return_value=mock_client, + ) + + +@pytest.mark.asyncio +async def test_fetch_models_prices_the_listing() -> None: + provider = TypeSafeUpstreamProvider(api_key="test") + payload = { + "models": [ + { + "name": "jev-latest", + "description": "The most recent stable release.", + "release_date": "2026-09-17", + }, + { + "name": "jev-preview", + "description": "The most recent release.", + "release_date": "2026-09-17", + }, + ] + } + + with _patch_client(_mock_models_response(payload)): + models = await provider.fetch_models() + + assert [m.id for m in models] == ["jev-latest", "jev-preview", "jev-1.13.0"] + for model in models: + # Input priced, output free; rates usable (zero is a real price). + assert model.pricing.prompt == pytest.approx(0.042 / 1_000_000) + assert model.pricing.completion == 0.0 + assert model.context_length == 64_000 + assert model.architecture.input_modalities == ["text"] + assert model.architecture.output_modalities == ["decisions"] + + +@pytest.mark.asyncio +async def test_fetch_models_handles_error() -> None: + provider = TypeSafeUpstreamProvider(api_key="test") + mock_response = MagicMock() + mock_response.raise_for_status = MagicMock( + side_effect=RuntimeError("upstream down") + ) + + with ( + _patch_client(mock_response), + patch("routstr.upstream.typesafe.logger") as mock_logger, + ): + models = await provider.fetch_models() + + assert models == [] + mock_logger.warning.assert_called_once() + + +@pytest.mark.asyncio +async def test_fetch_models_defaults_unknown_model_rates() -> None: + """A newly listed model still gets a usable default rate.""" + provider = TypeSafeUpstreamProvider(api_key="test") + payload = { + "models": [ + {"name": "jev-2.0", "description": "Future model", "release_date": None} + ] + } + + with _patch_client(_mock_models_response(payload)): + models = await provider.fetch_models() + + by_id = {m.id: m for m in models} + assert "jev-2.0" in by_id + assert by_id["jev-2.0"].pricing.prompt == pytest.approx(0.042 / 1_000_000) + assert by_id["jev-2.0"].created == 0 + + +@pytest.mark.asyncio +async def test_fetch_models_does_not_duplicate_listed_versioned_id() -> None: + provider = TypeSafeUpstreamProvider(api_key="test") + payload = { + "models": [ + { + "name": "jev-1.13.0", + "description": "Jev 1.13", + "release_date": "2026-09-17T00:00:00Z", + } + ] + } + + with _patch_client(_mock_models_response(payload)): + models = await provider.fetch_models() + + assert [m.id for m in models] == ["jev-1.13.0"] + assert models[0].description == "Jev 1.13" + assert models[0].created > 0