devplacepy/devplacepy/services/devii/actions/catalog/gateway.py

# retoor <retoor@molodetz.nl>

from __future__ import annotations

from ..spec import Action, Param
from ._shared import body, confirm, path


GATEWAY_ACTIONS: tuple[Action, ...] = (
    Action(
        name="gateway_providers",
        method="GET",
        path="/admin/gateway/providers",
        summary="List OpenAI gateway providers (admin only)",
        description=(
            "Returns JSON: the named upstream providers used by model routes, plus the implicit "
            "'default' provider read from the gateway service config."
        ),
        handler="http",
        requires_admin=True,
        read_only=True,
    ),
    Action(
        name="gateway_provider_set",
        method="POST",
        path="/admin/gateway/providers",
        summary="Create or update a gateway provider (admin only)",
        description=(
            "Adds or updates a named upstream provider. base_url is the OpenAI-compatible "
            "chat-completions endpoint; the embeddings and images endpoints are derived from it."
        ),
        handler="http",
        requires_admin=True,
        params=(
            body("name", "Provider name (letters, numbers, hyphen, underscore).", required=True),
            body("base_url", "Chat-completions endpoint URL."),
            body("api_key", "Upstream API key."),
            Param(
                name="is_active",
                location="body",
                description="Whether the provider is active ('1' or '0').",
                required=False,
                type="boolean",
            ),
        ),
    ),
    Action(
        name="gateway_provider_delete",
        method="DELETE",
        path="/admin/gateway/providers/{name}",
        summary="Delete a gateway provider (admin only, confirmation required)",
        handler="http",
        requires_admin=True,
        params=(path("name", "Provider name."), confirm()),
    ),
    Action(
        name="gateway_models",
        method="GET",
        path="/admin/gateway/models",
        summary="List OpenAI gateway model routes (admin only)",
        description=(
            "Returns JSON: every source-model route with its provider, target model, kind "
            "(chat/embed/image), optional vision model, context window, and per-model pricing economy "
            "(including any tiered/off-peak pricing configured on it)."
        ),
        handler="http",
        requires_admin=True,
        read_only=True,
    ),
    Action(
        name="gateway_model_set",
        method="POST",
        path="/admin/gateway/models",
        summary="Create or update a gateway model route (admin only)",
        description=(
            "Maps a requested source_model onto a provider + target_model, each with its own "
            "pricing. kind is 'chat', 'embed', or 'image'. A vision_model adds image-to-text augmentation "
            "for chat routes. Prices are USD per 1,000,000 tokens for chat/embed/vision; image routes "
            "use price_input_per_m as a flat USD per generated image. Optionally, a route can also "
            "charge a different (tier-2) rate once the request's input tokens exceed "
            "context_tier_threshold_tokens, and/or apply a percentage discount during a fixed "
            "UTC off-peak window - leave the tier2/off-peak fields unset to keep the flat rates "
            "above at all times. off_peak_start_minute and off_peak_end_minute must be set "
            "together (both or neither) or the call is rejected."
        ),
        handler="http",
        requires_admin=True,
        params=(
            body("source_model", "Model name clients request.", required=True),
            body("provider", "Provider name, or blank for the default upstream."),
            body("target_model", "Model name sent upstream.", required=True),
            body("kind", "Route kind: 'chat', 'embed', or 'image'."),
            body("vision_provider", "Provider for image description, or blank for the route provider."),
            body("vision_model", "Vision model name (blank disables the merge)."),
            Param(name="context_window", location="body", description="Max context tokens (0 = unknown).", required=False, type="integer"),
            Param(name="price_cache_hit_per_m", location="body", description="USD per 1M cache-hit input tokens.", required=False, type="number"),
            Param(name="price_cache_miss_per_m", location="body", description="USD per 1M cache-miss input tokens.", required=False, type="number"),
            Param(name="price_output_per_m", location="body", description="USD per 1M output tokens.", required=False, type="number"),
            Param(name="price_input_per_m", location="body", description="USD per 1M input tokens (embed/vision), or flat USD per image for image routes.", required=False, type="number"),
            Param(name="context_tier_threshold_tokens", location="body", description="Input tokens above which tier-2 rates apply (0 disables tiering).", required=False, type="integer"),
            Param(name="price_cache_hit_per_m_tier2", location="body", description="Tier-2 USD per 1M cache-hit input tokens (unset = keep tier-1 rate above threshold).", required=False, type="number"),
            Param(name="price_cache_miss_per_m_tier2", location="body", description="Tier-2 USD per 1M cache-miss input tokens (unset = keep tier-1 rate above threshold).", required=False, type="number"),
            Param(name="price_output_per_m_tier2", location="body", description="Tier-2 USD per 1M output tokens (unset = keep tier-1 rate above threshold).", required=False, type="number"),
            Param(name="price_input_per_m_tier2", location="body", description="Tier-2 USD per 1M input tokens, embed/vision (unset = keep tier-1 rate above threshold).", required=False, type="number"),
            Param(name="off_peak_start_minute", location="body", description="Off-peak window start, UTC minutes since midnight (0-1439). Must be set together with off_peak_end_minute.", required=False, type="integer"),
            Param(name="off_peak_end_minute", location="body", description="Off-peak window end, UTC minutes since midnight (0-1439). A value less than the start wraps past midnight.", required=False, type="integer"),
            Param(name="off_peak_discount_pct", location="body", description="Percentage discount (0-100) applied to the active tier's rates during the off-peak window.", required=False, type="number"),
            Param(name="is_active", location="body", description="Whether the route is active ('1' or '0').", required=False, type="boolean"),
        ),
    ),
    Action(
        name="gateway_model_delete",
        method="DELETE",
        path="/admin/gateway/models/{source_model}",
        summary="Delete a gateway model route (admin only, confirmation required)",
        handler="http",
        requires_admin=True,
        params=(path("source_model", "Source model name."), confirm()),
    ),
    Action(
        name="gateway_quota_rules",
        method="GET",
        path="/admin/gateway/quota-rules",
        summary="List AI gateway quota rules and the current global defaults (admin only)",
        description=(
            "Returns JSON: every quota rule (each scoped by any combination of role/owner_kind, "
            "a specific user uid, and an app_reference label) with its 24h limit, current 24h spend "
            "against that exact scope, active flag, and label, plus the global per-role default caps "
            "used when no rule matches a request."
        ),
        handler="http",
        requires_admin=True,
        read_only=True,
    ),
    Action(
        name="gateway_quota_rule_set",
        method="POST",
        path="/admin/gateway/quota-rules",
        summary="Create or update an AI gateway quota rule (admin only)",
        description=(
            "Caps rolling-24h USD spend on /openai/v1/*. Scope by any combination of owner_kind "
            "(internal/key/user/admin/anonymous), a specific owner_id (user uid), and app_reference "
            "(the X-App-Reference header apps send). At least one of the three must be set - an "
            "unscoped cap belongs in the global default fields on the gateway service config instead. "
            "Leaving a dimension blank makes it a wildcard: an app_reference-only rule pools spend "
            "across every caller using that app; an owner_kind-only rule pools spend across every "
            "caller of that role. Setting owner_id pins the rule to one specific caller. When several "
            "rules match one request, the MOST SPECIFIC one wins (most non-blank dimensions); ties "
            "break toward the smaller limit. limit_usd of 0 means unlimited for that rule. Pass uid "
            "to update an existing rule instead of creating a new one."
        ),
        handler="http",
        requires_admin=True,
        params=(
            body("uid", "Existing rule uid to update; omit to create a new rule."),
            body("owner_kind", "Role to scope by: internal, key, user, admin, or anonymous. Blank = any role."),
            body("owner_id", "Specific user uid to scope by. Blank = any caller of the matched role."),
            body("app_reference", "App label to scope by (the X-App-Reference header). Blank = any app."),
            Param(name="limit_usd", location="body", description="Rolling 24h USD cap for this rule. 0 = unlimited.", required=True, type="number"),
            Param(name="is_active", location="body", description="Whether the rule is enforced ('1' or '0'). Defaults to active.", required=False, type="boolean"),
            body("label", "Optional admin-facing note describing what this rule is for."),
        ),
    ),
    Action(
        name="gateway_quota_reset",
        method="POST",
        path="/admin/gateway/quota-resets",
        summary="Reset the rolling-24h AI gateway spend so a capped caller can call again (admin only)",
        description=(
            "Clears the counted spend for a scope without deleting any usage history, so the "
            "cost analytics on /admin/ai-usage stay intact. Scope it exactly like a quota rule: "
            "owner_kind (internal/key/user/admin/anonymous), a specific owner_id (user uid), and "
            "app_reference (the X-App-Reference header). Leaving all three blank resets every "
            "caller. A reset only clears spend recorded BEFORE it - new calls start counting "
            "again immediately against the same limit. Use this when an app is stuck on "
            "'AI gateway daily quota exceeded' and you want it running again without raising "
            "its cap."
        ),
        handler="http",
        requires_admin=True,
        params=(
            body("owner_kind", "Role to scope by: internal, key, user, admin, or anonymous. Blank = every role."),
            body("owner_id", "Specific user uid to scope by. Blank = every caller."),
            body("app_reference", "App label to scope by (the X-App-Reference header). Blank = every app."),
            confirm(),
        ),
    ),
    Action(
        name="gateway_quota_rule_delete",
        method="DELETE",
        path="/admin/gateway/quota-rules/{uid}",
        summary="Delete an AI gateway quota rule (admin only, confirmation required)",
        handler="http",
        requires_admin=True,
        params=(path("uid", "Quota rule uid."), confirm()),
    ),
)