v1alpha1

type DataLayerConfig struct {
	// +optional
	// InjectDefaults controls automatic injection of the default metrics source and extractor.
	// Defaults to true when omitted. Set to false to disable all default source injection.
	InjectDefaults *bool `json:"injectDefaults,omitempty"`
	// +optional
	// Sources is the list of sources to define to the DataLayer
	Sources []DataLayerSource `json:"sources,omitempty"`
	// +optional
	// Discovery specifies which EndpointDiscovery plugin to use for populating the
	// endpoint datastore. When set, the EPP bypasses Kubernetes CRD reconcilers and
	// relies entirely on the referenced plugin to enumerate and track inference
	// endpoints. This enables running the EPP without a Kubernetes cluster.
	// If omitted, the EPP uses the default Kubernetes-based discovery.
	Discovery *DiscoveryConfig `json:"discovery,omitempty"`
}

type DataLayerExtractor struct {
	// +required
	// +kubebuilder:validation:Required
	// PluginRef specifies a particular Plugin instance to be associated with
	// this Extractor. The reference is to the name of an entry of the Plugins
	// defined in the configuration's Plugins section
	PluginRef string `json:"pluginRef"`
}

type DataLayerSource struct {
	// +required
	// +kubebuilder:validation:Required
	// PluginRef specifies a particular Plugin instance to be associated with
	// this Source. The reference is to the name of an entry of the Plugins
	// defined in the configuration's Plugins section
	PluginRef string `json:"pluginRef"`

	// +required
	// +kubebuilder:validation:Required
	// Extractors specifies the list of Plugin instances to be associated with
	// this Source. The entries are references to the names of entries of the Plugins
	// defined in the configuration's Plugins section
	Extractors []DataLayerExtractor `json:"extractors"`
}

type DiscoveryConfig struct {
	// +required
	// +kubebuilder:validation:Required
	// PluginRef is the name of the plugin instance (from the Plugins list) that
	// implements EndpointDiscovery.
	PluginRef string `json:"pluginRef"`
}

type EndpointPickerConfig struct {
	metav1.TypeMeta `json:",inline"`

	// +optional
	// FeatureGates is a set of flags that enable various experimental features with the EPP.
	// If omitted none of these experimental features will be enabled.
	FeatureGates FeatureGates `json:"featureGates,omitempty"`

	// +required
	// +kubebuilder:validation:Required
	// Plugins is the list of plugins that will be instantiated.
	Plugins []PluginSpec `json:"plugins"`

	// +required
	// +kubebuilder:validation:Required
	// SchedulingProfiles is the list of named SchedulingProfiles
	// that will be created.
	SchedulingProfiles []SchedulingProfile `json:"schedulingProfiles"`

	// +optional
	// DataLayer configures the DataLayer. It is required if the new DataLayer is enabled.
	DataLayer *DataLayerConfig `json:"dataLayer"`

	// +optional
	// FlowControl configures the Flow Control layer.
	// This configuration is only respected if the "flowControl" FeatureGate is enabled.
	FlowControl *FlowControlConfig `json:"flowControl,omitempty"`

	// +optional
	// RequestHandler specifies the handling logic used by the EPP to process incoming requests.
	RequestHandler *RequestHandlerConfig `json:"requestHandler,omitempty"`

	// +optional
	// SaturationDetector specifies which saturation detector plugin to use.
	//
	// Deprecated: use flowControl.saturationDetector instead. If both are set, the new field is used.
	// Tracked in https://github.com/llm-d/llm-d-router/issues/1308
	SaturationDetector *SaturationDetectorConfig `json:"saturationDetector,omitempty"`

	// +optional
	// Parser specifies the parsing logic used by the EPP to process protocol messages.
	//
	// Deprecated: use requestHandler.parser instead. If both are set, the new field is used.
	// Tracked in https://github.com/llm-d/llm-d-router/issues/1308
	Parser *ParserConfig `json:"parser,omitempty"`
}

type FeatureGates []string

type FlowControlConfig struct {
	// +optional
	// MaxBytes defines the global maximum aggregate byte size of all active requests across all priority
	// levels. If this limit is exceeded, new requests will be rejected even if their specific
	// priority band has capacity.
	// Accepts standard Kubernetes resource quantities (e.g., "1Gi", "500M").
	// If omitted, no global byte limit is enforced.
	MaxBytes *resource.Quantity `json:"maxBytes,omitempty"`

	// +optional
	// MaxRequests defines the global maximum number of concurrent requests across all priority
	// levels. If this limit is exceeded, new requests will be rejected even if their specific
	// priority band has capacity.
	// Accepts standard Kubernetes resource quantities (e.g., "100", "1k").
	// If omitted, no global request limit is enforced.
	MaxRequests *resource.Quantity `json:"maxRequests,omitempty"`

	// +optional
	// DefaultRequestTTL serves as a fallback timeout for requests that do not specify their own
	// deadline.
	// It ensures that requests do not hang indefinitely in the queue.
	// If 0 or omitted, it defaults to the client context deadline, meaning requests may wait
	// indefinitely unless cancelled by the client.
	DefaultRequestTTL *metav1.Duration `json:"defaultRequestTTL,omitempty"`

	// +optional
	// DefaultPriorityBand allows you to define a template for handling traffic with priority levels
	// that are not explicitly configured in `PriorityBands`.
	// This ensures that unforeseen traffic classes are still managed (e.g., given a default capacity
	// limit) rather than being rejected or treated arbitrarily.
	// If not specified, a system-default template is used to dynamically provision bands for new
	// priority levels. This template cascades to the standard `PriorityBandConfig` defaults.
	DefaultPriorityBand *PriorityBandConfig `json:"defaultPriorityBand,omitempty"`

	// +optional
	// DefaultNegativePriorityBand allows you to define a separate template for priority levels
	// strictly below zero. This enables designating negative-priority traffic as sheddable by
	// setting lower capacity limits (e.g., maxBytes: "0" to drop immediately).
	// If not specified, negative priorities fall back to DefaultPriorityBand.
	DefaultNegativePriorityBand *PriorityBandConfig `json:"defaultNegativePriorityBand,omitempty"`

	// PriorityBands allows you to explicitly define policies (like capacity limits) for specific
	// priority levels. Traffic matching these priorities will be handled according to these rules.
	// If a priority band is not specified, it uses specific defaults.
	PriorityBands []PriorityBandConfig `json:"priorityBands,omitempty"`

	// +optional
	// UsageLimitPolicyPluginRef specifies the UsageLimitPolicy plugin to use for adaptive capacity management.
	// Must reference a named plugin instance defined in the top-level Plugins section.
	// If omitted, a default static policy (threshold=1.0, no gating) is used.
	UsageLimitPolicyPluginRef string `json:"usageLimitPolicyPluginRef,omitempty"`

	// +optional
	// SaturationDetector specifies which saturation detector plugin to use for both Admission and
	// Flow Control. If omitted, "utilization-detector" is used by default.
	SaturationDetector *SaturationDetectorConfig `json:"saturationDetector,omitempty"`
}

type ParserConfig struct {
	// +required
	// +kubebuilder:validation:Required
	// PluginRef specifies a particular Plugin instance to be associated with
	// this Parser. The reference is to the name of an entry of the Plugins
	// defined in the configuration's Plugins section
	// Default: openai-parser
	PluginRef string `json:"pluginRef"`
}

type PluginSpec struct {
	// +optional
	// Name provides a name for plugin entries to reference. If
	// omitted, the value of the Plugin's Type field will be used.
	Name string `json:"name"`

	// +required
	// +kubebuilder:validation:Required
	// Type specifies the plugin type to be instantiated.
	Type string `json:"type"`

	// +optional
	// Parameters are the set of parameters to be passed to the plugin's
	// factory function. The factory function is responsible
	// to parse the parameters.
	Parameters json.RawMessage `json:"parameters"`
}

type PriorityBandConfig struct {
	// Priority is the integer priority level for this band.
	// Higher values indicate higher priority.
	Priority int `json:"priority"`

	// +optional
	// MaxBytes is the maximum number of bytes allowed for this priority band.
	// Accepts standard Kubernetes resource quantities (e.g., "1Gi", "500M").
	// If omitted, the system default is used.
	MaxBytes *resource.Quantity `json:"maxBytes,omitempty"`

	// +optional
	// MaxRequests is the maximum number of concurrent requests allowed for this priority band.
	// Accepts standard Kubernetes resource quantities (e.g., "100", "1k").
	// If omitted, no request limit is enforced.
	MaxRequests *resource.Quantity `json:"maxRequests,omitempty"`

	// +optional
	// FairnessPolicyRef specifies the name of the policy that governs flow selection.
	// If omitted, the system default ("global-strict-fairness-policy") is used.
	FairnessPolicyRef string `json:"fairnessPolicyRef,omitempty"`

	// +optional
	// OrderingPolicyRef specifies the name of the policy that governs request selection within a flow.
	// If omitted, the system default ("fcfs-ordering-policy") is used.
	OrderingPolicyRef string `json:"orderingPolicyRef,omitempty"`
}

type RequestHandlerConfig struct {
	// +optional
	// Parsers specifies the parsing plugins used by the EPP to process protocol messages.
	// If unspecified, default parsing behavior will be applied.
	Parsers []ParserConfig `json:"parsers,omitempty"`
}

type SaturationDetectorConfig struct {
	// +optional
	// PluginRef specifies the name of the plugin instance to use for saturation detection.
	// The reference is to the name of an entry of the Plugins defined in the configuration's Plugins section.
	// If unspecified, "utilization-detector" is used by default.
	PluginRef string `json:"pluginRef,omitempty"`
}

type SchedulingPlugin struct {
	// +required
	// +kubebuilder:validation:Required
	// PluginRef specifies a particular Plugin instance to be associated with
	// this SchedulingProfile. The reference is to the name of an
	// entry of the Plugins defined in the configuration's Plugins
	// section
	PluginRef string `json:"pluginRef"`

	// +optional
	// Weight is the weight to be used if this plugin is a Scorer.
	Weight *float64 `json:"weight"`
}

type SchedulingProfile struct {
	// +kubebuilder:validation:Required
	// Name specifies the name of this SchedulingProfile
	Name string `json:"name"`

	// +required
	// +kubebuilder:validation:Required
	// Plugins is the list of plugins for this SchedulingProfile. They are assigned
	// to the appropriate "slots" based on their type.
	Plugins []SchedulingPlugin `json:"plugins"`
}