Files
ModelRouter/internal/config/config.go

185 lines
6.2 KiB
Go

// Package config loads the gateway YAML configuration plus a runtime overlay
// (web UI edits) and resolves them into sources with per-model priority.
package config
import (
"encoding/json"
"fmt"
"os"
"time"
"gopkg.in/yaml.v3"
)
// Config is the top-level gateway configuration.
type Config struct {
Listen string `yaml:"listen"`
GatewayKeys []string `yaml:"gateway_keys"`
DefaultModel string `yaml:"default_model"` // e.g. "AUTO" or a model id
AdapterDir string `yaml:"adapter_dir"`
RuntimeFile string `yaml:"runtime_file"`
MaxConcurrent int `yaml:"max_concurrent"` // global inflight cap, 0 = unlimited
Sources []Source `yaml:"sources"`
}
// Model is a single exposed model id bound to a source, with priority used by
// AUTO auto selection (higher number = preferred).
type Model struct {
ID string `yaml:"id" json:"id"`
Priority int `yaml:"priority" json:"priority"`
Kind string `yaml:"kind" json:"kind"` // "chat" (default) | "image"
Meta map[string]interface{} `yaml:"meta" json:"meta,omitempty"`
}
// Source describes a single upstream LLM provider.
type Source struct {
Name string `yaml:"name" json:"name"`
BaseURL string `yaml:"base_url" json:"base_url"`
APIKey string `yaml:"api_key" json:"api_key"`
APIKeyEnv string `yaml:"api_key_env,omitempty" json:"-"` // reference to an env var holding the key (overrides api_key)
Adapter string `yaml:"adapter" json:"adapter"`
Endpoint string `yaml:"endpoint" json:"endpoint,omitempty"` // chat endpoint override
ImageEndpoint string `yaml:"image_endpoint" json:"image_endpoint,omitempty"` // image endpoint override
Models []Model `yaml:"models" json:"models"`
Headers map[string]string `yaml:"headers" json:"headers,omitempty"`
Meta map[string]interface{} `yaml:"meta" json:"meta,omitempty"`
Temperature float64 `yaml:"temperature" json:"temperature,omitempty"`
MaxTokens int `yaml:"max_tokens" json:"max_tokens,omitempty"`
Timeout time.Duration `yaml:"timeout" json:"-"`
MaxConcurrent int `yaml:"max_concurrent" json:"max_concurrent"`
QueueTimeout time.Duration `yaml:"queue_timeout" json:"-"`
}
// Load reads and validates a config file.
func Load(path string) (*Config, error) {
data, err := os.ReadFile(path)
if err != nil {
return nil, err
}
var cfg Config
if err := yaml.Unmarshal(data, &cfg); err != nil {
return nil, fmt.Errorf("parse config: %w", err)
}
if err := cfg.ApplyDefaults(); err != nil {
return nil, err
}
return &cfg, nil
}
// ApplyDefaults sets missing values and validates the config.
func (c *Config) ApplyDefaults() error {
if c.Listen == "" {
c.Listen = ":8080"
}
if c.AdapterDir == "" {
c.AdapterDir = "adapters"
}
if c.RuntimeFile == "" {
c.RuntimeFile = "runtime.json"
}
if c.DefaultModel == "" {
c.DefaultModel = "AUTO"
}
seen := map[string]bool{}
modelOwners := map[string]string{}
for i := range c.Sources {
s := &c.Sources[i]
if s.Name == "" {
return fmt.Errorf("config: sources[%d] missing name", i)
}
if s.BaseURL == "" {
return fmt.Errorf("config: source %s missing base_url", s.Name)
}
if s.Adapter == "" {
s.Adapter = "openai"
}
if s.Timeout == 0 {
s.Timeout = 120 * time.Second
}
if s.QueueTimeout == 0 {
s.QueueTimeout = 60 * time.Second
}
if s.MaxConcurrent == 0 {
s.MaxConcurrent = 8
}
if seen[s.Name] {
return fmt.Errorf("config: duplicate source name %q", s.Name)
}
seen[s.Name] = true
for j := range s.Models {
m := &s.Models[j]
if m.ID == "" {
return fmt.Errorf("config: source %s has a model without id", s.Name)
}
if owner, ok := modelOwners[m.ID]; ok {
return fmt.Errorf("config: model %q defined by both %s and %s", m.ID, owner, s.Name)
}
modelOwners[m.ID] = s.Name
}
}
return nil
}
// RuntimeConfig is the persisted web-UI editable slice (sources added/edited).
type RuntimeConfig struct {
Sources []Source `json:"sources"`
DeletedSources []string `json:"deleted_sources,omitempty"`
DeletedAdapters []string `json:"deleted_adapters,omitempty"`
Keys []GWKey `json:"keys,omitempty"`
Auto []ModelScope `json:"auto,omitempty"`
}
// GWKey is a gateway API key persisted in the runtime store. Role is "admin"
// (full management) or "user" (sees only its own key); Models is the allowed
// model scope with per-model token quota (0 = unlimited).
type GWKey struct {
Key string `json:"key"`
Role string `json:"role"`
Name string `json:"name,omitempty"`
Models []ModelScope `json:"models,omitempty"`
Note string `json:"note,omitempty"`
CreatedAt int64 `json:"created_at,omitempty"`
}
// ModelScope is one allowed model for a key, or one AUTO scheduling slot,
// with an optional token quota and reset period. TokenQuota 0 = unlimited;
// Period "" = never resets; "hour"/"week"/"month" are fixed windows; "nhour"
// uses Hours as the window length in hours.
type ModelScope struct {
Model string `json:"model"`
Source string `json:"source,omitempty"` // optional: pin to one upstream source; "" = any source
Tier int `json:"tier,omitempty"`
TokenQuota int64 `json:"token_quota"`
Period string `json:"period,omitempty"`
Hours int64 `json:"hours,omitempty"`
}
// UnmarshalJSON accepts both the legacy "model-id" string form and the
// {"model":"...","token_quota":N} object form so old runtime files keep
// loading.
func (m *ModelScope) UnmarshalJSON(b []byte) error {
var s string
if err := json.Unmarshal(b, &s); err == nil {
m.Model = s
return nil
}
var o struct {
Model string `json:"model"`
Source string `json:"source"`
Tier int `json:"tier"`
TokenQuota int64 `json:"token_quota"`
Period string `json:"period"`
Hours int64 `json:"hours"`
}
if err := json.Unmarshal(b, &o); err != nil {
return err
}
m.Model = o.Model
m.Source = o.Source
m.Tier = o.Tier
m.TokenQuota = o.TokenQuota
m.Period = o.Period
m.Hours = o.Hours
return nil
}