provider

package
v0.4.21 Latest Latest
Warning

This package is not in the latest version of its module.

 Go to latest Published: Jun 23, 2026 License: MIT Imports: 16 Imported by: 0

Documentation

Index

Constants

View Source 
const (
	DefaultTimeoutDuration = 180 * time.Second
	DefaultMaxRetries      = 3
	DefaultRetryDelay      = 1 * time.Second
	MaxIdleConns           = 100
	MaxIdleConnsPerHost    = 10
	IdleConnTimeout        = 90 * time.Second
)

HTTP client configuration constants

View Source 
const LargeFileThreshold = 1024 * 1024 // 1MB

LargeFileThreshold is the size (bytes) above which files should use URL instead of base64

Variables

This section is empty.

Functions

func ConvertContentPartsToMap added in v0.4.19 

func ConvertContentPartsToMap(parts []types.ContentPart, config *ConvertConfig) []map[string]interface{}

ConvertContentPartsToMap converts content parts using the provider's configuration

func ConvertMessages 

func ConvertMessages(messages []types.Message) []map[string]interface{}

ConvertMessages converts internal Message type to OpenAI-compatible format

func ConvertMessagesForProvider added in v0.4.19 

func ConvertMessagesForProvider(messages []types.Message, bp *BaseProvider) []map[string]interface{}

ConvertMessagesForProvider converts messages using the provider's configuration

func ConvertMessagesWithConfig added in v0.4.19 

func ConvertMessagesWithConfig(messages []types.Message, config *ConvertConfig) []map[string]interface{}

ConvertMessagesWithConfig converts messages with custom config

func IsModelSupported added in v0.4.14 

func IsModelSupported(p Provider, model string) bool

IsModelSupported checks if a model is supported by the provider

func ModelSupportsVision added in v0.4.19 

func ModelSupportsVision(modelName string) bool

ModelSupportsVision checks if a model supports vision (image_url format)

func NewHTTPClient 

func NewHTTPClient(opts ...HTTPOption) *http.Client

NewHTTPClient creates a new HTTP client with optimized defaults

func NewVLLMProvider 

func NewVLLMProvider(baseURL, model string) *vLLMProvider

func ParseStreamResponse 

func ParseStreamResponse(body io.Reader, handler StreamHandler) error

ParseStreamResponse parses a standard OpenAI-compatible streaming response

func ParseStreamResponseWithTools 

func ParseStreamResponseWithTools(body io.Reader, handler StreamHandler) error

ParseStreamResponseWithTools parses streaming response handling tool calls

func ParseStreamWithParser 

func ParseStreamWithParser(body io.Reader, handler StreamHandler, parser StreamParser, config *StreamConfig) error

ParseStreamWithParser parses streaming response using a specific parser

func StreamWithTimeout 

func StreamWithTimeout(ctx context.Context, duration time.Duration, body io.Reader, handler StreamHandler) error

StreamWithTimeout performs streaming with automatic timeout

Types

type AnthropicProvider 

type AnthropicProvider struct {
	ConvertConfig *ConvertConfig
	// contains filtered or unexported fields
}

AnthropicProvider implements the Anthropic Claude API

func NewAnthropicProvider 

func NewAnthropicProvider(apiKey, model string) *AnthropicProvider

NewAnthropicProvider creates a new Anthropic provider

func (*AnthropicProvider) Chat 

func (p *AnthropicProvider) Chat(ctx context.Context, messages []Message) (*ChatResponse, error)

Chat implements the Provider interface

func (*AnthropicProvider) ChatWithTools 

func (p *AnthropicProvider) ChatWithTools(ctx context.Context, messages []Message, tools []map[string]interface{}) (*ChatResponse, error)

ChatWithTools implements the ToolCaller interface

func (*AnthropicProvider) GetCapabilities 

func (p *AnthropicProvider) GetCapabilities() *Capabilities

GetCapabilities returns the capabilities of Anthropic

func (*AnthropicProvider) GetConvertConfig added in v0.4.19 

func (p *AnthropicProvider) GetConvertConfig() *ConvertConfig

GetConvertConfig implements ConvertConfigProvider

func (*AnthropicProvider) Name 

func (p *AnthropicProvider) Name() string

func (*AnthropicProvider) SetConvertConfig added in v0.4.19 

func (p *AnthropicProvider) SetConvertConfig(config *ConvertConfig)

SetConvertConfig implements ConvertConfigProvider

func (*AnthropicProvider) Stream 

func (p *AnthropicProvider) Stream(ctx context.Context, messages []Message, handler StreamHandler) error

Stream implements the Streamer interface

func (*AnthropicProvider) StreamWithTools 

func (p *AnthropicProvider) StreamWithTools(ctx context.Context, messages []Message, tools []map[string]interface{}, handler StreamHandler) error

StreamWithTools implements the StreamingToolCaller interface

type AnthropicStreamParser 

type AnthropicStreamParser struct{}

AnthropicStreamParser parses Anthropic SSE streams

func (*AnthropicStreamParser) GetFormat 

func (p *AnthropicStreamParser) GetFormat() string

func (*AnthropicStreamParser) IsDone 

func (p *AnthropicStreamParser) IsDone(line string) bool

func (*AnthropicStreamParser) Parse 

func (p *AnthropicStreamParser) Parse(line string) (*StreamResponse, error)

Parse handles Anthropic's SSE format

type BaseProvider 

type BaseProvider struct {
	// HTTP client configuration
	Client     *http.Client
	BaseURL    string
	APIVersion string
	Timeout    time.Duration

	// Authentication
	APIKey       string
	APIKeyPrefix string // e.g., "Bearer" for OAuth or "sk-ant-api03" for Anthropic

	// Retry and resilience
	RetryConfig  *RetryConfig
	RetryEnabled bool

	// Health and metrics
	Metrics      *RequestMetrics
	HealthStatus HealthStatus

	// Configuration validation
	Validated bool

	// File conversion settings
	ConvertCfg *ConvertConfig
	// contains filtered or unexported fields
}

BaseProvider provides common HTTP functionality and infrastructure for all providers Inspired by Cortex Agent's Provider Resolution pattern with enhanced resilience features

func NewBaseProvider 

func NewBaseProvider(baseURL string) *BaseProvider

NewBaseProvider creates a new base provider with default configuration

func NewBaseProviderWithConfig 

func NewBaseProviderWithConfig(baseURL string, timeout time.Duration) *BaseProvider

NewBaseProviderWithConfig creates a base provider with custom configuration

func (*BaseProvider) DoRequest 

func (bp *BaseProvider) DoRequest(ctx context.Context, method, url string, body interface{}, headers map[string]string) ([]byte, int, error)

DoRequest performs an HTTP request with error handling and optional retry

func (*BaseProvider) DoStreamRequest 

func (bp *BaseProvider) DoStreamRequest(ctx context.Context, url string, body interface{}, headers map[string]string) (*http.Response, error)

DoStreamRequest performs a streaming HTTP request

func (*BaseProvider) GetHealthState 

func (bp *BaseProvider) GetHealthState() CircuitState

GetHealthState returns the current circuit breaker state

func (*BaseProvider) GetMetrics 

func (bp *BaseProvider) GetMetrics() (total, failed int64, avgLatency time.Duration, usage Usage)

GetMetrics returns a snapshot of provider metrics

func (*BaseProvider) IsHealthy 

func (bp *BaseProvider) IsHealthy() bool

IsHealthy returns whether the provider can accept requests

func (*BaseProvider) ParseAPIError 

func (bp *BaseProvider) ParseAPIError(body []byte, statusCode int) error

ParseAPIError parses API error responses with support for multiple formats

func (*BaseProvider) RecordFailure 

func (bp *BaseProvider) RecordFailure()

RecordFailure records a failed request for circuit breaker tracking

func (*BaseProvider) RecordSuccess 

func (bp *BaseProvider) RecordSuccess()

RecordSuccess records a successful request for circuit breaker tracking

func (*BaseProvider) ResetCircuitBreaker added in v0.4.14 

func (bp *BaseProvider) ResetCircuitBreaker()

ResetCircuitBreaker resets the circuit breaker to closed state

func (*BaseProvider) TransitionToHalfOpen 

func (bp *BaseProvider) TransitionToHalfOpen() bool

TransitionToHalfOpen attempts to transition from open to half-open state

func (*BaseProvider) Validate 

func (bp *BaseProvider) Validate() error

Validate checks if the provider is properly configured

func (*BaseProvider) WithAPIKey 

func (bp *BaseProvider) WithAPIKey(apiKey string) *BaseProvider

WithAPIKey sets the API key for authentication

func (*BaseProvider) WithAPIKeyPrefix 

func (bp *BaseProvider) WithAPIKeyPrefix(prefix string) *BaseProvider

WithAPIKeyPrefix sets the API key prefix (e.g., "Bearer", "sk-ant-api03")

func (*BaseProvider) WithConvertConfig added in v0.4.19 

func (bp *BaseProvider) WithConvertConfig(cfg *ConvertConfig) *BaseProvider

WithConvertConfig sets the file conversion configuration

func (*BaseProvider) WithRetryDisabled 

func (bp *BaseProvider) WithRetryDisabled() *BaseProvider

WithRetryDisabled disables automatic retries

type BaseStreamProvider 

type BaseStreamProvider struct {
	*OpenAICompatibleProvider
	StreamEndpoint string
}

BaseStreamProvider provides common streaming functionality for non-OpenAI-compatible providers

func NewBaseStreamProvider 

func NewBaseStreamProvider(name, apiKey, baseURL, model string) *BaseStreamProvider

NewBaseStreamProvider creates a new provider with streaming support

func (*BaseStreamProvider) Stream 

func (bp *BaseStreamProvider) Stream(ctx context.Context, messages []types.Message, handler StreamHandler) error

Stream implements the Streamer interface for base provider

func (*BaseStreamProvider) StreamWithTools 

func (bp *BaseStreamProvider) StreamWithTools(ctx context.Context, messages []types.Message, tools []map[string]interface{}, handler StreamHandler) error

StreamWithTools implements the StreamingToolCaller interface for base provider

type BufferedStreamReader 

type BufferedStreamReader struct {
	// contains filtered or unexported fields
}

BufferedStreamReader provides buffering for stream responses

func NewBufferedStreamReader 

func NewBufferedStreamReader(reader io.Reader, config *StreamConfig) *BufferedStreamReader

NewBufferedStreamReader creates a new buffered stream reader

func (*BufferedStreamReader) Close 

func (b *BufferedStreamReader) Close() error

Close closes the reader

func (*BufferedStreamReader) Read 

func (b *BufferedStreamReader) Read(p []byte) (n int, err error)

Read implements io.Reader

type Capabilities 

type Capabilities struct {
	ToolCalling    bool `json:"tool_calling"`
	Streaming      bool `json:"streaming"`
	StreamingTools bool `json:"streaming_tools"`
	MultiModal     bool `json:"multi_modal"`
	Vision         bool `json:"vision"`
}

Capabilities declares what features a provider supports

func DefaultCapabilities 

func DefaultCapabilities() *Capabilities

DefaultCapabilities returns default capabilities for OpenAI-compatible providers

func GetCapabilities 

func GetCapabilities(p Provider) *Capabilities

GetCapabilities returns the capabilities of a provider, or default if not specified

func NoStreamCapabilities 

func NoStreamCapabilities() *Capabilities

NoStreamCapabilities returns capabilities without streaming

type CapableProvider 

type CapableProvider interface {
	GetCapabilities() *Capabilities
}

CapableProvider is an optional interface for providers that declare their capabilities

type ChatResponse 

type ChatResponse struct {
	Content   string           `json:"content"`
	ToolCalls []types.ToolCall `json:"tool_calls,omitempty"`
	Usage     *Usage           `json:"usage,omitempty"`
}

ChatResponse represents a chat response with optional usage info

type CircuitState 

type CircuitState int

CircuitState represents the circuit breaker state 

const (
	CircuitClosed   CircuitState = iota // Normal operation
	CircuitOpen                         // Failing, reject requests
	CircuitHalfOpen                     // Testing recovery
)

func (CircuitState) String 

func (s CircuitState) String() string

type CohereProvider 

type CohereProvider struct {
	ConvertConfig *ConvertConfig
	// contains filtered or unexported fields
}

CohereProvider implements the Cohere API

func NewCohereProvider 

func NewCohereProvider(apiKey, baseURL, model string) *CohereProvider

NewCohereProvider creates a new Cohere provider

func (*CohereProvider) Chat 

func (p *CohereProvider) Chat(ctx context.Context, messages []Message) (*ChatResponse, error)

Chat implements the Provider interface

func (*CohereProvider) ChatWithTools 

func (p *CohereProvider) ChatWithTools(ctx context.Context, messages []Message, tools []map[string]interface{}) (*ChatResponse, error)

ChatWithTools implements the ToolCaller interface

func (*CohereProvider) GetCapabilities 

func (p *CohereProvider) GetCapabilities() *Capabilities

GetCapabilities returns the capabilities of Cohere

func (*CohereProvider) GetConvertConfig added in v0.4.19 

func (p *CohereProvider) GetConvertConfig() *ConvertConfig

GetConvertConfig implements ConvertConfigProvider

func (*CohereProvider) Name 

func (p *CohereProvider) Name() string

func (*CohereProvider) SetConvertConfig added in v0.4.19 

func (p *CohereProvider) SetConvertConfig(config *ConvertConfig)

SetConvertConfig implements ConvertConfigProvider

func (*CohereProvider) Stream 

func (p *CohereProvider) Stream(ctx context.Context, messages []Message, handler StreamHandler) error

Stream implements the Streamer interface

type ConfigManager 

type ConfigManager struct {
	// contains filtered or unexported fields
}

ConfigManager manages provider configurations with hot reload support

func NewConfigManager 

func NewConfigManager() *ConfigManager

NewConfigManager creates a new configuration manager

func (*ConfigManager) Close 

func (cm *ConfigManager) Close()

Close stops watching and releases resources

func (*ConfigManager) Delete 

func (cm *ConfigManager) Delete(name string)

Delete removes a provider configuration

func (*ConfigManager) EnableHotReload 

func (cm *ConfigManager) EnableHotReload(onChange func(string)) error

EnableHotReload enables watching the config file for changes

func (*ConfigManager) Get 

func (cm *ConfigManager) Get(name string) (*ProviderConfig, error)

Get returns a provider configuration

func (*ConfigManager) List 

func (cm *ConfigManager) List() []*ProviderConfig

List returns all provider configurations

func (*ConfigManager) LoadFromFile 

func (cm *ConfigManager) LoadFromFile(path string) error

LoadFromFile loads provider configurations from a JSON file

func (*ConfigManager) Save 

func (cm *ConfigManager) Save() error

Save saves the configuration to file

func (*ConfigManager) Set 

func (cm *ConfigManager) Set(pc *ProviderConfig) error

Set sets a provider configuration

type ConvertConfig added in v0.4.19 

type ConvertConfig struct {
	UploadURLPrefix string       // Public URL prefix for uploaded files
	Strategy        FileStrategy // File conversion strategy
	StrategyName    string       // String representation of strategy ("auto", "url", "base64")
	SupportVision   bool         // Whether the model supports vision (image_url format)
}

ConvertConfig holds conversion settings

func DefaultConvertConfig added in v0.4.19 

func DefaultConvertConfig() *ConvertConfig

DefaultConvertConfig returns default conversion config

type ConvertConfigProvider added in v0.4.19 

type ConvertConfigProvider interface {
	SetConvertConfig(cfg *ConvertConfig)
}

ConvertConfigProvider is an optional interface for providers that support file conversion config

type DashScopeProvider 

type DashScopeProvider struct {
	*BaseProvider
	// contains filtered or unexported fields
}

DashScopeProvider 阿里云DashScope (兼容OpenAI格式)

func NewDashScopeProvider 

func NewDashScopeProvider(apiKey, baseURL, model string) *DashScopeProvider

func (*DashScopeProvider) Chat 

func (p *DashScopeProvider) Chat(ctx context.Context, messages []Message) (*ChatResponse, error)

func (*DashScopeProvider) ChatWithTools 

func (p *DashScopeProvider) ChatWithTools(ctx context.Context, messages []Message, tools []map[string]interface{}) (*ChatResponse, error)

ChatWithTools implements the ToolCaller interface for DashScope

func (*DashScopeProvider) Name 

func (p *DashScopeProvider) Name() string

type DeepSeekProvider 

type DeepSeekProvider struct {
	*OpenAICompatibleProvider
}

DeepSeekProvider implements the DeepSeek API (OpenAI-compatible)

func NewDeepSeekProvider 

func NewDeepSeekProvider(apiKey, model string, userModels []ModelInfo) *DeepSeekProvider

NewDeepSeekProvider creates a new DeepSeek provider If userModels is provided (non-nil and non-empty), it will be used; otherwise defaults are loaded

func (*DeepSeekProvider) GetCapabilities 

func (p *DeepSeekProvider) GetCapabilities() *Capabilities

GetCapabilities returns the capabilities of DeepSeek

func (*DeepSeekProvider) Name 

func (p *DeepSeekProvider) Name() string

func (*DeepSeekProvider) Stream 

func (p *DeepSeekProvider) Stream(ctx context.Context, messages []types.Message, handler StreamHandler) error

Stream implements the Streamer interface (inherited from OpenAICompatibleProvider)

func (*DeepSeekProvider) StreamWithTools 

func (p *DeepSeekProvider) StreamWithTools(ctx context.Context, messages []types.Message, tools []map[string]interface{}, handler StreamHandler) error

StreamWithTools implements the StreamingToolCaller interface

type DoubaoProvider 

type DoubaoProvider struct {
	*OpenAICompatibleProvider
	// contains filtered or unexported fields
}

DoubaoProvider implements the Doubao (Volcengine) API using OpenAI-compatible format.

func NewDoubaoProvider 

func NewDoubaoProvider(apiKey, baseURL, model string) *DoubaoProvider

NewDoubaoProvider creates a new Doubao provider

func NewDoubaoProviderWithEndpoint 

func NewDoubaoProviderWithEndpoint(apiKey, endpointID string) *DoubaoProvider

NewDoubaoProviderWithEndpoint creates a Doubao provider with custom endpoint

func (*DoubaoProvider) GetCapabilities 

func (p *DoubaoProvider) GetCapabilities() *Capabilities

GetCapabilities returns the capabilities of Doubao

func (*DoubaoProvider) Name 

func (p *DoubaoProvider) Name() string

type ExtendedChatResponse 

type ExtendedChatResponse struct {
	Content   string        `json:"content"`
	ToolCalls []interface{} `json:"tool_calls,omitempty"`
	Usage     *Usage        `json:"usage,omitempty"`
}

ChatResponse extends types.ChatResponse with usage info

type FileStrategy added in v0.4.19 

type FileStrategy int

FileStrategy defines how files should be converted 

const (
	FileStrategyAuto   FileStrategy = iota // Auto-select based on file type
	FileStrategyURL                        // Prefer URL references (for large files)
	FileStrategyBase64                     // Always use base64
)

func ParseFileStrategy added in v0.4.19 

func ParseFileStrategy(s string) FileStrategy

ParseFileStrategy converts string to FileStrategy

type GeminiProvider 

type GeminiProvider struct {
	ConvertConfig *ConvertConfig
	// contains filtered or unexported fields
}

GeminiProvider implements the Google Gemini API

func NewGeminiProvider 

func NewGeminiProvider(apiKey, baseURL, model string) *GeminiProvider

NewGeminiProvider creates a new Gemini provider

func (*GeminiProvider) Chat 

func (p *GeminiProvider) Chat(ctx context.Context, messages []Message) (*ChatResponse, error)

Chat implements the Provider interface

func (*GeminiProvider) ChatWithTools 

func (p *GeminiProvider) ChatWithTools(ctx context.Context, messages []Message, tools []map[string]interface{}) (*ChatResponse, error)

ChatWithTools implements the ToolCaller interface

func (*GeminiProvider) GetCapabilities 

func (p *GeminiProvider) GetCapabilities() *Capabilities

GetCapabilities returns the capabilities of Gemini

func (*GeminiProvider) GetConvertConfig added in v0.4.19 

func (p *GeminiProvider) GetConvertConfig() *ConvertConfig

GetConvertConfig implements ConvertConfigProvider

func (*GeminiProvider) Name 

func (p *GeminiProvider) Name() string

func (*GeminiProvider) SetConvertConfig added in v0.4.19 

func (p *GeminiProvider) SetConvertConfig(config *ConvertConfig)

SetConvertConfig implements ConvertConfigProvider

func (*GeminiProvider) Stream 

func (p *GeminiProvider) Stream(ctx context.Context, messages []Message, handler StreamHandler) error

Stream implements the Streamer interface

type GroqProvider 

type GroqProvider struct {
	*OpenAICompatibleProvider
}

GroqProvider implements the Groq API (fast inference) using OpenAI-compatible format. Groq uses OpenAI-compatible format but with different base URL and specific models.

func NewGroqProvider 

func NewGroqProvider(apiKey, baseURL, model string) *GroqProvider

NewGroqProvider creates a new Groq provider

func (*GroqProvider) GetCapabilities 

func (p *GroqProvider) GetCapabilities() *Capabilities

GetCapabilities returns the capabilities of Groq

func (*GroqProvider) Name 

func (p *GroqProvider) Name() string

type HTTPOption 

type HTTPOption func(*http.Client)

HTTPOption is a functional option for HTTP client configuration

func WithTimeout 

func WithTimeout(d time.Duration) HTTPOption

WithTimeout sets the timeout for the HTTP client

func WithTransport 

func WithTransport(t *http.Transport) HTTPOption

WithTransport sets a custom transport for the HTTP client

type HealthCheck 

type HealthCheck struct {
	Provider  string
	URL       string
	LastCheck time.Time
	Healthy   bool
	Failures  int
	Successes int
	// contains filtered or unexported fields
}

HealthCheck represents a health check for a provider

type HealthStatus 

type HealthStatus struct {
	State       CircuitState
	Failures    int
	Successes   int
	LastFailure time.Time
	LastSuccess time.Time
	CoolingDown bool
	CooledAt    time.Time
}

HealthStatus tracks provider health state for circuit breaker

type HunyuanProvider 

type HunyuanProvider struct {
	*OpenAICompatibleProvider
}

HunyuanProvider implements the Tencent Hunyuan API using OpenAI-compatible format.

func NewHunyuanProvider 

func NewHunyuanProvider(apiKey, baseURL, model string) *HunyuanProvider

NewHunyuanProvider creates a new Hunyuan provider

func (*HunyuanProvider) GetCapabilities 

func (p *HunyuanProvider) GetCapabilities() *Capabilities

GetCapabilities returns the capabilities of Hunyuan

func (*HunyuanProvider) Name 

func (p *HunyuanProvider) Name() string

type HuoshanProvider 

type HuoshanProvider struct {
	*OpenAICompatibleProvider
}

HuoshanProvider 火山方舟 (字节跳动，兼容OpenAI格式)

func NewHuoshanProvider 

func NewHuoshanProvider(apiKey, baseURL, model string) *HuoshanProvider

NewHuoshanProvider creates a new Huoshan provider

func (*HuoshanProvider) GetCapabilities 

func (p *HuoshanProvider) GetCapabilities() *Capabilities

GetCapabilities returns the capabilities of Huoshan

func (*HuoshanProvider) Name 

func (p *HuoshanProvider) Name() string

type KimiProvider 

type KimiProvider struct {
	*OpenAICompatibleProvider
}

KimiProvider 月之暗面 Kimi (兼容OpenAI格式)

func NewKimiProvider 

func NewKimiProvider(apiKey, baseURL, model string) *KimiProvider

NewKimiProvider creates a new Kimi provider

func (*KimiProvider) GetCapabilities 

func (p *KimiProvider) GetCapabilities() *Capabilities

GetCapabilities returns the capabilities of Kimi

func (*KimiProvider) Name 

func (p *KimiProvider) Name() string

type LoadBalancingStrategy 

type LoadBalancingStrategy int

LoadBalancingStrategy defines how to select providers 

const (
	// RoundRobin selects providers in rotation
	RoundRobin LoadBalancingStrategy = iota
	// Random selects a random provider
	Random
	// Weighted selects based on configured weights
	Weighted
	// LeastUsed selects the provider with least usage
	LeastUsed
)

type Message 

type Message = types.Message

Message is an alias for types.Message

type MiMoProvider 

type MiMoProvider struct {
	*OpenAICompatibleProvider
}

MiMoProvider implements the Xiaomi MiMo API using OpenAI-compatible format.

func NewMiMoProvider 

func NewMiMoProvider(apiKey, baseURL, model string) *MiMoProvider

NewMiMoProvider creates a new MiMo provider

func (*MiMoProvider) GetCapabilities 

func (p *MiMoProvider) GetCapabilities() *Capabilities

GetCapabilities returns the capabilities of MiMo

func (*MiMoProvider) Name 

func (p *MiMoProvider) Name() string

type MiniMaxProvider 

type MiniMaxProvider struct {
	*OpenAICompatibleProvider
}

MiniMaxProvider MiniMax (兼容OpenAI格式)

func NewMiniMaxProvider 

func NewMiniMaxProvider(apiKey, baseURL, model string) *MiniMaxProvider

NewMiniMaxProvider creates a new MiniMax provider

func (*MiniMaxProvider) GetCapabilities 

func (p *MiniMaxProvider) GetCapabilities() *Capabilities

GetCapabilities returns the capabilities of MiniMax

func (*MiniMaxProvider) Name 

func (p *MiniMaxProvider) Name() string

type MistralProvider 

type MistralProvider struct {
	*OpenAICompatibleProvider
}

MistralProvider implements the Mistral AI API using OpenAI-compatible format.

func NewMistralProvider 

func NewMistralProvider(apiKey, baseURL, model string) *MistralProvider

NewMistralProvider creates a new Mistral AI provider

func (*MistralProvider) GetCapabilities 

func (p *MistralProvider) GetCapabilities() *Capabilities

GetCapabilities returns the capabilities of Mistral AI

func (*MistralProvider) Name 

func (p *MistralProvider) Name() string

type ModelInfo added in v0.4.14 

type ModelInfo struct {
	ID          string `json:"id"`          // Model ID used in API calls
	Name        string `json:"name"`        // Human-readable name
	Description string `json:"description"` // Model description
	ContextLen  int    `json:"context_len"` // Context window size (0 = unknown)
}

ModelInfo represents information about a supported model

func GetDefaultModels added in v0.4.14 

func GetDefaultModels(providerName string) []ModelInfo

GetDefaultModels returns the default models for known providers

type Modeler added in v0.4.14 

type Modeler interface {
	// SetModel sets the current model. Returns error if model is not supported.
	SetModel(model string) error
	// GetModel returns the current model ID.
	GetModel() string
	// ListModels returns the list of supported models.
	ListModels() []ModelInfo
}

Modeler is an optional interface for providers that support multiple models. Providers implementing this interface allow dynamic model switching.

func GetModeler added in v0.4.14 

func GetModeler(p Provider) (Modeler, bool)

GetModeler returns the Modeler interface if supported by the provider

type MoonshotProvider 

type MoonshotProvider struct {
	*OpenAICompatibleProvider
}

MoonshotProvider implements the Moonshot (Kimi) AI API using OpenAI-compatible format. Note: Moonshot and Kimi use the same API endpoint (https://api.moonshot.cn/v1). This provider is kept for backward compatibility with existing configurations.

func NewMoonshotProvider 

func NewMoonshotProvider(apiKey, baseURL, model string) *MoonshotProvider

NewMoonshotProvider creates a new Moonshot provider

func (*MoonshotProvider) GetCapabilities 

func (p *MoonshotProvider) GetCapabilities() *Capabilities

GetCapabilities returns the capabilities of Moonshot

func (*MoonshotProvider) Name 

func (p *MoonshotProvider) Name() string

type OllamaProvider 

type OllamaProvider struct {
	ConvertConfig *ConvertConfig
	// contains filtered or unexported fields
}

func NewOllamaProvider 

func NewOllamaProvider(baseURL, model string) *OllamaProvider

func (*OllamaProvider) Chat 

func (p *OllamaProvider) Chat(ctx context.Context, messages []Message) (*ChatResponse, error)

func (*OllamaProvider) ChatWithTools 

func (p *OllamaProvider) ChatWithTools(ctx context.Context, messages []Message, tools []map[string]interface{}) (*ChatResponse, error)

ChatWithTools implements the ToolCaller interface for Ollama

func (*OllamaProvider) GetConvertConfig added in v0.4.19 

func (p *OllamaProvider) GetConvertConfig() *ConvertConfig

GetConvertConfig implements ConvertConfigProvider

func (*OllamaProvider) Name 

func (p *OllamaProvider) Name() string

func (*OllamaProvider) SetConvertConfig added in v0.4.19 

func (p *OllamaProvider) SetConvertConfig(config *ConvertConfig)

SetConvertConfig implements ConvertConfigProvider

type OpenAICompatibleProvider 

type OpenAICompatibleProvider struct {
	*BaseProvider
	// contains filtered or unexported fields
}

OpenAICompatibleProvider provides OpenAI-compatible API functionality

func NewOpenAICompatibleProvider 

func NewOpenAICompatibleProvider(name, apiKey, baseURL, model string, userModels []ModelInfo) *OpenAICompatibleProvider

NewOpenAICompatibleProvider creates a new OpenAI-compatible provider If userModels is provided (non-nil), it will be used; otherwise defaults are loaded

func NewOpenAICompatibleProviderWithDefaults added in v0.4.14 

func NewOpenAICompatibleProviderWithDefaults(name, apiKey, baseURL, model string) *OpenAICompatibleProvider

NewOpenAICompatibleProviderWithDefaults creates a new OpenAI-compatible provider with default models This is a convenience function for providers that don't need custom model lists

func (*OpenAICompatibleProvider) Chat 

func (p *OpenAICompatibleProvider) Chat(ctx context.Context, messages []types.Message) (*ChatResponse, error)

Chat implements the Provider interface

func (*OpenAICompatibleProvider) ChatWithTools 

func (p *OpenAICompatibleProvider) ChatWithTools(ctx context.Context, messages []types.Message, tools []map[string]interface{}) (*ChatResponse, error)

ChatWithTools implements the ToolCaller interface

func (*OpenAICompatibleProvider) GetModel added in v0.4.14 

func (p *OpenAICompatibleProvider) GetModel() string

GetModel returns the current model ID.

func (*OpenAICompatibleProvider) ListModels added in v0.4.14 

func (p *OpenAICompatibleProvider) ListModels() []ModelInfo

ListModels returns the list of supported models.

func (*OpenAICompatibleProvider) Name 

func (p *OpenAICompatibleProvider) Name() string

Name returns the provider name

func (*OpenAICompatibleProvider) SetConvertConfig added in v0.4.19 

func (p *OpenAICompatibleProvider) SetConvertConfig(cfg *ConvertConfig)

SetConvertConfig implements ConvertConfigProvider

func (*OpenAICompatibleProvider) SetModel added in v0.4.14 

func (p *OpenAICompatibleProvider) SetModel(model string) error

SetModel sets the current model. If model is not in the list, it will be added.

func (*OpenAICompatibleProvider) SetModels added in v0.4.14 

func (p *OpenAICompatibleProvider) SetModels(models []ModelInfo)

SetModels sets the list of supported models.

func (*OpenAICompatibleProvider) Stream 

func (p *OpenAICompatibleProvider) Stream(ctx context.Context, messages []types.Message, handler StreamHandler) error

Stream implements the Streamer interface

func (*OpenAICompatibleProvider) StreamWithTools 

func (p *OpenAICompatibleProvider) StreamWithTools(ctx context.Context, messages []types.Message, tools []map[string]interface{}, handler StreamHandler) error

StreamWithTools implements the StreamingToolCaller interface

type OpenAIProvider 

type OpenAIProvider struct {
	*OpenAICompatibleProvider
}

OpenAIProvider implements the OpenAI API

func NewOpenAIProvider 

func NewOpenAIProvider(apiKey, baseURL, model string) *OpenAIProvider

NewOpenAIProvider creates a new OpenAI provider

func (*OpenAIProvider) GetCapabilities 

func (p *OpenAIProvider) GetCapabilities() *Capabilities

GetCapabilities returns the capabilities of OpenAI

func (*OpenAIProvider) Name 

func (p *OpenAIProvider) Name() string

type OpenAIStreamParser 

type OpenAIStreamParser struct{}

OpenAIStreamParser parses OpenAI-compatible SSE streams

func (*OpenAIStreamParser) GetFormat 

func (p *OpenAIStreamParser) GetFormat() string

func (*OpenAIStreamParser) IsDone 

func (p *OpenAIStreamParser) IsDone(line string) bool

func (*OpenAIStreamParser) Parse 

func (p *OpenAIStreamParser) Parse(line string) (*StreamResponse, error)

type OpenRouterProvider 

type OpenRouterProvider struct {
	*OpenAICompatibleProvider
}

OpenRouterProvider OpenRouter (兼容OpenAI格式)

func NewOpenRouterProvider 

func NewOpenRouterProvider(apiKey, baseURL, model string) *OpenRouterProvider

NewOpenRouterProvider creates a new OpenRouter provider

func (*OpenRouterProvider) GetCapabilities 

func (p *OpenRouterProvider) GetCapabilities() *Capabilities

GetCapabilities returns the capabilities of OpenRouter

func (*OpenRouterProvider) Name 

func (p *OpenRouterProvider) Name() string

type PerplexityProvider 

type PerplexityProvider struct {
	ConvertConfig *ConvertConfig
	// contains filtered or unexported fields
}

PerplexityProvider implements the Perplexity API

func NewPerplexityProvider 

func NewPerplexityProvider(apiKey, baseURL, model string) *PerplexityProvider

NewPerplexityProvider creates a new Perplexity provider

func (*PerplexityProvider) Chat 

func (p *PerplexityProvider) Chat(ctx context.Context, messages []Message) (*ChatResponse, error)

Chat implements the Provider interface

func (*PerplexityProvider) ChatWithTools 

func (p *PerplexityProvider) ChatWithTools(ctx context.Context, messages []Message, tools []map[string]interface{}) (*ChatResponse, error)

ChatWithTools implements the ToolCaller interface

func (*PerplexityProvider) GetCapabilities 

func (p *PerplexityProvider) GetCapabilities() *Capabilities

GetCapabilities returns the capabilities of Perplexity

func (*PerplexityProvider) GetConvertConfig added in v0.4.19 

func (p *PerplexityProvider) GetConvertConfig() *ConvertConfig

GetConvertConfig implements ConvertConfigProvider

func (*PerplexityProvider) Name 

func (p *PerplexityProvider) Name() string

func (*PerplexityProvider) SetConvertConfig added in v0.4.19 

func (p *PerplexityProvider) SetConvertConfig(config *ConvertConfig)

SetConvertConfig implements ConvertConfigProvider

func (*PerplexityProvider) Stream 

func (p *PerplexityProvider) Stream(ctx context.Context, messages []Message, handler StreamHandler) error

Stream implements the Streamer interface

type Provider 

type Provider interface {
	Chat(ctx context.Context, messages []Message) (*ChatResponse, error)
	Name() string
}

Provider is the interface for LLM providers.

type ProviderConfig 

type ProviderConfig struct {
	// Provider identification
	Name    string `json:"name"`
	Display string `json:"display"`

	// API configuration
	APIKey  string `json:"api_key,omitempty"`
	BaseURL string `json:"base_url,omitempty"`
	Model   string `json:"model,omitempty"`

	// Optional configuration
	APIVersion string            `json:"api_version,omitempty"`
	Extra      map[string]string `json:"extra,omitempty"`

	// Defaults
	DefaultModel string   `json:"default_model,omitempty"`
	Models       []string `json:"models,omitempty"`

	// Rate limiting
	MaxRPM int `json:"max_rpm,omitempty"` // requests per minute
	MaxTPM int `json:"max_tpm,omitempty"` // tokens per minute

	// Validation state
	Validated bool `json:"-"`

	// File path if loaded from file
	FilePath string `json:"-"`
}

ProviderConfig represents configuration for a provider

func (*ProviderConfig) Validate 

func (pc *ProviderConfig) Validate() error

Validate checks if the provider config is valid

type ProviderRouter 

type ProviderRouter struct {
	// contains filtered or unexported fields
}

ProviderRouter provides unified routing with model fallback, health checks, circuit breakers, and load balancing support

func NewProviderRouter 

func NewProviderRouter(config *RouterConfig) *ProviderRouter

NewProviderRouter creates a new enhanced provider router

func (*ProviderRouter) Close 

func (r *ProviderRouter) Close()

Close shuts down the router

func (*ProviderRouter) GetAllStats 

func (r *ProviderRouter) GetAllStats() map[string]RouterStats

GetAllStats returns stats for all providers

func (*ProviderRouter) GetHealthStatus 

func (r *ProviderRouter) GetHealthStatus(provider string) (healthy bool, lastCheck time.Time)

GetHealthStatus returns the health status of a provider

func (*ProviderRouter) GetProvider 

func (r *ProviderRouter) GetProvider(name string) (Provider, error)

GetProvider returns a provider by name

func (*ProviderRouter) GetProviderForModel 

func (r *ProviderRouter) GetProviderForModel(model string) (Provider, string, error)

GetProviderForModel returns the appropriate provider for a given model

func (*ProviderRouter) GetStats 

func (r *ProviderRouter) GetStats(provider string) (stats RouterStats)

GetStats returns routing statistics

func (*ProviderRouter) ListProviders 

func (r *ProviderRouter) ListProviders() []string

ListProviders returns all registered provider names

func (*ProviderRouter) Register 

func (r *ProviderRouter) Register(name string, provider Provider)

Register registers a provider with the router

func (*ProviderRouter) RegisterFallback 

func (r *ProviderRouter) RegisterFallback(provider string, fallbacks []string)

RegisterFallback registers fallback providers for a given provider

func (*ProviderRouter) RegisterModelRouting 

func (r *ProviderRouter) RegisterModelRouting(modelPrefix, provider string)

RegisterModelRouting registers model prefix -> provider mapping

func (*ProviderRouter) RouteWithFallback 

func (r *ProviderRouter) RouteWithFallback(ctx context.Context, model string,
	execute func(ctx context.Context, provider Provider) error) error

RouteWithFallback executes a request with automatic fallback

func (*ProviderRouter) SelectProvider 

func (r *ProviderRouter) SelectProvider(providers []string) string

SelectProvider selects a provider based on load balancing strategy

func (*ProviderRouter) SetLoadBalancing 

func (r *ProviderRouter) SetLoadBalancing(strategy LoadBalancingStrategy)

SetLoadBalancing sets the load balancing strategy

type ProviderToolFormat 

type ProviderToolFormat struct {
	Format     string // "openai", "anthropic", "gemini"
	UsesTools  bool   // Uses tools array in request
	UsesChoice bool   // Uses choice.tool_call format
}

ProviderToolFormat specifies the expected tool format for a provider

func GetProviderToolFormat 

func GetProviderToolFormat(provider string) ProviderToolFormat

GetProviderToolFormat returns the expected tool format for a provider

type Registry 

type Registry struct {
	// contains filtered or unexported fields
}

Registry manages provider instances.

func NewRegistry 

func NewRegistry() *Registry

NewRegistry creates a new provider registry

func (*Registry) Get 

func (r *Registry) Get(name string) (Provider, error)

Get returns a provider by name

func (*Registry) List 

func (r *Registry) List() []string

List returns all registered provider names

func (*Registry) Register 

func (r *Registry) Register(p Provider)

Register registers a provider in the registry

type RequestMetrics 

type RequestMetrics struct {
	TotalRequests  int64
	FailedRequests int64
	TotalLatency   time.Duration
	TokenUsage     Usage
	// contains filtered or unexported fields
}

RequestMetrics tracks request statistics

func (*RequestMetrics) AddFailure 

func (m *RequestMetrics) AddFailure()

AddFailure records a failed request

func (*RequestMetrics) AddRequest 

func (m *RequestMetrics) AddRequest(latency time.Duration, tokens *Usage)

AddRequest records a successful request

func (*RequestMetrics) GetStats 

func (m *RequestMetrics) GetStats() (total, failed int64, avgLatency time.Duration, usage Usage)

GetStats returns current metrics snapshot

type RetryConfig 

type RetryConfig struct {
	MaxRetries     int
	InitialDelay   time.Duration
	MaxDelay       time.Duration
	BackoffFactor  float64
	RetryableCodes []int // HTTP status codes that should be retried
	// Backoff strategy from retry package
	Backoff retry.BackoffStrategy
}

RetryConfig configures retry behavior - now uses internal/retry package

func DefaultRetryConfig 

func DefaultRetryConfig() *RetryConfig

DefaultRetryConfig returns sensible retry defaults using exponential backoff

type RouterConfig 

type RouterConfig struct {
	// Enable health checks
	HealthCheckEnabled bool
	// Health check interval
	HealthCheckInterval time.Duration
	// Timeout for health check requests
	HealthCheckTimeout time.Duration
	// Number of consecutive failures before marking unhealthy
	UnhealthyThreshold int
	// Number of consecutive successes before marking healthy
	HealthyThreshold int

	// Load balancing strategy
	LoadBalancing LoadBalancingStrategy

	// Fallback configuration
	EnableFallback bool
	FallbackDelay  time.Duration

	// Circuit breaker settings
	CircuitBreakerEnabled bool
	FailureThreshold      int
	RecoveryTimeout       time.Duration
}

RouterConfig configures router behavior

func DefaultRouterConfig 

func DefaultRouterConfig() *RouterConfig

DefaultRouterConfig returns sensible router defaults

type RouterStats 

type RouterStats struct {
	TotalRequests      int64
	FailedRequests     int64
	SuccessRequests    int64
	RequestsByProvider map[string]int64
	LastUsed           time.Time
	LastError          error
	// contains filtered or unexported fields
}

RouterStats tracks routing statistics

type StreamAccumulator 

type StreamAccumulator struct {
	// contains filtered or unexported fields
}

CreateStreamHandler creates a handler that accumulates responses

func NewStreamAccumulator 

func NewStreamAccumulator(handler StreamHandler) *StreamAccumulator

NewStreamAccumulator creates a new stream accumulator

func (*StreamAccumulator) GetContent 

func (sa *StreamAccumulator) GetContent() string

GetContent returns the accumulated content

func (*StreamAccumulator) GetToolCalls 

func (sa *StreamAccumulator) GetToolCalls() []types.ToolCall

GetToolCalls returns the accumulated tool calls

func (*StreamAccumulator) Handle 

func (sa *StreamAccumulator) Handle(resp *StreamResponse)

Handle processes a stream response

func (*StreamAccumulator) OnFinal 

func (sa *StreamAccumulator) OnFinal(handler StreamHandler) *StreamAccumulator

OnFinal sets a handler to call on stream completion

type StreamConfig 

type StreamConfig struct {
	// Timeout for reading chunks
	ReadTimeout time.Duration
	// Whether to accumulate content for final callback
	AccumulateContent bool
	// Buffer size for the reader
	BufferSize int
	// Heartbeat interval for keep-alive
	HeartbeatInterval time.Duration
}

StreamConfig configures streaming behavior

func DefaultStreamConfig 

func DefaultStreamConfig() *StreamConfig

DefaultStreamConfig returns sensible streaming defaults

type StreamContext 

type StreamContext struct {
	// contains filtered or unexported fields
}

StreamContext provides context-aware streaming with cancellation support

func NewStreamContext 

func NewStreamContext(parent context.Context) (*StreamContext, context.Context)

NewStreamContext creates a new streaming context with cancellation support

func (*StreamContext) Cancel 

func (sc *StreamContext) Cancel()

Cancel cancels the streaming operation

func (*StreamContext) Close 

func (sc *StreamContext) Close(err error)

Close marks the streaming as complete

func (*StreamContext) Wait 

func (sc *StreamContext) Wait() error

Wait waits for the streaming to complete

type StreamHandler 

type StreamHandler func(resp *StreamResponse)

StreamHandler is a callback function for handling streaming responses

func WrappedStreamHandler 

func WrappedStreamHandler(handler StreamHandler, accumulator *StreamAccumulator) StreamHandler

WrappedStreamHandler wraps a handler for accumulation

type StreamParser 

type StreamParser interface {
	// Parse parses a single SSE data line
	Parse(line string) (*StreamResponse, error)
	// IsDone checks if this line indicates stream completion
	IsDone(line string) bool
	// GetFormat returns the provider format name
	GetFormat() string
}

StreamParser handles SSE stream parsing for different provider formats

type StreamResponse 

type StreamResponse struct {
	Content   string           `json:"content,omitempty"`
	ToolCall  *types.ToolCall  `json:"tool_call,omitempty"`
	ToolCalls []types.ToolCall `json:"tool_calls,omitempty"`
	StreamTCs []StreamToolCall `json:"-"` // Internal use for delta tracking
	Done      bool             `json:"done"`
	Error     error            `json:"error,omitempty"`
	Usage     *Usage           `json:"usage,omitempty"`
}

StreamResponse represents a chunk of streaming response

type StreamToolCall 

type StreamToolCall struct {
	Index     int
	ID        string
	Type      string
	Name      string
	Arguments string
}

StreamToolCall represents a tool call with index for streaming delta

type Streamer 

type Streamer interface {
	Stream(ctx context.Context, messages []Message, handler StreamHandler) error
}

Streamer is an optional interface for providers that support streaming.

type StreamingToolCaller 

type StreamingToolCaller interface {
	StreamWithTools(ctx context.Context, messages []Message, tools []map[string]interface{}, handler StreamHandler) error
}

StreamingToolCaller is for providers that support both streaming and tool calling

type TogetherProvider 

type TogetherProvider struct {
	ConvertConfig *ConvertConfig
	// contains filtered or unexported fields
}

TogetherProvider implements the Together AI API

func NewTogetherProvider 

func NewTogetherProvider(apiKey, baseURL, model string) *TogetherProvider

NewTogetherProvider creates a new Together AI provider

func (*TogetherProvider) Chat 

func (p *TogetherProvider) Chat(ctx context.Context, messages []Message) (*ChatResponse, error)

Chat implements the Provider interface

func (*TogetherProvider) ChatWithTools 

func (p *TogetherProvider) ChatWithTools(ctx context.Context, messages []Message, tools []map[string]interface{}) (*ChatResponse, error)

ChatWithTools implements the ToolCaller interface

func (*TogetherProvider) GetCapabilities 

func (p *TogetherProvider) GetCapabilities() *Capabilities

GetCapabilities returns the capabilities of Together AI

func (*TogetherProvider) GetConvertConfig added in v0.4.19 

func (p *TogetherProvider) GetConvertConfig() *ConvertConfig

GetConvertConfig implements ConvertConfigProvider

func (*TogetherProvider) ListModels 

func (p *TogetherProvider) ListModels(ctx context.Context) ([]string, error)

ListModels returns available models from Together AI

func (*TogetherProvider) Name 

func (p *TogetherProvider) Name() string

func (*TogetherProvider) SetConvertConfig added in v0.4.19 

func (p *TogetherProvider) SetConvertConfig(config *ConvertConfig)

SetConvertConfig implements ConvertConfigProvider

func (*TogetherProvider) Stream 

func (p *TogetherProvider) Stream(ctx context.Context, messages []Message, handler StreamHandler) error

Stream implements the Streamer interface

type ToolCaller 

type ToolCaller interface {
	ChatWithTools(ctx context.Context, messages []Message, tools []map[string]interface{}) (*ChatResponse, error)
}

ToolCaller is an optional interface for providers that support tool calling.

type ToolConverter 

type ToolConverter struct{}

ToolConverter converts tool definitions between different provider formats

func NewToolConverter 

func NewToolConverter() *ToolConverter

NewToolConverter creates a new tool converter

func (*ToolConverter) ConvertToAnthropic 

func (tc *ToolConverter) ConvertToAnthropic(tools []map[string]interface{}) []map[string]interface{}

ConvertToAnthropic converts tools to Anthropic tool use format

func (*ToolConverter) ConvertToGemini 

func (tc *ToolConverter) ConvertToGemini(tools []map[string]interface{}) []map[string]interface{}

ConvertToGemini converts tools to Gemini function declaration format

func (*ToolConverter) ConvertToOpenAI 

func (tc *ToolConverter) ConvertToOpenAI(tools []map[string]interface{}) []map[string]interface{}

ConvertToOpenAI converts tools to OpenAI function format

func (*ToolConverter) ConvertToolResult 

func (tc *ToolConverter) ConvertToolResult(provider string, result interface{}, toolCallID string) (interface{}, error)

ConvertToolResult converts a tool execution result for different providers

func (*ToolConverter) NormalizeToolCall 

func (tc *ToolConverter) NormalizeToolCall(toolCall *types.ToolCall) error

NormalizeToolCall normalizes a tool call to internal format

func (*ToolConverter) ParseToolCall 

func (tc *ToolConverter) ParseToolCall(data interface{}) (*types.ToolCall, error)

ParseToolCall parses a tool call from various provider formats

func (*ToolConverter) ParseToolCalls 

func (tc *ToolConverter) ParseToolCalls(data interface{}) []types.ToolCall

ParseToolCalls parses multiple tool calls

type Usage 

type Usage struct {
	PromptTokens     int `json:"prompt_tokens"`
	CompletionTokens int `json:"completion_tokens"`
	TotalTokens      int `json:"total_tokens"`
}

Usage represents token usage statistics

type WenxinProvider 

type WenxinProvider struct {
	ConvertConfig *ConvertConfig
	// contains filtered or unexported fields
}

WenxinProvider implements the Baidu Wenxin (ERNIE) API

func NewWenxinProvider 

func NewWenxinProvider(apiKey, secretKey, model string) *WenxinProvider

NewWenxinProvider creates a new Wenxin provider

func (*WenxinProvider) Chat 

func (p *WenxinProvider) Chat(ctx context.Context, messages []Message) (*ChatResponse, error)

Chat implements the Provider interface

func (*WenxinProvider) ChatWithTools 

func (p *WenxinProvider) ChatWithTools(ctx context.Context, messages []Message, tools []map[string]interface{}) (*ChatResponse, error)

ChatWithTools implements the ToolCaller interface

func (*WenxinProvider) GetCapabilities 

func (p *WenxinProvider) GetCapabilities() *Capabilities

GetCapabilities returns the capabilities of Wenxin

func (*WenxinProvider) GetConvertConfig added in v0.4.19 

func (p *WenxinProvider) GetConvertConfig() *ConvertConfig

GetConvertConfig implements ConvertConfigProvider

func (*WenxinProvider) Name 

func (p *WenxinProvider) Name() string

func (*WenxinProvider) SetConvertConfig added in v0.4.19 

func (p *WenxinProvider) SetConvertConfig(config *ConvertConfig)

SetConvertConfig implements ConvertConfigProvider

func (*WenxinProvider) Stream 

func (p *WenxinProvider) Stream(ctx context.Context, messages []Message, handler StreamHandler) error

Stream implements the Streamer interface

type ZhipuProvider 

type ZhipuProvider struct {
	*OpenAICompatibleProvider
}

ZhipuProvider 智谱AI (兼容OpenAI格式)

func NewZhipuProvider 

func NewZhipuProvider(apiKey, baseURL, model string) *ZhipuProvider

NewZhipuProvider creates a new Zhipu provider

func (*ZhipuProvider) GetCapabilities 

func (p *ZhipuProvider) GetCapabilities() *Capabilities

GetCapabilities returns the capabilities of Zhipu

func (*ZhipuProvider) Name 

func (p *ZhipuProvider) Name() string

Source Files

View all Source files

Directories

Path Synopsis

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL
go.dev uses cookies from Google to deliver and enhance the quality of its services and to analyze traffic. Learn more.