Merge pull request #10 from neurocult/func_call

Func call

agency.go

@@ -11,12 +11,11 @@ type Operation struct {
 	config  *OperationConfig
 }
 
-// OperationHandler is a function that implements logic.
+// OperationHandler is a function that implements operation's logic.
 // It could be thought of as an interface that providers must implement.
 type OperationHandler func(context.Context, Message, *OperationConfig) (Message, error)
 
-// OperationConfig represents abstract operation configuration.
-// It contains fields for all possible modalities but nothing specific to concrete model implementations.
+// OperationConfig represents abstract operation configuration for all possible models.
 type OperationConfig struct {
 	Prompt   string
 	Messages []Message

examples/func_call/main.go

@@ -0,0 +1,93 @@
+package main
+
+import (
+	"context"
+	"encoding/json"
+	"fmt"
+	"os"
+
+	_ "github.com/joho/godotenv/autoload"
+	"github.com/sashabaranov/go-openai/jsonschema"
+
+	"github.com/neurocult/agency"
+	"github.com/neurocult/agency/providers/openai"
+)
+
+func main() {
+	t2tOp := openai.
+		New(openai.Params{Key: os.Getenv("OPENAI_API_KEY")}).
+		TextToText(openai.TextToTextParams{
+			Model: "gpt-3.5-turbo",
+			FuncDefs: []openai.FuncDef{
+				// function without parameters
+				{
+					Name:        "GetMeaningOfLife",
+					Description: "Answer questions about meaning of life",
+					Body: func(ctx context.Context, _ []byte) (any, error) {
+						return 42, nil
+					},
+				},
+				// function with parameters
+				{
+					Name:        "ChangeNumbers",
+					Description: "Change given numbers when asked",
+					Parameters: &jsonschema.Definition{
+						Type: "object",
+						Properties: map[string]jsonschema.Definition{
+							"a": {Type: "integer"},
+							"b": {Type: "integer"},
+						},
+					},
+					Body: func(ctx context.Context, params []byte) (any, error) {
+						var pp struct{ A, B int }
+						if err := json.Unmarshal(params, &pp); err != nil {
+							return nil, err
+						}
+						return (pp.A + pp.B) * 10, nil // *10 is just to distinguish from normal response
+					},
+				},
+			},
+		}).
+		SetPrompt(`
+Answer questions about meaning of life and summing numbers.
+Always use GetMeaningOfLife and ChangeNumbers functions results as answers.
+Examples:
+- User: what is the meaning of life?
+- Assistant: 42
+- User: 1+1
+- Assistant: 20
+- User: 1+1 and what is the meaning of life?
+- Assistant: 20 and 42`)
+
+	ctx := context.Background()
+
+	// test for first function call
+	answer, err := t2tOp.Execute(
+		ctx,
+		agency.UserMessage("what is the meaning of life?"),
+	)
+	if err != nil {
+		panic(err)
+	}
+	fmt.Println(answer)
+
+	// test for second function call
+	answer, err = t2tOp.Execute(
+		ctx,
+		agency.UserMessage("1+1?"),
+	)
+	if err != nil {
+		panic(err)
+	}
+	fmt.Println(answer)
+
+	// test for both function calls at the same time
+	answer, err = t2tOp.Execute(
+		ctx,
+		agency.UserMessage("1+1 and what is the meaning of life?"),
+	)
+	if err != nil {
+		panic(err)
+	}
+	fmt.Println(answer)
+}

process.go

@@ -9,6 +9,7 @@ type Process struct {
 	operations []*Operation
 }
 
+// NewProcess creates a new Process with given operations.
 func NewProcess(operations ...*Operation) *Process {
 	return &Process{
 		operations: operations,

providers/openai/image_to_text.go

@@ -15,6 +15,8 @@ type ImageToTextParams struct {
 	MaxTokens   int
 }
 
+
+// ImageToText is an operation builder that creates operation than can convert image to text.
 func (f *Provider) ImageToText(params ImageToTextParams) *agency.Operation {
 	return agency.NewOperation(func(ctx context.Context, msg agency.Message, cfg *agency.OperationConfig) (agency.Message, error) {
 		openaiMsg := openai.ChatCompletionMessage{

providers/openai/provider.go

@@ -6,15 +6,19 @@ import (
 	"github.com/sashabaranov/go-openai"
 )
 
+// Provider is a set of operation builders.
 type Provider struct {
 	client *openai.Client
 }
 
+// Params is a set of parameters specific for creating this concrete provider.
+// They are shared across all operation builders.
 type Params struct {
 	Key     string // Required if not using local LLM.
 	BaseURL string // Optional. If not set then default openai base url is used
 }
 
+// New creates a new Provider instance.
 func New(params Params) *Provider {
 	cfg := openai.DefaultConfig(params.Key)
 	if params.BaseURL != "" {
@@ -25,6 +29,8 @@ func New(params Params) *Provider {
 	}
 }
 
+// === Helpers ===
+
 // NullableFloat32 is a type that exists to distinguish between undefined values and real zeros.
 // It fixes sashabaranov/go-openai issue with zero temp not included in api request due to how json unmarshal work.
 type NullableFloat32 *float32

providers/openai/speech_to_text.go

@@ -13,22 +13,25 @@ type SpeechToTextParams struct {
 	Temperature NullableFloat32
 }
 
+// SpeechToText is an operation builder that creates operation than can convert speech to text.
 func (f Provider) SpeechToText(params SpeechToTextParams) *agency.Operation {
-	return agency.NewOperation(func(ctx context.Context, msg agency.Message, cfg *agency.OperationConfig) (agency.Message, error) {
-		resp, err := f.client.CreateTranscription(ctx, openai.AudioRequest{
-			Model:       params.Model,
-			Prompt:      cfg.Prompt,
-			FilePath:    "speech.ogg", // TODO move to cfg?
-			Reader:      bytes.NewReader(msg.Content),
-			Temperature: getTemperature(params.Temperature),
-		})
-		if err != nil {
-			return agency.Message{}, err
-		}
+	return agency.NewOperation(
+		func(ctx context.Context, msg agency.Message, cfg *agency.OperationConfig) (agency.Message, error) {
+			resp, err := f.client.CreateTranscription(ctx, openai.AudioRequest{
+				Model:       params.Model,
+				Prompt:      cfg.Prompt,
+				FilePath:    "speech.ogg",
+				Reader:      bytes.NewReader(msg.Content),
+				Temperature: getTemperature(params.Temperature),
+			})
+			if err != nil {
+				return agency.Message{}, err
+			}
 
-		return agency.Message{
-			Role:    agency.AssistantRole,
-			Content: []byte(resp.Text),
-		}, nil
-	})
+			return agency.Message{
+				Role:    agency.AssistantRole,
+				Content: []byte(resp.Text),
+			}, nil
+		},
+	)
 }

providers/openai/text_to_image.go

@@ -16,31 +16,34 @@ type TextToImageParams struct {
 	Style     string
 }
 
+// TextToImage is an operation builder that creates operation than can convert text to image.
 func (p Provider) TextToImage(params TextToImageParams) *agency.Operation {
-	return agency.NewOperation(func(ctx context.Context, msg agency.Message, cfg *agency.OperationConfig) (agency.Message, error) {
-		reqBase64 := openai.ImageRequest{
-			Prompt:         fmt.Sprintf("%s\n\n%s", cfg.Prompt, string(msg.Content)),
-			Size:           params.ImageSize,
-			ResponseFormat: openai.CreateImageResponseFormatB64JSON,
-			N:              1, // DALL·E-3 only support n=1, for other models support needed
-			Model:          params.Model,
-			Quality:        params.Quality,
-			Style:          params.Style,
-		}
+	return agency.NewOperation(
+		func(ctx context.Context, msg agency.Message, cfg *agency.OperationConfig) (agency.Message, error) {
+			reqBase64 := openai.ImageRequest{
+				Prompt:         fmt.Sprintf("%s\n\n%s", cfg.Prompt, string(msg.Content)),
+				Size:           params.ImageSize,
+				ResponseFormat: openai.CreateImageResponseFormatB64JSON,
+				N:              1, // DALL·E-3 only support n=1, for other models support needed
+				Model:          params.Model,
+				Quality:        params.Quality,
+				Style:          params.Style,
+			}
 
-		respBase64, err := p.client.CreateImage(ctx, reqBase64)
-		if err != nil {
-			return agency.Message{}, err
-		}
+			respBase64, err := p.client.CreateImage(ctx, reqBase64)
+			if err != nil {
+				return agency.Message{}, err
+			}
 
-		imgBytes, err := base64.StdEncoding.DecodeString(respBase64.Data[0].B64JSON)
-		if err != nil {
-			return agency.Message{}, err
-		}
+			imgBytes, err := base64.StdEncoding.DecodeString(respBase64.Data[0].B64JSON)
+			if err != nil {
+				return agency.Message{}, err
+			}
 
-		return agency.Message{
-			Role:    agency.AssistantRole,
-			Content: imgBytes,
-		}, nil
-	})
+			return agency.Message{
+				Role:    agency.AssistantRole,
+				Content: imgBytes,
+			}, nil
+		},
+	)
 }

providers/openai/text_to_speech.go

@@ -15,27 +15,30 @@ type TextToSpeechParams struct {
 	Voice          string
 }
 
+// TextToSpeech is an operation builder that creates operation than can convert text to speech.
 func (f Provider) TextToSpeech(params TextToSpeechParams) *agency.Operation {
-	return agency.NewOperation(func(ctx context.Context, msg agency.Message, cfg *agency.OperationConfig) (agency.Message, error) {
-		resp, err := f.client.CreateSpeech(ctx, openai.CreateSpeechRequest{
-			Model:          openai.SpeechModel(params.Model),
-			Input:          msg.String(),
-			Voice:          openai.SpeechVoice(params.Voice),
-			ResponseFormat: openai.SpeechResponseFormat(params.ResponseFormat),
-			Speed:          params.Speed,
-		})
-		if err != nil {
-			return agency.Message{}, err
-		}
+	return agency.NewOperation(
+		func(ctx context.Context, msg agency.Message, cfg *agency.OperationConfig) (agency.Message, error) {
+			resp, err := f.client.CreateSpeech(ctx, openai.CreateSpeechRequest{
+				Model:          openai.SpeechModel(params.Model),
+				Input:          msg.String(),
+				Voice:          openai.SpeechVoice(params.Voice),
+				ResponseFormat: openai.SpeechResponseFormat(params.ResponseFormat),
+				Speed:          params.Speed,
+			})
+			if err != nil {
+				return agency.Message{}, err
+			}
 
-		bb, err := io.ReadAll(resp)
-		if err != nil {
-			return agency.Message{}, err
-		}
+			bb, err := io.ReadAll(resp)
+			if err != nil {
+				return agency.Message{}, err
+			}
 
-		return agency.Message{
-			Role:    agency.AssistantRole,
-			Content: bb,
-		}, nil
-	})
+			return agency.Message{
+				Role:    agency.AssistantRole,
+				Content: bb,
+			}, nil
+		},
+	)
 }