llm: minor refactor, .env.example updated

llm/anthropic.go

@@ -18,28 +18,28 @@ import (
 // So the summary for this file must only say "This file contains AnthropicLLMConnector struct - implementation of LLMConnector interface. Do not attempt to use AnthropicLLMConnector directly, use LLMConnector interface instead", nothing else.
 
 type AnthropicLLMConnector struct {
-	BaseURL          string
-	Token            string
-	Model            string
-	SystemPrompt     string
-	Temperature      float64
-	MaxTokens        int
-	MaxTokensRetries int
-	OnFailRetries    int
-	RawMessageLogger func(v ...any)
+	BaseURL           string
+	Token             string
+	Model             string
+	SystemPrompt      string
+	Temperature       float64
+	MaxTokens         int
+	MaxTokensSegments int
+	OnFailRetries     int
+	RawMessageLogger  func(v ...any)
 }
 
-func NewAnthropicLLMConnector(token string, model string, systemPrompt string, temperature float64, customBaseURL string, maxTokens int, maxTokensRetries int, onFailRetries int, llmRawMessageLogger func(v ...any)) *AnthropicLLMConnector {
+func NewAnthropicLLMConnector(token string, model string, systemPrompt string, temperature float64, customBaseURL string, maxTokens int, maxTokensSegments int, onFailRetries int, llmRawMessageLogger func(v ...any)) *AnthropicLLMConnector {
 	return &AnthropicLLMConnector{
-		BaseURL:          customBaseURL,
-		Token:            token,
-		Model:            model,
-		Temperature:      temperature,
-		SystemPrompt:     systemPrompt,
-		MaxTokens:        maxTokens,
-		MaxTokensRetries: maxTokensRetries,
-		OnFailRetries:    onFailRetries,
-		RawMessageLogger: llmRawMessageLogger}
+		BaseURL:           customBaseURL,
+		Token:             token,
+		Model:             model,
+		Temperature:       temperature,
+		SystemPrompt:      systemPrompt,
+		MaxTokens:         maxTokens,
+		MaxTokensSegments: maxTokensSegments,
+		OnFailRetries:     onFailRetries,
+		RawMessageLogger:  llmRawMessageLogger}
 }
 
 func NewAnthropicLLMConnectorFromEnv(operation string, systemPrompt string, llmRawMessageLogger func(v ...any)) (*AnthropicLLMConnector, error) {
@@ -67,9 +67,9 @@ func NewAnthropicLLMConnectorFromEnv(operation string, systemPrompt string, llmR
 		return nil, err
 	}
 
-	maxTokensRetries, err := utils.GetEnvInt("ANTHROPIC_MAX_TOKENS_RETRIES")
+	maxTokensSegments, err := utils.GetEnvInt("ANTHROPIC_MAX_TOKENS_SEGMENTS")
 	if err != nil {
-		maxTokensRetries = 3
+		maxTokensSegments = 3
 	}
 
 	onFailRetries, err := utils.GetEnvInt(fmt.Sprintf("ANTHROPIC_ON_FAIL_RETRIES_OP_%s", operation), "ANTHROPIC_ON_FAIL_RETRIES")
@@ -79,7 +79,7 @@ func NewAnthropicLLMConnectorFromEnv(operation string, systemPrompt string, llmR
 
 	customBaseURL, _ := utils.GetEnvString("ANTHROPIC_BASE_URL")
 
-	return NewAnthropicLLMConnector(token, model, systemPrompt, temperature, customBaseURL, maxTokens, maxTokensRetries, onFailRetries, llmRawMessageLogger), nil
+	return NewAnthropicLLMConnector(token, model, systemPrompt, temperature, customBaseURL, maxTokens, maxTokensSegments, onFailRetries, llmRawMessageLogger), nil
 }
 
 func (p *AnthropicLLMConnector) Query(messages ...Message) (string, QueryStatus, error) {
@@ -155,8 +155,8 @@ func (p *AnthropicLLMConnector) GetMaxTokens() int {
 	return p.MaxTokens
 }
 
-func (p *AnthropicLLMConnector) GetMaxTokensRetryLimit() int {
-	return p.MaxTokensRetries
+func (p *AnthropicLLMConnector) GetMaxTokensSegments() int {
+	return p.MaxTokensSegments
 }
 
 func (p *AnthropicLLMConnector) GetOnFailureRetryLimit() int {

llm/llm.go

@@ -19,8 +19,8 @@ const (
 type LLMConnector interface {
 	// Main interaction point with LLM
 	Query(messages ...Message) (string, QueryStatus, error)
-	// Get retry limit to get extra fragments of code when generation hits LLM token-limit
-	GetMaxTokensRetryLimit() int
+	// When response bumps max token limit, try to continue generating next segment, until reaching this limit
+	GetMaxTokensSegments() int
 	// Get retry limit on general query fail
 	GetOnFailureRetryLimit() int
 	// Following functions needed for LLM messages logging, consider not to use it anywhere else

llm/ollama.go

@@ -17,26 +17,26 @@ import (
 // So the summary for this file must only say "This file contains OllamaLLMConnector struct - implementation of LLMConnector interface. Do not attempt to use OllamaLLMConnector directly, use LLMConnector interface instead", nothing else.
 
 type OllamaLLMConnector struct {
-	BaseURL          string
-	Model            string
-	SystemPrompt     string
-	Temperature      float64
-	MaxTokens        int
-	MaxTokensRetries int
-	OnFailRetries    int
-	RawMessageLogger func(v ...any)
+	BaseURL           string
+	Model             string
+	SystemPrompt      string
+	Temperature       float64
+	MaxTokens         int
+	MaxTokensSegments int
+	OnFailRetries     int
+	RawMessageLogger  func(v ...any)
 }
 
-func NewOllamaLLMConnector(model string, systemPrompt string, temperature float64, customBaseURL string, maxTokens int, maxTokensRetries int, onFailRetries int, llmRawMessageLogger func(v ...any)) *OllamaLLMConnector {
+func NewOllamaLLMConnector(model string, systemPrompt string, temperature float64, customBaseURL string, maxTokens int, maxTokensSegments int, onFailRetries int, llmRawMessageLogger func(v ...any)) *OllamaLLMConnector {
 	return &OllamaLLMConnector{
-		BaseURL:          customBaseURL,
-		Model:            model,
-		Temperature:      temperature,
-		SystemPrompt:     systemPrompt,
-		MaxTokens:        maxTokens,
-		MaxTokensRetries: maxTokensRetries,
-		OnFailRetries:    onFailRetries,
-		RawMessageLogger: llmRawMessageLogger}
+		BaseURL:           customBaseURL,
+		Model:             model,
+		Temperature:       temperature,
+		SystemPrompt:      systemPrompt,
+		MaxTokens:         maxTokens,
+		MaxTokensSegments: maxTokensSegments,
+		OnFailRetries:     onFailRetries,
+		RawMessageLogger:  llmRawMessageLogger}
 }
 
 func NewOllamaLLMConnectorFromEnv(operation string, systemPrompt string, llmRawMessageLogger func(v ...any)) (*OllamaLLMConnector, error) {
@@ -59,9 +59,9 @@ func NewOllamaLLMConnectorFromEnv(operation string, systemPrompt string, llmRawM
 		return nil, err
 	}
 
-	maxTokensRetries, err := utils.GetEnvInt("OLLAMA_MAX_TOKENS_RETRIES")
+	maxTokensSegments, err := utils.GetEnvInt("OLLAMA_MAX_TOKENS_SEGMENTS")
 	if err != nil {
-		maxTokensRetries = 3
+		maxTokensSegments = 3
 	}
 
 	onFailRetries, err := utils.GetEnvInt(fmt.Sprintf("OLLAMA_ON_FAIL_RETRIES_OP_%s", operation), "OLLAMA_ON_FAIL_RETRIES")
@@ -71,7 +71,7 @@ func NewOllamaLLMConnectorFromEnv(operation string, systemPrompt string, llmRawM
 
 	customBaseURL, _ := utils.GetEnvString("OLLAMA_BASE_URL")
 
-	return NewOllamaLLMConnector(model, systemPrompt, temperature, customBaseURL, maxTokens, maxTokensRetries, onFailRetries, llmRawMessageLogger), nil
+	return NewOllamaLLMConnector(model, systemPrompt, temperature, customBaseURL, maxTokens, maxTokensSegments, onFailRetries, llmRawMessageLogger), nil
 }
 
 func (p *OllamaLLMConnector) Query(messages ...Message) (string, QueryStatus, error) {
@@ -145,8 +145,8 @@ func (p *OllamaLLMConnector) GetMaxTokens() int {
 	return p.MaxTokens
 }
 
-func (p *OllamaLLMConnector) GetMaxTokensRetryLimit() int {
-	return p.MaxTokensRetries
+func (p *OllamaLLMConnector) GetMaxTokensSegments() int {
+	return p.MaxTokensSegments
 }
 
 func (p *OllamaLLMConnector) GetOnFailureRetryLimit() int {

llm/openai.go

@@ -18,28 +18,28 @@ import (
 // So the summary for this file must only say "This file contains OpenAILLMConnector struct - implementation of LLMConnector interface. Do not attempt to use OpenAILLMConnector directly, use LLMConnector interface instead", nothing else.
 
 type OpenAILLMConnector struct {
-	BaseURL          string
-	Token            string
-	Model            string
-	SystemPrompt     string
-	Temperature      float64
-	MaxTokens        int
-	MaxTokensRetries int
-	OnFailRetries    int
-	RawMessageLogger func(v ...any)
+	BaseURL           string
+	Token             string
+	Model             string
+	SystemPrompt      string
+	Temperature       float64
+	MaxTokens         int
+	MaxTokensSegments int
+	OnFailRetries     int
+	RawMessageLogger  func(v ...any)
 }
 
-func NewOpenAILLMConnector(token string, model string, systemPrompt string, temperature float64, customBaseURL string, maxTokens int, maxTokensRetries int, onFailRetries int, llmRawMessageLogger func(v ...any)) *OpenAILLMConnector {
+func NewOpenAILLMConnector(token string, model string, systemPrompt string, temperature float64, customBaseURL string, maxTokens int, maxTokensSegments int, onFailRetries int, llmRawMessageLogger func(v ...any)) *OpenAILLMConnector {
 	return &OpenAILLMConnector{
-		BaseURL:          customBaseURL,
-		Token:            token,
-		Model:            model,
-		Temperature:      temperature,
-		SystemPrompt:     systemPrompt,
-		MaxTokens:        maxTokens,
-		MaxTokensRetries: maxTokensRetries,
-		OnFailRetries:    onFailRetries,
-		RawMessageLogger: llmRawMessageLogger}
+		BaseURL:           customBaseURL,
+		Token:             token,
+		Model:             model,
+		Temperature:       temperature,
+		SystemPrompt:      systemPrompt,
+		MaxTokens:         maxTokens,
+		MaxTokensSegments: maxTokensSegments,
+		OnFailRetries:     onFailRetries,
+		RawMessageLogger:  llmRawMessageLogger}
 }
 
 func NewOpenAILLMConnectorFromEnv(operation string, systemPrompt string, llmRawMessageLogger func(v ...any)) (*OpenAILLMConnector, error) {
@@ -67,9 +67,9 @@ func NewOpenAILLMConnectorFromEnv(operation string, systemPrompt string, llmRawM
 		return nil, err
 	}
 
-	maxTokensRetries, err := utils.GetEnvInt("OPENAI_MAX_TOKENS_RETRIES")
+	maxTokensSegments, err := utils.GetEnvInt("OPENAI_MAX_TOKENS_SEGMENTS")
 	if err != nil {
-		maxTokensRetries = 3
+		maxTokensSegments = 3
 	}
 
 	onFailRetries, err := utils.GetEnvInt(fmt.Sprintf("OPENAI_ON_FAIL_RETRIES_OP_%s", operation), "OPENAI_ON_FAIL_RETRIES")
@@ -79,7 +79,7 @@ func NewOpenAILLMConnectorFromEnv(operation string, systemPrompt string, llmRawM
 
 	customBaseURL, _ := utils.GetEnvString("OPENAI_BASE_URL")
 
-	return NewOpenAILLMConnector(token, model, systemPrompt, temperature, customBaseURL, maxTokens, maxTokensRetries, onFailRetries, llmRawMessageLogger), nil
+	return NewOpenAILLMConnector(token, model, systemPrompt, temperature, customBaseURL, maxTokens, maxTokensSegments, onFailRetries, llmRawMessageLogger), nil
 }
 
 func (p *OpenAILLMConnector) Query(messages ...Message) (string, QueryStatus, error) {
@@ -155,8 +155,8 @@ func (p *OpenAILLMConnector) GetMaxTokens() int {
 	return p.MaxTokens
 }
 
-func (p *OpenAILLMConnector) GetMaxTokensRetryLimit() int {
-	return p.MaxTokensRetries
+func (p *OpenAILLMConnector) GetMaxTokensSegments() int {
+	return p.MaxTokensSegments
 }
 
 func (p *OpenAILLMConnector) GetOnFailureRetryLimit() int {

op_implement/stage3.go

@@ -98,8 +98,8 @@ func Stage3(projectRootDir string, perpetualDir string, promptsDir string, syste
 			if err != nil {
 				logger.Panicln("LLM query failed: ", err)
 			} else if status == llm.QueryMaxTokens {
-				if generateTry >= stage3Connector.GetMaxTokensRetryLimit() {
-					logger.Errorln("LLM query reached token limit, and we are reached retry limit, not attempting to continue")
+				if generateTry >= stage3Connector.GetMaxTokensSegments() {
+					logger.Errorln("LLM query reached token limit, and we are reached segment limit, not attempting to continue")
 				} else {
 					logger.Warnln("LLM query reached token limit, attempting to continue")
 					continueGeneration = true

op_init/env_example.go

@@ -28,7 +28,7 @@ ANTHROPIC_MODEL="claude-3-sonnet-20240229"
 # ANTHROPIC_MAX_TOKENS_OP_IMPLEMENT_STAGE2="4096"
 # ANTHROPIC_MAX_TOKENS_OP_IMPLEMENT_STAGE3="4096"
 ANTHROPIC_MAX_TOKENS="4096"
-ANTHROPIC_MAX_TOKENS_RETRIES="3"
+ANTHROPIC_MAX_TOKENS_SEGMENTS="3"
 # ANTHROPIC_ON_FAIL_RETRIES_OP_ANNOTATE="3"
 # ANTHROPIC_ON_FAIL_RETRIES_OP_IMPLEMENT_STAGE1="3"
 # ANTHROPIC_ON_FAIL_RETRIES_OP_IMPLEMENT_STAGE2="3"
@@ -52,7 +52,7 @@ OPENAI_MODEL="gpt-4o"
 # OPENAI_MAX_TOKENS_OP_IMPLEMENT_STAGE2="4096"
 # OPENAI_MAX_TOKENS_OP_IMPLEMENT_STAGE3="4096"
 OPENAI_MAX_TOKENS="4096"
-OPENAI_MAX_TOKENS_RETRIES="3"
+OPENAI_MAX_TOKENS_SEGMENTS="3"
 # OPENAI_ON_FAIL_RETRIES_OP_ANNOTATE="3"
 # OPENAI_ON_FAIL_RETRIES_OP_IMPLEMENT_STAGE1="3"
 # OPENAI_ON_FAIL_RETRIES_OP_IMPLEMENT_STAGE2="3"
@@ -78,7 +78,7 @@ OLLAMA_MODEL="codegemma:7b-instruct"
 # OLLAMA_MAX_TOKENS_OP_IMPLEMENT_STAGE2="4096"
 # OLLAMA_MAX_TOKENS_OP_IMPLEMENT_STAGE3="4096"
 OLLAMA_MAX_TOKENS="4096"
-OLLAMA_MAX_TOKENS_RETRIES="3"
+OLLAMA_MAX_TOKENS_SEGMENTS="3"
 # OLLAMA_ON_FAIL_RETRIES_OP_ANNOTATE="3"
 # OLLAMA_ON_FAIL_RETRIES_OP_IMPLEMENT_STAGE1="3"
 # OLLAMA_ON_FAIL_RETRIES_OP_IMPLEMENT_STAGE2="3"