refactor: json asserter

internal/version/VERSION

@@ -1 +1 @@
-v5.0.0-beta-2505221534
+v5.0.0-beta-2505221822

uixt/ai/asserter.go

@@ -54,8 +54,8 @@ func NewAsserter(ctx context.Context, modelConfig *ModelConfig) (*Asserter, erro
 	}
 
 	if modelConfig.ModelType == option.LLMServiceTypeUITARS {
-		asserter.systemPrompt += "\n\n" + uiTarsAssertionResponseFormat
+		asserter.systemPrompt += "\n" + uiTarsAssertionResponseFormat
-	} else if modelConfig.ModelType == option.LLMServiceTypeDoubaoVL {
+	} else {
 		// define output format
 		type OutputFormat struct {
 			Thought string `json:"thought"`
@@ -77,8 +77,6 @@ func NewAsserter(ctx context.Context, modelConfig *ModelConfig) (*Asserter, erro
 				Strict:      false,
 			},
 		}
-	} else {
-		asserter.systemPrompt += "\n\n" + defaultAssertionResponseJsonFormat
 	}
 
 	var err error
@@ -134,16 +132,16 @@ Here is the assertion. Please tell whether it is truthy according to the screens
 	// Call model service, generate response
 	logRequest(a.history)
 	startTime := time.Now()
-	resp, err := a.model.Generate(ctx, a.history)
+	message, err := a.model.Generate(ctx, a.history)
 	log.Info().Float64("elapsed(s)", time.Since(startTime).Seconds()).
 		Str("model", string(a.modelConfig.ModelType)).Msg("call model service for assertion")
 	if err != nil {
 		return nil, errors.Wrap(code.LLMRequestServiceError, err.Error())
 	}
-	logResponse(resp)
+	logResponse(message)
 
 	// Parse result
-	result, err := parseAssertionResult(resp.Content)
+	result, err := parseAssertionResult(message.Content)
 	if err != nil {
 		return nil, errors.Wrap(code.LLMParseAssertionResponseError, err.Error())
 	}
@@ -151,7 +149,7 @@ Here is the assertion. Please tell whether it is truthy according to the screens
 	// Append assistant message to history
 	a.history.Append(&schema.Message{
 		Role:    schema.Assistant,
-		Content: resp.Content,
+		Content: message.Content,
 	})
 
 	return result, nil

uixt/ai/asserter_prompts.go

@@ -3,15 +3,9 @@ package ai
 // Default assertion system prompt
 const defaultAssertionPrompt = `You are a senior testing engineer. User will give an assertion and a screenshot of a page. By carefully viewing the screenshot, please tell whether the assertion is truthy.`
 
-// Default assertion response format
-const defaultAssertionResponseJsonFormat = `Return in the following JSON format:
-{
-  pass: boolean, // whether the assertion is truthy
-  thought: string | null, // string, if the result is falsy, give the reason why it is falsy. Otherwise, put null.
-}`
-
 // UI-TARS assertion response format
-const uiTarsAssertionResponseFormat = `## Output Json String Format
+const uiTarsAssertionResponseFormat = `
+## Output Json String Format
 ` + "```" + `
 "{
   "pass": <<is a boolean value from the enum [true, false], true means the assertion is truthy>>,

uixt/ai/planner_parser.go

@@ -27,14 +27,10 @@ func NewLLMContentParser(modelType option.LLMServiceType) LLMContentParser {
 		return &UITARSContentParser{
 			systemPrompt: uiTarsPlanningPrompt,
 		}
-	case option.LLMServiceTypeDoubaoVL:
+	default:
 		return &JSONContentParser{
 			systemPrompt: defaultPlanningResponseJsonFormat,
 		}
-	default:
-		return &DefaultContentParser{
-			systemPrompt: defaultPlanningResponseStringFormat,
-		}
 	}
 }
 
@@ -433,50 +429,3 @@ func normalizeAction(action *ParsedAction) error {
 
 	return nil
 }
-
-// DefaultContentParser parses the response as string format
-type DefaultContentParser struct {
-	systemPrompt string
-}
-
-func (p *DefaultContentParser) SystemPrompt() string {
-	return p.systemPrompt
-}
-
-func (p *DefaultContentParser) Parse(content string, size types.Size) (*PlanningResult, error) {
-	content = strings.TrimSpace(content)
-	if strings.HasPrefix(content, "```json") && strings.HasSuffix(content, "```") {
-		content = strings.TrimPrefix(content, "```json")
-		content = strings.TrimSuffix(content, "```")
-	}
-	content = strings.TrimSpace(content)
-
-	var response PlanningResult
-	if err := json.Unmarshal([]byte(content), &response); err != nil {
-		return nil, fmt.Errorf("failed to parse VLM response: %v", err)
-	}
-
-	if response.Error != "" {
-		return nil, errors.New(response.Error)
-	}
-
-	if len(response.NextActions) == 0 {
-		return nil, errors.New("no actions returned from VLM")
-	}
-
-	// normalize actions
-	var normalizedActions []ParsedAction
-	for i := range response.NextActions {
-		// create a new variable, avoid implicit memory aliasing in for loop.
-		action := response.NextActions[i]
-		if err := normalizeAction(&action); err != nil {
-			return nil, errors.Wrap(err, "failed to normalize action")
-		}
-		normalizedActions = append(normalizedActions, action)
-	}
-
-	return &PlanningResult{
-		NextActions:   normalizedActions,
-		ActionSummary: response.ActionSummary,
-	}, nil
-}

uixt/ai/planner_prompts.go

@@ -38,7 +38,3 @@ Thought: ...
 Action: ...
 ` + "```" + `
 `
-
-const defaultPlanningResponseStringFormat = `
-You are a helpful assistant.
-`