fix: extracts JSON content from various formats in the response

internal/version/VERSION

@@ -1 +1 @@
-v5.0.0-beta-2506101319
+v5.0.0-beta-2506101402

uixt/ai/parser_default.go

@@ -4,6 +4,7 @@ import (
 	"fmt"
 	"strings"
 
+	"github.com/cloudwego/eino/schema"
 	"github.com/httprunner/httprunner/v5/internal/json"
 	"github.com/httprunner/httprunner/v5/uixt/option"
 	"github.com/httprunner/httprunner/v5/uixt/types"
@@ -45,13 +46,72 @@ func (p *JSONContentParser) SystemPrompt() string {
 	return p.systemPrompt
 }
 
+// extractJSONContent extracts JSON content from various formats in the response
+func (p *JSONContentParser) extractJSONContent(content string) string {
+	content = strings.TrimSpace(content)
+
+	// Case 1: Content wrapped in ```json ... ```
+	if strings.Contains(content, "```json") {
+		start := strings.Index(content, "```json")
+		if start != -1 {
+			start += 7 // length of "```json"
+			end := strings.Index(content[start:], "```")
+			if end != -1 {
+				jsonContent := strings.TrimSpace(content[start : start+end])
+				return jsonContent
+			}
+		}
+	}
+
+	// Case 2: Content wrapped in ``` ... ``` (without json specifier)
+	if strings.HasPrefix(content, "```") && strings.HasSuffix(content, "```") {
+		lines := strings.Split(content, "\n")
+		if len(lines) >= 3 {
+			// Remove first and last lines (the ``` markers)
+			jsonLines := lines[1 : len(lines)-1]
+			jsonContent := strings.Join(jsonLines, "\n")
+			jsonContent = strings.TrimSpace(jsonContent)
+			// Check if it looks like JSON
+			if strings.HasPrefix(jsonContent, "{") && strings.HasSuffix(jsonContent, "}") {
+				return jsonContent
+			}
+		}
+	}
+
+	// Case 3: Look for JSON object in the content
+	start := strings.Index(content, "{")
+	if start != -1 {
+		// Find the matching closing brace
+		braceCount := 0
+		for i := start; i < len(content); i++ {
+			if content[i] == '{' {
+				braceCount++
+			} else if content[i] == '}' {
+				braceCount--
+				if braceCount == 0 {
+					jsonContent := strings.TrimSpace(content[start : i+1])
+					return jsonContent
+				}
+			}
+		}
+	}
+
+	// Case 4: If content itself looks like JSON
+	if strings.HasPrefix(content, "{") && strings.HasSuffix(content, "}") {
+		return content
+	}
+
+	return ""
+}
+
 func (p *JSONContentParser) 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, "```")
+
+	// Extract JSON content from markdown code blocks
+	jsonContent := p.extractJSONContent(content)
+	if jsonContent == "" {
+		return nil, fmt.Errorf("no valid JSON content found in response")
 	}
-	content = strings.TrimSpace(content)
 
 	// Define a temporary struct to parse the expected JSON format
 	var jsonResponse struct {
@@ -60,7 +120,7 @@ func (p *JSONContentParser) Parse(content string, size types.Size) (*PlanningRes
 		Error   string   `json:"error"`
 	}
 
-	if err := json.Unmarshal([]byte(content), &jsonResponse); err != nil {
+	if err := json.Unmarshal([]byte(jsonContent), &jsonResponse); err != nil {
 		return nil, fmt.Errorf("failed to parse VLM response: %v", err)
 	}
 
@@ -68,7 +128,18 @@ func (p *JSONContentParser) Parse(content string, size types.Size) (*PlanningRes
 		return nil, errors.New(jsonResponse.Error)
 	}
 
+	// Handle cases where no actions are returned
 	if len(jsonResponse.Actions) == 0 {
+		// If there's a valid thought but no actions, this might be an informational response
+		// rather than an actionable UI task. Return the result with empty tool calls.
+		if jsonResponse.Thought != "" {
+			return &PlanningResult{
+				ToolCalls: []schema.ToolCall{}, // Empty tool calls for informational responses
+				Thought:   jsonResponse.Thought,
+				Content:   content, // Include the full response content
+				ModelName: string(p.modelType),
+			}, nil
+		}
 		return nil, errors.New("no actions returned from VLM")
 	}