feat(translator): improve Claude request handling with enhanced content processing

- Introduced helper functions (`appendTextContent`, `appendImageContent`, etc.) for structured content construction. - Refactored message generation logic for better clarity, supporting mixed content scenarios (text, images, and function calls). - Added `flushMessage` to ensure proper grouping of message contents.
2026-02-02 20:40:52 +08:00 · 2025-10-14 23:58:37 +08:00
parent 1cd275f4c1
commit c945e35983
2 changed files with 130 additions and 45 deletions
--- a/internal/translator/claude/openai/responses/claude_openai-responses_request.go
+++ b/internal/translator/claude/openai/responses/claude_openai-responses_request.go
@@ -143,21 +143,63 @@ func ConvertOpenAIResponsesRequestToClaude(modelName string, inputRawJSON []byte
 			}
 			switch typ {
 			case "message":
-				// Determine role from content type (input_text=user, output_text=assistant)
+				// Determine role and construct Claude-compatible content parts.
 				var role string
-				var text strings.Builder
+				var textAggregate strings.Builder
+				var partsJSON []string
+				hasImage := false
 				if parts := item.Get("content"); parts.Exists() && parts.IsArray() {
 					parts.ForEach(func(_, part gjson.Result) bool {
 						ptype := part.Get("type").String()
-						if ptype == "input_text" || ptype == "output_text" {
+						switch ptype {
+						case "input_text", "output_text":
 							if t := part.Get("text"); t.Exists() {
-								text.WriteString(t.String())
+								txt := t.String()
+								textAggregate.WriteString(txt)
+								contentPart := `{"type":"text","text":""}`
+								contentPart, _ = sjson.Set(contentPart, "text", txt)
+								partsJSON = append(partsJSON, contentPart)
 							}
 							if ptype == "input_text" {
 								role = "user"
-							} else if ptype == "output_text" {
+							} else {
 								role = "assistant"
 							}
+						case "input_image":
+							url := part.Get("image_url").String()
+							if url == "" {
+								url = part.Get("url").String()
+							}
+							if url != "" {
+								var contentPart string
+								if strings.HasPrefix(url, "data:") {
+									trimmed := strings.TrimPrefix(url, "data:")
+									mediaAndData := strings.SplitN(trimmed, ";base64,", 2)
+									mediaType := "application/octet-stream"
+									data := ""
+									if len(mediaAndData) == 2 {
+										if mediaAndData[0] != "" {
+											mediaType = mediaAndData[0]
+										}
+										data = mediaAndData[1]
+									}
+									if data != "" {
+										contentPart = `{"type":"image","source":{"type":"base64","media_type":"","data":""}}`
+										contentPart, _ = sjson.Set(contentPart, "source.media_type", mediaType)
+										contentPart, _ = sjson.Set(contentPart, "source.data", data)
+									}
+								} else {
+									contentPart = `{"type":"image","source":{"type":"url","url":""}}`
+									contentPart, _ = sjson.Set(contentPart, "source.url", url)
+								}
+								if contentPart != "" {
+									partsJSON = append(partsJSON, contentPart)
+									if role == "" {
+										role = "user"
+									}
+									hasImage = true
+								}
+							}
 						}
 						return true
 					})
@@ -174,14 +216,24 @@ func ConvertOpenAIResponsesRequestToClaude(modelName string, inputRawJSON []byte
 					}
 				}

-				if text.Len() > 0 || role == "system" {
+				if len(partsJSON) > 0 {
+					msg := `{"role":"","content":[]}`
+					msg, _ = sjson.Set(msg, "role", role)
+					if len(partsJSON) == 1 && !hasImage {
+						// Preserve legacy behavior for single text content
+						msg, _ = sjson.Delete(msg, "content")
+						textPart := gjson.Parse(partsJSON[0])
+						msg, _ = sjson.Set(msg, "content", textPart.Get("text").String())
+					} else {
+						for _, partJSON := range partsJSON {
+							msg, _ = sjson.SetRaw(msg, "content.-1", partJSON)
+						}
+					}
+					out, _ = sjson.SetRaw(out, "messages.-1", msg)
+				} else if textAggregate.Len() > 0 || role == "system" {
 					msg := `{"role":"","content":""}`
 					msg, _ = sjson.Set(msg, "role", role)
-					if text.Len() > 0 {
-						msg, _ = sjson.Set(msg, "content", text.String())
-					} else {
-						msg, _ = sjson.Set(msg, "content", "")
-					}
+					msg, _ = sjson.Set(msg, "content", textAggregate.String())
 					out, _ = sjson.SetRaw(out, "messages.-1", msg)
 				}

--- a/internal/translator/codex/claude/codex_claude_request.go
+++ b/internal/translator/codex/claude/codex_claude_request.go
@@ -68,36 +68,79 @@ func ConvertClaudeRequestToCodex(modelName string, inputRawJSON []byte, _ bool)

 		for i := 0; i < len(messageResults); i++ {
 			messageResult := messageResults[i]
+			messageRole := messageResult.Get("role").String()
+
+			newMessage := func() string {
+				msg := `{"type": "message","role":"","content":[]}`
+				msg, _ = sjson.Set(msg, "role", messageRole)
+				return msg
+			}
+
+			message := newMessage()
+			contentIndex := 0
+			hasContent := false
+
+			flushMessage := func() {
+				if hasContent {
+					template, _ = sjson.SetRaw(template, "input.-1", message)
+					message = newMessage()
+					contentIndex = 0
+					hasContent = false
+				}
+			}
+
+			appendTextContent := func(text string) {
+				partType := "input_text"
+				if messageRole == "assistant" {
+					partType = "output_text"
+				}
+				message, _ = sjson.Set(message, fmt.Sprintf("content.%d.type", contentIndex), partType)
+				message, _ = sjson.Set(message, fmt.Sprintf("content.%d.text", contentIndex), text)
+				contentIndex++
+				hasContent = true
+			}
+
+			appendImageContent := func(dataURL string) {
+				message, _ = sjson.Set(message, fmt.Sprintf("content.%d.type", contentIndex), "input_image")
+				message, _ = sjson.Set(message, fmt.Sprintf("content.%d.image_url", contentIndex), dataURL)
+				contentIndex++
+				hasContent = true
+			}

 			messageContentsResult := messageResult.Get("content")
 			if messageContentsResult.IsArray() {
 				messageContentResults := messageContentsResult.Array()
 				for j := 0; j < len(messageContentResults); j++ {
 					messageContentResult := messageContentResults[j]
-					messageContentTypeResult := messageContentResult.Get("type")
-					contentType := messageContentTypeResult.String()
+					contentType := messageContentResult.Get("type").String()

-					if contentType == "text" {
-						// Handle text content by creating appropriate message structure.
-						message := `{"type": "message","role":"","content":[]}`
-						messageRole := messageResult.Get("role").String()
-						message, _ = sjson.Set(message, "role", messageRole)
-
-						partType := "input_text"
-						if messageRole == "assistant" {
-							partType = "output_text"
+					switch contentType {
+					case "text":
+						appendTextContent(messageContentResult.Get("text").String())
+					case "image":
+						sourceResult := messageContentResult.Get("source")
+						if sourceResult.Exists() {
+							data := sourceResult.Get("data").String()
+							if data == "" {
+								data = sourceResult.Get("base64").String()
+							}
+							if data != "" {
+								mediaType := sourceResult.Get("media_type").String()
+								if mediaType == "" {
+									mediaType = sourceResult.Get("mime_type").String()
+								}
+								if mediaType == "" {
+									mediaType = "application/octet-stream"
+								}
+								dataURL := fmt.Sprintf("data:%s;base64,%s", mediaType, data)
+								appendImageContent(dataURL)
+							}
 						}
-
-						currentIndex := len(gjson.Get(message, "content").Array())
-						message, _ = sjson.Set(message, fmt.Sprintf("content.%d.type", currentIndex), partType)
-						message, _ = sjson.Set(message, fmt.Sprintf("content.%d.text", currentIndex), messageContentResult.Get("text").String())
-						template, _ = sjson.SetRaw(template, "input.-1", message)
-					} else if contentType == "tool_use" {
-						// Handle tool use content by creating function call message.
+					case "tool_use":
+						flushMessage()
 						functionCallMessage := `{"type":"function_call"}`
 						functionCallMessage, _ = sjson.Set(functionCallMessage, "call_id", messageContentResult.Get("id").String())
 						{
-							// Shorten tool name if needed based on declared tools
 							name := messageContentResult.Get("name").String()
 							toolMap := buildReverseMapFromClaudeOriginalToShort(rawJSON)
 							if short, ok := toolMap[name]; ok {
@@ -109,28 +152,18 @@ func ConvertClaudeRequestToCodex(modelName string, inputRawJSON []byte, _ bool)
 						}
 						functionCallMessage, _ = sjson.Set(functionCallMessage, "arguments", messageContentResult.Get("input").Raw)
 						template, _ = sjson.SetRaw(template, "input.-1", functionCallMessage)
-					} else if contentType == "tool_result" {
-						// Handle tool result content by creating function call output message.
+					case "tool_result":
+						flushMessage()
 						functionCallOutputMessage := `{"type":"function_call_output"}`
 						functionCallOutputMessage, _ = sjson.Set(functionCallOutputMessage, "call_id", messageContentResult.Get("tool_use_id").String())
 						functionCallOutputMessage, _ = sjson.Set(functionCallOutputMessage, "output", messageContentResult.Get("content").String())
 						template, _ = sjson.SetRaw(template, "input.-1", functionCallOutputMessage)
 					}
 				}
+				flushMessage()
 			} else if messageContentsResult.Type == gjson.String {
-				// Handle string content by creating appropriate message structure.
-				message := `{"type": "message","role":"","content":[]}`
-				messageRole := messageResult.Get("role").String()
-				message, _ = sjson.Set(message, "role", messageRole)
-
-				partType := "input_text"
-				if messageRole == "assistant" {
-					partType = "output_text"
-				}
-
-				message, _ = sjson.Set(message, "content.0.type", partType)
-				message, _ = sjson.Set(message, "content.0.text", messageContentsResult.String())
-				template, _ = sjson.SetRaw(template, "input.-1", message)
+				appendTextContent(messageContentsResult.String())
+				flushMessage()
 			}
 		}