feat: ASR语音转写管线 + 群聊身份混淆修复

- 新增ASR语音识别管线: QQ语音→下载音频→qwen3-asr-flash转录→注入用户消息 - 模型名称全部从models.json路由获取，无硬编码 - 修复群聊中AI将非管理员用户误称为管理员昵称(叶酱)的问题 - 助手回复缓存时标注[回复昵称 (UID)]，防止对话历史中身份混淆 - 群聊上下文指令改为肯定性表述，移除具体名称提及 - trace面板时间戳改为YYYY-MM-DD HH:MM:SS格式，耗时统一显示为秒 - 修复Go time.Duration纳秒值在前端显示问题(Duration/1e6转毫秒) - 新增video_tool插件模板 - 优化OpenAI adapter reasoning_content处理 Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-31 16:46:47 +08:00
parent d112fdd540
commit a9c79d7887
16 changed files with 780 additions and 67 deletions
@@ -321,6 +321,7 @@ func (a *Adapter) ToUnified(rawMessage interface{}) (*bridge.UnifiedMessage, err
 	}

 	var mentions []string
+	var replyToText string
 	if segments, ok := msg.Message.([]interface{}); ok {
 		for _, s := range segments {
 			if seg, ok := s.(map[string]interface{}); ok {
@@ -331,9 +332,24 @@ func (a *Adapter) ToUnified(rawMessage interface{}) (*bridge.UnifiedMessage, err
 						}
 					}
 				}
+				if seg["type"] == "reply" {
+					if data, ok := seg["data"].(map[string]interface{}); ok {
+						if t, ok := data["text"].(string); ok && t != "" {
+							replyToText = t
+						}
+						if id, ok := data["id"]; ok {
+							_ = id // message ID of the replied-to message
+						}
+					}
+				}
 			}
 		}
 	}
+
+	// Prepend reply context for the AI.
+	if replyToText != "" {
+		content = "【回复】" + truncateForReply(replyToText, 100) + "\n" + content
+	}
 	// Fallback: parse CQ at codes from string format (e.g. [CQ:at,qq=2254389756]).
 	if len(mentions) == 0 {
 		raw := msg.RawMessage
@@ -490,24 +506,76 @@ func (a *Adapter) ReadMessages(ctx context.Context, msgCh chan<- *OBv11Message)
 	}
 }

+// cqSimplifyMap maps CQ code types to simplified Chinese labels.
+var cqSimplifyMap = map[string]string{
+	"image":  "[图片]",
+	"reply":  "[回复]",
+	"face":   "[表情]",
+	"record": "[语音]",
+	"video":  "[视频]",
+	"file":   "[文件]",
+}
+
+// simplifyCQCodes replaces [CQ:type,...] codes with human-readable labels.
+func simplifyCQCodes(s string) string {
+	return cqAllRegex.ReplaceAllStringFunc(s, func(match string) string {
+		// match looks like "[CQ:image,file=xxx,url=xxx]"
+		// Extract the type (text between "CQ:" and the first "," or "]").
+		typ := match[4:] // strip "[CQ:"
+		for i, c := range typ {
+			if c == ',' || c == ']' {
+				typ = typ[:i]
+				break
+			}
+		}
+		if label, ok := cqSimplifyMap[typ]; ok {
+			return label
+		}
+		return "[" + typ + "]"
+	})
+}
+
 // extractText retrieves plain text from an OBv11 message.
+// CQ codes are converted to human-readable form where applicable (e.g. [CQ:at,qq=xxx] → @xxx).
 func extractText(msg *OBv11Message) string {
 	if msg.RawMessage != "" {
-		return msg.RawMessage
+		s := cqAtRegex.ReplaceAllString(msg.RawMessage, "@$1")
+		return simplifyCQCodes(s)
 	}
 	switch m := msg.Message.(type) {
 	case string:
-		return m
+		s := cqAtRegex.ReplaceAllString(m, "@$1")
+		return simplifyCQCodes(s)
 	case []interface{}:
 		var text string
 		for _, seg := range m {
 			if s, ok := seg.(map[string]interface{}); ok {
-				if s["type"] == "text" {
+				switch s["type"] {
+				case "text":
 					if data, ok := s["data"].(map[string]interface{}); ok {
 						if t, ok := data["text"].(string); ok {
 							text += t
 						}
 					}
+				case "at":
+					if data, ok := s["data"].(map[string]interface{}); ok {
+						if qq, ok := data["qq"].(string); ok {
+							text += "@" + qq
+						}
+					}
+				case "image":
+					text += "[图片]"
+				case "face":
+					text += "[表情]"
+				case "record":
+					text += "[语音]"
+				case "video":
+					text += "[视频]"
+				case "file":
+					text += "[文件]"
+				case "reply":
+					// Reply is handled separately in ToUnified with reply text.
+					text += "[回复]"
 				}
 			}
 		}
@@ -518,21 +586,50 @@ func extractText(msg *OBv11Message) string {

 var cqAtRegex = regexp.MustCompile(`\[CQ:at,qq=(\d+)\]`)
 var cqImageRegex = regexp.MustCompile(`\[CQ:image,[^\]]*\]`)
+var cqVideoRegex = regexp.MustCompile(`\[CQ:video,[^\]]*\]`)
+var cqRecordRegex = regexp.MustCompile(`\[CQ:record,[^\]]*\]`)
 var cqURLRegex = regexp.MustCompile(`\burl=([^,\]]+)`)
+var cqDurationRegex = regexp.MustCompile(`\bduration=(\d+)`)
+var cqAllRegex = regexp.MustCompile(`\[CQ:[^\]]+\]`)
 	var boldRegex = regexp.MustCompile(`\*\*(.+?)\*\*`)
 	var italicRegex = regexp.MustCompile(`\*(.+?)\*`)
 	var strikethroughRegex = regexp.MustCompile(`~~(.+?)~~`)

-// extractAttachments extracts image URLs from OBv11Message.
+func parseIntOr(s string, defaultVal int) int {
+	if s == "" {
+		return defaultVal
+	}
+	n := 0
+	for _, c := range s {
+		if c >= '0' && c <= '9' {
+			n = n*10 + int(c-'0')
+		} else {
+			return defaultVal
+		}
+	}
+	return n
+}
+
+// truncateForReply truncates reply preview text to keep messages readable.
+func truncateForReply(s string, maxLen int) string {
+	runes := []rune(s)
+	if len(runes) <= maxLen {
+		return s
+	}
+	return string(runes[:maxLen]) + "…"
+}
+
+// extractAttachments extracts image/video URLs from OBv11Message.
 // Handles both string format (CQ codes in raw_message) and array format (parsed segments).
 func extractAttachments(msg *OBv11Message) []bridge.Attachment {
 	var attachments []bridge.Attachment

-	// Array format: iterate segments looking for type="image".
+	// Array format: iterate segments looking for image and video.
 	if segments, ok := msg.Message.([]interface{}); ok {
 		for _, s := range segments {
 			if seg, ok := s.(map[string]interface{}); ok {
-				if seg["type"] != "image" {
+				segType, _ := seg["type"].(string)
+				if segType != "image" && segType != "video" && segType != "record" {
 					continue
 				}
 				data, _ := seg["data"].(map[string]interface{})
@@ -544,11 +641,17 @@ func extractAttachments(msg *OBv11Message) []bridge.Attachment {
 				if url == "" {
 					continue
 				}
-				attachments = append(attachments, bridge.Attachment{
-					Type:     "image",
+				att := bridge.Attachment{
+					Type:     segType,
 					URL:      url,
 					FileName: file,
-				})
+				}
+				if segType == "video" {
+					if d, ok := data["duration"].(float64); ok {
+						att.Duration = int(d)
+					}
+				}
+				attachments = append(attachments, att)
 			}
 		}
 		return attachments
@@ -561,14 +664,29 @@ func extractAttachments(msg *OBv11Message) []bridge.Attachment {
 			raw = s
 		}
 	}
-	matches := cqImageRegex.FindAllString(raw, -1)
-	for _, m := range matches {
+	// Images.
+	for _, m := range cqImageRegex.FindAllString(raw, -1) {
 		urlMatch := cqURLRegex.FindStringSubmatch(m)
 		if len(urlMatch) >= 2 {
-			attachments = append(attachments, bridge.Attachment{
-				Type: "image",
-				URL:  urlMatch[1],
-			})
+			attachments = append(attachments, bridge.Attachment{Type: "image", URL: urlMatch[1]})
+		}
+	}
+	// Videos.
+	for _, m := range cqVideoRegex.FindAllString(raw, -1) {
+		urlMatch := cqURLRegex.FindStringSubmatch(m)
+		if len(urlMatch) >= 2 {
+			dur := 0
+			if dm := cqDurationRegex.FindStringSubmatch(m); len(dm) >= 2 {
+				dur = parseIntOr(dm[1], 0)
+			}
+			attachments = append(attachments, bridge.Attachment{Type: "video", URL: urlMatch[1], Duration: dur})
+		}
+	}
+	// Records (voice messages).
+	for _, m := range cqRecordRegex.FindAllString(raw, -1) {
+		urlMatch := cqURLRegex.FindStringSubmatch(m)
+		if len(urlMatch) >= 2 {
+			attachments = append(attachments, bridge.Attachment{Type: "voice", URL: urlMatch[1]})
 		}
 	}
 	return attachments