@xdfnet/ispeak 1.6.1 → 1.6.2

package/README.md

@@ -1,6 +1,6 @@
 # iSpeak
 
-![Version](https://img.shields.io/badge/version-1.6.1-blue)
+![Version](https://img.shields.io/badge/version-1.6.2-blue)
 [![License: MIT](https://img.shields.io/badge/License-MIT-blue.svg)](https://opensource.org/licenses/MIT)
 [![Go Version](https://img.shields.io/badge/Go-1.26-blue)](https://golang.org/dl/)
 ![Platform](https://img.shields.io/badge/platform-macOS-green)
@@ -51,11 +51,12 @@ brew install ffmpeg
 git clone https://github.com/xdfnet/iSpeak.git && cd iSpeak && make install
 ```
 
-安装时手动输入 API Key，然后验证：
+安装后编辑 API Key，然后验证：
 
 ```bash
+open ~/.config/iSpeak/config.json
 ispeak status
-ispeak test "iSpeak 准备好了"
+ispeak "iSpeak 准备好了"
 ```
 
 ## 工作原理

package/configs/hook-speak.sh

@@ -11,52 +11,107 @@ LOG="$HOME/.config/iSpeak/hook.log"
 
 input=$(cat)
 
-# 从 stdin JSON 提取 transcript 路径和最后一条消息
-# 简单 JSON 解析（不依赖 python3）
-transcript=$(echo "$input" | sed -n 's/.*"transcript_path"\s*:\s*"\([^"]*\)".*/\1/p')
-last_msg=$(echo "$input" | sed -n 's/.*"last_assistant_message"\s*:\s*"\([^"]*\)".*/\1/p')
+json_value() {
+  local key="$1"
+  if command -v node >/dev/null 2>&1; then
+    printf "%s" "$input" | node -e '
+      const key = process.argv[1];
+      let input = "";
+      process.stdin.setEncoding("utf8");
+      process.stdin.on("data", chunk => input += chunk);
+      process.stdin.on("end", () => {
+        try {
+          const value = JSON.parse(input)[key];
+          if (typeof value === "string") process.stdout.write(value);
+        } catch (_) {}
+      });
+    ' "$key"
+    return
+  fi
 
-all_text="$last_msg"
+  printf "%s" "$input" | sed -n "s/.*\"$key\"[[:space:]]*:[[:space:]]*\"\([^\"]*\)\".*/\1/p"
+}
 
-# 如果有 transcript 文件，提取最近 30 秒内的所有 assistant 消息
-if [[ -n "$transcript" && -f "$transcript" ]]; then
-  # 计算 30 秒前的时间戳（毫秒）
-  cutoff=$(($(date +%s) * 1000 - 30000))
+extract_recent_assistant_text() {
+  local transcript="$1"
+  local cutoff="$2"
 
-  # 用 awk 解析 JSON lines，提取 role=assistant 且 timestamp > cutoff 的 text
-  extra=$(awk -v cutoff="$cutoff" '
+  if command -v node >/dev/null 2>&1; then
+    node -e '
+      const fs = require("fs");
+      const file = process.argv[1];
+      const cutoff = Number(process.argv[2]);
+      const out = [];
+
+      function collectText(content) {
+        if (typeof content === "string") {
+          out.push(content);
+          return;
+        }
+        if (!Array.isArray(content)) return;
+        for (const item of content) {
+          if (item && typeof item.text === "string") out.push(item.text);
+        }
+      }
+
+      for (const line of fs.readFileSync(file, "utf8").split(/\r?\n/)) {
+        if (!line.trim()) continue;
+        try {
+          const event = JSON.parse(line);
+          if (typeof event.timestamp === "number" && event.timestamp < cutoff) continue;
+          if (event.role === "assistant") collectText(event.content);
+          if (event.message && event.message.role === "assistant") collectText(event.message.content);
+        } catch (_) {}
+      }
+      process.stdout.write([...new Set(out.filter(Boolean))].join(" "));
+    ' "$transcript" "$cutoff" 2>/dev/null
+    return
+  fi
+
+  awk -v cutoff="$cutoff" '
     {
-      # 提取 timestamp
-      if (match($0, /"timestamp"\s*:\s*[0-9]+/)) {
+      if (match($0, /"timestamp"[[:space:]]*:[[:space:]]*[0-9]+/)) {
         ts = substr($0, RSTART, RLENGTH)
         gsub(/[^0-9]/, "", ts)
         ts = int(ts)
         if (ts < cutoff) next
       }
 
-      # 提取 role
-      if (match($0, /"role"\s*:\s*"assistant"/)) {
-        # 提取 content（可能是字符串或数组）
-        if (match($0, /"content"\s*:\s*\[/)) {
-          # 数组形式，提取所有 text 字段
+      if (match($0, /"role"[[:space:]]*:[[:space:]]*"assistant"/)) {
+        if (match($0, /"content"[[:space:]]*:[[:space:]]*\[/)) {
           gsub(/[^{]*\[/, "", $0)
           gsub(/\].*/, "", $0)
-          while (match($0, /"text"\s*:\s*"[^"]*"/)) {
+          while (match($0, /"text"[[:space:]]*:[[:space:]]*"[^"]*"/)) {
             t = substr($0, RSTART, RLENGTH)
-            gsub(/"text"\s*:\s*"/, "", t)
+            gsub(/"text"[[:space:]]*:[[:space:]]*"/, "", t)
             gsub(/"$/, "", t)
             if (t != "") print t
             $0 = substr($0, RSTART + RLENGTH)
           }
-        } else if (match($0, /"content"\s*:\s*"\([^"]*\)"/)) {
+        } else if (match($0, /"content"[[:space:]]*:[[:space:]]*"[^"]*"/)) {
           t = substr($0, RSTART, RLENGTH)
-          gsub(/"content"\s*:\s*"/, "", t)
+          gsub(/"content"[[:space:]]*:[[:space:]]*"/, "", t)
           gsub(/"$/, "", t)
           if (t != "") print t
         }
       }
     }
-  ' "$transcript" 2>/dev/null | sort -u | tr '\n' ' ')
+  ' "$transcript" 2>/dev/null | sort -u | tr '\n' ' '
+}
+
+# 从 stdin JSON 提取 transcript 路径和最后一条消息
+transcript=$(json_value "transcript_path")
+last_msg=$(json_value "last_assistant_message")
+
+all_text="$last_msg"
+
+# 如果有 transcript 文件，提取最近 30 秒内的所有 assistant 消息
+if [[ -n "$transcript" && -f "$transcript" ]]; then
+  # 计算 30 秒前的时间戳（毫秒）
+  cutoff=$(($(date +%s) * 1000 - 30000))
+
+  # 优先用 JSON parser，Node 不存在时回退到简易 awk。
+  extra=$(extract_recent_assistant_text "$transcript" "$cutoff")
 
   if [[ -n "$extra" ]]; then
     all_text="$extra"

package/main.go

@@ -387,6 +387,17 @@ func loadConfig() Config {
 		}
 		var cfg Config
 		if json.Unmarshal(data, &cfg) == nil && cfg.APIKey != "" {
+			if err := validateConfig(cfg); err != nil {
+				log.Printf("配置文件无效: %s err=%v", p, err)
+				configCacheMu.Lock()
+				if configCacheValid {
+					cached := configCache
+					configCacheMu.Unlock()
+					return cached
+				}
+				configCacheMu.Unlock()
+				return cfg
+			}
 			log.Printf("配置文件: %s", p)
 			if cfg.DefaultVoice != nil {
 				log.Printf("默认音色: %s (%s)", cfg.DefaultVoice.VoiceType, cfg.DefaultVoice.ResourceID)
@@ -779,8 +790,26 @@ func validateConfig(cfg Config) error {
 	if cfg.Endpoint == "" {
 		return fmt.Errorf("endpoint 未设置")
 	}
-	if cfg.DefaultVoice == nil || cfg.DefaultVoice.VoiceType == "" {
-		return fmt.Errorf("defaultVoice 未设置")
+	if err := validateVoiceInfo("defaultVoice", cfg.DefaultVoice); err != nil {
+		return err
+	}
+	for source, voice := range cfg.SourceVoices {
+		if err := validateVoiceInfo(fmt.Sprintf("sourceVoices.%s", source), voice); err != nil {
+			return err
+		}
+	}
+	return nil
+}
+
+func validateVoiceInfo(name string, voice *VoiceInfo) error {
+	if voice == nil {
+		return fmt.Errorf("%s 未设置", name)
+	}
+	if voice.VoiceType == "" {
+		return fmt.Errorf("%s.voice_type 未设置", name)
+	}
+	if voice.ResourceID == "" {
+		return fmt.Errorf("%s.resourceId 未设置", name)
 	}
 	return nil
 }
@@ -809,6 +838,10 @@ func handleConnection(conn net.Conn, engine *TaskEngine) {
 	defer conn.Close()
 
 	cfg := loadConfig()
+	if err := validateConfig(cfg); err != nil {
+		log.Printf("配置错误，跳过本次播报: %v", err)
+		return
+	}
 
 	var sb strings.Builder
 	scanner := bufio.NewScanner(conn)
@@ -816,6 +849,10 @@ func handleConnection(conn net.Conn, engine *TaskEngine) {
 	for scanner.Scan() {
 		sb.WriteString(scanner.Text())
 	}
+	if err := scanner.Err(); err != nil {
+		log.Printf("读取 socket 消息失败: %v", err)
+		return
+	}
 
 	text := strings.TrimSpace(sb.String())
 	if text == "" {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@xdfnet/ispeak",
-  "version": "1.6.1",
+  "version": "1.6.2",
   "description": "Local macOS TTS daemon for AI coding assistants, powered by Volcengine streaming TTS.",
   "license": "MIT",
   "homepage": "https://github.com/xdfnet/iSpeak#readme",

package/scripts/ispeak

@@ -2,7 +2,7 @@
 # ispeak — iSpeak 控制命令
 set -euo pipefail
 
-VERSION="1.6.1"
+VERSION="1.6.2"
 SOCK="$HOME/.config/iSpeak/ispeak.sock"
 PLIST="$HOME/Library/LaunchAgents/com.iSpeak.plist"
 CMD_NAME="$(basename "$0")"