@xdfnet/ispeak 1.6.5 → 1.6.7

package/configs/hook-speak.sh

@@ -1,122 +1,199 @@
 #!/bin/bash
-# Stop Hook: 只播报本次停止时的最后一条 assistant 回复
-# iAgent 调用 Claude 时设 ISPEAK_SKIP=1，此时跳过（iAgent 自己播）
+# Claude Code / Codex 共用播报 Hook：
+# 只取最后一条 assistant 回复，加 `{source:<name>}` 前缀后发给 ispeakd。
 [[ "$ISPEAK_SKIP" == "1" ]] && exit 0
 
-# 来源参数: claude 或 codex
 SOURCE="${1:-claude}"
-
 SOCK="$HOME/.config/iSpeak/ispeak.sock"
 LOG="$HOME/.config/iSpeak/hook.log"
+STATE_FILE="$HOME/.config/iSpeak/hook.last"
+
+# Codex `notify` 会把 JSON 作为最后一个参数传入；
+# Claude/Claude 风格 Stop Hook 会把 JSON 写到 stdin。
+input="${2:-}"
+if [[ -z "$input" ]]; then
+  input=$(cat)
+fi
+input_file=$(mktemp)
+trap 'rm -f "$input_file"' EXIT
+printf "%s" "$input" > "$input_file"
+
+result=$(SOURCE="$SOURCE" HOOK_INPUT_FILE="$input_file" HOOK_STATE_FILE="$STATE_FILE" node <<'NODE' 2>/dev/null
+const fs = require("fs");
+const crypto = require("crypto");
+
+(() => {
+  const input = readFile(process.env.HOOK_INPUT_FILE || "");
+  const payload = parseJSON(input) || {};
+  const source = process.env.SOURCE || "";
+  const stateFile = process.env.HOOK_STATE_FILE || "";
+  const result = source.startsWith("codex")
+    ? lastCodexAssistant(payload)
+    : lastClaudeAssistant(payload);
+
+  if (!result.text) {
+    return;
+  }
+
+  if (stateFile && result.turnId) {
+    if (isDuplicateTurn(stateFile, source, result.turnId)) {
+      return;
+    }
+    saveTurn(stateFile, source, result.turnId, result.text);
+  } else if (stateFile) {
+    saveTurn(stateFile, source, "", result.text);
+  }
 
-input=$(cat)
-
-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 (_) {}
-      });
-    ' "$key"
-    return
-  fi
-
-  printf "%s" "$input" | sed -n "s/.*\"$key\"[[:space:]]*:[[:space:]]*\"\([^\"]*\)\".*/\1/p"
+  process.stdout.write(result.text);
+})();
+
+function lastClaudeAssistant(payload) {
+  const direct = firstString(payload.last_assistant_message, payload.message);
+  if (direct) return { text: direct, turnId: extractTurnId(payload) };
+
+  const transcript = firstString(payload.transcript_path, payload.transcriptPath);
+  return transcript ? lastAssistantFromTranscript(transcript, "claude") : { text: "", turnId: extractTurnId(payload) };
 }
 
-extract_last_assistant_text() {
-  local transcript="$1"
-
-  if command -v node >/dev/null 2>&1; then
-    node -e '
-      const fs = require("fs");
-      const file = process.argv[1];
-      let last = "";
-
-      function collectText(content) {
-        const out = [];
-        if (typeof content === "string") {
-          return content;
-        }
-        if (!Array.isArray(content)) return "";
-        for (const item of content) {
-          if (item && typeof item.text === "string") out.push(item.text);
-        }
-        return out.join(" ");
-      }
+function lastCodexAssistant(payload) {
+  const direct = firstString(
+    payload["last-assistant-message"],
+    payload.last_assistant_message,
+    payload.lastAssistantMessage,
+    payload.message,
+    payload.lastMessage
+  );
+  if (direct) return { text: direct, turnId: extractTurnId(payload) };
+
+  const transcript = firstString(
+    payload.transcript_path,
+    payload.transcriptPath,
+    payload["transcript-path"]
+  );
+  return transcript ? lastAssistantFromTranscript(transcript, "codex") : { text: "", turnId: extractTurnId(payload) };
+}
+
+function readFile(file) {
+  try {
+    return fs.readFileSync(file, "utf8");
+  } catch {
+    return "";
+  }
+}
 
-      for (const line of fs.readFileSync(file, "utf8").split(/\r?\n/)) {
-        if (!line.trim()) continue;
-        try {
-          const event = JSON.parse(line);
-          if (event.role === "assistant") last = collectText(event.content) || last;
-          if (event.message && event.message.role === "assistant") last = collectText(event.message.content) || last;
-        } catch (_) {}
+function parseJSON(text) {
+  try {
+    return JSON.parse(text);
+  } catch {
+    return null;
+  }
+}
+
+function firstString(...values) {
+  for (const value of values) {
+    if (typeof value === "string" && value !== "") return value;
+  }
+  return "";
+}
+
+function collectText(content) {
+  if (typeof content === "string") return content;
+  if (!Array.isArray(content)) return "";
+  return content
+    .map(item => item && typeof item.text === "string" ? item.text : "")
+    .filter(Boolean)
+    .join(" ");
+}
+
+function lastAssistantFromTranscript(file, source) {
+  let data = "";
+  try {
+    data = fs.readFileSync(file, "utf8");
+  } catch {
+    return "";
+  }
+
+  let last = "";
+  let turnId = "";
+  for (const line of data.split(/\r?\n/)) {
+    if (!line.trim()) continue;
+    const event = parseJSON(line);
+    if (!event) continue;
+
+    if (source === "claude") {
+      if (event.role === "assistant") {
+        last = collectText(event.content) || last;
       }
-      process.stdout.write(last);
-    ' "$transcript" 2>/dev/null
-    return
-  fi
-
-  awk '
-    {
-      if (match($0, /"role"[[:space:]]*:[[:space:]]*"assistant"/)) {
-        if (match($0, /"content"[[:space:]]*:[[:space:]]*\[/)) {
-          gsub(/[^{]*\[/, "", $0)
-          gsub(/\].*/, "", $0)
-          msg = ""
-          while (match($0, /"text"[[:space:]]*:[[:space:]]*"[^"]*"/)) {
-            t = substr($0, RSTART, RLENGTH)
-            gsub(/"text"[[:space:]]*:[[:space:]]*"/, "", t)
-            gsub(/"$/, "", t)
-            if (t != "") msg = msg " " t
-            $0 = substr($0, RSTART + RLENGTH)
-          }
-          if (msg != "") last = msg
-        } else if (match($0, /"content"[[:space:]]*:[[:space:]]*"[^"]*"/)) {
-          t = substr($0, RSTART, RLENGTH)
-          gsub(/"content"[[:space:]]*:[[:space:]]*"/, "", t)
-          gsub(/"$/, "", t)
-          if (t != "") last = t
-        }
+      if (event.message && event.message.role === "assistant") {
+        last = collectText(event.message.content) || last;
       }
     }
-    END {
-      gsub(/^[[:space:]]+|[[:space:]]+$/, "", last)
-      if (last != "") print last
+
+    if (source === "codex" &&
+      event.type === "response_item" &&
+      event.payload &&
+      event.payload.type === "message" &&
+      event.payload.role === "assistant"
+    ) {
+      last = collectText(event.payload.content) || last;
+      turnId = turnId || extractTurnId(event) || extractTurnId(event.payload);
     }
-  ' "$transcript" 2>/dev/null
+  }
+  return { text: last, turnId };
 }
 
-# 从 stdin JSON 提取最后一条消息；没有时再从 transcript 取最后一条 assistant
-transcript=$(json_value "transcript_path")
-last_msg=$(json_value "last_assistant_message")
+function extractTurnId(payload) {
+  return firstString(
+    payload.turn_id,
+    payload.turnId,
+    payload["turn-id"],
+    payload.session_id,
+    payload.sessionId,
+    payload["session-id"],
+    payload.thread_id,
+    payload.threadId,
+    payload["thread-id"]
+  );
+}
+
+function isDuplicateTurn(stateFile, source, turnId) {
+  const current = `${source}:${turnId}`;
+  try {
+    return fs.readFileSync(stateFile, "utf8").trim() === current;
+  } catch {
+    return false;
+  }
+}
+
+function saveTurn(stateFile, source, turnId, text) {
+  const current = `${source}:${turnId || textHash(text)}`;
+  try {
+    fs.mkdirSync(require("path").dirname(stateFile), { recursive: true });
+    fs.writeFileSync(stateFile, current, "utf8");
+  } catch {
+    // 去重失败不影响播报。
+  }
+}
+
+function textHash(text) {
+  return crypto.createHash("sha1").update(text, "utf8").digest("hex");
+}
 
-all_text="$last_msg"
+NODE
+)
 
-# Claude Code 部分版本不提供 last_assistant_message，此时只取 transcript 最后一条 assistant。
-if [[ -z "$all_text" && -n "$transcript" && -f "$transcript" ]]; then
-  extra=$(extract_last_assistant_text "$transcript")
-  if [[ -n "$extra" ]]; then
-    all_text="$extra"
-  fi
+if [[ "$ISPEAK_HOOK_PRINT_TEXT" == "1" ]]; then
+  printf "%s" "$result"
+  exit 0
 fi
 
 echo "=== $(date) ===" >> "$LOG"
 echo "SOURCE: $SOURCE" >> "$LOG"
-echo "TEXT_LEN: ${#all_text}" >> "$LOG"
-echo "PREVIEW: ${all_text:0:150}" >> "$LOG"
+echo "TEXT_LEN: ${#result}" >> "$LOG"
+echo "PREVIEW: ${result:0:150}" >> "$LOG"
 
-if [[ -n "$all_text" && -S "$SOCK" ]]; then
-  printf "{source:%s}%s" "$SOURCE" "$all_text" | nc -U -w5 "$SOCK" 2>> "$LOG"
+if [[ -n "$result" && -S "$SOCK" ]]; then
+  printf "{source:%s}%s" "$SOURCE" "$result" | nc -U -w5 "$SOCK" 2>> "$LOG"
   echo "SPOKE: OK" >> "$LOG"
 else
   echo "SPOKE: SKIP" >> "$LOG"