#!/usr/bin/env python3 # -*- coding: utf-8 -*- """新思维2A工作台语音服务：静态文件 + /tts发音 + /asr识别 + /report上报 + /photo拍照上传""" import http.server, socketserver, subprocess, tempfile, os, json, urllib.parse, datetime, base64 ROOT = "/Users/zhangwen/.local/share/TeleAgent/TeleAgent的工作空间/学习工作台" WHISPER = ROOT + "/.temp/whisper-cpp/build/bin/whisper-cli" MODEL = ROOT + "/.temp/whisper-model/ggml-base.en.bin" REPORT_DIR = ROOT + "/reports" PHOTO_DIR = ROOT + "/photos" def ensure_report_dir(): os.makedirs(REPORT_DIR, exist_ok=True) def ensure_photo_dir(): os.makedirs(PHOTO_DIR, exist_ok=True) def load_report(date): ensure_report_dir() p = os.path.join(REPORT_DIR, date + ".json") if os.path.exists(p): try: with open(p, "r", encoding="utf-8") as f: return json.load(f) except Exception: return {"date": date, "subjects": {}} return {"date": date, "subjects": {}} def save_report(date, data): ensure_report_dir() p = os.path.join(REPORT_DIR, date + ".json") with open(p, "w", encoding="utf-8") as f: json.dump(data, f, ensure_ascii=False, indent=2) def mac_tts_to_wav(text, lang="en"): """用 macOS say + afconvert 生成 16kHz wav，返回字节；lang='zh' 用中文语音""" with tempfile.TemporaryDirectory() as td: aiff = os.path.join(td, "t.aiff") wav = os.path.join(td, "t.wav") voice = "Tingting" if lang == "zh" else "Samantha" subprocess.run(["say", "-o", aiff, "-v", voice, text[:300]], check=True, timeout=20, stderr=subprocess.DEVNULL, stdout=subprocess.DEVNULL) subprocess.run(["afconvert", "-f", "WAVE", "-d", "LEI16@16000", "-c", "1", aiff, wav], check=True, timeout=20) with open(wav, "rb") as f: return f.read() def whisper_asr(wav_bytes): """whisper-cli 转写，返回文本（先统一转 16kHz 单声道，兼容手机任意采样率）""" with tempfile.TemporaryDirectory() as td: wav = os.path.join(td, "a.wav") out = os.path.join(td, "b.wav") with open(wav, "wb") as f: f.write(wav_bytes) try: subprocess.run(["afconvert", "-f", "WAVE", "-d", "LEI16@16000", "-c", "1", wav, out], check=True, timeout=20, stderr=subprocess.DEVNULL, stdout=subprocess.DEVNULL) wav = out except Exception: pass # 转换失败就用原始文件 r = subprocess.run( [WHISPER, "-m", MODEL, "-f", wav, "-nt", "-np", "--no-prints"], capture_output=True, text=True, timeout=60) return (r.stdout or "").strip() class Handler(http.server.SimpleHTTPRequestHandler): def __init__(self, *a, **kw): super().__init__(*a, directory=ROOT, **kw) def log_message(self, fmt, *args): pass # 静默访问日志 def end_headers(self): self.send_header("Cross-Origin-Opener-Policy", "same-origin") self.send_header("Cross-Origin-Embedder-Policy", "require-corp") super().end_headers() def _send(self, code, ctype, data): self.send_response(code) self.send_header("Content-Type", ctype) self.send_header("Content-Length", str(len(data))) self.send_header("Cache-Control", "no-store") self.end_headers() self.wfile.write(data) def do_GET(self): try: parsed = urllib.parse.urlparse(self.path) if parsed.path == "/tts": q = urllib.parse.parse_qs(parsed.query) text = (q.get("text") or [""])[0].strip() lang = (q.get("lang") or ["en"])[0].strip() or "en" if not text: self._send(400, "text/plain", b"missing text") return data = mac_tts_to_wav(text, lang) self._send(200, "audio/wav", data) return except Exception as e: self._send(500, "application/json", json.dumps({"error": str(e)}).encode()) return super().do_GET() def do_POST(self): if self.path == "/report": self._handle_report() return if self.path == "/photo": self._handle_photo() return if self.path != "/asr": self._send(404, "text/plain", b"not found") return try: length = int(self.headers.get("Content-Length") or 0) if length <= 0 or length > 20 * 1024 * 1024: self._send(400, "text/plain", b"bad size") return data = self.rfile.read(length) text = whisper_asr(data) self._send(200, "application/json", json.dumps({"text": text}).encode()) except Exception as e: try: self._send(200, "application/json", json.dumps({"error": str(e)}).encode()) except Exception: pass def _handle_photo(self): """语文当日练习拍照上传：{base64, unit, date} → 存 Mac photos/ + 企微推送""" try: length = int(self.headers.get("Content-Length") or 0) if length <= 0 or length > 15 * 1024 * 1024: self._send(400, "text/plain", b"bad size") return raw = self.rfile.read(length) payload = json.loads(raw.decode("utf-8")) b64 = payload.get("base64", "") if not b64: self._send(400, "application/json", json.dumps({"error": "no image"}).encode()) return img = base64.b64decode(b64) date = payload.get("date", datetime.date.today().isoformat()) unit = payload.get("unit", "") ensure_photo_dir() ddir = os.path.join(PHOTO_DIR, date) os.makedirs(ddir, exist_ok=True) fname = datetime.datetime.now().strftime("%H%M%S") + "_" + (unit or "yu") + ".jpg" fpath = os.path.join(ddir, fname) with open(fpath, "wb") as f: f.write(img) # 企微推送（后台异步，不阻塞返回） try: subprocess.Popen(["python3", os.path.join(ROOT, ".temp", "push_photo.py"), fpath, date, unit], stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL) except Exception: pass self._send(200, "application/json", json.dumps({"ok": True, "path": fpath}).encode()) except Exception as e: try: self._send(200, "application/json", json.dumps({"error": str(e)}).encode()) except Exception: pass def _handle_report(self): """工作台打卡上报：{subject:'en|yu|ma', date:'YYYY-MM-DD', summary:{...}}""" try: length = int(self.headers.get("Content-Length") or 0) if length <= 0 or length > 2 * 1024 * 1024: self._send(400, "text/plain", b"bad size") return raw = self.rfile.read(length) payload = json.loads(raw.decode("utf-8")) subject = payload.get("subject", "") date = payload.get("date", datetime.date.today().isoformat()) summary = payload.get("summary") or {} if subject not in ("en", "yu", "sh") or not isinstance(summary, dict): self._send(400, "application/json", json.dumps({"error": "bad payload"}).encode()) return rep = load_report(date) rep["subjects"][subject] = summary save_report(date, rep) self._send(200, "application/json", json.dumps({"ok": True}).encode()) except Exception as e: try: self._send(200, "application/json", json.dumps({"error": str(e)}).encode()) except Exception: pass class Server(socketserver.ThreadingTCPServer): allow_reuse_address = True daemon_threads = True if __name__ == "__main__": print("voice server on :8000") Server(("0.0.0.0", 8000), Handler).serve_forever()