jjy-py/jjy.py

#!/usr/bin/env python3
# JJY.py: a port of jjy.js library and a standalone script
# to synchronize time on JJY-enabled wristwatches via headphones
# Depends on pyaudio and ntplib
# Created by Luxferre in 2024, released into public domain

import pyaudio, ntplib, math, array, time, datetime

OP_FREQ = 40000/3 # emitted frequency, Hz

# BCD and parity calculation helpers

def toBCD(val: int):
  return (val%10) + ((int(val/10)%10)<<4) + (int(val/100)<<8)

def calc_parity(val: int):
  i = 0
  while val != 0:
    i ^= (val & 1)
    val >>= 1
  return i

# internal time representation from unix time
# only fetches the fields necessary for JJY implementation
def intreptime(unixtm):
  res = {}
  tm = time.gmtime(unixtm)
  res['year'] = toBCD(tm.tm_year % 100)
  res['yday'] = toBCD(tm.tm_yday)
  res['hour'] = toBCD(tm.tm_hour)
  res['minute'] = toBCD(tm.tm_min)
  res['second'] = tm.tm_sec
  # in Python, Monday is 0; in JJY, Sunday is 0
  res['wday'] = (tm.tm_wday + 1) % 7
  res['unix'] = int(unixtm) # save the unix time representation
  return res

# time fetching part (returns JST)
def fetchtime(params = {}):
  delta = 0.0
  offset = 32400
  if 'delta' in params:
    delta = float(params['delta']) / 1000
  if 'offset' in params: # base offset from UTC in seconds
    offset = int(params['offset'])
  if 'server' in params and params['server'] is not None: # NTP server set
    ntpver = 3
    if 'version' in params:
      ntpver = params['version']
    c = ntplib.NTPClient()
    resp = c.request(params['server'], version=ntpver)
    unixtm = resp.tx_time
  else: # use current system time by default
    unixtm = time.time()
  unixtm += offset + delta # account for delta
  return intreptime(unixtm)

# timecode generation part
# accepts the result of fetchtime function
def gentimecode(ts):
  timecode = [2,0,0,0,0,0,0,0,0,2,0,0,0,0,0,0,0,0,0,2,0,0,0,0,0,0,0,
    0,0,2,0,0,0,0,0,0,0,0,0,2,0,0,0,0,0,0,0,0,0,2,0,0,0,0,0,0,0,0,0,2]
  # populate minute
  timecode[1] = (ts['minute']>>6)&1
  timecode[2] = (ts['minute']>>5)&1
  timecode[3] = (ts['minute']>>4)&1
  timecode[5] = (ts['minute']>>3)&1
  timecode[6] = (ts['minute']>>2)&1
  timecode[7] = (ts['minute']>>1)&1
  timecode[8] = ts['minute']&1
  # populate hour
  timecode[12] = (ts['hour']>>5)&1
  timecode[13] = (ts['hour']>>4)&1
  timecode[15] = (ts['hour']>>3)&1
  timecode[16] = (ts['hour']>>2)&1
  timecode[17] = (ts['hour']>>1)&1
  timecode[18] = ts['hour']&1
  # populate day number
  timecode[22] = (ts['yday']>>9)&1
  timecode[23] = (ts['yday']>>8)&1
  timecode[25] = (ts['yday']>>7)&1
  timecode[26] = (ts['yday']>>6)&1
  timecode[27] = (ts['yday']>>5)&1
  timecode[28] = (ts['yday']>>4)&1
  timecode[30] = (ts['yday']>>3)&1
  timecode[31] = (ts['yday']>>2)&1
  timecode[32] = (ts['yday']>>1)&1
  timecode[33] = ts['yday']&1
  # populate parity bits
  timecode[36] = calc_parity(ts['hour'])
  timecode[37] = calc_parity(ts['minute'])
  # populate year
  timecode[41] = (ts['year']>>7)&1
  timecode[42] = (ts['year']>>6)&1
  timecode[43] = (ts['year']>>5)&1
  timecode[44] = (ts['year']>>4)&1
  timecode[45] = (ts['year']>>3)&1
  timecode[46] = (ts['year']>>2)&1
  timecode[47] = (ts['year']>>1)&1
  timecode[48] = ts['year']&1
  # populate day of the week
  timecode[50] = (ts['wday']>>2)&1
  timecode[51] = (ts['wday']>>1)&1
  timecode[52] = ts['wday']&1
  return timecode

# generate an audio data chunk of specified duration
def gen_audio(duration, freq=OP_FREQ, sr=48000):
  smps = int(sr * duration)
  # create the sine wave array for the whole second
  rawdata = []
  for k in range(0, sr):
    v = math.sin(2 * math.pi * k * freq / sr)
    if k > smps: # reduced power mode
      v *= 0.1
    rawdata.append(int(v * 32767)) # max gain
  return array.array('h', rawdata).tobytes()

# global buffers for audio data and current position
curstream = b''
streampos = 0

# bitcode transmission callback
def bitcode_transmit(in_data, frame_count, time_info, status):
  global curstream, streampos
  framelen = frame_count << 1 # 2 bytes per frame as we're using int16
  framedata = curstream[streampos:streampos+framelen]
  streampos += framelen
  return (framedata, pyaudio.paContinue)

# main logic is here
def start_transmission(timeparams):
  global curstream
  p = pyaudio.PyAudio()
  sr = timeparams['sr']
  mins = timeparams['duration']
  jjy_bit_chunks = [ # pregenerate the chunks
    gen_audio(0.8, OP_FREQ, sr), # data bit 0
    gen_audio(0.5, OP_FREQ, sr), # data bit 1
    gen_audio(0.2, OP_FREQ, sr)  # marker bit
  ]
  ts = fetchtime(timeparams) # get the current timestamp
  print('Time fetched (Unix):', ts['unix'])
  bitcode = gentimecode(ts)[ts['second']+1:] # slice the rest of current minute
  nextmin = ts['unix'] - ts['second'] # rewind to start of the minute
  for i in range(0, mins): # generate bitcode for the next N minutes
    nextmin += 60 # calc the next minute
    bitcode += gentimecode(intreptime(nextmin))
  print("Transmitting... Press Ctrl+C to exit")
  # wait for the next second to start (roughly, with all the call overhead)
  time.sleep((1 - datetime.datetime.now().microsecond/1000000)/2)
  # open a PyAudio stream with callback
  stream = p.open(format=pyaudio.paInt16, channels=1, frames_per_buffer=16384,
            rate=sr, output=True, stream_callback=bitcode_transmit)
  curstream = jjy_bit_chunks[bitcode.pop(0)] # preload the first second
  while stream.is_active(): # wait for the stream to finish
    if len(bitcode) > 0: # feed the stream in parallel
      curstream += jjy_bit_chunks[bitcode.pop(0)]
    time.sleep(0.75) # feeding the stream should be faster than realtime
  # close audio
  stream.stop_stream()
  stream.close()
  p.terminate()
  print("Transmission ended")

if __name__ == '__main__':
  from argparse import ArgumentParser
  parser = ArgumentParser(description='JJY.py: an opensource longwave time synchronizer for JJY40-enabled watches and clocks', epilog='(c) Luxferre 2024 --- No rights reserved <https://unlicense.org>')
  parser.add_argument('-t', '--duration', type=int, default=30, help='Transmission duration (in minutes, default 30)')
  parser.add_argument('-d', '--delta', type=int, default=0, help='Manual delta correction (in ms, must be determined individually, 0 by default)')
  parser.add_argument('-o', '--tz-offset', type=float, default=9, help='Timezone offset from UTC to transmit (in hours, default 9 - corresponds to JST)')
  parser.add_argument('-r', '--sample-rate', type=int, default=48000, help='Transmission sampling rate (in Hz, default 48000)')
  parser.add_argument('-s', '--ntp-server', type=str, default=None, help='NTP server to sync from (if not specified then will sync from the local system time)')
  parser.add_argument('-n', '--ntp-version', type=int, default=4, help='NTP protocol version to use (default 4)')
  args = parser.parse_args()
  params = { # populate parameters from the command line
    'server': args.ntp_server, 'version': args.ntp_version,
    'delta': args.delta, 'offset': int(args.tz_offset * 3600),
    'sr': args.sample_rate, 'duration': args.duration
  }
  start_transmission(params)