import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import ipdb
import glob
from datetime import datetime, timedelta
import pytz

from multiprocessing import Pool
from os.path import join, splitext

from modules.datapipeline import datetime_to_ms

def get_flist(f_glob):
    f_list = sorted(glob.glob(f_glob, recursive=True))
    return f_list

def load_only_imu(f):
    try:
        df = pd.read_json(f, lines=True, compression='gzip')
    except EOFError:
        df = pd.read_json(splitext(f)[0], lines=True)
    if df.empty: return
    data_df = pd.DataFrame(df['data'].tolist())
    df = pd.concat([df.drop('data', axis=1), data_df], axis=1)

    mask = pd.isna(df['accelerometer'])
    na_inds = df.loc[mask, :].index.values
    not_na_inds = df.loc[~mask, :].index.values

    df_na = df.drop(index=not_na_inds)
    df_not_na = df.drop(index=na_inds)
    return df_not_na

def get_mean_fs(df):
    time = df['timestamp'].values
    diff = time[1:] - time[:-1]
    fs = 1/np.mean(diff)
    print(max(diff))
    print(min(diff))
    plt.plot(diff)
    plt.show()
    return fs

def fname_fs(f):
    imu_df = load_only_imu(f)
    if imu_df is None: return
    fs = get_mean_fs(imu_df)
    return f, fs

def imu_start_end_time(hdr_fname, data_fname):
    imu_hdr = pd.read_json(hdr_fname, orient='index')
    imu_hdr = imu_hdr.to_dict().pop(0)
    imu_df = load_only_imu(data_fname)

    iso_tz = imu_hdr['created']
    tzinfo = pytz.timezone(imu_hdr['timezone'])
    # adjust for UTC
    start_time = datetime.fromisoformat(iso_tz[:-1]) + timedelta(hours=11)
    imu_times = imu_df['timestamp'].values

    imu_datetimes = [start_time + timedelta(seconds=val) \
                     for val in imu_times]

    nbins = len(imu_times)
    est_end = datetime.fromtimestamp(imu_datetimes[0].timestamp() + nbins*(1/120))
    print("endtime: {0}\testimate: {1}".format(imu_datetimes[-1],
                                                       est_end))
    return imu_datetimes[0], imu_datetimes[-1]

def str_to_datetime(time_in):
    fmt ="%d/%m/%Y %H:%M:%S.%f" 
    dstr = datetime.strptime(time_in, fmt)
    return dstr

def harness_start_end_time(fname, fs=100):
    df = pd.read_csv(fname)
    t0 = str_to_datetime(df['Time'].iloc[0])
    t1 = str_to_datetime(df['Time'].iloc[-1])

    nbins = len(df)
    est_end = datetime.fromtimestamp(t0.timestamp() + nbins*(1/fs))
    print("endtime: {0}\testimate: {1}".format(t1,
                                                       est_end))
    return t0, t1

if __name__ == '__main__':
    data_dir = '/data/rqchia/aria_walk/Data/test-rest'
    f_glob = join(data_dir, '**', 'imudata.gz')
    h_glob = join(data_dir, '**', 'recording.g3')

    data_fname = glob.glob(f_glob)[0]
    hdr_fname = glob.glob(h_glob)[0]

    t0, t1 = imu_start_end_time(hdr_fname, data_fname)

    a_glob = join(data_dir, '**', '*_Accel.csv')
    b_glob = join(data_dir, '**', '*_Breathing.csv')
    s_glob = join(data_dir, '**', '*_SummaryEnhanced.csv')

    a_fname = glob.glob(a_glob)[0]
    a0, a1 = harness_start_end_time(a_fname, fs=100)

    b_fname = glob.glob(b_glob)[0]
    b0, b1 = harness_start_end_time(b_fname, fs=25)

    s_fname = glob.glob(s_glob)[0]
    s0, s1 = harness_start_end_time(s_fname, fs=1)

    print(f"imu {t0}\t{t1}\n"
          f"acc {a0}\t{a1}\n"
          f"bre {b0}\t{b1}\n"
          f"sum {s0}\t{s1}\n")

    # flist = get_flist(f_glob)
    # tmp = []
    # tmp = fname_fs(flist[0])
    # # with Pool(10) as p:
    # #     tmp = p.map(fname_fs, flist)

    # df = pd.DataFrame(tmp, columns=['fname', 'fs'])
    # df.dropna(inplace=True)
    # mask = df['fs'] > 10
    # df = df[mask]
    # print(df)