"""Parse data files from the International Earth Rotation Service.

See:
https://datacenter.iers.org/eop.php
ftp://cddis.gsfc.nasa.gov/pub/products/iers/readme.finals2000A

"""
import numpy as np
from ..constants import DAY_S

inf = float('inf')

# This regular expression must remain a plain string; attempting to
# compile it triggers a bug in older NumPy versions like 1.14.3:
# https://github.com/skyfielders/python-skyfield/issues/372
_R = (b'(?m)^......(.........) . '
      b'(.\\d.......)......... '
      b'(.\\d.......).........  '
      b'.(.\\d........)')

def parse_x_y_dut1_from_finals_all(f):
    return np.fromregex(f, _R, [
        ('utc_mjd', float),
        ('x_arcseconds', float),
        ('y_arcseconds', float),
        ('dut1', float),
    ])

def install_polar_motion_table(ts, finals_data):
    t = ts.utc(1858, 11, 17.0 + finals_data['utc_mjd'])
    ts.polar_motion_table = (
        t.tt,
        np.array(finals_data['x_arcseconds']),
        np.array(finals_data['y_arcseconds']),
    )

def build_timescale_arrays(utc_mjd, dut1):
    big_jumps = np.diff(dut1) > 0.9
    leap_second_mask = np.concatenate([[False], big_jumps])
    tt_minus_utc = np.cumsum(leap_second_mask) + 32.184 + 12.0

    daily_tt = utc_mjd + tt_minus_utc / DAY_S + 2400000.5
    daily_delta_t = (tt_minus_utc - dut1).round(7)
    leap_dates = utc_mjd[leap_second_mask]

    # Since "finals2000A.all" starts on 1973-01-02 and leaves out the
    # first two leap seconds, add them back.  (But check first, in case
    # the user loads a more complete data file than "finals2000A.all".)
    first_leap = leap_dates[0] if len(leap_dates) else 0
    more_leaps = [mjd for mjd in (41499.0, 41683.0) if first_leap > mjd]
    leap_dates = np.concatenate([more_leaps, leap_dates])

    leap_dates += 2400000.5
    leap_offsets = np.arange(11.0, len(leap_dates) + 11.0)
    return daily_tt, daily_delta_t, leap_dates, leap_offsets

# Compatibility with older Skyfield versions:

def parse_dut1_from_finals_all(f):
    data = parse_x_y_dut1_from_finals_all(f)
    return data['utc_mjd'], data['dut1']