import datetime
import re

class RUCAnalysis(object):
  def __init__(self, gsd):
    lines = gsd.split('\n')

    self.soundings = []

    self._parse_header(lines[:2])
    self._parse_other(lines[3:])

  def _parse_header(self, header_lines):
    pattern = r'(.*) analysis.*grid point (\d+\.\d+) nm / (\d+) deg from (.*):'
    match = re.match(pattern, header_lines[0])

    self.grid = {'source': match.group(4),
                 'offset_nm': float(match.group(2)),
                 'offset_deg': float(match.group(3))}

    time_bits = ' '.join(header_lines[1].split()[1:])
    
    self.valid = datetime.datetime.strptime(time_bits, '%H %d %b %Y')

  def _parse_other(self, other_lines):
    for line in other_lines:
      cols = line.split()
      
      if not (len(cols) == 7 and cols[0].isnumeric()):
        continue

      # Capture ValueError exceptions when trying to parse the strings to ints
      try:
        temp = int(cols[3])
        if temp == 99999:
          continue

        # Temp and dewpoint are in celsius * 10
        temp_c = temp / 10
        dewpoint_c = int(cols[4]) / 10

        # Altitude is in meters
        alt_ft = int(round(int(cols[2]) * 3.28084, 0))
        # Pressure is in decapascals
        pressure_mb = int(round(int(cols[1]) / 10, 0))

        self.soundings.append((alt_ft, pressure_mb, temp_c, dewpoint_c))

      except ValueError:
        continue