#!/usr/bin/python

#################################################################################
#										#
# $OPCSITE/python/convert_altimeter_hdf5_to_ascii.py				#
#										#
# Script to convert a NESDIS HDF5 altimeter file to the ascii required by 	#
# nmap2.									#
#										#
# A note on the h5py import errors:
#  import h5
#/usr/lib64/python2.6/site-packages/h5py/__init__.py:34: RuntimeWarning: numpy.dtype size changed, may indicate binary incompatibility
#  import h5, h5a, h5d, h5f, h5fd, h5g, h5l, h5o, h5i, h5p, h5r, h5s, h5t, h5z
#										#
# from the the Red Hat bugzilla page:
# 	The warning does not affect execution of the python module.
#	A temporary workaround is to downgrade numpy to numpy-1.3.0-6.2.el6.x86_64.
#										#
# Command line args:								#
#	-f /path/and/file/name		altimeter data in hdf5			#
#										#
# History:									#
# F.Achorn/OPC		10/24/2016	Script satarted
#################################################################################

import os
import sys
import getopt
import logging
import re
import numpy as np
import h5py
import datetime

def main(argv):
    # setup the logger
    set_log_file()    
    # import the logger
    this_script_name=os.path.basename(sys.argv[0])
    logger=logging.getLogger(this_script_name)
    
    infile=""
    try: 
        opts, args = getopt.getopt(argv,"f:s:",["file="])
    except getopt.GetoptError:
        logger.error("%s -f <input_hdf5_file_with_path> -s <satellite s3a or s3b>", this_script_name)
        finish_script(11)

    logger.info('Command line opts, args:') 
    if (len(opts) ==0):
        logger.error("No command line arguments given.\n%s -f <input_hdf5_file_with_path> -s <satellite s3a or s3b>", this_script_name)
        finish_script(11)
    for opt, arg in opts:
        logger.info('    %s %s', opt, arg)
        if opt == '-f':
            if (os.path.isfile(arg)):
                infile=arg
                logger.info('Input file exists: %s',infile)
            else:
                logger.error('Input file does not exist: %s',infile)
                finish_script(21)
        if opt == '-s':
            if (arg == "s3a" or arg == "s3b"):
                satellite=arg
                logger.info('Satellite provided: %s',satellite)
            else:
                logger.error('Satellite unknown or not provided: %s',infile)
                finish_script(21)

    decode_hdf5(infile, satellite)
    finish_script(0)

def set_log_file(log_base=''):
    #print ("in set_log_file")
    HOME=os.environ['HOME']
    logDir=HOME+"/logs"
    this_script_name=os.path.basename(sys.argv[0])
    logger=logging.getLogger(this_script_name)
    if (not os.path.isdir(logDir)):
        #make the directory
        #logger.info('Creating %s/logs', HOME)
        os.makedirs(logDir)
        print ('Creating %s/logs', HOME)
   
    #Check that the log_base is at least 4 characters long
    re_pattern=re.compile('\w{4}')
    if (not re_pattern.match(log_base)):
        #use the script name
        #this_script_no_ext=os.path.splitext(this_script_name)[0]
        log_base=os.path.splitext(os.path.basename(sys.argv[0]))[0]

    #Determine the day of the week
    day_of_week=datetime.datetime.today().weekday()+1
    # set log name
    log_name=logDir + '/' + log_base + "_" + str(day_of_week) + ".log"
    print ("log name = " +log_name)


    #Setup the logger
    logger=logging.getLogger(this_script_name)
    logger.setLevel(logging.DEBUG)
    
    #create the file handle 
    fh = logging.FileHandler(log_name)
    fh.setLevel(logging.INFO)
   
    #create the console logger to write to STDERR
    ch = logging.StreamHandler()
    ch.setLevel(logging.WARN)
    #ch.setLevel(logging.INFO)

    #set the log format for everything
    fh_formatter = logging.Formatter('%(process)d %(asctime)s %(levelname)s %(funcName)s| %(message)s')
    fh.setFormatter(fh_formatter) 
    ch_formatter = logging.Formatter('%(levelname)s %(message)s')
    ch.setFormatter(ch_formatter)
    
    # add the hadlers to the logger
    logger.addHandler(fh)
    logger.addHandler(ch)
    logger.info('-------------------------------------------')
    logger.info('-------------------------------------------')
    logger.info('Starting %s', this_script_name)
    #logger.warn("Log started: %s", log_name)
    #logger.debug('debug message')
    #logger.info('info message')
    #logger.warn('warn message')
    #logger.error('error message')
    #logger.critical('critical message')

    #cleanup log from 1 week ago
    tomorrow=datetime.datetime.today() + datetime.timedelta(days=1)
    tomorrow_day_of_week = tomorrow.weekday()+1
    tomorrow_log_name = logDir + '/' + log_base + "_" + str(tomorrow_day_of_week) + ".log"
    if (os.path.isfile(tomorrow_log_name)):
        os.remove(tomorrow_log_name)
        logger.warn("Removing tomorrow's log: $s", tomorrow_log_name) 

    return()

def finish_script(return_code):
    # import the logger
    this_script_name=os.path.basename(sys.argv[0])
    logger=logging.getLogger(this_script_name)

    logger.info('Done with %s', this_script_name)
    logging.shutdown()
    sys.exit(return_code)

def decode_hdf5(h5file, sat):   

    this_script_name=os.path.basename(sys.argv[0])
    logger=logging.getLogger(this_script_name)

    f = h5py.File(h5file, "r")
    # print f.name

    #root_group = f['/']
    # for name in root_group:
    #     print name
        
    # swh_dataset=f['/swh_ocean_01_ku']
    # get the date of each Sentinel ob in "days since 2000-01-01 00:00:00.0" as an ascii file
    UTC_day_dataset=f.get('/UTC_day_01')
    UTC_sec_dataset=f.get('/UTC_sec_01')
    
    # We probably don't have np.datetime64 available (starts in numpy v1.7
    # so the easiest way to do this is to convert the two integers into seconds since epoch
    # then string format the time later. 
    date_count_start=datetime.date(2000,01,01).strftime('%s')
    #UTC_days_since_2000 = np.array(UTC_day_dataset)*(24*60*60)
    # convert the days to seconds - use floats
    UTC_days_since_2000 = np.array(UTC_day_dataset)*24.0*60.0*60.0
    #print UTC_days_since_2000[0] 
    UTC_seconds=np.array(UTC_sec_dataset)
    ob_date_epoch=np.array(UTC_days_since_2000+UTC_seconds+float(date_count_start))
    #print UTC_seconds[0]
    #print date_count_start
    #print ob_date_epoch[0] 
    #print  datetime.datetime.utcfromtimestamp(ob_date_epoch[0])
    swh_dataset=f.get('/swh_ocean_01_plrm_ku')
    #swh_dataset=f.get('/swh_ocean_01_ku')
    #convert to the right SI units, then to ft and kt.
    swh =np.array(swh_dataset)/1000.0*3.2808
    wspd_dataset=f['/wind_speed_alt_01_plrm_ku']
    #wspd_dataset=f['/wind_speed_alt_01_ku']
    wspd=np.array(wspd_dataset)/100.0*1.9484
    lat_dataset=f['/lat_01']
    lat=np.array(lat_dataset)/1000000.0
    lon_dataset=f['/lon_01']
    lon=np.array(lon_dataset)/1000000.0
    swh_qc_dataset=f.get('/swh_ocean_qual_01_plrm_ku')
    swh_qc=np.array(swh_qc_dataset)
    wspd_qc_dataset=f['/sig0_ocean_qual_01_plrm_ku']
    wspd_qc=np.array(wspd_qc_dataset)
    
    # Longitude is in degrees East. Convert to positive for East and negatve for west.
    # find the mask where lon >180.
    # at those points, subtract from 360
    W_hemi_mask = lon > 180.0 
    lon[W_hemi_mask] = -360+lon[W_hemi_mask]
    
    #write out the entries
    obs_dict = {}
    logger.info("%d obs in the file", len(lon))
    for i in range(len(lon)):
        ob_date=datetime.datetime.utcfromtimestamp(ob_date_epoch[i])
        ob_yyyymmddhh=ob_date.strftime('%Y%m%d%H')
        #print UTC_days_since_2000[i]
        #print ob_yyyymmddhh
        # Sometimes, we get bad data with the sgwh and wspd 107.50 and 638.43 (respectively). That works backward to 0.0327664 or 32.7664. 
        if swh_qc[i] == 0:
            ob=" %4d %2d %2d %2d %2d %6.2f %7.2f  %6.2f %7.2f"% (ob_date.year, ob_date.month, ob_date.day, ob_date.hour, ob_date.minute, lat[i], lon[i], swh[i], wspd[i])
            #     print (" %04d %2d %2d %2d %2d %6.2f %7.2f  %6.2f %7.2f"% (ob_date.year, ob_date.month, ob_date.day, ob_date.hour, ob_date.minute, lat[i], lon[i], swh[i], wspd[i]))
            #print ob
            logger.debug(ob)
            # find and open the correct output file. Make any additions only if the exact ob isn't in the existing file
            if ob_yyyymmddhh in obs_dict:
                obs_dict[ob_yyyymmddhh].append(ob)
            else:
                obs_dict[ob_yyyymmddhh]=[ob]
        else:
            bad_ob=" %4d %2d %2d %2d %2d %6.2f %7.2f  %6.2f %7.2f"% (ob_date.year, ob_date.month, ob_date.day, ob_date.hour, ob_date.minute, lat[i], lon[i], swh[i], wspd[i])
            logger.info("QC'd ob. Not included: %s", bad_ob)
    
    # itterate through the obs dates we've found
    for date_key in obs_dict:
        # determin the nmap2 file name
        nmap2_file = sat + "_" + date_key + ".sgwh"
        # open the nmap2 file (should be in cwd
        if (os.path.isfile(nmap2_file)):
            logger.debug("reading %s", nmap2_file)
            with open(nmap2_file, 'r') as f:
                # read the data into a list.
                # f.readlines keeps the \n. Get rid of them
                old_list = f.read().splitlines()
        else:
            old_list = []

        logger.info("old list = %d obs", len(old_list))
        # concatenate the two obs lists
        new_list = old_list + obs_dict[date_key] 
        logger.info("new list = %d obs", len(new_list))
        # sort and uniq the list:
        new_sorted_list = sorted(set(new_list))
        logger.info("new sorted list = %d obs", len(new_sorted_list))
        # overwrite the file with the new list.
        with open(nmap2_file, 'w') as f:
            f.write('\n'.join(new_sorted_list)) 

    

if __name__ == "__main__":
    main(sys.argv[1:])