root/sodar/trunk/sodar/formattedData.py

#!/usr/bin/python
"""
Class to handle formatting sodar data samples.

A list of sample data.

A list of dictionaries, each with an item for header and body data.

A list of dictionaries of sample data, each with a two items:
    a header dictionary, and
    a body list of alitude dictionaries.
"""

__author__ = 'Chris Calloway'
__email__ = 'cbc@unc.edu'
__copyright__ = 'Copyright 2007 UNC-CH Department of Marine Science'
__license__ = 'GPL2'

import rawData
import numpy as n
import datetime

class FormattedData(list):
    
    """Processed daily sodar file data.
    
       A list of rawData.Sample deep copies.
    """
    
    def __init__(self, data):
        """Prepare raw sodar daily data for analysis."""
        super(FormattedData, self).__init__()
        self.extend([sample.data() for sample in data])
        self._formatt()
    
    def _format(self):
        """Clean up data for analysis."""
        self._convert()
        self._stamp()
        self._timeInterval()
        self._minAltitude()
        self._maxAltitude()
        self._numAltitudes()
        self._altInterval()
        # correct for missing times
        # correct for missing altitudes
        # mark maximum altitude with good values for each sample
        # mark minimum altitude with invalid values for each sample
        # convert direction to radians
        # compute u,v,c components
        # compute colorspecs
        # compute plotting parameters
    
    def _convert(self):
        """Convert data to numbers and correct for invalid values."""
        INVALID = "-9999"
        for sample in self:
            # convert header data to integers
            for key,value in sample['header'].items():
                try:
                    if value == INVALID:
                        raise ValueError
                    sample['header'][key] = int(value)
                except (ValueError, TypeError):
                    sample['header'][key] = n.NaN
            # convert body data to floats
            for altitude in sample['body']:
                for key,value in altitude.items():
                    try:
                        if value == INVALID:
                            raise ValueError
                        altitude[key] = float(value)
                    except (ValueError, TypeError):
                        altitude[key] = n.NaN
    
    def _stamp(self):
        """Add a datetime stamp item to each sample."""
        for sample in self:
            header = sample['header']
            sample['stamp'] = datetime.datetime(header['YEAR'],
                                                header['MONTH'],
                                                header['DAY'],
                                                header['HOUR'],
                                                header['MIN'])
    
    def _timeInterval(self):
        """Add a time interval attribute."""
        intervals = zip([sample['stamp'] for sample in self[:-1]],
                        [sample['stamp'] for sample in self[1:]])
        intervals = [interval[1] - interval[0] for interval in intervals]
        self.timeInterval = min(intervals)
    
    def _minAltitude(self):
        """Add an overall minimum altitude attribute."""
        altitudes = [sample['body'][0]['ALT'] for sample in self]
        self.minAltitude = min(altitudes)
    
    def _maxAltitude(self):
        """Add an overall maximum altitude attribute."""
        altitudes = [sample['body'][-1]['ALT'] for sample in self]
        self.maxAltitude = max(altitudes)
    
    def _numAltitudes(self):
        """Add an overall maximum number of altitudes attribute"""
        self.numAltitudes = max([len(sample['body']) for sample in self])
    
    def _altInterval(self):
        """Add an overall altitude interval attribute."""
        self.altInterval = (self.maxAltitude - self.minAltitude) / \
                           (self.numAltitudes - 1)


def _main():
    """Process as script from command line."""
    import urllib2
    try:
        rawDataHandle = urllib2.urlopen('http://nemo.isis.unc.edu/'\
                                        'data/nccoos/level0/dukeforest/sodar/'\
                                        'store/2007-06/20070601.dat')
        rawDataString = rawDataHandle.read()
    except:
        raise IOError("Failure to read raw test data")
    rawDataObject = rawData.RawData(rawDataString)
    formattedDataObject = FormattedData(rawDataObject)
    print "Time Interval =", formattedDataObject.timeInterval
    print "Minumum Altitude =", formattedDataObject.minAltitude
    print "Maximum Altitude =", formattedDataObject.maxAltitude
    print "Number of Altitudes =", formattedDataObject.numAltitudes
    print "Altitude Interval =", formattedDataObject.altInterval

if __name__ == "__main__":
    _main()