mantidlogs

#!/usr/bin/env python3
from __future__ import (absolute_import, division, print_function,
                        unicode_literals)
from glob import glob
import os
import pandas as pd
import sys
import time
__version__="0.0.1"

def getLogs(direc='~/.mantid'):
    direc = os.path.expanduser(direc)
    files = glob(os.path.join(direc, 'mantid.log*'))
    files.sort() # newest first by name
    return files

def listFiles(filenames):
    for name in filenames:
        timemod = time.localtime(os.path.getmtime(name))
        timemod = time.strftime('%Y-%m-%dT%H:%M:%S', timemod)

        filesize = os.path.getsize(name)
        if filesize > 1024:      # bigger than on 1kB
            filesize /= 1024
            if filesize >= 1024:  # bigger than 1MB
                filesize /= 1024
                filesize = '%.0fMB' % filesize
            else:
                filesize = '%.0fkB' % filesize
        else:
            filesize = '%dB' % filesize
        filesize = ' '*(5 - len(filesize)) + filesize
        print(timemod, filesize, name)

def toDateTime(data):
    (first, second) = string.split()[:2]
    return first+second

def parseLog(filename):
    data = pd.read_table(filename, header=None, names=['everything'])

    # split up the single column into many
    # WARNING!!!! this doesn't work at all for multi-line logs
    data = data['everything'].str.extract('^(?P<datetime>.{23})\s+\[(?P<proc>\d+)\]\s+(?P<level>\S+)\s+(?P<logname>\S+)\s+-\s+(?P<message>.+)')

    # convert to proper datetime objects
    data = data.assign(datetime=lambda x:
                     pd.to_datetime(x['datetime'],
                                    format='%Y-%m-%d %H:%M:%S,%f',
                                    errors='coerce'))

    # convert some fields to better types
    data['level'] = data['level'].astype('category')

    data.reset_index()

    return data

if __name__ == "__main__":
    import argparse
    parser = argparse.ArgumentParser(description='Explore ~/.mantid/mantid.log assuming default format')
    parser.add_argument('--list', action='store_true',
                        help='List log files available')
    parser.add_argument('--dir', default='~/.mantid',
                        help="Directory with log files DEFAULT='%(default)s'")

    args=parser.parse_args()
    logfilenames = getLogs(args.dir)

    if args.list:
        listFiles(logfilenames)
        sys.exit(0)

    # only load the most recent
    data = parseLog(logfilenames[0])

    print("--------------------------")
    print(data.describe())
    print("----------")
    print(data.info())
    print("----------")
    print(data.columns)
    print("----------")
    print(data.head())
    print("----------")
    print(data.iloc[0])
    print("----------")
    print(data.iloc[-1]['message'])