Concatenates all log files in directory into one file, sorted by timestamp.Usage: logsplat.py directory

logsplat.py

"""
Concatenates all log files in directory into one file, sorted by timestamp.

Usage: logsplat.py directory
"""

import collections
import glob
import os
from operator import attrgetter
import re
import sys

STAMP_REGEX = re.compile(r"(?P<timestamp>[\d\/]{10} [\d:\.]{12}) (?P<content>.*)")
FILENAME = "splat.log"

Line = collections.namedtuple("Line", ["timestamp", "content", "file"])

prettify_line = "[{0.file}] {0.timestamp} {0.content}".format

def get_log_files():
    log_files = glob.iglob('*.log')
    
    try:
        log_files.remove(FILENAME)
    except ValueError:
        pass
    
    return log_files

def read_lines(log_file):
    
    def construct_line_from_match(match):
        start_of_filename = log_file[:8]
        return Line(file=start_of_filename, **match.groupdict()

    with open(log_file) as open_log:
        lines = list(open_log)
    
    line_regex_matches = []
    
    for line in lines:
        match = STAMP_REGEX.match(line)
        
        if match:
            line_regex_matches.append(match)
    
    return (construct_line_from_match(match) for match in line_regex_matches)
        

def main():
    directory = sys.argv[1]
    os.chdir(directory)
    
    log_files = get_log_files()
    
    lines = [read_lines(log_file) for log_file in log_files]

    lines = sorted(lines, key=attrgetter('timestamp'))

    with open(FILENAME, 'w') as splat_file:
        splat_file.write("\n".join(map(prettify_line, lines)))

    print("Wrote {} lines to {}".format(len(lines), FILENAME))

if __name__ == '__main__':
    main()