#!/usr/bin/python
import sys
import gzip
import bz2
from optparse import OptionParserclass LogProcessor(object):'''Process a combined log format.This processor handles log files in a combined format,objects that act on the results are passed in tothe init method as a series of methods.'''def __init__(self, call_chain=None):"""Setup parserSave the call chain. Each time we process a log ,we'll run the list of callbacks with the processedlog results."""if call_chain is None:call_chain = []self._call_chain = call_chaindef split(self, line):"""Split a log file.Initially,we just want size and requested file name . sowe'll split on spaces and pull the data out."""parts = line.split()return {'size': 0 if parts[9] == '-' else int(parts[9]),'file_requested': parts[6]}def parse(self, handle):"""Parses the log file.Returns a dictionary composed of log entry valuesfor easy data summation"""for line in handle:fields = self.split(line)for func in self._call_chain:func(fields)class ColumnLogProcessor(LogProcessor):def split(self, line):parts = line.split()return {'size': int(parts[1]),'file_requested': parts[0]}class MaxSizeHandler(object):"""Check a file's size."""def __init__(self, size):self.size = sizedef process(self, fields):"""Looks at each line individually.Looks at each parsed log line individually andperforms a size calculation. If it's bigger thanour self.size, we just print a warning."""if fields['size'] > self.size:#print ('Warning: %s exceeds $d bytes (%s) !' % (fields['file_requested'], str(self.size), fields['size']))print ('Warning: {0} exceeds {1} bytes {2} !'.format (fields['file_requested'], str(self.size), fields['size']))
def get_stream(path):"""Detect compression.If the file name ends in a compression suffix, we'll openusing the correct algorith. If not, we just return a standardfile object."""_open = openif path.endswith(',gz'):_open = gzip.openelif path.endswith('.bz2'):_open = bz2.open()return _open(path)
if __name__ == '__main__':parser = OptionParser()parser.add_option('-s', '--size', dest = "size",help = "Maximum File Size Allowed",default = 0, type = "int")parser.add_option('-f', '--file', dest = "file",help = "Path to Web Log File",default = "-")opts,args = parser.parse_args()call_chain = []if opts.file == '-':file_stream = sys.stdinelse:try:#file_stream = open(opts.file, 'r')file_stream = get_stream(opts.file)except IOError as e:print (sys.stderr,str(e))sys.exit(-1)size_check = MaxSizeHandler(opts.size)call_chain.append(size_check.process)processor = LogProcessor(call_chain)processor.parse(file_stream)#processorC = ColumnLogProcessor(call_chain)#processorC.parse(file_stream)