#!/usr/bin/env python
import sys
sys.path.append('../')
from logparser import LenMa
input_dir = '../logs/HDFS/' # The input directory of log file
output_dir = 'Lenma_result/' # The output directory of parsing results
log_file = 'HDFS_2k.log' # The input log file name
log_format = '<Date> <Time> <Pid> <Level> <Component>: <Content>' # HDFS log format
threshold = 0.9 # TODO description (default: 0.9)
regex = [] # Regular expression list for optional preprocessing (default: [])
parser = LenMa.LogParser(input_dir,
output_dir,
log_format,
threshold=threshold,
rex=regex)
parser.parse(log_file)
'log_format':
'<Month> <Date> <Time> <User> <Component>\[<PID>\]( \(<Address>\))?: <Content>',
'regex': [r'([\w-]+\.){2,}[\w-]+'],
'threshold': 0.86
},
}
bechmark_result = []
for dataset, setting in benchmark_settings.items():
print('\n=== Evaluation on %s ===' % dataset)
indir = os.path.join(input_dir, os.path.dirname(setting['log_file']))
log_file = os.path.basename(setting['log_file'])
parser = LenMa.LogParser(log_format=setting['log_format'],
indir=indir,
outdir=output_dir,
rex=setting['regex'],
threshold=setting['threshold'])
parser.parse(log_file)
accuracy_PA, accuracy_exact_string_matching, edit_distance_result, edit_distance_result_median = evaluator.evaluate(
groundtruth=os.path.join(indir, log_file + '_structured.csv'),
parsedresult=os.path.join(output_dir, log_file + '_structured.csv'))
bechmark_result.append([
dataset, accuracy_PA, accuracy_exact_string_matching,
edit_distance_result, edit_distance_result_median
])
print('\n=== Overall evaluation results ===')
df_result = pd.DataFrame(bechmark_result,
columns=[
import collections
import os
import pandas as pd
from benchmark.Lenma_benchmark import benchmark_settings
from logparser import LenMa
from File_Info import get_file_size
dataset = 'BGL'
output_dir = 'LenMa_result/' # The output directory of parsing results
one_setting = benchmark_settings[dataset]
log_file = os.path.basename(one_setting['log_file'])
input_dir = os.path.join('../logs/', os.path.dirname(one_setting['log_file']))
file_size = get_file_size(dataset)
results = collections.defaultdict(list)
for size, file in file_size.items():
parser = LenMa.LogParser(
input_dir,
output_dir,
log_format=one_setting['log_format'],
threshold=one_setting['threshold'],
rex=one_setting['regex'],
)
time_elapsed = parser.parse(file)
results['size'].append(size)
results['time'].append(time_elapsed.total_seconds())
print(results['time'])
pd.DataFrame(results).to_csv('./LenMa_%s.csv' % dataset)