python解析smart结构数据

python编程解析如下smart结构数据，得到一行smart信息

run: smartctl -a /dev/sda
out: smartctl 6.3 2014-07-26 r3976 [x86_64-linux-2.6.18-164.el5] (local build)
out: Copyright (C) 2002-14, Bruce Allen, Christian Franke, www.smartmontools.org
out:
out: === START OF INFORMATION SECTION ===
out: Vendor:               TOSHIBA
out: Product:              MBF2300RC
out: Revision:             0109
out: User Capacity:        300,000,000,000 bytes [300 GB]
out: Logical block size:   512 bytes
out: Rotation Rate:        10025 rpm
out: Form Factor:          2.5 inches
out: Logical Unit id:      0x50000393d84b42bc
out: Serial number:        EB00PC208HFC
out: Device type:          disk
out: Transport protocol:   SAS (SPL-3)
out: Local Time is:        Tue Dec 30 00:10:03 2014 CST
out: SMART support is:     Available - device has SMART capability.
out: SMART support is:     Enabled
out: Temperature Warning:  Enabled
out:
out: === START OF READ SMART DATA SECTION ===
out: SMART Health Status: OK
out:
out: Current Drive Temperature:     28 C
out: Drive Trip Temperature:        65 C
out:
out: Manufactured in week 08 of year 2012
out: Specified cycle count over device lifetime:  50000
out: Accumulated start-stop cycles:  21
out: Specified load-unload count over device lifetime:  200000
out: Accumulated load-unload cycles:  69
out: Elements in grown defect list: 0
out:
out: Error counter log:
out:            Errors Corrected by           Total   Correction     Gigabytes    Total
out:                ECC          rereads/    errors   algorithm      processed    uncorrected
out:            fast | delayed   rewrites  corrected  invocations   [10^9 bytes]  errors
out: read:          0        0         0         0          0     300744.962           0
out: write:         0        0         0         0          0      10841.446           0
out:
out: Non-medium error count:        0
out:
out: No self-tests have been logged
out:
out: 

python文件如下：

#!/bin/env python

import os,time,re,sys
import logging

logging.basicConfig(filename = os.path.join(os.getcwd(), 'load.log'), level = logging.INFO, format = '%(asctime)s - %(levelname)s: %(message)s')

class attribute:
    pattern = ''
    value = '-1'

if __name__ == '__main__':
    log_file = os.path.join(os.getcwd(), sys.argv[1])
    if os.path.exists(log_file):
        logging.info('start loading %s...' % (log_file))
    else:
        logging.error('%s not exists' % (log_file))

    update_time = time.strftime('%Y-%m-%d %H:%M:%S', time.localtime(os.stat(log_file).st_ctime))
    block_list = []
    attrs = {}

    #information section
    attrs['serial_number'] = attribute()
    attrs['serial_number'].pattern = 'Serial number:\s*(\w*)'
    attrs['vendor'] = attribute()
    attrs['vendor'].pattern = 'Vendor:\s*(\w*)'
    attrs['product'] = attribute()
    attrs['product'].pattern = 'Product:\s*(\w*).'
    attrs['revision'] = attribute()
    attrs['revision'].pattern = 'Revision:\s*(\w*)'
    attrs['compliance'] = attribute()
    attrs['compliance'].pattern = 'Compliance:\s*(\w*)'
    attrs['user_capacity'] = attribute()
    attrs['user_capacity'].pattern = 'User Capacity:.*\[(\w*)'
    attrs['logical_block_size'] = attribute()
    attrs['logical_block_size'].pattern = 'Logical block size:\s*(\w*)'
    attrs['rotation_rate'] = attribute()
    attrs['rotation_rate'].pattern = 'Rotation Rate:\s*(\w*)'
    attrs['form_factor'] = attribute()
    attrs['form_factor'].pattern = 'Form Factor:\s*([\w\.]*)'
    attrs['logical_unit_id'] = attribute()
    attrs['logical_unit_id'].pattern = 'Logical Unit id:\s*(\w*)'
    attrs['device_type'] = attribute()
    attrs['device_type'].pattern = 'Device type:\s*(\w*)'
    attrs['transport_protocol'] = attribute()
    attrs['transport_protocol'].pattern = 'Transport protocol:\s*(.*)'
    attrs['smart_support'] = attribute()
    attrs['smart_support'].pattern = 'SMART support is:\s*(\w*)'
    attrs['smart_enable'] = attribute()
    attrs['smart_enable'].pattern = 'SMART support is:\s*(Enabled|Disabled)'
    attrs['temperature_warning'] = attribute()
    attrs['temperature_warning'].pattern = 'Temperature Warning:\s*(\w*)'
    attrs['ip'] = attribute()
    attrs['ip'].pattern = '\[([\w\.]*)'

    #smart data section
    attrs['smart_health_status'] = attribute()
    attrs['smart_health_status'].pattern = 'SMART Health Status:\s*(\w*)'
    attrs['current_drive_temperature'] = attribute()
    attrs['current_drive_temperature'].pattern = 'Current Drive Temperature:\s*(\w*)'
    attrs['drive_trip_temperature'] = attribute()
    attrs['drive_trip_temperature'].pattern = 'Drive Trip Temperature:\s*(\w*)'
    attrs['elements_in_grown_defect_list'] = attribute()
    attrs['elements_in_grown_defect_list'].pattern = 'Elements in grown defect list:\s*(\w*)'
    attrs['manufactured_time'] = attribute()
    attrs['manufactured_time'].pattern = 'Manufactured in (.*)'
    attrs['cycle_count'] = attribute()
    attrs['cycle_count'].pattern = 'Specified cycle count over device lifetime:\s*(\w*)'
    attrs['start_stop_cycles'] = attribute()
    attrs['start_stop_cycles'].pattern = 'Accumulated start-stop cycles:\s*(\w*)'
    attrs['load_unload_count'] = attribute()
    attrs['load_unload_count'].pattern = 'Specified load-unload count over device lifetime:\s*(\w*)'
    attrs['load_unload_cycles'] = attribute()
    attrs['load_unload_cycles'].pattern = 'Accumulated load-unload cycles:\s*(\w*)'
    attrs['blocks_sent_to_initiator'] = attribute()
    attrs['blocks_sent_to_initiator'].pattern = 'Blocks sent to initiator =\s*(\w*)'
    attrs['blocks_received_from_initiator'] = attribute()
    attrs['blocks_received_from_initiator'].pattern = 'Blocks received from initiator =\s*(\w*)'
    attrs['blocks_read_from_cache'] = attribute()
    attrs['blocks_read_from_cache'].pattern = 'Blocks read from cache and sent to initiator =\s*(\w*)'
    attrs['num_commands_size_not_larger_than_segment_size'] = attribute()
    attrs['num_commands_size_not_larger_than_segment_size'].pattern = '<= segment size =\s*(\w*)'
    attrs['num_commands_size_larger_than_segment_size'] = attribute()
    attrs['num_commands_size_larger_than_segment_size'].pattern = '> segment size=\s*(\w*)'
    attrs['num_hours_powered_up'] = attribute()
    attrs['num_hours_powered_up'].pattern = 'number of hours powered up =\s*(\w*)'
    attrs['num_minutes_next_test'] = attribute()
    attrs['num_minutes_next_test'].pattern = 'number of minutes until next internal SMART test =\s*(\w*)'
    attrs['non_medium_error_count'] = attribute()
    attrs['non_medium_error_count'].pattern = 'Non-medium error count:\s*(\w*)'

    new_information_count = 0
    insert_smart_count = 0
    fail_count = 0

    for line in open(log_file):
        if line.find('run:') != -1 or not line.strip(): #contains 'run' or blank line
            block = '\n'.join(block_list)
	    if block and re.search('smartctl 6.3', block):
                for (k, v) in attrs.items():
                    attrs[k].value = '-1'
                    match = re.search(attrs[k].pattern, block)
                    if match:
                        attrs[k].value = match.group(1)
                if attrs['vendor'].value == 'LSI':
                    block_list = []
                    logging.error('ip with LSI vendor: %s' % (attrs['ip'].value));
                    continue

                #insert information section
                if attrs['serial_number'].value == '-1':
                    block_list = []
                    fail_count = fail_count + 1
                    logging.info('invalid ip without serial number(-1): %s' % (attrs['ip'].value))
                    continue

                #print 'hive-xdf-information'+";"+attrs['serial_number'].value+";"+ update_time+";"+ attrs['vendor'].value+";"+ attrs['product'].value+";"+ attrs['revision'].value+";"+attrs['compliance'].value+";"+ attrs['user_capacity'].value+";"+ attrs['logical_block_size'].value+";"+ attrs['rotation_rate'].value+";"+ attrs['form_factor'].value+";"+attrs['logical_unit_id'].value+";"+ attrs['device_type'].value+";"+ attrs['transport_protocol'].value+";"+ attrs['smart_support'].value+";"+ attrs['smart_enable'].value+";"+attrs['temperature_warning'].value+";"+ attrs['ip'].value+";"+ update_time
                #insert smart data section

                match = re.search('read:\s+(\d+)\s+(\d+)\s+(\d+)\s+(\d+)\s+(\d+)\s+([\d\.]+)\s+(\d+)', block)
                if match:
                    read_corrected_ecc_fast = match.group(1)
                    read_corrected_ecc_delayed = match.group(2)
                    read_corrected_re = match.group(3)
                    read_total_errors_corrected = match.group(4)
                    read_correction_algo_invocations = match.group(5)
                    read_gigabytes_processed = match.group(6)
                    read_total_uncorrected_errors = match.group(7)
                else:
                    read_corrected_ecc_fast = bytes(-1)
                    read_corrected_ecc_delayed = bytes(-1)
                    read_corrected_re = bytes(-1)
                    read_total_errors_corrected = bytes(-1)
                    read_correction_algo_invocations = bytes(-1)
                    read_gigabytes_processed = bytes(-1)
                    read_total_uncorrected_errors = bytes(-1)

                match = re.search('write:\s+(\d+)\s+(\d+)\s+(\d+)\s+(\d+)\s+(\d+)\s+([\d\.]+)\s+(\d+)', block)
                if match:
                    write_corrected_ecc_fast = match.group(1)
                    write_corrected_ecc_delayed = match.group(2)
                    write_corrected_re = match.group(3)
                    write_total_errors_corrected = match.group(4)
                    write_correction_algo_invocations = match.group(5)
                    write_gigabytes_processed = match.group(6)
                    write_total_uncorrected_errors = match.group(7)
                else:
                    write_corrected_ecc_fast = bytes(-1)
                    write_corrected_ecc_delayed = bytes(-1)
                    write_corrected_re = bytes(-1)
                    write_total_errors_corrected = bytes(-1)
                    write_correction_algo_invocations = bytes(-1)
                    write_gigabytes_processed = bytes(-1)
                    write_total_uncorrected_errors = bytes(-1)

                match = re.search('verify:\s+(\d+)\s+(\d+)\s+(\d+)\s+(\d+)\s+(\d+)\s+([\d\.]+)\s+(\d+)', block)
                if match:
                    verify_corrected_ecc_fast = match.group(1)
                    verify_corrected_ecc_delayed = match.group(2)
                    verify_corrected_re = match.group(3)
                    verify_total_errors_corrected = match.group(4)
                    verify_correction_algo_invocations = match.group(5)
                    verify_gigabytes_processed = match.group(6)
                    verify_total_uncorrected_errors = match.group(7)
                else:
                    verify_corrected_ecc_fast = bytes(-1)
                    verify_corrected_ecc_delayed = bytes(-1)
                    verify_corrected_re = bytes(-1)
                    verify_total_errors_corrected = bytes(-1)
                    verify_correction_algo_invocations = bytes(-1)
                    verify_gigabytes_processed = bytes(-1)
                    verify_total_uncorrected_errors = bytes(-1)

                insert_smart_count = insert_smart_count + 1
                print sys.argv[2]+"@@"+attrs['serial_number'].value+";"+ update_time+";"+ attrs['smart_health_status'].value+";"+ attrs['current_drive_temperature'].value+";"+ attrs['drive_trip_temperature'].value+";"+attrs['elements_in_grown_defect_list'].value+";"+ attrs['manufactured_time'].value+";"+ attrs['cycle_count'].value+";"+ attrs['start_stop_cycles'].value+";"+ attrs['load_unload_count'].value+";"+attrs['load_unload_cycles'].value+";"+ attrs['blocks_sent_to_initiator'].value+";"+ attrs['blocks_received_from_initiator'].value+";"+ attrs['blocks_read_from_cache'].value+";"+attrs['num_commands_size_not_larger_than_segment_size'].value+";"+attrs['num_commands_size_larger_than_segment_size'].value+";"+ attrs['num_hours_powered_up'].value+";"+ attrs['num_minutes_next_test'].value+";"+ attrs['non_medium_error_count'].value+';'+read_corrected_ecc_fast+';'+ read_corrected_ecc_delayed+';'+read_corrected_re+';'+ read_total_errors_corrected+';'+ read_correction_algo_invocations+';'+ read_gigabytes_processed+';'+ read_total_uncorrected_errors+';'+  write_corrected_ecc_fast+';'+ write_corrected_ecc_delayed+';'+ write_corrected_re+';'+ write_total_errors_corrected+';'+ write_correction_algo_invocations+';'+ write_gigabytes_processed+';'+ write_total_uncorrected_errors+';'+ verify_corrected_ecc_fast+';'+ verify_corrected_ecc_delayed+';'+ verify_corrected_re+';'+verify_total_errors_corrected+';'+ verify_correction_algo_invocations+';'+ verify_gigabytes_processed+';'+ verify_total_uncorrected_errors

		block_list = []
        elif line.find('out:') != -1:
            block_list.append(line.strip())

解析结果如下：

hive-xdf-smart_data@@EB00PC208HFC;2015-06-23 18:56:09;OK;28;65;0;week 08 of year 2012;50000;21;200000;69;-1;-1;-1;-1;-1;-1;-1;0;0;0;0;0;0;300744.962;0;0;0;0;0;0;10841.446;0;-1;-1;-1;-1;-1;-1;-1