parse_smart.py

#!/usr/bin/env python3

import re, os
from subprocess import getoutput

keys = [
        "ECC corrected fast",
        "ECC corrected delayed",
        "ECC reread/rewrites",
        "Total errors corrected",
        "Correction algorithm invocations",
        "Gigabytes processed [10^9 bytes]",
        "Total uncorrected errors",
]

def log(type, disk, message):
        # print("!!LOG", type, disk, message)
        pass

def get_all_scsi_path():
        disks = os.listdir('/dev/disk/by-id/')
        out = []
        for disk in disks:
                if disk.startswith('scsi') and 'part' not in disk:
                        out.append('/dev/disk/by-id/'+disk)
        return out

def call_smart(disks):
        out = {}
        for disk in disks:
                content = getoutput('smartctl -d scsi -a {}'.format(disk))
                print(disk, content)
                content = parse_out(content, disk)
                out[disk] = content

        return out

def save_status(stats):
        for disk in status:
                for key in ['read', 'write', 'verify']:
                        if len(disk[key]) == 7:
                                db.add_status(disk, key, disk[key], disk['Non-medium error count'])
                        else:
                                log('missing', disk, '{} is missing values'.fotmat(key))


def parse_out(content, disk=''):
        out = {
                'read': {},
                'write': {},
                'verify': {},
                'Non-medium error count': 0,
                'Elements in grown defect list': 0,
                'SMART Health Status': ''
        }

        try:
                out['SMART Health Status'] = re.search(
                        r'SMART Health Status:\s+(?P<status>.+)', content
                ).group('status')
        except AttributeError as error:
                log('missing', disk, "SMART Health Status not found")

        try:
                out['Elements in grown defect list'] = re.search(
                        r'Elements in grown defect list:\s+(?P<number>\d+)', content
                ).group('number')
        except AttributeError as error:
                log('missing', disk, "Elements in grown defect list not found")

        try:
                out['Non-medium error count'] = re.search(
                        r'Non-medium error count:\s+(?P<number>\d+)', content
                ).group('number')
        except AttributeError as error:
                log('missing', disk, "Non-medium error count not found")

        try:
                content = content.split("Error counter log:")[1]
        except IndexError:
                log('failed', disk, 'Missing error information section')
                return {}

        for line in content.split('\n'):
                if line.startswith('read:'):
                        line = re.sub(r'^[a-z:\s]+', '', line)
                        line = re.split(r'\s+', line)
                        out['read'] = dict(zip(keys, line))
                        continue

                if line.startswith('write:'):
                        line = re.sub(r'^[a-z:\s]+', '', line)
                        line = re.split(r'\s+', line)
                        out['write'] = dict(zip(keys, line))
                        continue

                if line.startswith('verify:'):
                        line = re.sub(r'^[a-z:\s]+', '', line)
                        line = re.split(r'\s+', line)
                        out['verify'] = dict(zip(keys, line))
                        continue

        return out

if __name__ == "__main__":
        import json
        disks = get_all_scsi_path()
        parsed = call_smart(disks)
        print(json.dumps(parsed, indent=4, sort_keys=True))
        print(*[(i,parsed[i].get('SMART Health Status')) for i in parsed], sep='\n')