def logs_folder_files(args, task):
uri = BROWSE_FOLDER_FORMAT.format(logfetch_base.base_uri(args), task)
files_json = get_json_response(uri, args, {'path' : '{0}/logs'.format(task)})
if 'files' in files_json:
files = files_json['files']
return [f['name'] for f in files if logfetch_base.is_in_date_range(args, f['mtime'])]
else:
return [f['path'].rsplit('/')[-1] for f in files_json if logfetch_base.is_in_date_range(args, f['mtime'])]
def logs_folder_files(args, task):
uri = BROWSE_FOLDER_FORMAT.format(logfetch_base.base_uri(args), task)
files_json = logfetch_base.get_json_response(uri, args, {"path": "{0}/logs".format(task)}, True)
if "files" in files_json:
files = files_json["files"]
return [f["name"] for f in files if logfetch_base.is_in_date_range(args, f["mtime"])]
else:
return [f["path"].rsplit("/")[-1] for f in files_json if logfetch_base.is_in_date_range(args, f["mtime"])]
def log_file_in_date_range(args, log_file):
if 'startTime' in log_file:
if 'endTime' in log_file:
return logfetch_base.date_range_overlaps(args, int(str(log_file['startTime'])[0:-3]), int(str(log_file['endTime'])[0:-3]))
else:
return logfetch_base.date_range_overlaps(args, int(str(log_file['startTime'])[0:-3]), int(str(log_file['lastModified'])[0:-3]))
elif 'endTime' in log_file:
return logfetch_base.is_in_date_range(args, int(str(log_file['endTime'])[0:-3]))
else:
return logfetch_base.is_in_date_range(args, int(str(log_file['lastModified'])[0:-3]))
def logs_folder_files(args, task):
uri = BROWSE_FOLDER_FORMAT.format(logfetch_base.base_uri(args), task)
files_json = get_json_response(uri, {'path': '{0}/logs'.format(task)})
if 'files' in files_json:
files = files_json['files']
return [
f['name'] for f in files
if logfetch_base.is_in_date_range(args, f['mtime'])
]
else:
return [
f['path'].rsplit('/')[-1] for f in files_json
if logfetch_base.is_in_date_range(args, f['mtime'])
]
def in_date_range(args, filename):
timestamps = re.findall(r"\d{13}", filename)
if timestamps:
return logfetch_base.is_in_date_range(args,
int(str(timestamps[-1])[0:-3]))
else:
return True
def in_date_range(args, filename):
timestamps = re.findall(r"-\d{13}-", filename)
if timestamps:
return is_in_date_range(
args, int(str(timestamps[-1]).replace("-", "")[0:-3]))
else:
return True
def log_file_in_date_range(args, log_file):
if 'startTime' in log_file:
if 'endTime' in log_file:
return logfetch_base.date_range_overlaps(
args, int(str(log_file['startTime'])[0:-3]),
int(str(log_file['endTime'])[0:-3]))
else:
return logfetch_base.date_range_overlaps(
args, int(str(log_file['startTime'])[0:-3]),
int(str(log_file['lastModified'])[0:-3]))
elif 'endTime' in log_file:
return logfetch_base.is_in_date_range(
args, int(str(log_file['endTime'])[0:-3]))
else:
return logfetch_base.is_in_date_range(
args, int(str(log_file['lastModified'])[0:-3]))
def download_s3_logs(args):
sys.stderr.write(colored('Checking for S3 log files', 'cyan') + '\n')
logs = logs_for_all_requests(args)
async_requests = []
zipped_files = []
all_logs = []
for log_file in logs:
filename = log_file['key'].rsplit("/", 1)[1]
if logfetch_base.is_in_date_range(args, time_from_filename(filename)):
if not already_downloaded(args.dest, filename):
async_requests.append(
grequests.AsyncRequest('GET', log_file['getUrl'], callback=generate_callback(log_file['getUrl'], args.dest, filename, args.chunk_size, args.verbose))
)
else:
if args.verbose:
sys.stderr.write(colored('Log already downloaded {0}'.format(filename), 'magenta') + '\n')
all_logs.append('{0}/{1}'.format(args.dest, filename.replace('.gz', '.log')))
zipped_files.append('{0}/{1}'.format(args.dest, filename))
else:
if args.verbose:
sys.stderr.write(colored('Excluding {0}, not in date range'.format(filename), 'magenta') + '\n')
if async_requests:
sys.stderr.write(colored('Starting S3 Downloads with {0} parallel fetches'.format(args.num_parallel_fetches), 'cyan'))
grequests.map(async_requests, stream=True, size=args.num_parallel_fetches)
else:
sys.stderr.write(colored('No S3 logs to download', 'cyan'))
sys.stderr.write(colored('\nUnpacking S3 logs\n', 'cyan'))
all_logs = all_logs + logfetch_base.unpack_logs(args, zipped_files)
sys.stderr.write(colored('All S3 logs up to date', 'cyan') + '\n')
return all_logs
def download_s3_logs(args):
if not args.silent:
sys.stderr.write(colored('Checking for S3 log files', 'cyan') + '\n')
callbacks.progress = 0
logs = logs_for_all_requests(args)
async_requests = []
all_logs = []
for log_file in logs:
filename = log_file['key'].rsplit("/", 1)[1]
if logfetch_base.is_in_date_range(args, int(str(log_file['lastModified'])[0:-3])):
if not args.logtype or log_matches(args, filename):
logfetch_base.log(colored('Including log {0}'.format(filename), 'blue') + '\n', args, True)
if not already_downloaded(args.dest, filename):
async_requests.append(
grequests.AsyncRequest('GET', log_file['getUrl'], callback=callbacks.generate_callback(log_file['getUrl'], args.dest, filename, args.chunk_size, args.verbose, args.silent), headers=args.headers)
)
else:
logfetch_base.log(colored('Log already downloaded {0}'.format(filename), 'blue') + '\n', args, True)
all_logs.append('{0}/{1}'.format(args.dest, filename))
else:
logfetch_base.log(colored('Excluding {0} log does not match logtype argument {1}'.format(filename, args.logtype), 'magenta') + '\n', args, True)
else:
logfetch_base.log(colored('Excluding {0}, not in date range'.format(filename), 'magenta') + '\n', args, True)
if async_requests:
logfetch_base.log(colored('Starting {0} S3 Downloads with {1} parallel fetches\n'.format(len(async_requests), args.num_parallel_fetches), 'cyan'), args, False)
callbacks.goal = len(async_requests)
grequests.map(async_requests, stream=True, size=args.num_parallel_fetches)
else:
logfetch_base.log(colored('No S3 logs to download\n', 'cyan'), args, False)
logfetch_base.log(colored('All S3 logs up to date\n', 'cyan'), args, False)
all_logs = modify_download_list(all_logs)
return all_logs
def valid_logfile(args, fileData):
is_in_range = logfetch_base.is_in_date_range(args, fileData['mtime'])
not_a_directory = not fileData['mode'].startswith('d')
is_a_logfile = fnmatch.fnmatch(
fileData['name'], '*.log') or fnmatch.fnmatch(
fileData['name'], '*.out') or fnmatch.fnmatch(
fileData['name'], '*.err')
return is_in_range and not_a_directory and is_a_logfile
def download_s3_logs(args):
if not args.silent:
sys.stderr.write(colored("Checking for S3 log files", "cyan") + "\n")
callbacks.progress = 0
logs = logs_for_all_requests(args)
async_requests = []
all_logs = []
for log_file in logs:
filename = log_file["key"].rsplit("/", 1)[1]
if logfetch_base.is_in_date_range(args, int(str(log_file["lastModified"])[0:-3])):
if not args.logtype or log_matches(args, filename):
logfetch_base.log(colored("Including log {0}".format(filename), "blue") + "\n", args, True)
if not already_downloaded(args.dest, filename):
async_requests.append(
grequests.AsyncRequest(
"GET",
log_file["getUrl"],
callback=callbacks.generate_callback(
log_file["getUrl"], args.dest, filename, args.chunk_size, args.verbose, args.silent
),
headers=args.headers,
)
)
else:
logfetch_base.log(colored("Log already downloaded {0}".format(filename), "blue") + "\n", args, True)
all_logs.append("{0}/{1}".format(args.dest, filename))
else:
logfetch_base.log(
colored(
"Excluding {0} log does not match logtype argument {1}".format(filename, args.logtype),
"magenta",
)
+ "\n",
args,
True,
)
else:
logfetch_base.log(
colored("Excluding {0}, not in date range".format(filename), "magenta") + "\n", args, True
)
if async_requests:
logfetch_base.log(
colored(
"Starting {0} S3 Downloads with {1} parallel fetches\n".format(
len(async_requests), args.num_parallel_fetches
),
"cyan",
),
args,
False,
)
callbacks.goal = len(async_requests)
grequests.map(async_requests, stream=True, size=args.num_parallel_fetches)
else:
logfetch_base.log(colored("No S3 logs to download\n", "cyan"), args, False)
logfetch_base.log(colored("All S3 logs up to date\n", "cyan"), args, False)
return all_logs
def valid_logfile(args, fileData):
is_in_range = logfetch_base.is_in_date_range(args, fileData["mtime"])
not_a_directory = not fileData["mode"].startswith("d")
is_a_logfile = (
fnmatch.fnmatch(fileData["name"], "*.log")
or fnmatch.fnmatch(fileData["name"], "*.out")
or fnmatch.fnmatch(fileData["name"], "*.err")
)
return is_in_range and not_a_directory and is_a_logfile
def download_s3_logs(args):
sys.stderr.write(colored('Checking for S3 log files', 'cyan') + '\n')
logs = logs_for_all_requests(args)
async_requests = []
zipped_files = []
all_logs = []
for log_file in logs:
filename = log_file['key'].rsplit("/", 1)[1]
if logfetch_base.is_in_date_range(
args, int(str(log_file['lastModified'])[0:-3])):
if not args.logtype or log_matches(args, filename):
if not already_downloaded(args.dest, filename):
async_requests.append(
grequests.AsyncRequest('GET',
log_file['getUrl'],
callback=generate_callback(
log_file['getUrl'],
args.dest, filename,
args.chunk_size,
args.verbose),
headers=args.headers))
else:
if args.verbose:
sys.stderr.write(
colored(
'Log already downloaded {0}'.format(filename),
'magenta') + '\n')
all_logs.append('{0}/{1}'.format(
args.dest, filename.replace('.gz', '.log')))
zipped_files.append('{0}/{1}'.format(args.dest, filename))
else:
if args.verbose:
sys.stderr.write(
colored(
'Excluding {0} log does not match logtype argument {1}'
.format(filename, args.logtype), 'magenta') + '\n')
else:
if args.verbose:
sys.stderr.write(
colored(
'Excluding {0}, not in date range'.format(filename),
'magenta') + '\n')
if async_requests:
sys.stderr.write(
colored(
'Starting S3 Downloads with {0} parallel fetches'.format(
args.num_parallel_fetches), 'cyan'))
grequests.map(async_requests,
stream=True,
size=args.num_parallel_fetches)
else:
sys.stderr.write(colored('No S3 logs to download', 'cyan'))
sys.stderr.write(colored('\nUnpacking S3 logs\n', 'cyan'))
all_logs = all_logs + logfetch_base.unpack_logs(args, zipped_files)
sys.stderr.write(colored('All S3 logs up to date', 'cyan') + '\n')
return all_logs
def download_s3_logs(args):
sys.stderr.write(colored('Checking for S3 log files', 'cyan') + '\n')
logs = logs_for_all_requests(args)
async_requests = []
all_logs = []
for log_file in logs:
filename = log_file['key'].rsplit("/", 1)[1]
if logfetch_base.is_in_date_range(args, time_from_filename(filename)):
if not already_downloaded(args.dest, filename):
async_requests.append(
grequests.AsyncRequest('GET', log_file['getUrl'], callback=generate_callback(log_file['getUrl'], args.dest, filename, args.chunk_size))
)
all_logs.append('{0}/{1}'.format(args.dest, filename.replace('.gz', '.log')))
if async_requests:
sys.stderr.write(colored('Starting S3 Downloads', 'cyan'))
grequests.map(async_requests, stream=True, size=args.num_parallel_fetches)
zipped_files = ['{0}/{1}'.format(args.dest, log_file['key'].rsplit("/", 1)[1]) for log_file in logs]
sys.stderr.write(colored('Unpacking S3 logs\n', 'cyan'))
logfetch_base.unpack_logs(zipped_files)
sys.stderr.write(colored('All S3 logs up to date', 'cyan') + '\n')
return all_logs
def is_valid_live_log(args, file_data):
is_in_range = logfetch_base.is_in_date_range(args, file_data['mtime'])
has_data = logfetch_base.logfile_has_data(file_data)
return is_in_range and has_data and logfetch_base.is_valid_log(file_data)
def valid_logfile(args, fileData):
is_in_range = logfetch_base.is_in_date_range(args, fileData['mtime'])
not_a_directory = not fileData['mode'].startswith('d')
is_a_logfile = fnmatch.fnmatch(fileData['name'], '*.log') or fnmatch.fnmatch(fileData['name'], '*.out') or fnmatch.fnmatch(fileData['name'], '*.err')
return is_in_range and not_a_directory and is_a_logfile
def in_date_range(args, filename):
timestamps = re.findall(r"-\d{13}-", filename)
if timestamps:
return is_in_date_range(args, int(str(timestamps[-1]).replace("-", "")[0:-3]))
else:
return True
def is_valid_live_log(args, file_data):
is_in_range = logfetch_base.is_in_date_range(args, file_data['mtime'])
has_data = logfetch_base.logfile_has_data(file_data)
return is_in_range and has_data and logfetch_base.is_valid_log(file_data)
def in_date_range(args, filename):
timestamps = re.findall(r"\d{13}", filename)
if timestamps:
return logfetch_base.is_in_date_range(args, int(str(timestamps[-1])[0:-3]))
else:
return True
