def update(self, amount):
'''Increment the number of files processed by one, and update the progress bar accordingly
'''
self.done += amount
elapsed = time.perf_counter() - self.started
estimate = elapsed * self.total / self.done
remains = (estimate - elapsed) * 1.1 # overestimate by 10%
progress = self.done / self.total
sys.stdout.write('\r[{0}{1}]{2}% ~{3} left '.format(
'#' * int(self.width * progress),
' ' * int(self.width * (1 - progress)),
int(progress * 100),
utils.time_diff(int(remains)))
)
sys.stdout.flush()
def main(argv):
start = time.time()
min_readers = 1
max_readers = 20
# NOTE: the defaults for begin time and end time
# Begin : 00010101
# End : 99991231
# Parse Args -------------------------------------------------------------
parser = argparse.ArgumentParser(prog='ncreview', description='Compare netCDF files between two directories or summarize from single directory.',
epilog='''Note that if --begin and --end are unspecified when comparing datastreams,
the time span chosen will be the intersection of the time periods spanned by both datastreams.''')
parser.add_argument('--version', '-v', action='version', version='%(prog)s v' + ncr.__version__)
parser.add_argument('old_dir', help='Old netCDF files directory')
parser.add_argument('new_dir', nargs='?', default=None,
help='New netCDF files directory, exclude to simply summarize a single directory.')
parser.add_argument('--begin', '-b', default='00010101',
metavar='YYYYMMDD', help='Ignore files before YYYYMMDD')
parser.add_argument('--end', '-e', default='99991231',
metavar='YYYYMMDD', help='Ignore files after YYYYMMDD')
parser.add_argument('--sample_interval', '-t', default=None,
help='Time interval to average data over in HH-MM-SS. If not provided, ' +
'defaults to 1 day if more than 10 days are being processed, otherwise defaults to hourly samples')
parser.add_argument('--metadata_only', '-m', action='store_true', default=False,
help='Review only metadata, ignoring variable data. Much faster than standard review.')
parser.add_argument('--write_dir', '-w', default=None, metavar='DIR',
help='write output data files to specified directory')
parser.add_argument('--name', '-n', default=None,
help='Specify custom name to be used for the run. Will be the directory name where the ' +
'summary files ncreview creates are stored as well as the URL suffix.')
parser.add_argument('--readers', type=int, default=10,
help='Specify number of concurrent file readers. Will accept a number between %d and %d (inclusive).' % (min_readers, max_readers))
args = parser.parse_args()
# Get absolute directory paths...
# This will be important for the webpage to know where the datastreams
# came from.
args.old_dir = os.path.abspath(args.old_dir)
if args.new_dir:
args.new_dir = os.path.abspath(args.new_dir)
if args.write_dir:
args.write_dir = os.path.abspath(args.write_dir)
if not os.path.exists(os.path.dirname(args.write_dir)):
raise ValueError(
"Error: write directory %s does not exist\n" % os.path.dirname(args.write_dir))
args.begin = dt.datetime.strptime(args.begin, '%Y%m%d')
args.end = dt.datetime.strptime(args.end, '%Y%m%d')
if args.readers < min_readers or args.readers > max_readers:
raise ValueError("Error: number of readers must be between %d and %d (inclusive)." % (
min_readers, max_readers))
try:
if args.sample_interval is not None:
h, m, s = args.sample_interval.split('-')
args.sample_interval = int(h) * 60 * 60 + int(m) * 60 + int(s)
# if interval is more than 10 days
elif args.end - args.begin > dt.timedelta(days=10):
args.sample_interval = 24 * 60 * 60 # set interval to 24 hr
else:
args.sample_interval = 60 * 60 # set interval to 1 hr
except:
raise ValueError("Error: chunk time %s is invalid.\n" %
args.sample_interval)
if args.sample_interval <= 0: # if user specified non-positive sample interval
raise ValueError(
'Error: sample interval must be a positive number, not ' + str(args.sample_interval))
# Review Data ------------------------------------------------------------
def is_valid(fname):
t = utils.file_time(fname)
return t is not None and args.begin <= t <= args.end
args.new_dir = os.path.abspath(
args.new_dir) if args.new_dir else args.new_dir
args.old_dir = os.path.abspath(
args.old_dir) if args.old_dir else args.old_dir
jdata = None
if args.new_dir:
new_files = sorted(filter(is_valid, os.listdir(args.new_dir)))
old_files = sorted(filter(is_valid, os.listdir(args.old_dir)))
if not new_files:
raise RuntimeError(
args.new_dir + ' contains no netCDF files in the specified time period.')
if not old_files:
raise RuntimeError(
args.new_dir + ' contains no netCDF files in the specified time period.')
# Get the latest begin and earliest end
new_times = list(map(utils.file_time, new_files))
old_times = list(map(utils.file_time, old_files))
# These values are hardcoded to match the default dates
# If user passed in start/end times, show the entire timeline at those dates
# Otherwise the program defaults to only showing overlap
if str(args.begin) != '0001-01-01 00:00:00':
args.begin = min(min(new_times), min(old_times)).replace(
hour=0, minute=0, second=0, microsecond=0)
else:
args.begin = max(min(new_times), min(old_times)).replace(
hour=0, minute=0, second=0, microsecond=0)
if str(args.end) != '9999-12-31 00:00:00':
args.end = max(max(new_times), max(old_times)).replace(
hour=23, minute=59, second=59, microsecond=999)
pass
else:
args.end = min(max(new_times), max(old_times)).replace(
hour=23, minute=59, second=59, microsecond=999)
# Re-filter the files with the new time bounds
new_files = sorted(filter(is_valid, new_files))
old_files = sorted(filter(is_valid, old_files))
if not new_files or not old_files:
raise RuntimeError('Old and New directories do not appear to have overlapping measurement ' +
'times in the specified time period. Cannot determine a comparison interval.')
print('Scanning directories...')
total_size = 0
for (which, path, files) in (('old', args.old_dir, old_files), ('new', args.new_dir, new_files)):
for f in files:
total_size += os.stat('%s/%s' % (path, f)).st_size
progress_bar = ProgressBar(total_size)
print('Reading data...')
old_ds = Datastream(is_plottable(args.old_dir), args.sample_interval)
new_ds = Datastream(is_plottable(args.new_dir), args.sample_interval)
progress_bar.start()
with ProcessPoolExecutor(max_workers=args.readers) as executor:
for s in executor.map(summarize, map(lambda f: ('%s/%s' % (args.old_dir, f), args.sample_interval, args.metadata_only), old_files)):
old_ds.add(s)
progress_bar.update(os.stat(s['path']).st_size)
for s in executor.map(summarize, map(lambda f: ('%s/%s' % (args.new_dir, f), args.sample_interval, args.metadata_only), new_files)):
new_ds.add(s)
progress_bar.update(os.stat(s['path']).st_size)
progress_bar.complete()
print('Comparing...')
dsdiff = DatastreamDiff(old_ds, new_ds)
jdata = dsdiff.jsonify()
else:
path = args.old_dir
files = sorted(filter(is_valid, os.listdir(path)))
if not files:
raise RuntimeError(
path + ' contains no netCDF files in the specified time period.')
print('Scanning directory...')
total_size = 0
for f in files:
total_size += os.stat('%s/%s' % (path, f)).st_size
progress_bar = ProgressBar(total_size)
print('Reading data...')
ds = Datastream(is_plottable(path), args.sample_interval)
progress_bar.start()
with ProcessPoolExecutor(max_workers=args.readers) as executor:
for s in executor.map(summarize, map(lambda f: ('%s/%s' % (path, f), args.sample_interval, args.metadata_only), files)):
ds.add(s)
progress_bar.update(os.stat(s['path']).st_size)
progress_bar.complete()
jdata = ds.jsonify()
# Write out the data -----------------------------------------------------
def unique_name(format_str, path):
'''Produce a unique directory name at the specified path'''
ID = 1
while os.path.exists(path + '/' + format_str.format(ID)):
ID += 1
return format_str.format(ID)
wpath = '/data/tmp/ncreview/'
if args.write_dir is not None:
wpath = args.write_dir
if not os.path.exists(wpath):
os.mkdir(wpath)
format_str = ''
if args.name:
format_str = args.name
if os.path.exists(wpath + '/' + args.name):
# if the directory already exists, add a unique id
format_str += '.{0}'
elif args.write_dir:
format_str = '.ncr.' + dt.datetime.now().strftime('%y%m%d.%H%M%S')
if os.path.exists(format_str):
# if the directory already exists, add a unique id
format_str += '.{0}'
else:
format_str = '%s.%s.{0}' % (os.environ['USER'], os.environ['HOST'])
jdata_dir = unique_name(format_str, wpath)
jdata_path = wpath + '/' + jdata_dir + '/'
os.mkdir(jdata_path)
def separate_data(obj, n=1):
to_separate = []
if obj['type'] in ['plot', 'timeline', 'fileTimeline', 'timelineDiff']:
to_separate = ['data']
elif obj['type'] in ['plotDiff', 'fileTimelineDiff']:
to_separate = ['old_data', 'new_data']
for key in to_separate:
# Generate a unique csv file name
while os.path.isfile(jdata_path + 'ncreview.{0}.csv'.format(n)):
n += 1
# Write out the data as csv
with open(jdata_path + 'ncreview.{0}.csv'.format(n), 'w', newline='') as csvfile:
writer = csv.writer(csvfile, quoting=csv.QUOTE_NONNUMERIC)
for row in obj[key]:
writer.writerow(row)
# Make what was the data a reference to the file
obj[key] = n
if 'contents' in obj:
for c in obj['contents']:
separate_data(c, n)
separate_data(jdata)
with open(jdata_path + 'ncreview.json', 'w') as jfile:
jfile.write(json.dumps(jdata, default=utils.JEncoder))
first_dir, user, *_ = os.path.realpath(__file__).split('/')[1:]
location = '/~' + user + '/dsutil' if first_dir == 'home' else ''
url_string = jdata_dir
if args.write_dir: # if custom write location, put full path
url_string = jdata_path
# reads csv file
# finds new/old values that are different
"""different_times = []
# the .5.csv is the file that has
file_path = '/data/tmp/ncreview/' + url_string + '/ncreview.5.csv'
with open(file_path, newline='') as csvfile:
reading = csv.reader(csvfile, delimiter = ',')
counter = 0
for row in reading:
old, new, begin, end = row
#dont need end
if counter == 0:
counter += 1
continue
if old != new:
#date = time.ctime(int(begin))
date = dt.datetime.fromtimestamp(int(begin)).strftime('%Y-%m-%d')
print('date: {}\told: {}\tnew: {}\tdifference: {}'.format(date, old, new, int(new) - int(old)))"""
print("")
print("Complete! Took %s." % utils.time_diff(time.time() - start))
print("------------------------------------------------------------------------")
print('https://engineering.arm.gov' +
location + '/ncreview/?' + url_string)
print("")
return 0
