# python summarize.py mnist-search.txt

# val_accuracy	depth	width	activation	regularization_function	regularization_penalty	dropout	optimizer	learning_rate	batch_size	tuner/epochs	tuner/initial_epoch	tuner/bracket	tuner/round
# 0.9850000143051147	3	512	swish	l1_l2	1e-05	0.2	adam	0.01	512	32	11	2	2
# ...

import re
import sys
hyperparameters = {}
metric_name = None
data = []
for line in open(sys.argv[1], 'r'):
    line = line.strip('\r\n')
    match = re.match(r'^(\S+)\s+\|(\S+)\s+\|(\S+)$', line)
    if match:
        hyperparameters[match.group(3)] = match.group(1)
    else:
        match = re.match(r'^(val_\S+): (\S+)$', line)
        if match:
            if metric_name is None:
                metric_name = match.group(1)
            data.append((match.group(2), hyperparameters))
            hyperparameters = {}
output = []
for k in data[-1][1].keys():
    output.append(k)
print(metric_name, '\t'.join(output), sep = '\t')
for metric_value,dictionary in sorted(data, key = lambda x: x[0], reverse = True):
    output = []
    for k in data[-1][1].keys():
        if k in dictionary:
            output.append(dictionary[k])
        else:
            output.append('')
    print(metric_value, '\t'.join(output), sep = '\t')