def plot_index_size():
compile_command = compile_base_command + " " + cpp_ipmt_source
run(compile_command, print_output=True)
for textfile_name in textfiles_names:
textfile_path = textfiles_dir + textfile_name
textfile_idx_path = textfiles_dir + textfile_name + ".idx"
plot_values = []
plot_labels = []
for compression_algorithm in compression_algorithms:
for index_algorithm in index_algorithms:
plot_labels += [compression_algorithm + '-' + index_algorithm]
for compression_algorithm in compression_algorithms:
for index_algorithm in index_algorithms:
index_command = "./a.out index -v %s --compression=%s --indextype=%s" % (
textfile_path, compression_algorithm, index_algorithm)
run(index_command)
index_size = os.path.getsize(textfile_idx_path)
plot_values += [index_size]
PlotUtils.bar_plot(plot_values=plot_values,
plot_labels=plot_labels,
colors=colors,
ylabel="size in bytes",
title="Index size of " + textfile_name,
save_filename='plots/index-size-' + textfile_name)
def plot_index_time():
compile_command = compile_base_command + " " + cpp_ipmt_source
run(compile_command, print_output=True)
for textfile_name in textfiles_names:
textfile_path = textfiles_dir + textfile_name
plot_values = []
plot_labels = []
for compression_algorithm in compression_algorithms:
for index_algorithm in index_algorithms:
plot_labels += [compression_algorithm + '-' + index_algorithm]
for compression_algorithm in compression_algorithms:
for index_algorithm in index_algorithms:
index_command = "./a.out index -v %s --compression=%s --indextype=%s" % (
textfile_path, compression_algorithm, index_algorithm)
r = functools.partial(run, index_command, print_output=True)
run_time = get_run_time(r, runs=num_of_runs)
plot_values += [run_time]
# plot_data[textfile_name] += [gzip_compressed_size]
PlotUtils.bar_plot(plot_values=plot_values,
plot_labels=plot_labels,
colors=colors,
ylabel="time in seconds",
title="Index time of " + textfile_name,
save_filename='plots/index-time-' + textfile_name)
def plot_decompression_time():
compile_command = compile_base_command + " " + cpp_compression_source
run(compile_command, print_output=True)
for textfile_name in textfiles_names:
textfile_path = textfiles_dir + textfile_name
textfile_zip_path = textfile_path + ".myzip"
plot_values = []
plot_labels = []
for algorithm in compression_algorithms:
plot_labels += [algorithm]
plot_labels += ['gzip']
for algorithm in compression_algorithms:
compression_command = run_base_command + ' %s %s compress' % (
algorithm, textfile_path)
decompression_command = run_base_command + ' %s %s decompress' % (
algorithm, textfile_zip_path)
run(compression_command, print_output=True)
r = functools.partial(run,
decompression_command,
print_output=True)
run_time = get_run_time(r, runs=num_of_runs)
plot_values += [run_time]
textfile_gzip_path = textfile_path + ".gz"
gzip_compress_command = "gzip -k -f " + textfile_path
gzip_decompress_command = "gzip -k -f -d " + textfile_gzip_path
run(gzip_compress_command)
r = functools.partial(run, gzip_decompress_command, print_output=True)
run_time = get_run_time(r, runs=num_of_runs)
plot_values += [run_time]
print(plot_values)
PlotUtils.bar_plot(plot_values=plot_values,
plot_labels=plot_labels,
colors=colors,
ylabel="time in seconds",
title="Decompression time of " + textfile_name,
save_filename='plots/decompress-time-' +
textfile_name)
def plot_compression_size():
compile_command = '%s %s' % (compile_base_command, cpp_compression_source)
run(compile_command, print_output=True)
for textfile_name in textfiles_names:
textfile_path = textfiles_dir + textfile_name
textfile_zip_path = textfile_path + ".myzip"
plot_values = []
plot_labels = []
for algorithm in compression_algorithms:
plot_labels += [algorithm]
plot_labels += ['gzip']
for algorithm in compression_algorithms:
run_command = run_base_command + ' %s %s compress' % (
algorithm, textfile_path)
if os.path.exists(textfile_zip_path):
os.remove(textfile_zip_path)
run(run_command, print_output=True)
compressed_size = os.path.getsize(textfile_zip_path)
plot_values += [compressed_size]
gzip_command = "gzip -k -f " + textfile_path
run(gzip_command)
textfile_gzip_path = textfile_path + ".gz"
gzip_size = os.path.getsize(textfile_gzip_path)
plot_values += [gzip_size]
print(plot_values)
PlotUtils.bar_plot(plot_values,
plot_labels=plot_labels,
colors=colors,
ylabel="size in bytes",
title="Compression size of " + textfile_name,
save_filename='plots/compress-size-' +
textfile_name)
