def analyse_2d_density(
exp_list='experiments_completed.txt',
res_dir='',
out_dir='',
source_filter='',
min_values='3',
xmetric='throughput',
ymetric='tcprtt',
variables='',
out_name='',
xmin='0',
xmax='0',
ymin='0',
ymax='0',
lnames='',
group_by='aqm',
replot_only='0',
pdf_dir='',
stime='0.0',
etime='0.0',
ts_correct='1',
smoothed='1',
link_len='0',
plot_params='',
plot_script='',
xstat_index='',
ystat_index='',
dupacks='0',
cum_ackseq='1',
merge_data='0',
#sburst='1', eburst='0', test_id_prefix='[0-9]{8}\-[0-9]{6}_experiment_',
sburst='1',
eburst='0',
test_id_prefix='exp_[0-9]{8}\-[0-9]{6}_',
slowest_only='0',
query_host=''):
"2d density / ellipse plot for different experiments"
test_id_pfx = ''
check = get_metric_params(xmetric, smoothed, ts_correct)
if check == None:
abort('Unknown metric %s specified with xmetric' % xmetric)
check = get_metric_params(ymetric, smoothed, ts_correct)
if check == None:
abort('Unknown metric %s specified with ymetric' % ymetric)
#if source_filter == '':
# abort('Must specify at least one source filter')
if len(source_filter.split(';')) > 12:
abort('Cannot have more than 12 filters')
# XXX more param checking
# make sure res_dir has valid form (out_dir is handled by extract methods)
res_dir = valid_dir(res_dir)
# Initialise source filter data structure
sfil = SourceFilter(source_filter)
# read test ids
experiments = read_experiment_ids(exp_list)
# get path based on first experiment id
dir_name = get_first_experiment_path(experiments)
# if we haven' got the extracted data run extract method(s) first
if res_dir == '':
for experiment in experiments:
(ex_function,
kwargs) = get_extract_function(xmetric,
link_len,
xstat_index,
sburst=sburst,
eburst=eburst,
slowest_only=slowest_only,
query_host=query_host)
(dummy, out_files,
out_groups) = ex_function(test_id=experiment,
out_dir=out_dir,
source_filter=source_filter,
replot_only=replot_only,
ts_correct=ts_correct,
**kwargs)
(ex_function,
kwargs) = get_extract_function(ymetric,
link_len,
ystat_index,
sburst=sburst,
eburst=eburst,
slowest_only=slowest_only,
query_host=query_host)
(dummy, out_files,
out_groups) = ex_function(test_id=experiment,
out_dir=out_dir,
source_filter=source_filter,
replot_only=replot_only,
ts_correct=ts_correct,
**kwargs)
if out_dir == '' or out_dir[0] != '/':
res_dir = dir_name + '/' + out_dir
else:
res_dir = out_dir
else:
if res_dir[0] != '/':
res_dir = dir_name + '/' + res_dir
# make sure we have trailing slash
res_dir = valid_dir(res_dir)
if pdf_dir == '':
pdf_dir = res_dir
else:
if pdf_dir[0] != '/':
pdf_dir = dir_name + '/' + pdf_dir
pdf_dir = valid_dir(pdf_dir)
# if pdf_dir specified create if it doesn't exist
mkdir_p(pdf_dir)
#
# build match string from variables
#
(match_str, match_str2) = build_match_strings(experiments[0], variables,
test_id_prefix)
#
# filter out the experiments to plot, generate x-axis labels, get test id prefix
#
(fil_experiments, test_id_pfx,
dummy) = filter_experiments(experiments, match_str, match_str2)
#
# get groups based on group_by variable
#
group_idx = 1
levels = {}
groups = []
leg_names = []
_experiments = []
for experiment in fil_experiments:
level = ''
add_exp = True
for g in group_by.split(';'):
p = experiment.find(g)
if p > -1:
s = experiment.find('_', p)
s += 1
e = experiment.find('_', s)
level += g + ':' + experiment[s:e] + ' '
else:
add_exp = False
break
# remove the final space from the string
level = level[:-1]
if add_exp == True:
_experiments.append(experiment)
#print('level: ' + level)
if level not in levels:
levels[level] = group_idx
group_idx += 1
leg_names.append(level)
if merge_data == '1':
groups.append(levels[level])
else:
for i in range(len(source_filter.split(';'))):
groups.append(levels[level])
fil_experiments = _experiments
#
# get metric parameters and list of data files
#
# get the metric parameter for both x and y
x_axis_params = get_metric_params(xmetric, smoothed, ts_correct,
xstat_index, dupacks, cum_ackseq,
slowest_only)
y_axis_params = get_metric_params(ymetric, smoothed, ts_correct,
ystat_index, dupacks, cum_ackseq,
slowest_only)
x_ext = x_axis_params[0]
y_ext = y_axis_params[0]
# if we merge responders make sure we only use the merged files
if merge_data == '1':
# reset source filter so we match the merged file
sfil.clear()
sfil = SourceFilter('S_0.0.0.0_0')
x_files = []
y_files = []
for experiment in fil_experiments:
_x_files = []
_y_files = []
_x_ext = x_ext
_y_ext = y_ext
_files = get_testid_file_list('', experiment, _x_ext, 'LC_ALL=C sort',
res_dir)
if merge_data == '1':
_x_ext += '.all'
_files = merge_data_files(_files)
_x_files += _files
_files = get_testid_file_list('', experiment, _y_ext, 'LC_ALL=C sort',
res_dir)
if merge_data == '1':
_y_ext += '.all'
_files = merge_data_files(_files)
_y_files += _files
match_str = '.*_([0-9\.]*_[0-9]*_[0-9\.]*_[0-9]*)[0-9a-z_.]*' + _x_ext
for f in _x_files:
#print(f)
res = re.search(match_str, f)
#print(res.group(1))
if res and sfil.is_in(res.group(1)):
# only add file if enough data points
rows = int(
local('wc -l %s | awk \'{ print $1 }\'' % f, capture=True))
if rows > int(min_values):
x_files.append(f)
match_str = '.*_([0-9\.]*_[0-9]*_[0-9\.]*_[0-9]*)[0-9a-z_.]*' + _y_ext
for f in _y_files:
# print(f)
res = re.search(match_str, f)
if res and sfil.is_in(res.group(1)):
# only add file if enough data points
rows = int(
local('wc -l %s | awk \'{ print $1 }\'' % f, capture=True))
if rows > int(min_values):
y_files.append(f)
yindexes = [str(x_axis_params[2]), str(y_axis_params[2])]
yscalers = [str(x_axis_params[3]), str(y_axis_params[3])]
aggr_flags = [x_axis_params[5], y_axis_params[5]]
diff_flags = [x_axis_params[6], y_axis_params[6]]
if lnames != '':
lnames_arr = lnames.split(';')
if len(lnames_arr) != len(leg_names):
abort(
'Number of legend names must be qual to the number of source filters'
)
leg_names = lnames_arr
print(x_files)
print(y_files)
print(groups)
print(leg_names)
#
# pass the data files and auxilary info to plot function
#
if out_name != '':
oprefix = out_name + '_' + test_id_pfx + '_' + xmetric + '_' + ymetric
else:
oprefix = test_id_pfx + '_' + xmetric + '_' + ymetric
title = oprefix
plot_2d_density(title, x_files, y_files, x_axis_params[1],
y_axis_params[1], yindexes, yscalers, 'pdf', oprefix,
pdf_dir, x_axis_params[4], y_axis_params[4], aggr_flags,
diff_flags, xmin, xmax, ymin, ymax, stime, etime, groups,
leg_names, plot_params, plot_script)
# done
puts('\n[MAIN] COMPLETED analyse_2d_density %s \n' % test_id_pfx)
def analyse_cmpexp(
exp_list='experiments_completed.txt',
res_dir='',
out_dir='',
source_filter='',
min_values='3',
omit_const='0',
metric='throughput',
ptype='box',
variables='',
out_name='',
ymin='0',
ymax='0',
lnames='',
group_by_prefix='0',
omit_const_xlab_vars='0',
replot_only='0',
pdf_dir='',
stime='0.0',
etime='0.0',
ts_correct='1',
smoothed='1',
link_len='0',
plot_params='',
plot_script='',
stat_index='',
dupacks='0',
cum_ackseq='1',
merge_data='0',
sburst='1',
#eburst='0', test_id_prefix='[0-9]{8}\-[0-9]{6}_experiment_',
eburst='0',
test_id_prefix='exp_[0-9]{8}\-[0-9]{6}_',
slowest_only='0',
res_time_mode='0',
query_host=''):
"Compare metrics for different experiments"
if ptype != 'box' and ptype != 'mean' and ptype != 'median':
abort('ptype must be either box, mean or median')
check = get_metric_params(metric, smoothed, ts_correct)
if check == None:
abort('Unknown metric %s specified' % metric)
if source_filter == '':
abort('Must specify at least one source filter')
if len(source_filter.split(';')) > 12:
abort('Cannot have more than 12 filters')
# prevent wrong use of res_time_mode
if metric != 'restime' and res_time_mode != '0':
res_time_mode = '0'
if ptype == 'box' and res_time_mode == '2':
res_time_mode = '0'
# XXX more param checking
# Initialise source filter data structure
sfil = SourceFilter(source_filter)
# read test ids
experiments = read_experiment_ids(exp_list)
# get path based on first experiment id
dir_name = get_first_experiment_path(experiments)
# if we haven' got the extracted data run extract method(s) first
if res_dir == '':
for experiment in experiments:
(ex_function,
kwargs) = get_extract_function(metric,
link_len,
stat_index,
sburst=sburst,
eburst=eburst,
slowest_only=slowest_only,
query_host=query_host)
(dummy, out_files,
out_groups) = ex_function(test_id=experiment,
out_dir=out_dir,
source_filter=source_filter,
replot_only=replot_only,
ts_correct=ts_correct,
**kwargs)
if out_dir == '' or out_dir[0] != '/':
res_dir = dir_name + '/' + out_dir
else:
res_dir = out_dir
else:
if res_dir[0] != '/':
res_dir = dir_name + '/' + res_dir
# make sure we have trailing slash
res_dir = valid_dir(res_dir)
if pdf_dir == '':
pdf_dir = res_dir
else:
if pdf_dir[0] != '/':
pdf_dir = dir_name + '/' + pdf_dir
pdf_dir = valid_dir(pdf_dir)
# if pdf_dir specified create if it doesn't exist
mkdir_p(pdf_dir)
#
# build match string from variables
#
(match_str, match_str2) = build_match_strings(experiments[0], variables,
test_id_prefix)
#
# filter out the experiments to plot, generate x-axis labels, get test id prefix
#
(fil_experiments, test_id_pfx,
xlabs) = filter_experiments(experiments, match_str, match_str2)
#
# get out data files based on filtered experiment list and source_filter
#
(ext, ylab, yindex, yscaler, sep, aggr,
diff) = get_metric_params(metric, smoothed, ts_correct, stat_index,
dupacks, cum_ackseq, slowest_only)
if res_time_mode == '1':
plot_params += ' NOMINAL_RES_TIME="1"'
if res_time_mode == '2':
if ptype == 'median':
ylab = 'Median resp time / nominal resp time'
elif ptype == 'mean':
ylab = 'Mean resp time / nominal resp time'
plot_params += ' RATIO_RES_TIME="1"'
leg_names = source_filter.split(';')
# if we merge responders make sure we only use the merged files
if merge_data == '1':
# set label to indicate merged data
leg_names = ['Merged data']
# reset source filter so we match the merged file
sfil.clear()
source_filter = 'S_0.0.0.0_0'
sfil = SourceFilter(source_filter)
file_names = []
for experiment in fil_experiments:
out_files = {}
_ext = ext
files = get_testid_file_list('', experiment, '%s' % _ext,
'LC_ALL=C sort', res_dir)
if merge_data == '1':
# change extension
_ext += '.all'
files = merge_data_files(files)
#print(files)
match_str = '.*_([0-9\.]*_[0-9]*_[0-9\.]*_[0-9]*)[0-9a-z_.]*' + _ext
for f in files:
# print(f)
res = re.search(match_str, f)
#print(res.group(1))
if res and sfil.is_in(res.group(1)):
# only add file if enough data points
rows = int(
local('wc -l %s | awk \'{ print $1 }\'' % f, capture=True))
if rows > int(min_values):
out_files[res.group(1)] = f
#print(out_files)
#print(leg_names)
if len(out_files) < len(leg_names):
abort(
'No data files for some of the source filters for experiment %s'
% experiment)
sorted_files = sort_by_flowkeys(out_files, source_filter)
for name, file_name in sorted_files:
file_names.append(file_name)
if group_by_prefix == '1':
# group by test prefix (and flow)
# first, get all test id prefixes
test_id_pfxs = {}
for experiment in fil_experiments:
res = re.search(match_str2, experiment)
if res:
test_id_pfxs[res.group(1)] = 1
# second, sort files so that same parameter combinations for different
# prefixes are together
# if we have multiple prefixes, create legend entry for each
# prefix+flow combination
_file_names = [''] * len(file_names)
_leg_names = []
pfx_cnt = len(test_id_pfxs)
i = 0
j = -1
last_pfx = ''
for name in file_names:
for p in test_id_pfxs:
if name.find(p) > -1:
curr_pfx = p
break
if curr_pfx != last_pfx:
i = 0
j += 1
for l in leg_names:
_leg_names.append(curr_pfx + '-' + l)
_file_names[i * pfx_cnt + j] = name
i += 1
last_pfx = curr_pfx
file_names = _file_names
leg_names = _leg_names
# remove duplicates in the x-axis labels
xlabs = list(set(xlabs))
if lnames != '':
lnames_arr = lnames.split(';')
if len(lnames_arr) != len(leg_names):
abort(
'Number of legend names must be qual to the number of source filters'
)
leg_names = lnames_arr
# filter out unchanged variables in the x labels (need at least 2 labels)
if omit_const_xlab_vars == '1' and len(xlabs) > 1:
xlabs_arrs = {}
xlabs_changed = {}
for i in range(len(xlabs)):
xlabs_arrs[i] = xlabs[i].split('\n')
for i in range(len(xlabs_arrs[0])):
changed = False
xlab_var = xlabs_arrs[0][i]
for j in range(1, len(xlabs)):
if xlabs_arrs[j][i] != xlab_var:
changed = True
break
xlabs_changed[i] = changed
for i in range(len(xlabs)):
tmp = []
for j in range(len(xlabs_arrs[i])):
if xlabs_changed[j]:
tmp.append(xlabs_arrs[i][j].replace('_', ' ', 1))
xlabs[i] = '\n'.join(tmp)
print(leg_names)
print(file_names)
#
# pass the data files and auxilary info to plot function
#
if out_name != '':
oprefix = out_name + '_' + test_id_pfx + '_' + metric + '_' + ptype
else:
oprefix = test_id_pfx + '_' + metric + '_' + ptype
title = oprefix
plot_cmpexp(title, file_names, xlabs, ylab, yindex, yscaler, 'pdf',
oprefix, pdf_dir, sep, aggr, diff, omit_const, ptype, ymin,
ymax, leg_names, stime, etime, plot_params, plot_script)
# done
puts('\n[MAIN] COMPLETED analyse_cmpexp %s \n' % test_id_pfx)