update and save

This commit is contained in:
Ta-Shun Su
2019-08-10 12:52:46 +08:00
parent 4ed89630f7
commit 2ff06e2142
2 changed files with 281 additions and 40 deletions
+1
View File
@@ -0,0 +1 @@
*.txt
+280 -40
View File
@@ -1,51 +1,291 @@
import pandas as pd
import numpy as np
"""
purpose:
in order to analyze the data loss of bluetooth, a small code is needed.
Guess that there are three type of data loss. In order to prove this idea,
several tests are needed
types of data loss:
1. packet loss:
bluetooth notification does not need any negotiation.
if there is any packet loss during the connection,
it will not be lost forever
2. program delay causes loss in client(headstage):
too many effort on data packaging, it would cause
data delay
3. program delay causes loss in host(controller):
...todo and need to define
Assume that we have lost several bytes data. We want to analyze the type of data lost.
method:
compare ramp data with time stamp, there are several conditions
if data_delta * 1 / sampling_rate == time_delta:
this should be packet loss
else
this should be program delay
method to coped with the corresponding problem:
1. try 'Indication' to check data loss is reduced or not.
2. modified the procedure of data packaging
3. after the upper two problems are excluded, this problem should be left over
"""
import os.path
import sys
from collections import Counter
from typing import List, Optional, Dict, Set, Any
def main():
'''
purpose:
in order to analyze the data loss of bluetooth, a small code is needed.
Guess that there are three type of data loss. In order to prove this idea,
several tests are needed
class Options:
def __init__(self):
self.skip = False
self.round = 2
self.overflow_pow = 10
types of data loss:
1. packet loss:
bluetooth notification does not need any negotiation.
if there is any packet loss during the connection,
it will not be lost forever
2. program delay causes loss in client(headstage):
too many effort on data packaging, it would cause
data delay
3. program delay causes loss in host(controller):
...todo and need to define
Assume that we have lost several bytes data. We want to analyze the type of data lost.
method:
compare ramp data with time stamp, there are several conditions
if data_delta * 1 / sampling rate == time_delta:
this should be packet loss
else
this should be program delay
class Result:
def __init__(self, options: Options):
# data file information
self.date_time: Optional[str] = None
self.device_name: Optional[str] = None
self.device_address: Optional[str] = None
self.parameter: Dict[str, str] = {}
self.channel: Set[int] = set()
self.total_duration: float = 0.0
method to coped with the corresponding problem:
1. try 'Indication' to check data loss is reduced or not.
2. modified the procedure of data packaging
3. after the upper two problems are excluded, this problem should be left over
# analysis data
self.data_count = 0
self.time_delta_counter: Dict[float, int] = Counter()
self.value_delta_counter: Dict[int, int] = Counter()
# analysis options
self.round = options.round
self.overflow = 2 ** options.overflow_pow
# temp data
self._prev_time: Optional[float] = None
self._prev_value: Optional[int] = None
self._close = False
@property
def sample_rate(self) -> int:
try:
sample_rate = self.parameter['SAMPLE_RATE']
except KeyError as e:
raise RuntimeError('data file do not contain SAMPLE_RATE') from e
try:
return int(sample_rate)
except ValueError as e:
raise RuntimeError('data file SAMPLE_RATE value not a value : ' + str(sample_rate)) from e
def update(self, time: float, channel: int, value: int):
if self._close:
raise RuntimeError('closed')
self.data_count += 1
if self._prev_time is not None:
delta = round(time - self._prev_time, self.round)
self.time_delta_counter[delta] += 1
if self._prev_value is not None:
delta = self._prev_value - value
self.value_delta_counter[delta] += 1
self._prev_time = time
self.channel.add(channel)
self._prev_value = value
def close(self):
self._close = True
def calculate(txt_file: str, options: Options) -> Result:
if not os.path.exists(txt_file):
raise FileNotFoundError(txt_file)
result = Result(options)
parsing_state = 0
with open(txt_file) as _file:
for line, content in enumerate(_file): # type: int, str
content = content.strip()
if len(content) == 0:
continue
if parsing_state == 0:
if content.startswith('#'):
if result.date_time is None:
result.date_time = content[1:].strip()
elif content.startswith('# device_name'):
result.device_name = content[len('# device_name'):].strip()
elif content.startswith('# mac_address'):
result.device_address = content[len('# mac_address'):].strip()
elif content.startswith('# parameter'):
parsing_state = 1
elif parsing_state == 1:
if content.startswith('#'):
if content.startswith('# time_stamp'):
parsing_state = 2
else:
part = content[2:].split(' ', maxsplit=2)
result.parameter[part[0]] = part[1]
elif parsing_state == 2:
if content.startswith('#'):
continue
part = content.split(' ', maxsplit=3)
try:
time = part[0]
channel = part[1]
value = part[2]
except IndexError:
print('@%d' % (line + 1), 'incomplete data', ':', content)
continue
check_pass = True
try:
time = float(time)
except ValueError:
print('@%d' % (line + 1), 'incomplete time', ':', time)
check_pass = False
try:
channel = int(channel)
except ValueError:
print('@%d' % (line + 1), 'incomplete channel', ':', channel)
check_pass = False
try:
value = int(value)
except ValueError:
print('@%d' % (line + 1), 'incomplete value', ':', value)
check_pass = False
if check_pass:
result.update(time, channel, value)
result.close()
return result
def print_result(result: Result, options: Options):
def print_table(counter: Dict[Any, Any], key_format='%10d', value_format='%d'):
f = ' %s : %s' % (key_format, value_format)
for k in sorted(counter.keys()):
print(f % (k, counter[k]))
print('data_count', result.data_count)
print('channel', ' '.join(list(map(str, sorted(result.channel)))))
try:
sample_rate = result.sample_rate
print('sample_rate [1/s]', sample_rate)
print('1/sample_rate [ms]', 1000 / sample_rate)
except RuntimeError:
print('sample_rate', '(E)')
print('time_delta')
print_table(result.time_delta_counter, '%10.2f')
print('value_delta')
print_table(result.value_delta_counter)
def print_help():
prg = sys.argv[0]
print(prg, '[OPTIONS]', 'FILE', '...')
print()
print('OPTIONS:')
print(' -h, --help : print help')
print(' --skip : skip if FILE not found')
print(' --round VALUE : floating number round')
print(' --overflow VALUE : overflow number. 2^VALUE')
print()
print('ARGUMENTS:')
print(' FILE : txt data file')
print()
def main(argv: List[str]):
"""
:param argv:
:return:
'''
"""
# df = pd.read_csv("C:/Users/yichin/Downloads/last-2019-07-18-16-30-15-7.csv", sep=",", skiprows=12, header=None)
# sel_df = pd.DataFrame(df)
# column, row = sel_df.shape
# sel_df = sel_df.fillna(value=0)
# delta_df = sel_df.diff(periods=1)
# delta_df = delta_df.fillna(value=0)
# delta_df = delta_df.drop(column - 1)
# delta_df.set_axis(['time_delta', 'data_delta'], axis='columns', inplace=True)
# skip header information
df = pd.read_csv("C:/Users/yichin/Downloads/last-2019-07-18-16-30-15-7.csv", sep=",", skiprows=12, header=None)
sel_df = pd.DataFrame(df)
column, row = sel_df.shape
sel_df = sel_df.fillna(value=0)
delta_df = sel_df.diff(periods=1)
delta_df = delta_df.fillna(value=0)
delta_df = delta_df.drop(column-1)
delta_df.set_axis(['time_delta', 'data_delta'], axis='columns', inplace=True)
o = Options()
i = 0
while i < len(argv):
arg = argv[i]
if arg in ('-h', '--help'):
print_help()
return
elif arg == '--skip':
o.skip = True
i += 1
elif arg == '--round':
o.round = int(argv[i + 1])
i += 2
elif arg.startswith('--round='):
o.round = int(arg[len('--round='):])
i += 1
elif arg == '--overflow':
o.overflow_pow = int(argv[i + 1])
i += 2
elif arg.startswith('--overflow='):
o.overflow_pow = int(arg[len('--overflow='):])
i += 1
elif arg.startswith('-'):
raise ValueError('unknown options : ' + arg)
else:
i += 1
try:
result = calculate(arg, o)
except FileNotFoundError:
if o.skip:
continue
else:
raise
else:
print_result(result, o)
if __name__ == '__main__':
main()
main(sys.argv[1:])