diff options
Diffstat (limited to 'python')
-rw-r--r-- | python/vyos/accel_ppp.py | 74 |
1 files changed, 74 insertions, 0 deletions
diff --git a/python/vyos/accel_ppp.py b/python/vyos/accel_ppp.py new file mode 100644 index 000000000..bfc8ee5a9 --- /dev/null +++ b/python/vyos/accel_ppp.py @@ -0,0 +1,74 @@ +#!/usr/bin/env python3 +# +# Copyright (C) 2022 VyOS maintainers and contributors +# +# This program is free software; you can redistribute it and/or modify +# it under the terms of the GNU General Public License version 2 or later as +# published by the Free Software Foundation. +# +# This program is distributed in the hope that it will be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +# GNU General Public License for more details. +# +# You should have received a copy of the GNU General Public License +# along with this program. If not, see <http://www.gnu.org/licenses/>. +# + +import sys + +import vyos.opmode +from vyos.util import rc_cmd + + +def get_server_statistics(accel_statistics, pattern, sep=':') -> dict: + import re + + stat_dict = {'sessions': {}} + + cpu = re.search(r'cpu(.*)', accel_statistics).group(0) + # Find all lines with pattern, for example 'sstp:' + data = re.search(rf'{pattern}(.*)', accel_statistics, re.DOTALL).group(0) + session_starting = re.search(r'starting(.*)', data).group(0) + session_active = re.search(r'active(.*)', data).group(0) + + for entry in {cpu, session_starting, session_active}: + if sep in entry: + key, value = entry.split(sep) + if key in ['starting', 'active', 'finishing']: + stat_dict['sessions'][key] = value.strip() + continue + stat_dict[key] = value.strip() + return stat_dict + + +def accel_cmd(port: int, command: str) -> str: + _, output = rc_cmd(f'/usr/bin/accel-cmd -p{port} {command}') + return output + + +def accel_out_parse(accel_output: list[str]) -> list[dict[str, str]]: + """ Parse accel-cmd show sessions output """ + data_list: list[dict[str, str]] = list() + field_names: list[str] = list() + + field_names_unstripped: list[str] = accel_output.pop(0).split('|') + for field_name in field_names_unstripped: + field_names.append(field_name.strip()) + + while accel_output: + if '|' not in accel_output[0]: + accel_output.pop(0) + continue + + current_item: list[str] = accel_output.pop(0).split('|') + item_dict: dict[str, str] = {} + + for field_index in range(len(current_item)): + field_name: str = field_names[field_index] + field_value: str = current_item[field_index].strip() + item_dict[field_name] = field_value + + data_list.append(item_dict) + + return data_list |