Implement message parser and average latency calculator. (#38)
* Implement message parser and average latency calculator. * Fix node_id type parsing, add more stats. * Add more stats, slight refactor. * Remove unnecessary class * Improve naming. * Adapt to new log format
This commit is contained in:
parent
5adb0a1ad6
commit
424585dd2a
|
@ -1,3 +1,4 @@
|
|||
.venv/
|
||||
__pycache__/
|
||||
*.py[cod]
|
||||
*$py.class
|
||||
|
@ -10,3 +11,4 @@ Cargo.lock
|
|||
simlib/**/target
|
||||
simlib/**/Cargo.lock
|
||||
simlib/test.json
|
||||
*.ignore*
|
||||
|
|
|
@ -0,0 +1,130 @@
|
|||
# !/usr/bin/env python
|
||||
import json
|
||||
import sys
|
||||
from collections.abc import Iterable
|
||||
from typing import Dict, Optional
|
||||
import statistics
|
||||
import argparse
|
||||
|
||||
from json_stream.base import TransientStreamingJSONObject
|
||||
|
||||
JsonStream = Iterable[TransientStreamingJSONObject]
|
||||
|
||||
|
||||
class Message:
|
||||
def __init__(self, message_id: str, step_a: Optional[int]):
|
||||
self.id = message_id
|
||||
self.step_a = int(step_a) if step_a is not None else None
|
||||
self.step_b = None
|
||||
|
||||
def __hash__(self):
|
||||
return self.id
|
||||
|
||||
def __repr__(self):
|
||||
return f"[{self.id}] {self.step_a} -> {self.step_b}"
|
||||
|
||||
@property
|
||||
def latency(self) -> Optional[int]:
|
||||
if self.step_a is not None and self.step_b is not None:
|
||||
return abs(self.step_a - self.step_b)
|
||||
|
||||
|
||||
MessageStorage = Dict[str, Message]
|
||||
|
||||
|
||||
def compute_results(message_storage: MessageStorage, step_duration: int) -> str:
|
||||
latencies = [message_record.latency for message_record in message_storage.values()]
|
||||
valued_latencies = [latency for latency in latencies if latency is not None]
|
||||
incomplete_latencies = sum((1 for latency in latencies if latency is None))
|
||||
|
||||
total_messages = len(latencies)
|
||||
total_messages_full_latency = len(valued_latencies)
|
||||
total_messages_incomplete_latency = incomplete_latencies
|
||||
latency_average_steps = statistics.mean(valued_latencies)
|
||||
latency_average_ms = "{:.2f}ms".format(latency_average_steps * step_duration)
|
||||
latency_median_steps = statistics.median(valued_latencies)
|
||||
latency_median_ms = "{:.2f}ms".format(latency_median_steps * step_duration)
|
||||
max_latency_steps = max(valued_latencies)
|
||||
max_latency_ms = "{:.2f}ms".format(max_latency_steps * step_duration)
|
||||
min_latency_steps = min(valued_latencies)
|
||||
min_latency_ms = "{:.2f}ms".format(min_latency_steps * step_duration)
|
||||
|
||||
return f"""[Results]
|
||||
- Total messages: {total_messages}
|
||||
- Full latencies: {total_messages_full_latency}
|
||||
- Incomplete latencies: {total_messages_incomplete_latency}
|
||||
- Averages
|
||||
- Steps: {latency_average_steps}
|
||||
- Duration: {latency_average_ms}
|
||||
- Median
|
||||
- Steps: {latency_median_steps}
|
||||
- Duration: {latency_median_ms}
|
||||
- Max
|
||||
- Steps: {max_latency_steps}
|
||||
- Duration: {max_latency_ms}
|
||||
- Min
|
||||
- Steps: {min_latency_steps}
|
||||
- Duration: {min_latency_ms}"""
|
||||
|
||||
|
||||
def parse_record_stream(record_stream: Iterable[str]) -> MessageStorage:
|
||||
storage: MessageStorage = {}
|
||||
|
||||
for record in record_stream:
|
||||
json_record = json.loads(record)
|
||||
payload_id = json_record["payload_id"]
|
||||
step_id = json_record["step_id"]
|
||||
|
||||
if (stored_message := storage.get(payload_id)) is None:
|
||||
storage[payload_id] = Message(payload_id, step_id)
|
||||
else:
|
||||
stored_message.step_b = step_id
|
||||
|
||||
return storage
|
||||
|
||||
|
||||
def line_to_json_stream(record_stream: Iterable[str]) -> Iterable[str]:
|
||||
for record in record_stream:
|
||||
bracket_pos = record.rfind("{")
|
||||
yield record[bracket_pos:]
|
||||
|
||||
|
||||
def get_pipe_stream() -> Iterable[str]:
|
||||
yield from sys.stdin
|
||||
|
||||
|
||||
def get_file_stream(input_filename) -> Iterable[str]:
|
||||
with open(input_filename, "r") as file:
|
||||
yield from file
|
||||
|
||||
|
||||
def get_input_stream(input_filename: Optional[str]) -> Iterable[str]:
|
||||
stream = get_file_stream(input_filename) if input_filename is not None else get_pipe_stream()
|
||||
return line_to_json_stream(stream)
|
||||
|
||||
|
||||
def build_argument_parser() -> argparse.ArgumentParser:
|
||||
parser = argparse.ArgumentParser(description="Log analysis for nomos-simulations.")
|
||||
parser.add_argument(
|
||||
"--step-duration",
|
||||
type=int,
|
||||
default=100,
|
||||
help="Duration (in ms) of each step in the simulation."
|
||||
)
|
||||
parser.add_argument(
|
||||
"input_file",
|
||||
nargs="?",
|
||||
help="The file to parse. If not provided, input will be read from stdin."
|
||||
)
|
||||
return parser
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
argument_parser = build_argument_parser()
|
||||
arguments = argument_parser.parse_args()
|
||||
|
||||
input_stream = get_input_stream(arguments.input_file)
|
||||
messages = parse_record_stream(input_stream)
|
||||
|
||||
results = compute_results(messages, arguments.step_duration)
|
||||
print(results)
|
Loading…
Reference in New Issue