bb-log-slow-tasks

#! /usr/bin/env python3

# Given a bitbake log, identify what tasks are the slowest
#
# Licensed under the MIT license


import enum
import sys
import re
import datetime

states = enum.Enum("State", "Started Succeeded Failed")

def log_parser(stream, timestamps=False):
    if timestamps:
        task_re = re.compile(r"(?P<timestamp>[0-9-]+ [0-9:]+).*NOTE: recipe (?P<recipe>.+): task (?P<task>.+): (?P<state>\w+)")
    else:
        task_re = re.compile(r"NOTE: recipe (?P<recipe>.+): task (?P<task>.+): (?P<state>\w+)")

    for line in stream:
        m = task_re.search(line)
        if m:
            # TODO: really need to return a proper object
            time = datetime.datetime.fromisoformat(m.group("timestamp"))
            yield time, states[m.group("state")], m.group("recipe"), m.group("task")

running = {}
tasks = {}
for timestamp, state, recipe, task in log_parser(open(sys.argv[1], encoding="utf-8"), timestamps=True):
    name = recipe + ":" + task
    if state == states.Started:
        running[name] = timestamp
    elif state in (states.Succeeded, states.Failed):
        start_time = running.pop(name)
        duration = timestamp - start_time
        tasks[name] = duration
    else:
        print(f"ERROR: Unhandled state {state}.")
        break

for task, duration in sorted(tasks.items(), key=lambda a: a[1])[-20:]:
    print(f"{task}: {duration}")