Execution time splitter for flame graphs

etsplit.py

#!/usr/bin/env python3

"""
Execution time splitter for flame graphs.

The script takes a flame graph SVG [1] and splits the samples by execution
time: a sample can be either compile time, or a run time. Useful for profiling
VMs employing JIT compilation.

Currently, functions with names starting with "llvm::" are compile time,
everything else is run time.

Usage: etsplit <graph.svg>

[1]: https://github.com/brendangregg/FlameGraph
"""

from collections import namedtuple
from xml.dom import minidom
import os
import re
import sys


ExecTimeSplit = namedtuple('ExecTimeSplit', ('compile_time', 'run_time'))


def is_compile_time_name(name):
    return name.startswith('llvm::')


def gs_ranges(gs):
    "X-axis ranges for all `g` elements in the input iterable."
    return ((float(a.get('x').value),
             float(a.get('x').value) + float(a.get('width').value))
            for a in (g.childNodes[2].attributes for g in gs))


def compile_time_ranges(svg):
    "X-axis ranges for all compile-time functions."
    return gs_ranges(g
                     for g in svg.getElementsByTagName('g')
                     if is_compile_time_name(
                             g.childNodes[1].firstChild.nodeValue))


def total_length(svg):
    "Total length of the SVG canvas."
    x1s, x2s = zip(*gs_ranges(svg.getElementsByTagName('g')))
    return max(x2s) - min(x1s)


def ranges_length(ranges):
    "Total length of x-axis covered by ranges."

    ranges = sorted(ranges)

    if not ranges:
        return 0

    x, length = ranges[0][0], 0

    for x1, x2 in ranges:
        length += x2 - max(x, x1)
        x = x2

    return length


def exec_time_split(svg):
    total_time = total_length(svg)
    compile_time = ranges_length(compile_time_ranges(svg))
    run_time = total_time - compile_time
    return ExecTimeSplit(compile_time / total_time, run_time / total_time)


def total_samples(svg):
    "Total amount of samples in the report."

    all_title = [title
                 for title in (g.childNodes[1].firstChild.nodeValue
                               for g in svg.getElementsByTagName('g'))
                 if title.startswith('all (')]
    assert(len(all_title) == 1)
    all_title = all_title[0]

    return float(re.match('^.*\(([\d,.]+) .*$', all_title)
                 .group(1)
                 .replace(',', ''))


if __name__ == '__main__':
    if len(sys.argv) != 2:
        sys.stderr.write('Usage: {} <graph.svg>\n'.format(sys.argv[0]))
        sys.exit(os.EX_USAGE)

    svg = minidom.parse(sys.argv[1])
    compile_time, run_time = exec_time_split(svg)
    samples = total_samples(svg)
    print('Compile time:\t{:6.1f} samples ({:5.2f}%)\n'
          'Run time:\t{:6.1f} samples ({:5.2f}%)'.format(
              compile_time * samples,
              compile_time * 100,
              run_time * samples,
              run_time * 100))