Programmatically Creating A FlameGraph .htm file

[grantlouisherman]

Hey all,

So right now Im just experimenting with memray and really liking it so far. However, I really wanted handle some stuff programmatically instead of using the command line.

I have this snippet of code
`import subprocess

from memray import FileReader, Tracker
from memray.reporters.flamegraph import FlameGraphReporter

subprocess.run(["rm", "-f", "memray_output.bin"], check=True)
output_file = "memray_output.bin"
with Tracker(output_file):
subprocess.run(["python3", "test.py"], check=True)
print("read file")
reader = FileReader(output_file).get_high_watermark_allocation_records()
reporter = FlameGraphReporter.from_snapshot(
reader, memory_records=[], native_traces=False
)
reporter.render(
outfile="flamegraph.html",
metadata=None,
show_memory_leaks=False,
merge_threads=False,
inverted=False,
)
print(f"Profile saved to: {output_file}")
print(f"\nMemray output saved to: {memray_output}")
print(f"Generate flamegraph: memray flamegraph {memray_output}")
print(f"Generate table: memray table {memray_output}")`

Where Im using the Tracker and then trying to create the FlameGraph. I mainly pulled this code from looking at unit tests and playing around. However, where Im stuck is the .render function. It fails because I dont have a proper Metadata object. I could create one by hand, but I wanted to see if there was a utility function off of the Filereader or somewhere where I could get this info.

[godlygeek]

Memray does not yet have a stable, supported API for generating reports. The only documented API that we expose is the tracking API - memray.Tracker, along with the memray.FileDestination and memray.SocketDestination and memray.FileFormat that are used with it. We've always intended to eventually provide a public API for parsing and processing the raw capture files, but we've never gotten around to it. All of which is to say: please don't use memray.reporters.flamegraph.FlameGraphReporter.render to produce a flame graph, please instead use subprocess.run(["memray", "flamegraph", ...]) to generate the flame graph.

Also, this is unrelated to your question, but

with Tracker(output_file):
    subprocess.run(["python3", "test.py"], check=True)

is probably not doing what you want it to do. That's not tracking the memory usage of test.py, it's tracking the memory usage only of subprocess.run() itself. It's tracking memory usage within the current process, not memory usage by the newly spawned subprocess. If you want to track memory usage of a different script, you don't want the Tracker API at all, you want

subprocess.run([sys.executable, "-m", "memray", "run", "-o", output_file, "test.py"])