dom/media/driftcontrol/plot.py

   1 #! /usr/bin/env python3
   2 #
   3 # This Source Code Form is subject to the terms of the Mozilla Public
   4 # License, v. 2.0. If a copy of the MPL was not distributed with this
   5 # file, You can obtain one at http://mozilla.org/MPL/2.0/.
   6 #
   7 # This scripts plots graphs produced by our drift correction code.
   8 #
   9 # Install dependencies with:
  10 #   > pip install bokeh pandas
  11 #
  12 # Generate the csv data file with the DriftControllerGraphs log module:
  13 #   > MOZ_LOG=raw,sync,DriftControllerGraphs:5 \
  14 #   > MOZ_LOG_FILE=/tmp/driftcontrol.csv       \
  15 #   > ./mach gtest '*AudioDrift*StepResponse'
  16 #
  17 # Generate the graphs with this script:
  18 #   > ./dom/media/driftcontrol/plot.py /tmp/driftcontrol.csv.moz_log
  19 #
  20 # The script should produce a file plot.html in the working directory and
  21 # open it in the default browser.
  22
  23 import argparse
  24 from collections import OrderedDict
  25
  26 import pandas
  27 from bokeh.io import output_file, show
  28 from bokeh.layouts import gridplot
  29 from bokeh.models import TabPanel, Tabs
  30 from bokeh.plotting import figure
  31
  32
  33 def main():
  34     parser = argparse.ArgumentParser(
  35         prog="plot.py for DriftControllerGraphs",
  36         description="""Takes a csv file of DriftControllerGraphs data
  37 (from a single DriftController instance) and plots
  38 them into plot.html in the current working directory.
  39
  40 The easiest way to produce the data is with MOZ_LOG:
  41 MOZ_LOG=raw,sync,DriftControllerGraphs:5 \
  42 MOZ_LOG_FILE=/tmp/driftcontrol.csv       \
  43 ./mach gtest '*AudioDrift*StepResponse'""",
  44     )
  45     parser.add_argument("csv_file", type=str)
  46     args = parser.parse_args()
  47
  48     all_df = pandas.read_csv(args.csv_file)
  49
  50     # Filter on distinct ids to support multiple plotting sources
  51     tabs = []
  52     for id in list(OrderedDict.fromkeys(all_df["id"])):
  53         df = all_df[all_df["id"] == id]
  54
  55         t = df["t"]
  56         buffering = df["buffering"]
  57         avgbuffered = df["avgbuffered"]
  58         desired = df["desired"]
  59         buffersize = df["buffersize"]
  60         inlatency = df["inlatency"]
  61         outlatency = df["outlatency"]
  62         inframesavg = df["inframesavg"]
  63         outframesavg = df["outframesavg"]
  64         inrate = df["inrate"]
  65         outrate = df["outrate"]
  66         steadystaterate = df["steadystaterate"]
  67         nearthreshold = df["nearthreshold"]
  68         corrected = df["corrected"]
  69         hysteresiscorrected = df["hysteresiscorrected"]
  70         configured = df["configured"]
  71
  72         output_file("plot.html")
  73
  74         fig1 = figure()
  75         # Variables with more variation are plotted after smoother variables
  76         # because latter variables are drawn on top and so visibility of
  77         # individual values in the variables with more variation is improved
  78         # (when both variables are shown).
  79         fig1.line(
  80             t, inframesavg, color="violet", legend_label="Average input packet size"
  81         )
  82         fig1.line(
  83             t, outframesavg, color="purple", legend_label="Average output packet size"
  84         )
  85         fig1.line(t, inlatency, color="hotpink", legend_label="In latency")
  86         fig1.line(t, outlatency, color="firebrick", legend_label="Out latency")
  87         fig1.line(t, desired, color="goldenrod", legend_label="Desired buffering")
  88         fig1.line(
  89             t, avgbuffered, color="orangered", legend_label="Average buffered estimate"
  90         )
  91         fig1.line(t, buffering, color="dodgerblue", legend_label="Actual buffering")
  92         fig1.line(t, buffersize, color="seagreen", legend_label="Buffer size")
  93         fig1.varea(
  94             t,
  95             [d - h for (d, h) in zip(desired, nearthreshold)],
  96             [d + h for (d, h) in zip(desired, nearthreshold)],
  97             alpha=0.2,
  98             color="goldenrod",
  99             legend_label='"Near" band (won\'t reduce desired buffering outside)',
 100         )
 101
 102         slowConvergenceSecs = 30
 103         adjustmentInterval = 1
 104         slowHysteresis = 1
 105         avgError = avgbuffered - desired
 106         absAvgError = [abs(e) for e in avgError]
 107         slow_offset = [e / slowConvergenceSecs - slowHysteresis for e in absAvgError]
 108         fast_offset = [e / adjustmentInterval for e in absAvgError]
 109         low_offset, high_offset = zip(
 110             *[
 111                 (s, f) if e >= 0 else (-f, -s)
 112                 for (e, s, f) in zip(avgError, slow_offset, fast_offset)
 113             ]
 114         )
 115
 116         fig2 = figure(x_range=fig1.x_range)
 117         fig2.varea(
 118             t,
 119             steadystaterate + low_offset,
 120             steadystaterate + high_offset,
 121             alpha=0.2,
 122             color="goldenrod",
 123             legend_label="Deadband (won't change in rate within)",
 124         )
 125         fig2.line(t, inrate, color="hotpink", legend_label="Nominal in sample rate")
 126         fig2.line(t, outrate, color="firebrick", legend_label="Nominal out sample rate")
 127         fig2.line(
 128             t,
 129             steadystaterate,
 130             color="orangered",
 131             legend_label="Estimated in rate with drift",
 132         )
 133         fig2.line(
 134             t, corrected, color="dodgerblue", legend_label="Corrected in sample rate"
 135         )
 136         fig2.line(
 137             t,
 138             hysteresiscorrected,
 139             color="seagreen",
 140             legend_label="Hysteresis-corrected in sample rate",
 141         )
 142         fig2.line(
 143             t, configured, color="goldenrod", legend_label="Configured in sample rate"
 144         )
 145
 146         fig1.legend.location = "top_left"
 147         fig2.legend.location = "top_right"
 148         for fig in (fig1, fig2):
 149             fig.legend.background_fill_alpha = 0.6
 150             fig.legend.click_policy = "hide"
 151
 152         tabs.append(TabPanel(child=gridplot([[fig1, fig2]]), title=str(id)))
 153
 154     show(Tabs(tabs=tabs))
 155
 156
 157 if __name__ == "__main__":
 158     main()