path: root/bin/analyze-archive.py

#!/usr/bin/env python3
"""
analyze-archive - generate PTA energy model from dfatool benchmark traces

analyze-archive generates a PTA energy model from one or more annotated
traces generated by dfatool. By default, it does nothing else.

Cross-Validation help:
    If <method> is "montecarlo": Randomly divide data into 2/3 training and 1/3
    validation, <count> times. Reported model quality is the average of all
    validation runs. Data is partitioned without regard for parameter values,
    so a specific parameter combination may be present in both training and
    validation sets or just one of them.

    If <method> is "kfold": Perform k-fold cross validation with k=<count>.
    Divide data into 1-1/k training and 1/k validation, <count> times.
    In the first set, items 0, k, 2k, ... ard used for validation, in the
    second set, items 1, k+1, 2k+1, ... and so on.
    validation, <count> times. Reported model quality is the average of all
    validation runs. Data is partitioned without regard for parameter values,
    so a specific parameter combination may be present in both training and
    validation sets or just one of them.

Trace Export:
    Each JSON file lists all occurences of the corresponding state/transition in the
    benchmark's PTA trace. Each occurence contains the corresponding PTA
    parameters (if any) in 'parameter' and measurement results in 'offline'.
    As measurements are typically run repeatedly, 'offline' is in turn a list
    of measurements: offline[0]['uW'] is the power trace of the first
    measurement of this state/transition, offline[1]['uW'] corresponds t the
    second measurement, etc. Values are provided in microwatts.
    For example, TX.json[0].offline[0].uW corresponds to the first measurement
    of the first TX state in the benchmark, and TX.json[5].offline[2].uW
    corresponds to the third measurement of the sixth TX state in the benchmark.
    WARNING: Several GB of RAM and disk space are required for complex measurements.
             (JSON files may grow very large -- we trade efficiency for easy handling)
"""

import argparse
import json
import logging
import random
import re
import sys
from dfatool import plotter
from dfatool.loader import RawData, pta_trace_to_aggregate
from dfatool.functions import (
    gplearn_to_function,
    SplitFunction,
    AnalyticFunction,
    SubstateFunction,
    StaticFunction,
)
from dfatool.model import PTAModel
from dfatool.validation import CrossValidator
from dfatool.utils import (
    filter_aggregate_by_param,
    detect_outliers_in_aggregate,
    NpEncoder,
    is_numeric,
)
from dfatool.automata import PTA


def print_model_quality(results):
    for state_or_tran in results.keys():
        print()
        for key, result in results[state_or_tran].items():
            if "smape" in result:
                print(
                    "{:20s} {:15s} {:.2f}% / {:.0f}".format(
                        state_or_tran, key, result["smape"], result["mae"]
                    )
                )
            else:
                print("{:20s} {:15s} {:.0f}".format(state_or_tran, key, result["mae"]))


def format_quality_measures(result):
    if "smape" in result:
        return "{:6.2f}% / {:9.0f}".format(result["smape"], result["mae"])
    else:
        return "{:6}    {:9.0f}".format("", result["mae"])


def model_quality_table(header, result_lists, info_list):
    print(
        "{:20s} {:15s}       {:19s}       {:19s}       {:19s}".format(
            "key",
            "attribute",
            header[0].center(19),
            header[1].center(19),
            header[2].center(19),
        )
    )
    for state_or_tran in result_lists[0].keys():
        for key in result_lists[0][state_or_tran].keys():
            buf = "{:20s} {:15s}".format(state_or_tran, key)
            for i, results in enumerate(result_lists):
                info = info_list[i]
                buf += "  |||  "
                if (
                    info is None
                    or (
                        key != "energy_Pt"
                        and type(info(state_or_tran, key)) is not StaticFunction
                    )
                    or (
                        key == "energy_Pt"
                        and (
                            type(info(state_or_tran, "power")) is not StaticFunction
                            or type(info(state_or_tran, "duration"))
                            is not StaticFunction
                        )
                    )
                ):
                    result = results[state_or_tran][key]
                    buf += format_quality_measures(result)
                else:
                    buf += "{:7}----{:8}".format("", "")
            print(buf)


def model_summary_table(result_list):
    buf = "transition duration"
    for results in result_list:
        if len(buf):
            buf += "  |||  "
        buf += format_quality_measures(results["duration_by_trace"])
    print(buf)
    buf = "total energy       "
    for results in result_list:
        if len(buf):
            buf += "  |||  "
        buf += format_quality_measures(results["energy_by_trace"])
    print(buf)
    buf = "rel total energy   "
    for results in result_list:
        if len(buf):
            buf += "  |||  "
        buf += format_quality_measures(results["rel_energy_by_trace"])
    print(buf)
    buf = "state-only energy  "
    for results in result_list:
        if len(buf):
            buf += "  |||  "
        buf += format_quality_measures(results["state_energy_by_trace"])
    print(buf)
    buf = "transition timeout "
    for results in result_list:
        if len(buf):
            buf += "  |||  "
        buf += format_quality_measures(results["timeout_by_trace"])
    print(buf)


def print_text_model_data(model, pm, pq, lm, lq, am, ai, aq):
    print("")
    print(r"key attribute $1 - \frac{\sigma_X}{...}$")
    for state_or_tran in model.by_name.keys():
        for attribute in model.attributes(state_or_tran):
            print(
                "{} {} {:.8f}".format(
                    state_or_tran,
                    attribute,
                    model.attr_by_name[state_or_tran][
                        attr_by_name
                    ].stats.generic_param_dependence_ratio(),
                )
            )

    print("")
    print(r"key attribute parameter $1 - \frac{...}{...}$")
    for state_or_tran in model.by_name.keys():
        for attribute in model.attributes(state_or_tran):
            for param in model.parameters:
                print(
                    "{} {} {} {:.8f}".format(
                        state_or_tran,
                        attribute,
                        param,
                        model.attr_by_name[state_or_tran][
                            attribute
                        ].stats.param_dependence_ratio(param),
                    )
                )
            if state_or_tran in model._num_args:
                for arg_index in range(model._num_args[state_or_tran]):
                    print(
                        "{} {} {:d} {:.8f}".format(
                            state_or_tran,
                            attribute,
                            arg_index,
                            model.attr_by_name[state_or_tran][
                                attribute
                            ].stats.arg_dependence_ratio(arg_index),
                        )
                    )


def print_html_model_data(raw_data, model, pm, pq, lm, lq, am, ai, aq):
    state_attributes = model.attributes(model.states[0])
    trans_attributes = model.attributes(model.transitions[0])

    print("# Setup")
    print("* Input files: `", " ".join(raw_data.filenames), "`")
    print(
        f"""* Number of usable / performed measurements: {raw_data.preprocessing_stats["num_valid"]}/{raw_data.preprocessing_stats["num_runs"]}"""
    )
    print(f"""* State duration: {raw_data.setup_by_fileno[0]["state_duration"]} ms""")
    print()
    print("# States")

    for state in model.states:
        print()
        print(f"## {state}")
        print()
        for param in model.parameters:
            print(
                "* {} ∈ {}".format(
                    param,
                    model.attr_by_name[state][
                        "power"
                    ].stats.distinct_values_by_param_name[param],
                )
            )
        for attribute in state_attributes:
            unit = ""
            if attribute == "power":
                unit = "µW"
            static_quality = pq[state][attribute]["smape"]
            print(
                f"* {attribute} mean: {pm(state, attribute):.0f} {unit} (± {static_quality:.1f}%)"
            )
            if ai(state, attribute):
                analytic_quality = aq[state][attribute]["smape"]
                fstr = ai(state, attribute)["function"].model_function
                fstr = fstr.replace("0 + ", "", 1)
                for i, marg in enumerate(ai(state, attribute)["function"].model_args):
                    fstr = fstr.replace(f"regression_arg({i})", str(marg))
                fstr = fstr.replace("+ -", "-")
                print(f"* {attribute} function: {fstr} (± {analytic_quality:.1f}%)")

    print()
    print("# Transitions")

    for trans in model.transitions:
        print()
        print(f"## {trans}")
        print()
        for param in model.parameters:
            print(
                "* {} ∈ {}".format(
                    param,
                    model.attr_by_name[trans][
                        "duration"
                    ].stats.distinct_values_by_param_name[param],
                )
            )
        for attribute in trans_attributes:
            unit = ""
            if attribute == "duration":
                unit = "µs"
            elif attribute in ["energy", "rel_energy_prev"]:
                unit = "pJ"
            static_quality = pq[trans][attribute]["smape"]
            print(
                f"* {attribute} mean: {pm(trans, attribute):.0f} {unit} (± {static_quality:.1f}%)"
            )
            if ai(trans, attribute):
                analytic_quality = aq[trans][attribute]["smape"]
                fstr = ai(trans, attribute)["function"].model_function
                fstr = fstr.replace("0 + ", "", 1)
                for i, marg in enumerate(ai(trans, attribute)["function"].model_args):
                    fstr = fstr.replace(f"regression_arg({i})", str(marg))
                fstr = fstr.replace("+ -", "-")
                print(f"* {attribute} function: {fstr} (± {analytic_quality:.1f}%)")

    print(
        "<table><tr><th>state</th><th>"
        + "</th><th>".join(state_attributes)
        + "</th></tr>"
    )
    for state in model.states:
        print("<tr>", end="")
        print("<td>{}</td>".format(state), end="")
        for attribute in state_attributes:
            unit = ""
            if attribute == "power":
                unit = "µW"
            print(
                "<td>{:.0f} {} ({:.1f}%)</td>".format(
                    pm(state, attribute), unit, pq[state][attribute]["smape"]
                ),
                end="",
            )
        print("</tr>")
    print("</table>")

    trans_attributes = model.attributes(model.transitions[0])
    if "rel_energy_prev" in trans_attributes:
        trans_attributes.remove("rel_energy_next")

    print(
        "<table><tr><th>transition</th><th>"
        + "</th><th>".join(trans_attributes)
        + "</th></tr>"
    )
    for trans in model.transitions:
        print("<tr>", end="")
        print("<td>{}</td>".format(trans), end="")
        for attribute in trans_attributes:
            unit = ""
            if attribute == "duration":
                unit = "µs"
            elif attribute in ["energy", "rel_energy_prev"]:
                unit = "pJ"
            print(
                "<td>{:.0f} {} ({:.1f}%)</td>".format(
                    pm(trans, attribute), unit, pq[trans][attribute]["smape"]
                ),
                end="",
            )
        print("</tr>")
    print("</table>")


def get_kconfig(model):
    buf = str()
    for param_name in model.parameters:
        unique_values = set()
        is_relevant = False
        for name in model.names:
            unique_values.update(
                model.attr_by_name[name]["power"].stats.distinct_values_by_param_name[
                    param_name
                ]
            )
            for attr in model.attr_by_name[name].values():
                # FIXME this indicates whether it might depend on the parameter, not whether it actually uses it (there's no API for that yet)
                if attr.stats.depends_on_param(param_name):
                    is_relevant = True
        unique_values.discard(None)
        if not unique_values or not is_relevant:
            # unused by the model
            continue

        buf += f"config {param_name}\n"
        buf += f'  prompt "{param_name}"\n'
        if unique_values == {0, 1}:
            buf += "  bool\n"
        elif all(map(is_numeric, unique_values)):
            buf += "  int\n"
            buf += f"  range {min(unique_values)} {max(unique_values)}\n"
        else:
            buf += "  string\n"
            buf += f"  #!accept [{unique_values}]\n"

    return buf


def plot_traces(preprocessed_data, sot_name):
    traces = list()
    timestamps = list()
    for trace in preprocessed_data:
        for state_or_transition in trace["trace"]:
            if state_or_transition["name"] == sot_name:
                timestamps.extend(
                    map(lambda x: x["plot"][0], state_or_transition["offline"])
                )
                traces.extend(
                    map(lambda x: x["plot"][1], state_or_transition["offline"])
                )
    if len(traces) == 0:
        print(
            f"""Did not find traces for state or transition {sot_name}. Abort.""",
            file=sys.stderr,
        )
        sys.exit(2)

    if len(traces) > 40:
        print(f"""Truncating plot to 40 of {len(traces)} traces (random sample)""")
        indexes = random.sample(range(len(traces)), 40)
        timestamps = [timestamps[i] for i in indexes]
        traces = [traces[i] for i in indexes]

    plotter.plot_xy(
        timestamps, traces, xlabel="t [s]", ylabel="P [W]", title=sot_name, family=True
    )


def print_static(model, static_model, name, attribute):
    unit = "  "
    if attribute == "power":
        unit = "µW"
    elif attribute == "duration":
        unit = "µs"
    elif attribute == "substate_count":
        unit = "su"
    print(
        "{:10s}: {:.0f} {:s}  ({:.2f})".format(
            name,
            static_model(name, attribute),
            unit,
            model.attr_by_name[name][attribute].stats.generic_param_dependence_ratio(),
        )
    )
    for param in model.parameters:
        print(
            "{:10s}  dependence on {:15s}: {:.2f}".format(
                "",
                param,
                model.attr_by_name[name][attribute].stats.param_dependence_ratio(param),
            )
        )


def print_analyticinfo(prefix, info):
    empty = ""
    print(f"{prefix}: {info.model_function}")
    print(f"{empty:{len(prefix)}s}  {info.model_args}")


def print_splitinfo(param_names, info, prefix=""):
    if type(info) is SplitFunction:
        for k, v in info.child.items():
            if info.param_index < len(param_names):
                param_name = param_names[info.param_index]
            else:
                param_name = f"arg{info.param_index - len(param_names)}"
            print_splitinfo(param_names, v, f"{prefix} {param_name}={k}")
    elif type(info) is AnalyticFunction:
        print_analyticinfo(prefix, info)
    elif type(info) is StaticFunction:
        print(f"{prefix}: {info.value}")
    else:
        print(f"{prefix}: UNKNOWN")


def _mogrify(function_str, parameter_names, regression_args):
    for i in range(len(regression_args)):
        function_str = function_str.replace(
            f"regression_arg({i})", str(regression_args[i])
        )
    for parameter_name in parameter_names:
        function_str = function_str.replace(
            f"parameter({parameter_name})", f"""param["{parameter_name}"]"""
        )
    for arg_num in range(10):
        function_str = function_str.replace(
            f"function_arg({arg_num})", f"args[{arg_num}]"
        )
    function_str = function_str.replace("np.", "Math.")
    return "#![modelfunction]" + function_str


def mogrify(model):
    if type(model) is dict:
        if "functionStr" in model:
            model["function"] = _mogrify(
                model["functionStr"], model["parameterNames"], model["regressionModel"]
            )
            model.pop("functionStr")
            model.pop("parameterNames")
            model.pop("regressionModel")
        else:
            for k, v in model.items():
                model[k] = mogrify(v)
    elif type(model) is list:
        for i, elem in enumerate(model):
            model[i] = mogrify(model[i])
    return model


if __name__ == "__main__":

    ignored_trace_indexes = []
    safe_functions_enabled = False
    function_override = {}
    show_models = []
    show_quality = []
    pta = None
    energymodel_export_file = None
    trace_export_dir = None
    xv_method = None
    xv_count = 10

    parser = argparse.ArgumentParser(
        formatter_class=argparse.RawDescriptionHelpFormatter, description=__doc__
    )
    parser.add_argument(
        "--info",
        action="store_true",
        help="Show state duration and (for each state and transition) number of measurements and parameter values)",
    )
    parser.add_argument(
        "--no-cache", action="store_true", help="Do not load cached measurement results"
    )
    parser.add_argument(
        "--plot-unparam",
        metavar="<name>:<attribute>:<Y axis label>[;<name>:<attribute>:<label>;...]",
        type=str,
        help="Plot all mesurements for <name> <attribute> without regard for parameter values. "
        "X axis is measurement number/id.",
    )
    parser.add_argument(
        "--plot-param",
        metavar="<name> <attribute> <parameter> [gplearn function][;<name> <attribute> <parameter> [function];...])",
        type=str,
        help="Plot measurements for <name> <attribute> by <parameter>. "
        "X axis is parameter value. "
        "Plots the model function as one solid line for each combination of non-<parameter> parameters. "
        "Also plots the corresponding measurements. "
        "If gplearn function is set, it is plotted using dashed lines.",
    )
    parser.add_argument(
        "--plot-traces",
        metavar="NAME",
        type=str,
        help="Plot power trace for state or transition NAME. X axis is wrong for non-MIMOSA measurements",
    )
    parser.add_argument(
        "--remove-outliers",
        action="store_true",
        help="Remove outliers exceeding the configured z score (default: 10)",
    )
    parser.add_argument(
        "--z-score",
        type=int,
        default=10,
        help="Configure z score for outlier detection (and optional removel)",
    )
    parser.add_argument(
        "--show-models",
        choices=["static", "paramdetection", "param", "all", "tex", "html"],
        action="append",
        default=list(),
        help="static: show static model values as well as parameter detection heuristic.\n"
        "paramdetection: show stddev of static/lut/fitted model\n"
        "param: show parameterized model functions and regression variable values\n"
        "all: all of the above\n"
        "tex: print tex/pgfplots-compatible model data on stdout\n"
        "html: print model and quality data as HTML table on stdout",
    )
    parser.add_argument(
        "--show-quality",
        choices=["table", "summary", "all", "tex", "html"],
        action="append",
        default=list(),
        help="table: show static/fitted/lut SMAPE and MAE for each name and attribute.\n"
        "summary: show static/fitted/lut SMAPE and MAE for each attribute, averaged over all states/transitions.\n"
        "all: all of the above.\n"
        "tex: print tex/pgfplots-compatible model quality data on stdout.",
    )
    parser.add_argument(
        "--ignored-trace-indexes",
        metavar="<i1,i2,...>",
        type=str,
        help="Specify traces which should be ignored due to bogus data. "
        "1 is the first trace, 2 the second, and so on.",
    )
    parser.add_argument(
        "--function-override",
        metavar="<name> <attribute> <function>[;<name> <attribute> <function>;...]",
        type=str,
        help="Manually specify the function to fit for <name> <attribute>. "
        "A function specified this way bypasses parameter detection: "
        "It is always assigned, even if the model seems to be independent of the parameters it references.",
    )
    parser.add_argument(
        "--export-traces",
        metavar="DIRECTORY",
        type=str,
        help="Export power traces of all states and transitions to DIRECTORY. "
        "Creates a JSON file for each state and transition.",
    )
    parser.add_argument(
        "--export-dref",
        metavar="FILE",
        type=str,
        help="Export model and model quality to LaTeX dataref file",
    )
    parser.add_argument(
        "--filter-param",
        metavar="<parameter name>=<parameter value>[,<parameter name>=<parameter value>...]",
        type=str,
        help="Only consider measurements where <parameter name> is <parameter value>. "
        "All other measurements (including those where it is None, that is, has not been set yet) are discarded. "
        "Note that this may remove entire function calls from the model.",
    )
    parser.add_argument(
        "--log-level",
        metavar="LEVEL",
        choices=["debug", "info", "warning", "error"],
        default="warning",
        help="Set log level",
    )
    parser.add_argument(
        "--cross-validate",
        metavar="<method>:<count>",
        type=str,
        help="Perform cross validation when computing model quality. "
        "Only works with --show-quality=table at the moment.",
    )
    parser.add_argument(
        "--with-safe-functions",
        action="store_true",
        help="Include 'safe' functions (safe_log, safe_inv, safe_sqrt) which are also defined for 0 and -1. "
        "This allows a greater range of functions to be tried during fitting.",
    )
    parser.add_argument(
        "--hwmodel",
        metavar="FILE",
        type=str,
        help="Load DFA hardware model from JSON or YAML FILE",
    )
    parser.add_argument(
        "--export-dot",
        metavar="FILE",
        type=str,
        help="Export PTA representation suitable for Graphviz dot to FILE",
    )
    parser.add_argument(
        "--export-energymodel",
        metavar="FILE",
        type=str,
        help="Export JSON energy model to FILE. Works out of the box for v1+, requires --hwmodel for v0",
    )
    parser.add_argument(
        "--export-webconf",
        metavar="FILE",
        type=str,
        help="Export KConfig model to FILE.Kconfig and energy model to FILE.js. Works out of the box for v1+, requires --hwmodel for v0",
    )
    parser.add_argument(
        "--with-substates",
        metavar="PELT_CONFIG",
        type=str,
        help="Perform substate analysis",
    )
    parser.add_argument("measurement", nargs="+")

    args = parser.parse_args()

    if args.log_level:
        numeric_level = getattr(logging, args.log_level.upper(), None)
        if not isinstance(numeric_level, int):
            print(f"Invalid log level: {args.log_level}", file=sys.stderr)
            sys.exit(1)
        logging.basicConfig(level=numeric_level)

    if args.ignored_trace_indexes:
        ignored_trace_indexes = list(map(int, args.ignored_trace_indexes.split(",")))
        if 0 in ignored_trace_indexes:
            logging.error("arguments to --ignored-trace-indexes start from 1")

    if args.function_override:
        for function_desc in args.function_override.split(";"):
            state_or_tran, attribute, *function_str = function_desc.split(" ")
            function_override[(state_or_tran, attribute)] = " ".join(function_str)

    show_models = args.show_models
    show_quality = args.show_quality

    if args.cross_validate:
        xv_method, xv_count = args.cross_validate.split(":")
        xv_count = int(xv_count)

    if args.filter_param:
        args.filter_param = list(
            map(lambda x: x.split("="), args.filter_param.split(","))
        )
    else:
        args.filter_param = list()

    if args.with_safe_functions is not None:
        safe_functions_enabled = True

    if args.hwmodel:
        pta = PTA.from_file(args.hwmodel)

    raw_data = RawData(
        args.measurement,
        with_traces=(
            args.export_traces is not None
            or args.plot_traces is not None
            or args.with_substates is not None
        ),
        skip_cache=args.no_cache,
    )

    if args.info:
        print(" ".join(raw_data.filenames) + ":")
        data_source = "???"
        if raw_data.ptalog:
            options = " --".join(
                map(lambda kv: f"{kv[0]}={str(kv[1])}", raw_data.ptalog["opt"].items())
            )
            print(f"    Options: --{options}")
        if raw_data.version <= 1:
            data_source = "MIMOSA"
        elif raw_data.version == 2:
            if raw_data.ptalog and "sync" in raw_data.ptalog["opt"]["energytrace"]:
                data_source = "MSP430 EnergyTrace, sync={}".format(
                    raw_data.ptalog["opt"]["energytrace"]["sync"]
                )
            else:
                data_source = "MSP430 EnergyTrace"
        elif raw_data.version == 3:
            data_source = "Keysight"
        print(f"    Data source ID: {raw_data.version} ({data_source})")

    preprocessed_data = raw_data.get_preprocessed_data()

    if args.info:
        print(
            f"""    Valid Runs: {raw_data.preprocessing_stats["num_valid"]}/{raw_data.preprocessing_stats["num_runs"]}"""
        )
        state_durations = map(
            lambda x: str(x["state_duration"]), raw_data.setup_by_fileno
        )
        print(f"""    State Duration: {" / ".join(state_durations)} ms""")

    if args.export_traces:
        uw_per_sot = dict()
        for trace in preprocessed_data:
            for state_or_transition in trace["trace"]:
                name = state_or_transition["name"]
                if name not in uw_per_sot:
                    uw_per_sot[name] = list()
                for elem in state_or_transition["offline"]:
                    elem["plot"] = list(elem["plot"])
                uw_per_sot[name].append(state_or_transition)
        for name, data in uw_per_sot.items():
            target = f"{args.export_traces}/{name}.json"
            print(f"exporting {target} ...")
            with open(target, "w") as f:
                json.dump(data, f)

    if args.with_substates is not None:
        arg_dict = dict()
        if args.with_substates != "":
            for kv in args.with_substates.split(","):
                k, v = kv.split("=")
                try:
                    arg_dict[k] = float(v)
                except ValueError:
                    arg_dict[k] = v
        args.with_substates = arg_dict

    if args.plot_traces:
        plot_traces(preprocessed_data, args.plot_traces)

    if raw_data.preprocessing_stats["num_valid"] == 0:
        print("No valid data available. Abort.", file=sys.stderr)
        sys.exit(2)

    if pta is None and raw_data.pta is not None:
        pta = PTA.from_json(raw_data.pta)

    by_name, parameters, arg_count = pta_trace_to_aggregate(
        preprocessed_data, ignored_trace_indexes
    )

    filter_aggregate_by_param(by_name, parameters, args.filter_param)
    detect_outliers_in_aggregate(
        by_name, z_limit=args.z_score, remove_outliers=args.remove_outliers
    )

    model = PTAModel(
        by_name,
        parameters,
        arg_count,
        traces=preprocessed_data,
        function_override=function_override,
        pta=pta,
        pelt=args.with_substates,
    )

    if xv_method:
        xv = CrossValidator(PTAModel, by_name, parameters, arg_count)

    if args.info:
        for state in model.states:
            print("{}:".format(state))
            print(f"""    Number of Measurements: {len(by_name[state]["power"])}""")
            for param in model.parameters:
                print(
                    "    Parameter {} ∈ {}".format(
                        param,
                        model.attr_by_name[state][
                            "power"
                        ].stats.distinct_values_by_param_name[param],
                    )
                )
        for transition in model.transitions:
            print("{}:".format(transition))
            print(
                f"""    Number of Measurements: {len(by_name[transition]["duration"])}"""
            )
            for param in model.parameters:
                print(
                    "    Parameter {} ∈ {}".format(
                        param,
                        model.attr_by_name[transition][
                            "duration"
                        ].stats.distinct_values_by_param_name[param],
                    )
                )
            for i in range(model._num_args[transition]):
                print(
                    "    Argument  {} ∈ {}".format(
                        i,
                        model.attr_by_name[transition][
                            "duration"
                        ].stats.distinct_values_by_param_index[
                            len(model.parameters) + i
                        ],
                    )
                )

    if args.plot_unparam:
        for kv in args.plot_unparam.split(";"):
            state_or_trans, attribute, ylabel = kv.split(":")
            fname = "param_y_{}_{}.pdf".format(state_or_trans, attribute)
            plotter.plot_y(
                model.by_name[state_or_trans][attribute],
                xlabel="measurement #",
                ylabel=ylabel,
                output=fname,
            )

    if len(show_models):
        print("--- simple static model ---")
    static_model = model.get_static()
    if "static" in show_models or "all" in show_models:
        for state in model.states:
            for attribute in model.attributes(state):
                print_static(model, static_model, state, attribute)
        if args.with_substates:
            for submodel in model.submodel_by_name.values():
                for substate in submodel.states:
                    for subattribute in submodel.attributes(substate):
                        print_static(
                            submodel, submodel.get_static(), substate, subattribute
                        )

        for trans in model.transitions:
            if "energy" in model.attributes(trans):
                try:
                    print(
                        "{:10s}: {:.0f} / {:.0f} / {:.0f} pJ  ({:.2f} / {:.2f} / {:.2f})".format(
                            trans,
                            static_model(trans, "energy"),
                            static_model(trans, "rel_energy_prev"),
                            static_model(trans, "rel_energy_next"),
                            model.attr_by_name[trans][
                                "energy"
                            ].stats.generic_param_dependence_ratio(),
                            model.attr_by_name[trans][
                                "rel_energy_prev"
                            ].stats.generic_param_dependence_ratio(),
                            model.attr_by_name[trans][
                                "rel_energy_next"
                            ].stats.generic_param_dependence_ratio(),
                        )
                    )
                except KeyError:
                    print(
                        "{:10s}: {:.0f} pJ  ({:.2f})".format(
                            trans,
                            static_model(trans, "energy"),
                            model.attr_by_name[trans][
                                "energy"
                            ].stats.generic_param_dependence_ratio(),
                        )
                    )
            else:
                try:
                    print(
                        "{:10s}: {:.0f} / {:.0f} / {:.0f} pJ  (E=P·t)".format(
                            trans,
                            static_model(trans, "power")
                            * static_model(trans, "duration"),
                            static_model(trans, "rel_power_prev")
                            * static_model(trans, "duration"),
                            static_model(trans, "rel_power_next")
                            * static_model(trans, "duration"),
                        )
                    )
                except KeyError:
                    print(
                        "{:10s}: {:.0f} pJ  (E=P·t)".format(
                            trans,
                            static_model(trans, "power")
                            * static_model(trans, "duration"),
                        )
                    )
            print(
                "{:10s}: {:.0f} µs  ({:.2f})".format(
                    trans,
                    static_model(trans, "duration"),
                    model.attr_by_name[trans][
                        "duration"
                    ].stats.generic_param_dependence_ratio(),
                )
            )
            try:
                print(
                    "{:10s}: {:.0f} / {:.0f} / {:.0f} µW  ({:.2f} / {:.2f} / {:.2f})".format(
                        trans,
                        static_model(trans, "power"),
                        static_model(trans, "rel_power_prev"),
                        static_model(trans, "rel_power_next"),
                        model.attr_by_name[trans][
                            "power"
                        ].stats.generic_param_dependence_ratio(),
                        model.attr_by_name[trans][
                            "rel_power_prev"
                        ].stats.generic_param_dependence_ratio(),
                        model.attr_by_name[trans][
                            "rel_power_next"
                        ].stats.generic_param_dependence_ratio(),
                    )
                )
            except KeyError:
                print(
                    "{:10s}: {:.0f} pJ  ({:.2f})".format(
                        trans,
                        static_model(trans, "power"),
                        model.attr_by_name[trans][
                            "power"
                        ].stats.generic_param_dependence_ratio(),
                    )
                )

    if xv_method == "montecarlo":
        static_quality = xv.montecarlo(lambda m: m.get_static(), xv_count)
    elif xv_method == "kfold":
        static_quality = xv.kfold(lambda m: m.get_static(), xv_count)
    else:
        static_quality = model.assess(static_model)

    if len(show_models):
        print("--- LUT ---")
    lut_model = model.get_param_lut()

    if xv_method == "montecarlo":
        lut_quality = xv.montecarlo(lambda m: m.get_param_lut(fallback=True), xv_count)
    elif xv_method == "kfold":
        lut_quality = xv.kfold(lambda m: m.get_param_lut(fallback=True), xv_count)
    else:
        lut_quality = model.assess(lut_model)

    if len(show_models):
        print("--- param model ---")

    # get_fitted_sub -> with sub-state detection and modeling
    param_model, param_info = model.get_fitted(
        safe_functions_enabled=safe_functions_enabled
    )

    if "paramdetection" in show_models or "all" in show_models:
        for name in model.names:
            for attribute in model.attributes(name):
                info = param_info(name, attribute)
                print(
                    "{:10s} {:10s} non-param stddev {:f}".format(
                        name,
                        attribute,
                        model.attr_by_name[name][attribute].stats.std_static,
                    )
                )
                print(
                    "{:10s} {:10s} param-lut stddev {:f}".format(
                        name,
                        attribute,
                        model.attr_by_name[name][attribute].stats.std_param_lut,
                    )
                )
                for param in sorted(
                    model.attr_by_name[name][attribute].stats.std_by_param.keys()
                ):
                    print(
                        "{:10s} {:10s} {:10s} stddev {:f}".format(
                            name,
                            attribute,
                            param,
                            model.attr_by_name[name][attribute].stats.std_by_param[
                                param
                            ],
                        )
                    )
                for arg_index in range(model.attr_by_name[name][attribute].arg_count):
                    print(
                        "{:10s} {:10s} {:10s} stddev {:f}".format(
                            name,
                            attribute,
                            f"arg{arg_index}",
                            model.attr_by_name[name][attribute].stats.std_by_arg[
                                arg_index
                            ],
                        )
                    )
                if type(info) is AnalyticFunction:
                    for param_name in sorted(info.fit_by_param.keys(), key=str):
                        param_fit = info.fit_by_param[param_name]["results"]
                        for function_type in sorted(param_fit.keys()):
                            function_rmsd = param_fit[function_type]["rmsd"]
                            print(
                                "{:10s} {:10s} {:10s} mean {:10s} RMSD {:.0f}".format(
                                    name,
                                    attribute,
                                    str(param_name),
                                    function_type,
                                    function_rmsd,
                                )
                            )

    if "param" in show_models or "all" in show_models:
        for state in model.states:
            for attribute in model.attributes(state):
                info = param_info(state, attribute)
                if type(info) is AnalyticFunction:
                    print_analyticinfo(f"{state:10s} {attribute:15s}", info)
                elif type(info) is SplitFunction:
                    print_splitinfo(
                        model.parameters, info, f"{state:10s} {attribute:15s}"
                    )
                elif type(info) is SubstateFunction:
                    print(f"{state:10s} {attribute:15s}: Substate (TODO)")
        for trans in model.transitions:
            for attribute in model.attributes(trans):
                info = param_info(trans, attribute)
                if type(info) is AnalyticFunction:
                    print_analyticinfo(f"{trans:10s} {attribute:15s}", info)
                elif type(info) is SplitFunction:
                    print_splitinfo(
                        model.parameters, info, f"{trans:10s} {attribute:15s}"
                    )
                elif type(info) is SubstateFunction:
                    print(f"{state:10s} {attribute:15s}: Substate (TODO)")
        if args.with_substates:
            for submodel in model.submodel_by_name.values():
                sub_param_model, sub_param_info = submodel.get_fitted()
                for substate in submodel.states:
                    for subattribute in submodel.attributes(substate):
                        info = sub_param_info(substate, subattribute)
                        if type(info) is AnalyticFunction:
                            print(
                                "{:10s} {:15s}: {}".format(
                                    substate, subattribute, info.model_function
                                )
                            )
                            print("{:10s} {:15s}  {}".format("", "", info.model_args))

    if args.with_substates:
        for state in model.states:
            if (
                type(model.attr_by_name[state]["power"].model_function)
                is SubstateFunction
            ):
                # sub-state models need to know the duration of the state / transition. only needed for eval.
                model.attr_by_name[state]["power"].model_function.static_duration = (
                    raw_data.setup_by_fileno[0]["state_duration"] * 1e3
                )

    if xv_method == "montecarlo":
        analytic_quality = xv.montecarlo(lambda m: m.get_fitted()[0], xv_count)
    elif xv_method == "kfold":
        analytic_quality = xv.kfold(lambda m: m.get_fitted()[0], xv_count)
    else:
        analytic_quality = model.assess(param_model)

    if "tex" in show_models or "tex" in show_quality:
        print_text_model_data(
            model,
            static_model,
            static_quality,
            lut_model,
            lut_quality,
            param_model,
            param_info,
            analytic_quality,
        )

    if "html" in show_models or "html" in show_quality:
        print_html_model_data(
            raw_data,
            model,
            static_model,
            static_quality,
            lut_model,
            lut_quality,
            param_model,
            param_info,
            analytic_quality,
        )

    if "table" in show_quality or "all" in show_quality:
        model_quality_table(
            ["static", "parameterized", "LUT"],
            [static_quality, analytic_quality, lut_quality],
            [None, param_info, None],
        )
        if args.with_substates:
            for submodel in model.submodel_by_name.values():
                sub_static_model = submodel.get_static()
                sub_static_quality = submodel.assess(sub_static_model)
                sub_lut_model = submodel.get_param_lut()
                sub_lut_quality = submodel.assess(sub_lut_model)
                sub_param_model, sub_param_info = submodel.get_fitted()
                sub_analytic_quality = submodel.assess(sub_param_model)
                model_quality_table(
                    ["static", "parameterized", "LUT"],
                    [sub_static_quality, sub_analytic_quality, sub_lut_quality],
                    [None, sub_param_info, None],
                )

    if "overall" in show_quality or "all" in show_quality:
        print("overall state static/param/lut MAE assuming equal state distribution:")
        print(
            "    {:6.1f}  /  {:6.1f}  /  {:6.1f}  µW".format(
                model.assess_states(static_model),
                model.assess_states(param_model),
                model.assess_states(lut_model),
            )
        )
        distrib = dict()
        num_states = len(model.states)
        p95_state = None
        for state in model.states:
            distrib[state] = 1.0 / num_states

        if "STANDBY1" in model.states:
            p95_state = "STANDBY1"
        elif "SLEEP" in model.states:
            p95_state = "SLEEP"

        if p95_state is not None:
            for state in distrib.keys():
                distrib[state] = 0.05 / (num_states - 1)
            distrib[p95_state] = 0.95

            print(f"overall state static/param/lut MAE assuming 95% {p95_state}:")
            print(
                "    {:6.1f}  /  {:6.1f}  /  {:6.1f}  µW".format(
                    model.assess_states(static_model, distribution=distrib),
                    model.assess_states(param_model, distribution=distrib),
                    model.assess_states(lut_model, distribution=distrib),
                )
            )

    if "summary" in show_quality or "all" in show_quality:
        model_summary_table(
            [
                model.assess_on_traces(static_model),
                model.assess_on_traces(param_model),
                model.assess_on_traces(lut_model),
            ]
        )

    if args.plot_param:
        for kv in args.plot_param.split(";"):
            try:
                state_or_trans, attribute, param_name, *function = kv.split(" ")
            except ValueError:
                print(
                    "Usage: --plot-param='state_or_trans attribute param_name [additional function spec]'",
                    file=sys.stderr,
                )
                sys.exit(1)
            if len(function):
                function = gplearn_to_function(" ".join(function))
            else:
                function = None
            plotter.plot_param(
                model,
                state_or_trans,
                attribute,
                model.param_index(param_name),
                extra_function=function,
            )

    if args.export_dref:
        dref = raw_data.to_dref()
        dref.update(model.to_dref(static_quality, lut_quality, analytic_quality))
        with open(args.export_dref, "w") as f:
            for k, v in dref.items():
                if type(v) is not tuple:
                    v = (v, None)
                if v[1] is None:
                    prefix = r"\drefset{"
                else:
                    prefix = r"\drefset" + f"[unit={v[1]}]" + "{"
                print(f"{prefix}{k}" + "}{" + str(v[0]) + "}", file=f)

    if args.export_webconf:
        if not pta:
            print(
                "Note: v0 measurements do not embed the PTA used for benchmark generation. Estimating PTA from recorded observations."
            )
        json_model = model.to_json()
        mogrify(json_model)
        json_model_str = json.dumps(json_model, indent=2, sort_keys=True, cls=NpEncoder)
        json_model_out = ""
        for line in json_model_str.split("\n"):
            match = re.fullmatch(r"(.*)\"#!\[modelfunction\](.*\")(.*)", line)
            if match:
                line = (
                    match.group(1)
                    + "(param, args) => "
                    + json.loads('"' + match.group(2))
                    + match.group(3)
                )
            json_model_out += line + "\n"
        with open(f"{args.export_webconf}.json", "w") as f:
            f.write(json_model_out)
        with open(f"{args.export_webconf}.kconfig", "w") as f:
            f.write(get_kconfig(model))

    if args.export_energymodel:
        if not pta:
            print(
                "Note: v0 measurements do not embed the PTA used for benchmark generation. Estimating PTA from recorded observations."
            )
        json_model = model.to_json()
        with open(args.export_energymodel, "w") as f:
            json.dump(json_model, f, indent=2, sort_keys=True, cls=NpEncoder)

    if args.export_dot:
        if not pta:
            print(
                "Note: v0 measurements do not embed the PTA used for benchmark generation. Estimating PTA from recorded observations."
            )
        json_model = model.to_json()
        with open(args.export_dot, "w") as f:
            f.write(model.to_dot())

    sys.exit(0)