source: doc/theses/mike_brooks_MMath/plots/common.py@ a17f496

import pandas as pd
import numpy as np
import sys
import os

def parseTestCorpus(dt):
    dt[['test-slug',
        'sut-platform',
        'operation',
        'sut-cfa-level',
        'sut-cfa-sharing',
        'op-alloc']] = dt['test'].str.strip().str.split('-', expand=True)
    dt['sut'] = dt[['sut-platform',
                        'sut-cfa-level',
                        'sut-cfa-sharing',
                        'op-alloc']].agg('-'.join, axis=1)

    dt[['corpus-basename',
        'corpus-ext']] = dt['corpus'].str.strip().str.rsplit('.', n=1, expand=True)
    dt[['corpus-slug',
        'corpus-nstrs-tgt',
        'corpus-meanlen-tgt',
        'corpus-runid']] = dt['corpus-basename'].str.strip().str.split('-', expand=True)
    dt["corpus-nstrs-tgt"] = pd.to_numeric(dt["corpus-nstrs-tgt"])
    dt["corpus-meanlen-tgt"] = pd.to_numeric(dt["corpus-meanlen-tgt"])
    dt[['corpus-relscale',
        'corpus-seed',
        'corpus-offset-instr']] = dt['corpus-runid'].str.strip().str.split('+', expand=True)
    dt["corpus-relscale"] = pd.to_numeric(dt["corpus-relscale"])

def loadParseTimingData( infileLocal, xClasNames=[], xClasDtypes={}, xFactNames=[], xFactDtypes={} ):

    infile = os.path.dirname(os.path.abspath(__file__)) + '/../benchmarks/string/' + infileLocal

    timings = pd.read_csv(
        infile,
        names=['test', 'corpus'] + xClasNames + [ 'concatsPerReset', 'corpusItemCountAct', 'corpusMeanLenCharsAct', 'concatDoneActualCount', 'execTimeActualSec'] + xFactNames,
        dtype={**xClasDtypes, **xFactDtypes, **{
            'test':                  str,
            'corpus':                str,
            'concatsPerReset':       'Int64', # allows missing; https://stackoverflow.com/a/70626154
            'corpusItemCountAct':    np.int64,
            'corpusMeanLenCharsAct': np.float64,
            'concatDoneActualCount': np.int64,
            'execTimeActualSec':     np.float64,
            'Q': np.int64}},

        na_values=['xxx'],
    )
    # print(timings.head())


    # project: parse executable and corpus names

    parseTestCorpus(timings)

    # project: calculate fact

    timings['op-duration-s'] = timings['execTimeActualSec'] / timings['concatDoneActualCount']
    timings['op-duration-ns'] = timings['op-duration-s'] * 1000 * 1000 * 1000

    return timings

def loadParseSizingData( infileLocal, xClasNames=[], xClasDtypes={}, xFactNames=[], xFactDtypes={} ):

    infile = os.path.dirname(os.path.abspath(__file__)) + '/../benchmarks/string/' + infileLocal

    sizings = pd.read_csv(
        infile,
        sep=' ',
        names=['test', 'corpus'] + xClasNames + ['ppid', 'pid', 'malloc_count', 'free_count',
            'calloc_count', 'realloc_count',
            'requsted_mem(B)', 'current_req_mem(B)', 'hw_cur_req_mem(B)', 'text', 'heap', 'mmap_so', 'mmap',
            'stack', 'vvar', 'vdso', 'vsyscall', 'unfigured', 'total_dynamic',
            'epoch_timestamp(ms)'] + xFactNames,
        dtype={**xClasDtypes, **xFactDtypes, **{
            'test':                  str,
            'corpus':                str,
            'ppid': np.int64, 'pid': np.int64, 'malloc_count': np.int64, 'free_count': np.int64,
            'calloc_count': np.int64, 'realloc_count': np.int64,
            'requsted_mem(B)': np.int64, 'current_req_mem(B)': np.int64,
            'hw_cur_req_mem(B)': np.int64, 'text': np.int64, 'heap': np.int64,
            'mmap_so': np.int64, 'mmap': np.int64,
            'stack': np.int64, 'vvar': np.int64, 'vdso': np.int64, 'vsyscall': np.int64, 'unfigured': np.int64, 'total_dynamic': np.int64,
            'epoch_timestamp(ms)': np.int64}}
    )

    parseTestCorpus(sizings)

    return sizings


def loadParseAttribData( infileLocal ):

    infile = os.path.dirname(os.path.abspath(__file__)) + '/../benchmarks/string/' + infileLocal

    attribs = pd.read_csv(
        infile,
        sep=' ',
        names=[
            "test", "corpus", "expansion", "category", "samples_in_category", "total_samples",
            "fraction", "sources"],
        dtype={
            "test": str,
            "corpus": str,
            "expansion": np.float64,
            "category": str,
            "samples_in_category": np.int64,
            "total_samples": np.int64,
            "fraction": np.float64,
            "sources": str}
    )

    parseTestCorpus(attribs)

    return attribs