benchmark.py

import sys
import time
import hotshot
import hotshot.stats
import ipdb

import openpyxl

def get_process_rss():
    procstat = open("/proc/self/status").readlines()
    for line in procstat:
        if line.startswith("VmSize:"):
            return int(line.split()[1]) * 1024
    else:
        return 0


def highest_row(sheet):
    return sheet.get_highest_row()


def highest_column(sheet):
    try:
        return max(
            openpyxl.cell.column_index_from_string(cell.column)
            for cell in sheet._cells.itervalues()
        ) - 1
    except ValueError:
        return 1


def main(args):
    start = time.time()
    start_mem = get_process_rss()
    print "start", start, start_mem

    wb = openpyxl.load_workbook(args[1])

    opened = time.time()
    opened_mem = get_process_rss()
    print "open", opened-start, opened_mem-start_mem


    # We are going to go and count the numberof cells with values,
    # the number of cells within our bounding box, and the number of
    # cell.value accesses we do.

    values = 0
    cells = 0
    accesses = 0

    for sheet in wb.worksheets:
        rows = highest_row(sheet)
        columns = highest_column(sheet)

        for r in xrange(rows):
            blanks = 0

            for c in xrange(columns):
                cell = sheet.cell(row=r, column=c)

                if cell.value is not None:
                    values += 1
                else:
                    blanks += 1

                cell.offset(row=1, column=1).value

                accesses += 2
                cells += 1

            if blanks == c+1:
                break

    end = time.time()
    total = end-start
    done_mem = get_process_rss()

    print "read", total, cells, values, accesses, done_mem-start_mem

    cells = float(cells)
    accesses = float(accesses)
    values = float(values)

    print (total/accesses)*1000, "ms per cell access"
    print (done_mem-start_mem) / cells, "bytes per cell"
    print (done_mem-start_mem) / values, "bytes per value"


if 0:
    prof = hotshot.Profile("speed.prof")
    prof.runcall(main, sys.argv)
    prof.close()

    stats = hotshot.stats.load("speed.prof")
    stats.strip_dirs()
    stats.sort_stats('time', 'calls')

    ipdb.set_trace()
else:
    main(sys.argv)