Benchmark for/else loop vs function for date parsing.

datefmt_bench.py

#!/usr/bin/env python3

import pytz
import time
import random
import numpy as np
import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt

from functools import wraps
from datetime import datetime

HUMAN_DATETIME    = "%a %b %d %H:%M:%S %Y %z"
HUMAN_DATE        = "%b %d, %Y"
HUMAN_TIME        = "%I:%M:%S %p"
HUMAN_HOUR_TZ     = "%-I:%M%p %Z"
JSON_DATETIME     = "%Y-%m-%dT%H:%M:%S.%fZ" # Must be UTC
ISO8601_DATETIMEX = "%Y-%m-%dT%H:%M:%S"
ISO8601_DATETIME  = "%Y-%m-%dT%H:%M:%S%z"
RFC3339_DATETIME  = ISO8601_DATETIME
ISO8601_DATE      = "%Y-%m-%d"
ISO8601_TIME      = "%H:%M:%S"
COMMON_DATETIME   = "%d/%b/%Y:%H:%M:%S %z"
WEB_UTC_DATETIME  = "%a, %b %d, %Y at %H:%M UTC"
PG_DATE_FMT       = "%Y-%m-%d"
PG_TIMESTAMP_FMT  = "%Y-%m-%d %H:%M:%S" # Must be UTC

DATETIME_FORMATS = (
    HUMAN_DATETIME,
    HUMAN_DATE,
    HUMAN_TIME,
    JSON_DATETIME,
    ISO8601_DATETIMEX,
    ISO8601_DATETIME,
    RFC3339_DATETIME,
    ISO8601_DATE,
    ISO8601_TIME,
    COMMON_DATETIME,
    WEB_UTC_DATETIME,
    PG_DATE_FMT,
    PG_TIMESTAMP_FMT,
)

def timeit(func):
    @wraps(func)
    def timer_wrapper(*args, **kwargs):
        """
        Inner function that uses the Timer context object
        """
        start = time.time()
        func(*args, **kwargs)
        return time.time() - start
    return timer_wrapper


def generate_dateset(n=100):
    return [
        pytz.utc.localize(datetime.utcnow()).strftime(random.choice(DATETIME_FORMATS))
        for _ in range(n)
    ]


@timeit
def benchmark_function(dateset):
    def parse_timestamp(ts):
        for fmt in DATETIME_FORMATS:
            try:
                return datetime.strptime(ts, fmt)
            except ValueError:
                continue

        raise ValueError("could not parse timestamp")

    for ts in dateset:
        ts = parse_timestamp(ts)


@timeit
def benchmark_forelse(dateset):
    for ts in dateset:
        # Try to parse a timestamp with a bunch of formats
        for fmt in DATETIME_FORMATS:
            try:
                ts = datetime.strptime(ts, fmt)
                break
            except ValueError:
                continue
        else:
            # Could not parse with any of the formats required
            raise ValueError("could not parse timestamp")


if __name__ == '__main__':
    index = []
    funcb = []
    foreb = []

    for n in np.linspace(1000, 500000, 10):
        n = int(n)
        index.append(n)
        dateset = generate_dateset(n)
        funcb.append(benchmark_function(dateset))
        foreb.append(benchmark_forelse(dateset))

        print("{} parses".format(n))

    print("benchmark complete")
    funcb = pd.Series(funcb, index=index, name="function")
    foreb = pd.Series(foreb, index=index, name="for/else")

    sns.set_style('whitegrid')
    sns.set_context('notebook')

    _, ax = plt.subplots(figsize=(9,6))
    funcb.plot(ax=ax, c='b')
    foreb.plot(ax=ax, c='g')

    ax.legend(frameon=True)
    ax.set_ylabel("time in seconds")
    ax.set_xlabel("number of dates parsed")
    plt.savefig("2018-05-17-benchmark.png")