splash multi-page benchmark script

benchmarks.py

#!/usr/bin/env python

import argparse

"""
Benchmark script.
"""

from splash.tests.stress import benchmark_png
import pandas as pd
from base64 import standard_b64decode
from PIL import Image
from cStringIO import StringIO

parser = argparse.ArgumentParser(description=__doc__)
parser.add_argument('out_file', nargs='?', default='bench.pkl')


def main():
    args = parser.parse_args()
    out_file = args.out_file
    if not out_file.endswith('.pkl'):
        out_file += '.pkl'

    results = []
    for url in [
            "http://wikipedia.org",
            "http://google.com",
            "http://w3.org",
            "http://w3.org/TR/2010/REC-xhtml-basic-20101123/",
            "http://blog.pinterest.com",
            "http://imgur.com",
    ]:
        for (width, height) in [
                (None, None), (None, 500), (500, None), (500, 500),
                (2000, None), (2000, 2000), (2100, None)]:
            result = {'url': url, 'width': width, 'height': height}
            # if result != {'url': "http://w3.org/TR/2010/REC-xhtml-basic-20101123/",
            #               'width': None, 'height': None}: continue
            try:
                result.update(benchmark_png(url=url, width=width,
                                            height=height, timeout=180.))
                img = Image.open(StringIO(standard_b64decode(result['png'])))
                del result['png']
                result['imgsize'] = "%dx%d" % img.size
            except Exception:
                import traceback
                print traceback.format_exc()
            finally:
                results.append(result)

    df = pd.DataFrame(results).set_index(['url', 'width', 'height'])
    print df
    df.to_pickle(args.out_file)


if __name__ == '__main__':
    main()