Simple LayoutTests global baseline optimization

find_dupes.py

from __future__ import print_function
from collections import defaultdict
import hashlib
import re


def main():
    baselines = defaultdict(dict)
    f = open('baselines', 'r')
    regexp_head = re.compile(r'^\./(platform/[^/]+/)')
    regexp_tail = re.compile(r'-expected\.(png|txt|wav)$')
    for line in f.readlines():
        filename = line.strip()
        testname = regexp_head.sub('', filename)
        testname = regexp_tail.sub('.html', testname)
        sha1sum = hashlib.sha1(open(filename, 'rb').read()).digest()
        if sha1sum not in baselines[testname]:
            baselines[testname][sha1sum] = [filename]
        else:
            baselines[testname][sha1sum].append(filename)

    for testname in baselines:
        found_dupe = False
        for sha1sum in baselines[testname]:
            if len(baselines[testname][sha1sum]) == 1:
                continue
            found_dupe = True
            break
        if found_dupe:
            print(testname)

if __name__ == '__main__':
    main()

find_dupes.sh

#!/bin/bash

# Must run from third_party/WebKit/LayoutTests

find . -name "*-expected.txt" -or -name "*-expected.png" -or -name "*-expected.wav" > baselines
python2 find_dupes.py > dupes
cat dupes | xargs ../Tools/Scripts/webkit-patch optimize-baselines