test_cypher.py

import time
import os

from cypher import identify
from dev import LANG_INFO

supported = [s.lower() for s in LANG_INFO.keys()]
data = '' # path to benchmark data (https://github.com/nbraud/benchmarksgame/tree/master/bench)
name2ext = {
    'csharp': 'C#', 'gcc': 'C', 'gpp': 'C++', 'ghc': 'Haskell',
    'jruby': 'Ruby', 'python3': 'Python', 'hack': 'PHP', 'yarv': 'Ruby',
    'C-sharp': 'C#'
}

count = 0.0
correct = 0
before =  time.time()
for subdir, _, files in os.walk(data):
    for f in files:
        path = os.path.join(subdir, f)
        ext = path.split(".")[-1]
        if ext in name2ext:
            ext = name2ext.get(ext).lower()
        if ext in supported:
            count += 1
            guess = identify(path).lower()
            if guess == ext:
                correct += 1
            else:
                print(guess, path)
                
print("{} ({} / {})".format(round(correct / count, 3), correct, count))
print("Time: {}".format(time.time() - before))