Rooba
/
RapidFuzz
mirror of https://github.com/maxbachmann/RapidFuzz.git
1
1
Fork 0
Code Issues Packages Projects Releases Wiki Activity

remove benchmarks already covered by new benchmark suite

This commit is contained in:
Max Bachmann 2022-12-24 13:46:02 +01:00 committed by Max Bachmann
parent fa90076b65
commit 171c7d5ff4
6 changed files with 0 additions and 368 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

51
bench/benchmark_damerau_levenshtein.py
View File

@ -1,51 +0,0 @@
# todo combine benchmarks of scorers into common code base
import timeit
import pandas
def benchmark(name, func, setup, lengths, count):
print(f"starting {name}")
start = timeit.default_timer()
results = []
from tqdm import tqdm
for length in tqdm(lengths):
test = timeit.Timer(func, setup=setup.format(length, count))
results.append(min(test.timeit(number=1) for _ in range(7)) / count)
stop = timeit.default_timer()
print(f"finished {name}, Runtime: ", stop - start)
return results
setup = """
from rapidfuzz.distance.DamerauLevenshtein import distance
from jellyfish import damerau_levenshtein_distance
import string
import random
random.seed(18)
characters = string.ascii_letters + string.digits + string.whitespace + string.punctuation
a = ''.join(random.choice(characters) for _ in range({0}))
b_list = [''.join(random.choice(characters) for _ in range({0})) for _ in range({1})]
"""
lengths = list(range(1, 256, 2))
count = 1000
time_rapidfuzz = benchmark(
"rapidfuzz", "[distance(a, b) for b in b_list]", setup, lengths, count
)
time_jellyfish = benchmark(
"jellyfish",
"[damerau_levenshtein_distance(a, b) for b in b_list]",
setup,
lengths,
count,
)
df = pandas.DataFrame(
data={"length": lengths, "rapidfuzz": time_rapidfuzz, "jellyfish": time_jellyfish}
)
df.to_csv("results/levenshtein_damerau.csv", sep=",", index=False)

62
bench/benchmark_indel_levenshtein.py
View File

@ -1,62 +0,0 @@
# todo combine benchmarks of scorers into common code base
import timeit
import numpy as np
import pandas
def benchmark(name, func, setup, lengths, count):
print(f"starting {name}")
start = timeit.default_timer()
results = []
for length in lengths:
test = timeit.Timer(func, setup=setup.format(length, count))
results.append(min(test.timeit(number=1) for _ in range(7)) / count)
stop = timeit.default_timer()
print(f"finished {name}, Runtime: ", stop - start)
return results
setup = """
from rapidfuzz import string_metric, process, fuzz
import Levenshtein
import string
import random
random.seed(18)
characters = string.ascii_letters + string.digits + string.whitespace + string.punctuation
a = ''.join(random.choice(characters) for _ in range({0}))
b_list = [''.join(random.choice(characters) for _ in range({0})) for _ in range({1})]
"""
lengths = list(range(1, 512, 2))
count = 3000
time_rapidfuzz = benchmark(
"rapidfuzz",
"[string_metric.levenshtein(a, b, weights=(1,1,2)) for b in b_list]",
setup,
lengths,
count,
)
# this gets very slow, so only benchmark it for smaller values
time_python_levenshtein = (
benchmark(
"python-Levenshtein",
"[Levenshtein.ratio(a, b) for b in b_list]",
setup,
list(range(1, 256, 2)),
count,
)
+ [np.NaN] * 128
)
df = pandas.DataFrame(
data={
"length": lengths,
"rapidfuzz": time_rapidfuzz,
"python-Levenshtein": time_python_levenshtein,
}
)
df.to_csv("results/levenshtein_indel.csv", sep=",", index=False)

60
bench/benchmark_jaro.py
View File

@ -1,60 +0,0 @@
# todo combine benchmarks of scorers into common code base
import timeit
import numpy as np
import pandas
def benchmark(name, func, setup, lengths, count):
print(f"starting {name}")
start = timeit.default_timer()
results = []
from tqdm import tqdm
for length in tqdm(lengths):
test = timeit.Timer(func, setup=setup.format(length, count))
results.append(min(test.timeit(number=1) for _ in range(7)) / count)
stop = timeit.default_timer()
print(f"finished {name}, Runtime: ", stop - start)
return results
setup = """
from rapidfuzz.distance import Jaro
import jellyfish
import string
import random
random.seed(18)
characters = string.ascii_letters + string.digits + string.whitespace + string.punctuation
a = ''.join(random.choice(characters) for _ in range({0}))
b_list = [''.join(random.choice(characters) for _ in range({0})) for _ in range({1})]
"""
lengths = list(range(1, 256, 4))
count = 4000
time_rapidfuzz = benchmark(
"rapidfuzz", "[Jaro.similarity(a, b) for b in b_list]", setup, lengths, count
)
# this gets very slow, so only benchmark it for smaller values
time_jellyfish = (
benchmark(
"jellyfish",
"[jellyfish.jaro_similarity(a, b) for b in b_list]",
setup,
list(range(1, 128, 4)),
count,
)
+ [np.NaN] * 32
)
df = pandas.DataFrame(
data={
"length": lengths,
"rapidfuzz": time_rapidfuzz,
"jellyfish": time_jellyfish,
}
)
df.to_csv("results/jaro.csv", sep=",", index=False)

60
bench/benchmark_jaro_winkler.py
View File

@ -1,60 +0,0 @@
# todo combine benchmarks of scorers into common code base
import timeit
import numpy as np
import pandas
def benchmark(name, func, setup, lengths, count):
print(f"starting {name}")
start = timeit.default_timer()
results = []
from tqdm import tqdm
for length in tqdm(lengths):
test = timeit.Timer(func, setup=setup.format(length, count))
results.append(min(test.timeit(number=1) for _ in range(7)) / count)
stop = timeit.default_timer()
print(f"finished {name}, Runtime: ", stop - start)
return results
setup = """
from rapidfuzz.distance import JaroWinkler
import jellyfish
import string
import random
random.seed(18)
characters = string.ascii_letters + string.digits + string.whitespace + string.punctuation
a = ''.join(random.choice(characters) for _ in range({0}))
b_list = [''.join(random.choice(characters) for _ in range({0})) for _ in range({1})]
"""
lengths = list(range(1, 256, 4))
count = 4000
time_rapidfuzz = benchmark(
"rapidfuzz", "[JaroWinkler.similarity(a, b) for b in b_list]", setup, lengths, count
)
# this gets very slow, so only benchmark it for smaller values
time_jellyfish = (
benchmark(
"jellyfish",
"[jellyfish.jaro_winkler_similarity(a, b) for b in b_list]",
setup,
list(range(1, 128, 4)),
count,
)
+ [np.NaN] * 32
)
df = pandas.DataFrame(
data={
"length": lengths,
"rapidfuzz": time_rapidfuzz,
"jellyfish": time_jellyfish,
}
)
df.to_csv("results/jaro_winkler.csv", sep=",", index=False)

56
bench/benchmark_osa.py
View File

@ -1,56 +0,0 @@
# todo combine benchmarks of scorers into common code base
import timeit
import numpy as np
import pandas
def benchmark(name, func, setup, lengths, count):
print(f"starting {name}")
start = timeit.default_timer()
results = []
from tqdm import tqdm
for length in tqdm(lengths):
test = timeit.Timer(func, setup=setup.format(length, count))
results.append(min(test.timeit(number=1) for _ in range(7)) / count)
stop = timeit.default_timer()
print(f"finished {name}, Runtime: ", stop - start)
return results
setup = """
from rapidfuzz.distance.OSA import distance
from pyxdameraulevenshtein import damerau_levenshtein_distance
import string
import random
random.seed(18)
characters = string.ascii_letters + string.digits + string.whitespace + string.punctuation
a = ''.join(random.choice(characters) for _ in range({0}))
b_list = [''.join(random.choice(characters) for _ in range({0})) for _ in range({1})]
"""
lengths = list(range(1, 256, 2))
count = 1000
time_rapidfuzz = benchmark(
"rapidfuzz", "[distance(a, b) for b in b_list]", setup, lengths, count
)
time_pyxdameraulevenshtein = benchmark(
"pyxdameraulevenshtein",
"[damerau_levenshtein_distance(a, b) for b in b_list]",
setup,
list(range(1, 16, 2)),
count,
) + [np.NaN] * int((256 - 16) / 2)
df = pandas.DataFrame(
data={
"length": lengths,
"rapidfuzz": time_rapidfuzz,
"pyxdameraulevenshtein": time_pyxdameraulevenshtein,
}
)
df.to_csv("results/osa.csv", sep=",", index=False)

79
bench/benchmark_uniform_levenshtein.py
View File

@ -1,79 +0,0 @@
import timeit
import numpy as np
import pandas
def benchmark(name, func, setup, lengths, count):
print(f"starting {name}")
start = timeit.default_timer()
results = []
for length in lengths:
test = timeit.Timer(func, setup=setup.format(length, count))
results.append(min(test.timeit(number=1) for _ in range(7)) / count)
stop = timeit.default_timer()
print(f"finished {name}, Runtime: ", stop - start)
return results
setup = """
from rapidfuzz import string_metric
import Levenshtein
import polyleven
import edlib
import editdistance
import string
import random
random.seed(18)
characters = string.ascii_letters + string.digits + string.whitespace + string.punctuation
a = ''.join(random.choice(characters) for _ in range({0}))
b_list = [''.join(random.choice(characters) for _ in range({0})) for _ in range({1})]
"""
lengths = list(range(1, 512, 2))
count = 2000
time_rapidfuzz = benchmark(
"rapidfuzz",
"[string_metric.levenshtein(a, b) for b in b_list]",
setup,
lengths,
count,
)
time_polyleven = benchmark(
"polyleven", "[polyleven.levenshtein(a, b) for b in b_list]", setup, lengths, count
)
# this gets very slow, so only benchmark it for smaller values
time_python_levenshtein = (
benchmark(
"python-Levenshtein",
"[Levenshtein.distance(a, b) for b in b_list]",
setup,
list(range(1, 256, 2)),
count,
)
+ [np.NaN] * 128
)
time_edlib = benchmark(
"edlib", "[edlib.align(a, b) for b in b_list]", setup, lengths, count
)
time_editdistance = benchmark(
"editdistance", "[editdistance.eval(a, b) for b in b_list]", setup, lengths, count
)
df = pandas.DataFrame(
data={
"length": lengths,
"rapidfuzz": time_rapidfuzz,
"polyleven": time_polyleven,
"python-Levenshtein": time_python_levenshtein,
"edlib": time_edlib,
"editdistance": time_editdistance,
}
)
df.to_csv("results/levenshtein_uniform.csv", sep=",", index=False)