Up to 4x faster than Series.str.contains / Series.eq - can handle Unicode!

pip install a-pandas-ex-fast-string

from a_pandas_ex_fast_string import pd_add_fast_string

import pandas as pd



pd_add_fast_string()



df2 = pd.read_csv(

    "https://raw.githubusercontent.com/pandas-dev/pandas/main/doc/data/titanic.csv",

    dtype="string",

)



# To check if it can handle unicode strings

df2.Name.iloc[0] += "ö"

df2.Name.iloc[10] += "ä"

df2.Name.iloc[20] += "ü"



# converts the whole dataframe

df900 = pd.Q_convert_to_fast_string(df2.copy())





dfone = df2.copy()

# converts one column

dfone.Cabin.ds_update_fast_string()



# Let's create some DataFrames of different sizes

df9000 = pd.Q_convert_to_fast_string(

    pd.concat([df2.copy() for _ in range(10)], ignore_index=True)

)

df90000 = pd.Q_convert_to_fast_string(

    pd.concat([df2.copy() for _ in range(100)], ignore_index=True)

)

df900000 = pd.Q_convert_to_fast_string(

    pd.concat([df2.copy() for _ in range(1000)], ignore_index=True)

)

df9000000 = pd.Q_convert_to_fast_string(

    pd.concat([df2.copy() for _ in range(10000)], ignore_index=True)

)







%timeit df900.loc[df900.Name.s_string_contains('y') | df900.Name.s_string_is('Montvila, Rev. Juozas')]

%timeit df900.loc[df900.Name.str.contains('y',regex=False) | (df900.Name == 'Montvila, Rev. Juozas')]

604 µs ± 9.09 µs per loop (mean ± std. dev. of 7 runs, 1,000 loops each)

997 µs ± 13.2 µs per loop (mean ± std. dev. of 7 runs, 1,000 loops each)





%timeit df9000.loc[df9000.Name.s_string_contains('y') | df9000.Name.s_string_is('Montvila, Rev. Juozas')]

%timeit df9000.loc[df9000.Name.str.contains('y',regex=False) | (df9000.Name == 'Montvila, Rev. Juozas')]

1.15 ms ± 15.2 µs per loop (mean ± std. dev. of 7 runs, 1,000 loops each)

2.77 ms ± 11.2 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)





%timeit df90000.loc[df90000.Name.s_string_contains('y') | df90000.Name.s_string_is('Montvila, Rev. Juozas')]

%timeit df90000.loc[df90000.Name.str.contains('y',regex=False) | (df90000.Name == 'Montvila, Rev. Juozas')]

6.45 ms ± 77.4 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)

20.7 ms ± 166 µs per loop (mean ± std. dev. of 7 runs, 10 loops each)





%timeit df900000.loc[df900000.Name.s_string_contains('y') | df900000.Name.s_string_is('Montvila, Rev. Juozas')]

%timeit df900000.loc[df900000.Name.str.contains('y',regex=False) | (df900000.Name == 'Montvila, Rev. Juozas')]

60.5 ms ± 853 µs per loop (mean ± std. dev. of 7 runs, 10 loops each)

206 ms ± 840 µs per loop (mean ± std. dev. of 7 runs, 1 loop each)





%timeit df9000000.loc[df9000000.Name.s_string_contains('y') | df9000000.Name.s_string_is('Montvila, Rev. Juozas')]

%timeit df9000000.loc[df9000000.Name.str.contains('y',regex=False) | (df9000000.Name == 'Montvila, Rev. Juozas')]

596 ms ± 11.8 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)

2.06 s ± 2.5 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)





# Good news: it can handle unicode characters! 

df9000.loc[df9000.Name.s_string_contains('ö')].Name

Out[14]: 

0       Braund, Mr. Owen Harrisö

891     Braund, Mr. Owen Harrisö

1782    Braund, Mr. Owen Harrisö

2673    Braund, Mr. Owen Harrisö

3564    Braund, Mr. Owen Harrisö

4455    Braund, Mr. Owen Harrisö

5346    Braund, Mr. Owen Harrisö

6237    Braund, Mr. Owen Harrisö

7128    Braund, Mr. Owen Harrisö

8019    Braund, Mr. Owen Harrisö

Name: Name, dtype: string





# Bad news: every time you modify a Series, you have to update it: 



df9000.loc[df9000.Name.s_string_contains('ö')].Name

0       Braund, Mr. Owen Harrisö

891     Braund, Mr. Owen Harrisö

1782    Braund, Mr. Owen Harrisö

2673    Braund, Mr. Owen Harrisö

3564    Braund, Mr. Owen Harrisö





df9000.loc[df9000.Name.s_string_contains('ö'), "Name"] = df9000.loc[df9000.Name.s_string_contains('ö'), "Name"] + 'Ä' # updating 



df9000.Name

0                               Braund, Mr. Owen HarrisöÄ

1       Cumings, Mrs. John Bradley (Florence Briggs Th...

2                                  Heikkinen, Miss. Laina



df9000.loc[df9000.Name.s_string_contains('ö'), "Name"]  # Exception because ds_update_fast_string was not called



Traceback (most recent call last):

  File "C:\Users\Gamer\anaconda3\envs\dfdir\lib\site-packages\IPython\core\interactiveshell.py", line 3398, in run_code

    exec(code_obj, self.user_global_ns, self.user_ns)

  File "<ipython-input-7-2b0dfaf8b41c>", line 1, in <cell line: 1>

    df9000.loc[df9000.Name.s_string_contains('ö'), "Name"]

  File "C:/Users/Gamer/anaconda3/envs/dfdir/a_pandas_string_search.py", line 133, in search_contains

    wordtosearchbin, columntosearch = _get_col_word(

  File "C:/Users/Gamer/anaconda3/envs/dfdir/a_pandas_string_search.py", line 103, in _get_col_word

    return wordtosearchbin, series._stringser.__array__()

AttributeError: 'NoneType' object has no attribute '__array__'



df9000.Name.ds_update_fast_string() # Necessary after changing a Series

# you can also update the whole DataFrame: df9000 = df9000.ds_update_fast_string()

# Be careful: df9000.Name.ds_update_fast_string() returns None (inplace) 

# df9000.ds_update_fast_string() returns a DataFrame



df9000.loc[df9000.Name.s_string_contains('ö'), "Name"]  # Now it is working!



0       Braund, Mr. Owen HarrisöÄ

891     Braund, Mr. Owen HarrisöÄ

1782    Braund, Mr. Owen HarrisöÄ

2673    Braund, Mr. Owen HarrisöÄ