stocks splitting api google or yahoo

This is how the quantmod R package does it. The split information is in the "Dividend Only" CSV:
http://ichart.finance.yahoo.com/x?s=IBM&a=00&b=2&c=1962&d=04&e=25&f=2011&g=v&y=0&z=30000

You can do it easily in python 3 with the help of the pandas datareader package. Starting defining a function which will return the split history as a dataframe:

def split_history(stock, date_start, date_end, limit_denominator=1000):
    from decimal import Decimal
    from fractions import Fraction
    from pandas_datareader import data as web
    df = web.DataReader(stock, data_source='yahoo-actions', start=date_start, end=date_end)
    is_split = df['action']=='SPLIT'
    df = df[is_split]
    ratios = []
    for index, row in df.iterrows():
        # Taking the inverse of the row['value'] as it is Yahoo finance convention
        ratio = Fraction(1/Decimal(row['value'])).limit_denominator(limit_denominator)
        ratios.append("{num} for {denom}".\
                            format(num=ratio.numerator, denom=ratio.denominator))
    df['ratio'] = ratios
    return df

Now we can get the splits of Microsoft ('MSFT') as an example:

stock = 'MSFT'
date_start = '1987-01-01'
date_end = '2020-07-22'
split_history(stock, date_start, date_end)

            action  value       ratio
2003-02-18  SPLIT   0.500000    2 for 1
1999-03-29  SPLIT   0.500000    2 for 1
1998-02-23  SPLIT   0.500000    2 for 1
1996-12-09  SPLIT   0.500000    2 for 1
1994-05-23  SPLIT   0.500000    2 for 1
1992-06-15  SPLIT   0.666667    3 for 2
1991-06-27  SPLIT   0.666667    3 for 2
1990-04-16  SPLIT   0.500000    2 for 1
1987-09-21  SPLIT   0.500000    2 for 1

It handles also properly the reverse stock splits:

stock = 'PHM.MC'
split_history(stock, date_start, date_end)

                action  value   ratio
 2020-07-22     SPLIT   12.0    1 for 12

ps: probably there are better ways to input the dates. ps2: also, the limit_denominator is there to avoid wrong roundings. You can extend it in rare split ratio cases.