def get_report_ru():
"""
Get data from Wikipedia page with
COVID-19 statistics for each state of Russia
:return: Pandas DataFrame
"""
url = 'https://en.wikipedia.org/wiki/COVID-19_pandemic_in_Russia'
df = utils.get_wiki_table_df(url, 'Per 1 million')
df = pd.DataFrame(df.values[:, 1:5],
columns=['state', 'confirmed', 'recovered', 'deaths'])
df.drop(df[df['state'].str.len() > 40].index, inplace=True)
df['state'] = df['state'].apply(lambda x: utils.clean_territory_name(x))
df = utils.wiki_table_df_numeric_column_clean(
df, ['confirmed', 'deaths', 'recovered'])
df['country'] = 'Russia'
check_report(df)
return df
def get_report_us():
"""
Get COVID-19 data from Wikipedia page with
COVID-19 statistics for each state of USA
:return: Pandas DataFrame
"""
url = 'https://en.wikipedia.org/wiki/Timeline_of_the_COVID-19_pandemic_in_the_United_States'
df = utils.get_wiki_table_df(
url, 'COVID-19 pandemic in the United States by state and territory')
df = pd.DataFrame(df.values[:, 1:5],
columns=['state', 'confirmed', 'deaths', 'recovered'])
df.drop(df[df['state'].str.len() > 40].index, inplace=True)
df['state'] = df['state'].apply(lambda x: utils.clean_territory_name(x))
df = utils.wiki_table_df_numeric_column_clean(
df, ['confirmed', 'deaths', 'recovered'])
df['country'] = 'United States'
check_report(df)
return df
def get_report_countries():
"""
Get data from Wikipedia page with
COVID-19 statistics for each country
:return: Pandas DataFrame
"""
url = 'https://en.wikipedia.org/wiki/COVID-19_pandemic_by_country_and_territory'
df = utils.get_wiki_table_df(url, 'Location[b]')
df = pd.DataFrame(df.values[:, 1:5],
columns=['country', 'confirmed', 'deaths', 'recovered'])
df = df[~df['country'].isna()]
df['country'] = df['country'].apply(
lambda x: utils.clean_territory_name(x))
df.drop(df[df['country'].str.len() > 40].index, inplace=True)
df = utils.wiki_table_df_numeric_column_clean(
df, ['confirmed', 'deaths', 'recovered'])
df['state'] = None
check_report(df)
return df