Update python-holidays integration (#2379)

python/prophet/make_holidays.py

@@ -6,13 +6,32 @@
 
 from __future__ import absolute_import, division, print_function
 
-import warnings
-
 import numpy as np
 import pandas as pd
 
-import prophet.hdays as hdays_part2
-import holidays as hdays_part1
+import holidays
+
+
+def get_country_holidays_class(country):
+    """Get class for a supported country.
+
+    Parameters
+    ----------
+    country: country code
+
+    Returns
+    -------
+    A valid country holidays class
+    """
+    substitutions = {
+        "TU": "TR",  # For compatibility with Turkey as 'TU' cases.
+    }
+
+    country = substitutions.get(country, country)
+    if not hasattr(holidays, country):
+        raise AttributeError(f"Holidays in {country} are not currently supported!")
+
+    return getattr(holidays, country)
 
 
 def get_holiday_names(country):
@@ -26,18 +45,8 @@ def get_holiday_names(country):
     -------
     A set of all possible holiday names of given country
     """
-    years = np.arange(1995, 2045)
-    try:
-        with warnings.catch_warnings():
-            warnings.simplefilter("ignore")
-            holiday_names = getattr(hdays_part2, country)(years=years).values()
-    except AttributeError:
-        try:
-            holiday_names = getattr(hdays_part1, country)(years=years).values()
-        except AttributeError as e:
-            raise AttributeError(f"Holidays in {country} are not currently supported!") from e
-
-    return set(holiday_names)
+    country_holidays = get_country_holidays_class(country)
+    return set(country_holidays(language="en_US", years=np.arange(1995, 2045)).values())
 
 
 def make_holidays_df(year_list, country, province=None, state=None):
@@ -53,16 +62,15 @@ def make_holidays_df(year_list, country, province=None, state=None):
     Dataframe with 'ds' and 'holiday', which can directly feed
     to 'holidays' params in Prophet
     """
-    try:
-        holidays = getattr(hdays_part2, country)(years=year_list, expand=False)
-    except AttributeError:
-        try:
-            holidays = getattr(hdays_part1, country)(prov=province, state=state, years=year_list, expand=False)
-        except AttributeError as e:
-            raise AttributeError(f"Holidays in {country} are not currently supported!") from e
+    country_holidays = get_country_holidays_class(country)
+    holidays = country_holidays(expand=False, language="en_US", subdiv=province, years=year_list)
 
-    holidays_df = pd.DataFrame([(date, holidays.get_list(date)) for date in holidays], columns=['ds', 'holiday'])
-    holidays_df = holidays_df.explode('holiday')
+    holidays_df = pd.DataFrame(
+        [(date, holidays.get_list(date)) for date in holidays],
+        columns=["ds", "holiday"],
+    )
+    holidays_df = holidays_df.explode("holiday")
     holidays_df.reset_index(inplace=True, drop=True)
-    holidays_df['ds'] = pd.to_datetime(holidays_df['ds'])
-    return (holidays_df)
+    holidays_df["ds"] = pd.to_datetime(holidays_df["ds"])
+
+    return holidays_df

python/pyproject.toml

@@ -19,7 +19,7 @@ dependencies = [
   "pandas>=1.0.4",
   "LunarCalendar>=0.0.9",
   "convertdate>=2.1.2",
-  "holidays>=0.14.2",
+  "holidays>=0.25",
   "python-dateutil>=2.8.0",
   "tqdm>=4.36.1",
   "importlib_resources",

python/scripts/generate_holidays_file.py

@@ -8,72 +8,72 @@ from __future__ import division
 from __future__ import print_function
 from __future__ import unicode_literals
 
-import inspect
+import re
 import unicodedata
-import warnings
 
 import pandas as pd
 import numpy as np
 
-import holidays as hdays_part1
-import prophet.hdays as hdays_part2
+from holidays import list_supported_countries
 from prophet.make_holidays import make_holidays_df
 
 
-def utf8_to_ascii(text):
-    """Holidays often have utf-8 characters. These are not allowed in R
-    package data (they generate a NOTE).
+def utf8_to_ascii(text: str) -> str:
+    """Holidays often have utf-8 characters. These are not allowed in R package data (they generate a NOTE).
     TODO: revisit whether we want to do this lossy conversion.
     """
-    ascii_text = (
-        unicodedata.normalize('NFD', text)
-        .encode('ascii', 'ignore')
-        .decode('ascii')
-        .strip()
-    )
+    ascii_text = unicodedata.normalize("NFD", text).encode("ascii", "ignore").decode("ascii")
+    # Remove trailing empty brackets and spaces.
+    ascii_text = re.sub(r"\(\)$", "", ascii_text).strip()
+
     # Check if anything converted
-    if sum(1 for x in ascii_text if x not in [' ', '(', ')', ',']) == 0:
-        return 'FAILED_TO_PARSE'
+    if sum(1 for x in ascii_text if x not in [" ", "(", ")", ","]) == 0:
+        return "FAILED_TO_PARSE"
     else:
         return ascii_text
 
 
-def generate_holidays_file():
-    """Generate csv file of all possible holiday names, ds,
-     and countries, year combination
-    """
-    year_list = np.arange(1995, 2045, 1).tolist()
+def generate_holidays_df() -> pd.DataFrame:
+    """Generate csv file of all possible holiday names, ds, and countries, year combination."""
+    country_codes = set(list_supported_countries().keys())
+
+    # For compatibility with Turkey as 'TU' cases.
+    country_codes.add("TU")
+
     all_holidays = []
-    # class names in holiday packages which are not countries
-    # Also cut out countries with utf-8 holidays that don't parse to ascii
-    class_to_exclude = {'rd', 'BY', 'BG', 'JP', 'RS', 'UA', 'KR'}
-
-    class_list2 = inspect.getmembers(hdays_part2, inspect.isclass)
-    country_set = {name for name in list(zip(*class_list2))[0] if len(name) == 2}
-    class_list1 = inspect.getmembers(hdays_part1, inspect.isclass)
-    country_set1 = {name for name in list(zip(*class_list1))[0] if len(name) == 2}
-    country_set.update(country_set1)
-    country_set -= class_to_exclude
-
-    for country in country_set:
-        df = make_holidays_df(year_list=year_list, country=country)
-        df['country'] = country
+    for country_code in country_codes:
+        df = make_holidays_df(
+            year_list=np.arange(1995, 2045, 1).tolist(),
+            country=country_code,
+        )
+        df["country"] = country_code
         all_holidays.append(df)
 
     generated_holidays = pd.concat(all_holidays, axis=0, ignore_index=True)
-    generated_holidays['year'] = generated_holidays.ds.dt.year
-    generated_holidays.sort_values(['country', 'ds', 'holiday'], inplace=True)
+    generated_holidays["year"] = generated_holidays.ds.dt.year
+    generated_holidays.sort_values(["country", "ds", "holiday"], inplace=True)
 
     # Convert to ASCII, and drop holidays that fail to convert
-    generated_holidays['holiday'] = generated_holidays['holiday'].apply(utf8_to_ascii)
-    failed_countries = generated_holidays.loc[generated_holidays['holiday'] == 'FAILED_TO_PARSE', 'country'].unique()
+    generated_holidays["holiday"] = generated_holidays["holiday"].apply(utf8_to_ascii)
+    failed_countries = generated_holidays.loc[
+        generated_holidays["holiday"] == "FAILED_TO_PARSE", "country"
+    ].unique()
     if len(failed_countries) > 0:
         print("Failed to convert UTF-8 holidays for:")
-        print('\n'.join(failed_countries))
-    assert 'FAILED_TO_PARSE' not in generated_holidays['holiday'].unique()
-    generated_holidays.to_csv("../R/data-raw/generated_holidays.csv", index=False)
+        print("\n".join(failed_countries))
+    assert "FAILED_TO_PARSE" not in generated_holidays["holiday"].unique()
+    return generated_holidays
 
 
 if __name__ == "__main__":
-    # execute only if run as a script
-    generate_holidays_file()
+    import argparse
+    import pathlib
+
+    if not pathlib.Path.cwd().stem == "python":
+        raise RuntimeError("Run script from prophet/python directory")
+    OUT_CSV_PATH = pathlib.Path(".") / ".." / "R/data-raw/generated_holidays.csv"
+    parser = argparse.ArgumentParser()
+    parser.add_argument("-o", "--outfile", default=OUT_CSV_PATH)
+    args = parser.parse_args()
+    df = generate_holidays_df()
+    df.to_csv(args.outfile, index=False)

python_shim/fbprophet/hdays.py

@@ -1,6 +0,0 @@
-# Copyright (c) Facebook, Inc. and its affiliates.
-
-# This source code is licensed under the MIT license found in the
-# LICENSE file in the root directory of this source tree.
-
-from prophet.hdays import *