AudioSaladTemplate.py

Go to the documentation of this file.

import re
import awswrangler as wr
 
class AudioSaladTemplate:    
    def preprocessing(self, filename, features, session, rel_col):
        """Loads file and do some fixes to fit it to our standard and make it able to be processed
 
        Args:
            filename (str): s3 full path of filename
            features (dict): contains delimiter, skip_rows and encoding required for current file
            session (boto3 obj): AWS client connection
            rel_col (str): column name used to identify upc/release_id
        Returns: df (pandas dataframe)
        """
        delimiter = features["delimiter"]
        skip_rows = features["skip_rows"]
        encoding = features["encoding"]
        df = wr.s3.read_csv(filename, sep=delimiter, skiprows=skip_rows, encoding=encoding, dtype={rel_col:str}, low_memory=False, boto3_session=session)
        return df
 
    def date(self, filename, df=None):
        """Sets date column given the filename (it contains the date)
           Sometimes just rename the date column to fit to the otto mapping
           For example AudioSalad_2023_-_M7_GYROstream_Sales_Export.csv
        Args:
            filename (str): current filename
            df (pandas dataframe): dataframe where changes will applied
        Returns: date_str (str)
        """
        if str(type(df)) != "<class 'NoneType'>":
            df["as_iso_date"] = df["sale date"]
            return df["as_iso_date"]
        else:
            year = re.findall(r'\d{4}', filename)[0]
            month = re.findall(r'M\d{1,2}', filename)[0]
            month = month.replace("M", "")
            if len(month) == 1:
                month = "0"+month
                date_str = year+month
            return date_str
 
    def serviceFinder(self, service):
        """Maps the service name with the service specified by the client according their requirements
        Args:
            service (str): service name
        Returns: (service renamed)
        """
        service = str(service)
        if "Red" in service:
            return "YouTube Music"
        elif "UMA" in service:
            return "UMA"
        elif "Google" in service:
            return "Google Play"
        elif "Youtube Ads" in service:
            return "YouTube Content ID"
        elif "Amazon" in service:
            return "Amazon"
        elif "Spotify" in service:
            return "Spotify"
        elif "Yandex" in service:
            return "Yandex"
        elif "Soundtrack Your Brand" in service:
            return "Soundtrack Your Brand"
        elif "Apple" in service or "iTunes" in service:
            return "iTunes"
        elif "UMA" in service:
            return "UMA"
        elif "NetEase" in service:
            return "Netease"
        elif "Napster" in service:
            return "Napster"
        elif "Tidal" in service:
            return "Tidal"
        elif "MFS" in service:
            return "MFS"
        elif "Facebook" in service:
            return "facebook"
        elif "Audio tier" in service:
            return "Youtube Audio Tier"
        elif "Tik Tok" in service:
            return "TikTok"
        elif "Slacker" in service:
            return "Slacker"
        elif "Soundcloud" in service:
            return "Soundcloud"
        return service
 
    def type(self, Type, configuration):
        """Maps the sale type with the name specified by the client according their requirements
        Args:
            Type (str): sale type shown in the original file
            configuration (str):  
        Returns: (sale type in a single one letter)
        """
        if Type == "Release" and configuration == "Download":
            return "A"
        elif Type == "Track" and configuration == "Download":
            return "T"
        return "S"
 
    def assetType(self, df):
        """Applies serviceFinder and type functions to current dataframe
        Args:
            df (pandas dataframe): dataframe where changes will applied
        Returns: df (pandas dataframe)
        """
        df["getservice"] = df.apply(lambda Row: self.serviceFinder(Row["sub source"]), axis=1)
        df["gettypeaudiosalad"] = df.apply(lambda Row: self.type(Row["type"], Row["configuration"]), axis=1)
        return df