etl.py

import os
import glob
import psycopg2
import pandas as pd
from sql_queries import *
import datetime


def process_song_file(cur, filepath):
    """
    This procedure processes a song file whose filepath has been provided as an arugment.
    It extracts the song information in order to store it into the songs table.
    Then it extracts the artist information in order to store it into the artists table.

    INPUTS: 
    * cur : the cursor variable
    * filepath : the file path to the song file
    
    RETURNS: NONE
    """
    # open song file
    df = pd.read_json(filepath, lines=True)

    # insert song record
    song_data = list(df[['song_id','title','artist_id','year','duration']].values[0])
    cur.execute(song_table_insert, song_data)
    
    # insert artist record
    artist_data = list(df[['artist_id','artist_name','artist_location','artist_latitude','artist_longitude']].values[0])
    cur.execute(artist_table_insert, artist_data)


def process_log_file(cur, filepath):
    """
    This procedure processes a log file whose filepath has been provided as an arugment.
    It extracts the time information and convert to datetime datatype in order to store it into the time table.
    Then it extracts the user information in order to store it into the users table.
    Finally, It finds song_id and artist_id data from song and artist table by specific columns (song, artist, length).
    Then It inserts song play record data into songplays table.
    
    INPUTS: 
    * cur : the cursor variable
    * filepath : the file path to the log file
    
    RETURNS: NONE
    """
    # open log file
    df = pd.read_json(filepath, lines=True)

    # filter by NextSong action
    df = df.loc[df['page']=="NextSong"]

    # convert timestamp column to datetime
    t = df.assign(ts=[datetime.datetime.fromtimestamp(ts/1000) for ts in list(df['ts'])])
    
    # insert time data records
    time_data = (t['ts'].astype(int) // 10**6, t['ts'].dt.hour, t['ts'].dt.day, t['ts'].dt.week, t['ts'].dt.month, t['ts'].dt.year, t['ts'].dt.weekday)
    column_labels = ("timestamp", "hour", "day", "week_of_year", "month", "year", "weekday")
    time_df = pd.DataFrame({column_labels[num]:time_data[num] for num in range(len(column_labels))})

    for i, row in time_df.iterrows():
        cur.execute(time_table_insert, list(row))

    # load user table
    user_df = df[['userId','firstName','lastName','gender','level']]

    # insert user records
    for i, row in user_df.iterrows():
        cur.execute(user_table_insert, row)

    # insert songplay records
    for index, row in df.iterrows():
        
        # get songid and artistid from song and artist tables
        cur.execute(song_select, (row.song, row.artist, row.length))
        results = cur.fetchone()
        
        if results:
            songid, artistid = results
        else:
            songid, artistid = None, None

        # insert songplay record
        songplay_data = (row.ts, row.userId, row.level, songid, artistid, row.sessionId, row.location, row.userAgent)
        cur.execute(songplay_table_insert, songplay_data)


def process_data(cur, conn, filepath, func):
    """
    This procedure handle song/log files before processing query jobs.
    
    INPUTS: 
    * cur : the cursor variable
    * conn : the psycopg2.connect variable
    * filepath : the file path to the song/log file
    * func : the function that proceeding to process song/log file
    
    RETURNS: NONE
    """
    # get all files matching extension from directory
    all_files = []
    for root, dirs, files in os.walk(filepath):
        files = glob.glob(os.path.join(root,'*.json'))
        for f in files :
            all_files.append(os.path.abspath(f))

    # get total number of files found
    num_files = len(all_files)
    print('{} files found in {}'.format(num_files, filepath))

    # iterate over files and process
    for i, datafile in enumerate(all_files, 1):
        func(cur, datafile)
        conn.commit()
        print('{}/{} files processed.'.format(i, num_files))


def main():
    """
    main
    """
    conn = psycopg2.connect("host=127.0.0.1 dbname=sparkifydb user=student password=student")
    cur = conn.cursor()

    process_data(cur, conn, filepath='data/song_data', func=process_song_file)
    process_data(cur, conn, filepath='data/log_data', func=process_log_file)

    conn.close()


if __name__ == "__main__":
    main()