-
Notifications
You must be signed in to change notification settings - Fork 0
/
etl.py
183 lines (145 loc) · 6.71 KB
/
etl.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
import configparser
from datetime import datetime
import os
from pyspark.sql import SparkSession
from pyspark.sql.types import DateType, TimestampType
from pyspark.sql.functions import udf, col
from pyspark.sql.functions import year, month, dayofmonth, hour, weekofyear, date_format
config = configparser.ConfigParser()
config.read('dl.cfg')
os.environ["AWS_ACCESS_KEY_ID"]= config['AWS']['AWS_ACCESS_KEY_ID']
os.environ["AWS_SECRET_ACCESS_KEY"]= config['AWS']['AWS_SECRET_ACCESS_KEY']
def create_spark_session():
"""
Description:
- Instantiates a SparkSession
Returns:
- Spark instance object
"""
spark = SparkSession \
.builder \
.config("spark.jars.packages", "org.apache.hadoop:hadoop-aws:2.7.0") \
.getOrCreate()
return spark
def process_song_data(spark, input_data, output_data):
"""
Description:
- Extract song meta data from JSON files stored in S3 bucket
- Transforms song meta data into two separate DataFrames; songs_table and artists_table
- Loads them back into s3 as parquet files stored in a separate s3-bucket for analytical purposes
Arguments:
- Parameter spark: the instantiated SparkSession
- Parameter input_data: input path
- Parameter output_data: output path
Returns:
- None
"""
# get filepath to song data file
song_data = input_data + "song_data/*/*/*/*.json"
# read song data file
df = spark.read.json(song_data)
# extract columns to create songs table
songs_table = df.selectExpr(["song_id", "title", "artist_id", "year", "duration"])
# write songs table to parquet files partitioned by year and artist
songs_table.write.mode("overwrite").partitionBy("year", "artist_id").parquet(output_data + "songs.parquet")
# extract columns to create artists table
artists_table = df.selectExpr(["artist_id" ,\
"artist_name as name" ,\
"artist_location as location" ,\
"artist_latitude as latitude" ,\
"artist_longitude as longitude"]).dropDuplicates(["artist_id"])
# write artists table to parquet files
artists_table.write.mode("overwrite").parquet(output_data + "artists.parquet")
def process_log_data(spark, input_data, output_data):
"""
Description:
- Extract log data from JSON files stored in S3 bucket
- Transforms log data into three separate DataFrames; users_table, time_table and songplays_table
- Loads them back into s3 as parquet files stored in a separate s3-bucket for analytical purposes
Arguments:
- Parameter spark: the instantiated SparkSession
- Parameter input_data: input path
- Parameter output_data: output path
Returns:
- None
"""
# get filepath to log data file
log_data = input_data + "log_data/*.json"
# read log data file
df = spark.read.json(log_data)
# filter by actions for song plays
df = df.where(df["page"] == "NextSong")
# extract columns for users table
users_table = df.selectExpr(["userId as user_id" ,\
"firstName as first_name" ,\
"lastName as last_name" ,\
"gender" ,\
"level"]).dropDuplicates()
# write users table to parquet files
users_table.write.mode("overwrite").parquet(output_data + "users.parquet")
# create timestamp column from original timestamp column
get_timestamp = udf(lambda epoch_time: datetime.fromtimestamp(epoch_time / 1000), TimestampType())
df = df.withColumn('timestamp', get_timestamp(col('ts')))
# create datetime column from original timestamp column
get_datetime = udf(lambda epoch_time: datetime.fromtimestamp(epoch_time / 1000), DateType())
df = df.withColumn('datetime', get_datetime(col('ts')))
# extract columns to create time table
time_table = df.select([hour("timestamp").alias("hour") ,\
dayofmonth("timestamp").alias("day") ,\
weekofyear("timestamp").alias("week") ,\
month("timestamp").alias("month") ,\
year("timestamp").alias("year") ,\
date_format("timestamp", 'E').alias("weekday")]).dropDuplicates()
# write time table to parquet files partitioned by year and month
time_table.write.mode("overwrite").partitionBy("year", "month").parquet(output_data + "time.parquet")
# read in song data to use for songplays table
song_df = spark.read.parquet(output_data + "songs.parquet")
# extract columns from joined song and log datasets to create songplays table
songplays_table = song_df.join(df, df.song == song_df.title)\
.selectExpr(["timestamp as start_time" ,\
"userid as user_id" ,\
"level" ,\
"song_id" ,\
"artist_id" ,\
"sessionid as session_id" ,\
"location" ,\
"useragent as user_agent"]) \
.withColumn("year", year("start_time")) \
.withColumn("month", month("start_time")).dropDuplicates()
# write songplays table to parquet files partitioned by year and month
songplays_table.createOrReplaceTempView("songplays")
spark.sql("""
SELECT row_number() over (order by start_time asc) as songplay_id,
start_time,
user_id,
level,
song_id,
artist_id,
session_id,
location,
user_agent,
year,
month
FROM songplays
""").write.mode("overwrite").partitionBy("year", "month").parquet(output_data + "songplays.parquet")
def main():
"""
Description:
- Creates spark session
- Creates i/o path variables
- Calls the process_song_data function
- Calls the process_log_data function
Arguments:
- Parameter spark: the instantiated SparkSession
- Parameter input_data: input path
- Parameter output_data: output path
Returns:
- None
"""
spark = create_spark_session()
input_data = "s3a://udacity-dend/"
output_data = "s3a://aletje/"
process_song_data(spark, input_data, output_data)
process_log_data(spark, input_data, output_data)
if __name__ == "__main__":
main()