Created
July 21, 2015 17:58
-
-
Save ldacosta/5616c1da57bcc6dd60aa to your computer and use it in GitHub Desktop.
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
private def templateForATest(dir: String, sql: SQLContext) = { | |
// whatever structure we want to test | |
val good = CleanData(reportDate = Timestamp.valueOf("2015-08-01 19:20:21"), // yyyy-[m]m-[d]d hh:mm:ss | |
date = Timestamp.valueOf("2015-08-01 19:20:21"), | |
impressions = 10, | |
clicks = 20, | |
totalConversions = 33, | |
impressionsConvergenceProbability = Some(0.99), | |
clicksConvergenceProbability = Some(0.99), | |
totalConversionsConvergenceProbability = Some(0.99)) | |
val l = List(good) | |
// let's make it an RDD | |
val rdd: RDD[CleanData] = sc.parallelize(l) | |
// where do I save it? | |
val fileNameAsString = s"/tmp/afile.parquet" // s"${dir}/afile.parquet" | |
HDFS.rm(fileNameAsString) | |
// blahblah :-) | |
info(s"Generating Parquet file here: ${fileNameAsString}. Should contain ${rdd.count()} elements") | |
// OK, SAVE!!! | |
val df = sql.createDataFrame(rdd) | |
df.saveAsParquetFile(fileNameAsString) | |
// NOW, READ IT: | |
val dfP = DataFrameParquet(sql.parquetFile(fileNameAsString)) | |
val wrappedRDD = { | |
DataFrameReader.toRDD[CleanData](dfP.unwrap) match { | |
case Success(anRDD) => anRDD | |
case Failure(fs) => | |
sys.error(s"FAILURE: ${fs.shows}") // sc.emptyRDD[CleanData] | |
} | |
} | |
if (wrappedRDD.isEmpty()) info("EEEEEEEMPTY!!") else info(s"This puppy has ${wrappedRDD.count()} elements") | |
} | |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment