|
| 1 | +package net.sansa_stack.examples.spark.query |
| 2 | + |
| 3 | +import scala.collection.mutable |
| 4 | + |
| 5 | +import net.sansa_stack.query.spark.query._ |
| 6 | +import org.apache.jena.riot.Lang |
| 7 | +import org.apache.spark.sql.SparkSession |
| 8 | + |
| 9 | +/** |
| 10 | + * Run SPARQL queries over Spark using Data Lake approach. |
| 11 | + */ |
| 12 | +object DataLake { |
| 13 | + |
| 14 | + def main(args: Array[String]) { |
| 15 | + parser.parse(args, Config()) match { |
| 16 | + case Some(config) => |
| 17 | + run(config.queryFile, config.mappingsFile, config.configFile) |
| 18 | + case None => |
| 19 | + println(parser.usage) |
| 20 | + } |
| 21 | + } |
| 22 | + |
| 23 | + def run(queryFile: String, mappingsFile: String, configFile: String): Unit = { |
| 24 | + |
| 25 | + println("======================================") |
| 26 | + println("| Sparqlify example |") |
| 27 | + println("======================================") |
| 28 | + |
| 29 | + val spark = SparkSession.builder |
| 30 | + .appName(s"DataLake (CSV) example") |
| 31 | + .master("local[*]") |
| 32 | + .getOrCreate() |
| 33 | + |
| 34 | + val result = spark.sparqlDL(queryFile, mappingsFile, configFile) |
| 35 | + result.show() |
| 36 | + |
| 37 | + spark.stop |
| 38 | + |
| 39 | + } |
| 40 | + |
| 41 | + case class Config( |
| 42 | + queryFile: String = getClass.getResource("/datalake/queries/Q1.sparql").getPath, |
| 43 | + mappingsFile: String = getClass.getResource("/datalake/config_csv-only").getPath, |
| 44 | + configFile: String = getClass.getResource("/datalake/mappings_csv-only.ttl").getPath) |
| 45 | + |
| 46 | + val parser = new scopt.OptionParser[Config]("Sparqlify example") { |
| 47 | + |
| 48 | + head(" DataLake (CSV) example") |
| 49 | + |
| 50 | + opt[String]('f', "queryFile").valueName("<queryFile>"). |
| 51 | + action((x, c) => c.copy(queryFile = x)). |
| 52 | + text("a file containing SPARQL queries or a single query, default: /queries/Q1.sparql") |
| 53 | + |
| 54 | + opt[String]('m', "mappingsFile").valueName("<mappingsFile>"). |
| 55 | + action((x, c) => c.copy(mappingsFile = x)). |
| 56 | + text("the mappings to the target sources, default: /config_csv-only") |
| 57 | + |
| 58 | + opt[String]('c', "configFile").optional().valueName("<configFile>"). |
| 59 | + action((x, c) => c.copy(configFile = x)). |
| 60 | + text("configuration file for different data sources, default: /mappings_csv-only.ttl") |
| 61 | + |
| 62 | + help("help").text("prints this usage text") |
| 63 | + } |
| 64 | + |
| 65 | +} |
| 66 | + |
0 commit comments