Downstream changes

README.md

@@ -118,34 +118,37 @@ To build the executable, run
 The program can accept a number of command line options:
 
     $ bin/eventsim --help
-        -a, --attrition-rate  <arg>    annual user attrition rate (as a fraction of
-                                       current, so 1% => 0.01) (default = 0.0)
-        -c, --config  <arg>            config file
-            --continuous               continuous output
-            --nocontinuous             run all at once
-        -e, --end-time  <arg>          end time for data
-                                       (default = 2015-08-12T14:56:25.006)
-        -f, --from  <arg>              from x days ago (default = 15)
-            --generate-counts          generate listen counts file then stop
-            --nogenerate-counts        run normally
-            --generate-similars        generate similar song file then stop
-            --nogenerate-similars      run normally
-        -g, --growth-rate  <arg>       annual user growth rate (as a fraction of
-                                       current, so 1% => 0.01) (default = 0.0)
-            --kafkaBrokerList  <arg>   kafka broker list
-        -k, --kafkaTopic  <arg>        kafka topic
-        -n, --nusers  <arg>            initial number of users (default = 1)
-        -r, --randomseed  <arg>        random seed
-        -s, --start-time  <arg>        start time for data
-                                       (default = 2015-08-05T14:56:25.040)
-            --tag  <arg>               tag applied to each line (for example, A/B test
-                                       group)
-        -t, --to  <arg>                to y days ago (default = 1)
-        -u, --userid  <arg>            first user id (default = 1)
-            --help                     Show help message
-
-       trailing arguments:
-        output-file (not required)   File name
+          -a, --attrition-rate  <arg>    annual user attrition rate (as a fraction of
+                                         current, so 1% => 0.01) (default = 0.0)
+          -c, --config  <arg>            config file
+              --continuous               continuous output
+              --nocontinuous             run all at once
+          -e, --end-time  <arg>          end time for data
+                                         (default = 2016-01-06T15:02:33.785)
+          -f, --from  <arg>              from x days ago (default = 15)
+              --generate-counts          generate listen counts file then stop
+              --nogenerate-counts        run normally
+              --generate-similars        generate similar song file then stop
+              --nogenerate-similars      run normally
+          -g, --growth-rate  <arg>       annual user growth rate (as a fraction of
+                                         current, so 1% => 0.01) (default = 0.0)
+          -k, --kafkaBrokerList  <arg>   kafka broker list
+          -n, --nusers  <arg>            initial number of users (default = 1)
+          -r, --randomseed  <arg>        random seed
+          -s, --start-time  <arg>        start time for data
+                                         (default = 2015-12-30T15:02:33.839)
+              --tag  <arg>               tag applied to each line (for example, A/B test
+                                         group)
+          -t, --to  <arg>                to y days ago (default = 1)
+              --useAvro                  output data as Avro
+              --nouseAvro                output data as JSON
+          -u, --userid  <arg>            first user id (default = 1)
+          -v, --verbose                  verbose output (not implemented yet)
+              --noverbose                silent mode
+              --help                     Show help message
+
+         trailing arguments:
+          output-dir (not required)   Directory for output files
 
 Only the config file is required.
 

bin/eventsim

@@ -1,2 +1,2 @@
 #! /bin/bash
-java -XX:+AggressiveOpts -XX:+UseG1GC -XX:+UseStringDeduplication -Xmx8G -jar target/scala-2.10/eventsim-assembly-1.0.jar $*
+java -XX:+AggressiveOpts -XX:+UseG1GC -XX:+UseStringDeduplication -Xmx8G -jar target/scala-2.11/eventsim-assembly-2.0.jar $*

build.sbt

@@ -1,15 +1,27 @@
 name := "eventsim"
 
-version := "1.0"
+version := "2.0"
 
-libraryDependencies += "org.apache.commons" % "commons-math3" % "3.5"
+scalaVersion := "2.11.6"
+
+// resolvers += Resolver.url("confluent", url("http://packages.confluent.io/maven"))
+
+libraryDependencies += "org.apache.avro" % "avro" % "1.7.7"
+
+libraryDependencies += "org.apache.commons" % "commons-math3" % "3.6"
 
 libraryDependencies += "de.jollyday" % "jollyday" % "0.5.1"
 
-libraryDependencies += "org.rogach" %% "scallop" % "0.9.5"
+libraryDependencies += "org.rogach" % "scallop_2.11" % "0.9.5"
+
+libraryDependencies += "com.fasterxml.jackson.core" % "jackson-core" % "2.7.0"
+
+libraryDependencies += "com.fasterxml.jackson.core" % "jackson-databind" % "2.7.0"
+
+libraryDependencies += "org.apache.kafka" % "kafka-clients" % "0.9.0.0"
 
-libraryDependencies += "com.fasterxml.jackson.core" % "jackson-core" % "2.6.1"
+libraryDependencies += "org.scala-lang.modules" % "scala-parser-combinators_2.11" % "1.0.4"
 
-libraryDependencies += "com.fasterxml.jackson.core" % "jackson-databind" % "2.6.1"
+// libraryDependencies += "io.confluent" % "kafka-avro-serializer" % "2.0.0"
 
-libraryDependencies += "org.apache.kafka" % "kafka_2.10" % "0.8.2.1"
+seq( sbtavro.SbtAvro.avroSettings : _*)