aboutsummaryrefslogtreecommitdiffstats
path: root/scalding/src
diff options
context:
space:
mode:
Diffstat (limited to 'scalding/src')
-rw-r--r--scalding/src/main/scala/sandcrawler/HBaseCountJob.scala6
-rw-r--r--scalding/src/main/scala/sandcrawler/HBaseRowCountJob.scala6
-rw-r--r--scalding/src/test/scala/sandcrawler/HBaseMimeCountTest.scala8
-rw-r--r--scalding/src/test/scala/sandcrawler/HBaseRowCountTest.scala8
-rw-r--r--scalding/src/test/scala/sandcrawler/HBaseStatusCountTest.scala8
5 files changed, 18 insertions, 18 deletions
diff --git a/scalding/src/main/scala/sandcrawler/HBaseCountJob.scala b/scalding/src/main/scala/sandcrawler/HBaseCountJob.scala
index 53ce325..863c16f 100644
--- a/scalding/src/main/scala/sandcrawler/HBaseCountJob.scala
+++ b/scalding/src/main/scala/sandcrawler/HBaseCountJob.scala
@@ -24,9 +24,9 @@ class HBaseCountJob(args: Args, colSpec: String) extends JobBase(args) with HBas
}
object HBaseCountJob {
- def getHBaseSource(hbase_table: String, zookeeper_hosts: String, colSpec: String) = HBaseBuilder.build(
- hbase_table, // HBase Table Name
- zookeeper_hosts, // HBase Zookeeper server (to get runtime config info; can be array?)
+ def getHBaseSource(hbaseTable: String, zookeeperHosts: String, colSpec: String) = HBaseBuilder.build(
+ hbaseTable, // HBase Table Name
+ zookeeperHosts, // HBase Zookeeper server (to get runtime config info; can be array?)
List(colSpec),
SourceMode.SCAN_ALL)
}
diff --git a/scalding/src/main/scala/sandcrawler/HBaseRowCountJob.scala b/scalding/src/main/scala/sandcrawler/HBaseRowCountJob.scala
index 7f5c84a..334b7a6 100644
--- a/scalding/src/main/scala/sandcrawler/HBaseRowCountJob.scala
+++ b/scalding/src/main/scala/sandcrawler/HBaseRowCountJob.scala
@@ -23,10 +23,10 @@ class HBaseRowCountJob(args: Args) extends JobBase(args) with HBasePipeConversio
object HBaseRowCountJob {
// eg, "wbgrp-journal-extract-0-qa", "mtrcs-zk1.us.archive.org:2181"
- def getHBaseSource(hbase_table: String, zookeeper_hosts: String) : HBaseSource = {
+ def getHBaseSource(hbaseTable: String, zookeeperHosts: String) : HBaseSource = {
return HBaseBuilder.build(
- hbase_table,
- zookeeper_hosts,
+ hbaseTable,
+ zookeeperHosts,
List("file:size"),
SourceMode.SCAN_ALL)
}
diff --git a/scalding/src/test/scala/sandcrawler/HBaseMimeCountTest.scala b/scalding/src/test/scala/sandcrawler/HBaseMimeCountTest.scala
index fcd9087..fde2290 100644
--- a/scalding/src/test/scala/sandcrawler/HBaseMimeCountTest.scala
+++ b/scalding/src/test/scala/sandcrawler/HBaseMimeCountTest.scala
@@ -16,7 +16,7 @@ import scala._
class HBaseMimeCountTest extends FunSpec with TupleConversions {
val output = "/tmp/testOutput"
- val (test_table, test_host) = ("test-table", "dummy-host:2181")
+ val (testTable, testHost) = ("test-table", "dummy-host:2181")
val log = LoggerFactory.getLogger(this.getClass.getName)
@@ -41,10 +41,10 @@ class HBaseMimeCountTest extends FunSpec with TupleConversions {
.arg("test", "")
.arg("app.conf.path", "app.conf")
.arg("output", output)
- .arg("hbase-table", test_table)
- .arg("zookeeper-hosts", test_host)
+ .arg("hbase-table", testTable)
+ .arg("zookeeper-hosts", testHost)
.arg("debug", "true")
- .source[Tuple](HBaseCountJob.getHBaseSource(test_table, test_host, "file:mime"),
+ .source[Tuple](HBaseCountJob.getHBaseSource(testTable, testHost, "file:mime"),
sampleData.map(l => new Tuple(l.map(s => {new ImmutableBytesWritable(Bytes.toBytes(s))}):_*)))
.sink[Tuple](Tsv(output)) {
outputBuffer =>
diff --git a/scalding/src/test/scala/sandcrawler/HBaseRowCountTest.scala b/scalding/src/test/scala/sandcrawler/HBaseRowCountTest.scala
index a5c35a2..3424a36 100644
--- a/scalding/src/test/scala/sandcrawler/HBaseRowCountTest.scala
+++ b/scalding/src/test/scala/sandcrawler/HBaseRowCountTest.scala
@@ -19,7 +19,7 @@ import scala._
class HBaseRowCountTest extends FunSpec with TupleConversions {
val output = "/tmp/testOutput"
- val (test_table, test_host) = ("test-table", "dummy-host:2181")
+ val (testTable, testHost) = ("test-table", "dummy-host:2181")
val log = LoggerFactory.getLogger(this.getClass.getName)
@@ -38,10 +38,10 @@ class HBaseRowCountTest extends FunSpec with TupleConversions {
.arg("test", "")
.arg("app.conf.path", "app.conf")
.arg("output", output)
- .arg("hbase-table", test_table)
- .arg("zookeeper-hosts", test_host)
+ .arg("hbase-table", testTable)
+ .arg("zookeeper-hosts", testHost)
.arg("debug", "true")
- .source[Tuple](HBaseRowCountJob.getHBaseSource(test_table, test_host),
+ .source[Tuple](HBaseRowCountJob.getHBaseSource(testTable, testHost),
sampleData.map(l => new Tuple(l.map(s => {new ImmutableBytesWritable(Bytes.toBytes(s))}):_*)))
.sink[Tuple](Tsv(output)) {
outputBuffer =>
diff --git a/scalding/src/test/scala/sandcrawler/HBaseStatusCountTest.scala b/scalding/src/test/scala/sandcrawler/HBaseStatusCountTest.scala
index 1877d6f..35a4177 100644
--- a/scalding/src/test/scala/sandcrawler/HBaseStatusCountTest.scala
+++ b/scalding/src/test/scala/sandcrawler/HBaseStatusCountTest.scala
@@ -16,7 +16,7 @@ import scala._
class HBaseStatusCountTest extends FunSpec with TupleConversions {
val output = "/tmp/testOutput"
- val (test_table, test_host) = ("test-table", "dummy-host:2181")
+ val (testTable, testHost) = ("test-table", "dummy-host:2181")
val log = LoggerFactory.getLogger(this.getClass.getName)
@@ -41,10 +41,10 @@ class HBaseStatusCountTest extends FunSpec with TupleConversions {
.arg("test", "")
.arg("app.conf.path", "app.conf")
.arg("output", output)
- .arg("hbase-table", test_table)
- .arg("zookeeper-hosts", test_host)
+ .arg("hbase-table", testTable)
+ .arg("zookeeper-hosts", testHost)
.arg("debug", "true")
- .source[Tuple](HBaseCountJob.getHBaseSource(test_table, test_host, "grobid0:status"),
+ .source[Tuple](HBaseCountJob.getHBaseSource(testTable, testHost, "grobid0:status"),
sampleData.map(l => new Tuple(l.map(s => {new ImmutableBytesWritable(Bytes.toBytes(s))}):_*)))
.sink[Tuple](Tsv(output)) {
outputBuffer =>