Merge branch 'refactoring' into 'master'

Refactoring to add, use, and test class HBaseBuilder to eliminate duplicated code and facilitate HBaseSource creation

See merge request webgroup/sandcrawler!1

1 files changed, 43 insertions, 0 deletions

diff --git a/scalding/src/test/scala/sandcrawler/HBaseBuilderTest.scala b/scalding/src/test/scala/sandcrawler/HBaseBuilderTest.scala
new file mode 100644
index 0000000..b45751d
--- /dev/null
+++ b/scalding/src/test/scala/sandcrawler/HBaseBuilderTest.scala
@@ -0,0 +1,43 @@
+package example
+
+import cascading.tuple.Fields
+import org.scalatest._
+import sandcrawler.HBaseBuilder
+
+class HBaseBuilderTest extends FlatSpec with Matchers {
+  "parseColSpec()" should "work on legal nontrivial input" in {
+    val (fams, fields) = HBaseBuilder.parseColSpec(List("file:size", "file:cdx", "match0:status"))
+    fams should have length 2
+    fields should have length 2
+    val fileIndex = fams.indexOf("file")
+    fileIndex should not be -1
+    fields(fileIndex) should be (new Fields("size", "cdx"))
+    val match0Index = fams.indexOf("match0")
+    match0Index should not be -1
+    fields(match0Index) should be (new Fields("status"))
+  }
+
+  it should "work on empty input" in {
+    val (fams, fields) = HBaseBuilder.parseColSpec(List())
+    fams should have length 0
+    fields should have length 0
+  }
+
+  it should "throw IllegalArgumentException on malformed input" in {
+    a [IllegalArgumentException] should be thrownBy {
+      HBaseBuilder.parseColSpec(List("file_size"))
+    }
+  }
+
+  it should "throw IllegalArgumentException on nonexistent family" in {
+    a [IllegalArgumentException] should be thrownBy {
+      HBaseBuilder.parseColSpec(List("foo:bar"))
+    }
+  }
+
+  it should "throw IllegalArgumentException on nonexistent column" in {
+    a [IllegalArgumentException] should be thrownBy {
+      HBaseBuilder.parseColSpec(List("file:bar"))
+    }
+  }
+}