2 files changed, 139 insertions, 0 deletions
diff --git a/scalding/src/main/scala/sandcrawler/DumpUnGrobidedJob.scala b/scalding/src/main/scala/sandcrawler/DumpUnGrobidedJob.scala
new file mode 100644
index 0000000..7fd3ce0
--- /dev/null
+++ b/scalding/src/main/scala/sandcrawler/DumpUnGrobidedJob.scala
@@ -0,0 +1,67 @@
+package sandcrawler
+
+import java.util.Properties
+
+import cascading.property.AppProps
+import cascading.tuple.Fields
+import com.twitter.scalding._
+import com.twitter.scalding.typed.TDsl._
+import parallelai.spyglass.base.JobBase
+import parallelai.spyglass.hbase.HBaseConstants.SourceMode
+import parallelai.spyglass.hbase.HBasePipeConversions
+import parallelai.spyglass.hbase.HBaseSource
+
+// Filters for HBase rows which have not had GROBID run on them, but do have
+// full CDX metadata, and dumps to a TSV for later extraction by the
+// "extraction-ungrobided" job.
+//
+// Does the same horrible join thing that DumpUnGrobidedJob does.
+class DumpUnGrobidedJob(args: Args) extends JobBase(args) with HBasePipeConversions {
+
+  val output = args("output")
+
+  val allKeys : TypedPipe[(String,String,String,String)] = DumpUnGrobidedJob.getHBaseKeySource(
+    args("hbase-table"),
+    args("zookeeper-hosts"))
+    .read
+    .fromBytesWritable('key, 'c, 'mime, 'cdx)
+    .toTypedPipe[(String,String,String,String)]('key, 'c, 'mime, 'cdx)
+
+  val existingKeys : TypedPipe[(String,Boolean)] = DumpUnGrobidedJob.getHBaseColSource(
+    args("hbase-table"),
+    args("zookeeper-hosts"))
+    .read
+    .fromBytesWritable('key)
+    .toTypedPipe[String]('key)
+    .map{ key => (key, true) }
+
+  val missingKeys : TypedPipe[(String,String,String,String)] = allKeys
+    .groupBy(_._1)
+    .leftJoin(existingKeys.groupBy(_._1))
+    .toTypedPipe
+    .collect { case (key, ((_, c, mime, cdx), None)) => (key, c, mime, cdx) }
+
+  missingKeys
+    .write(TypedTsv[(String,String,String,String)](output))
+
+}
+
+object DumpUnGrobidedJob {
+
+  // eg, "wbgrp-journal-extract-0-qa",7 "mtrcs-zk1.us.archive.org:2181"
+  def getHBaseColSource(hbaseTable: String, zookeeperHosts: String) : HBaseSource = {
+    HBaseBuilder.build(
+      hbaseTable,
+      zookeeperHosts,
+      List("grobid0:status_code"),
+      SourceMode.SCAN_ALL)
+  }
+
+  def getHBaseKeySource(hbaseTable: String, zookeeperHosts: String) : HBaseSource = {
+    HBaseBuilder.build(
+      hbaseTable,
+      zookeeperHosts,
+      List("f:c", "file:mime", "file:cdx"),
+      SourceMode.SCAN_ALL)
+  }
+}
diff --git a/scalding/src/test/scala/sandcrawler/DumpUnGrobidedJobTest.scala b/scalding/src/test/scala/sandcrawler/DumpUnGrobidedJobTest.scala
new file mode 100644
index 0000000..8dda5c8
--- /dev/null
+++ b/scalding/src/test/scala/sandcrawler/DumpUnGrobidedJobTest.scala
@@ -0,0 +1,72 @@
+package sandcrawler
+
+import cascading.tuple.Fields
+import cascading.tuple.Tuple
+import com.twitter.scalding.JobTest
+import com.twitter.scalding.Tsv
+import com.twitter.scalding.TupleConversions
+import com.twitter.scalding.TypedTsv
+import org.apache.hadoop.hbase.io.ImmutableBytesWritable
+import org.apache.hadoop.hbase.util.Bytes
+import org.junit.runner.RunWith
+import org.scalatest.FunSpec
+import org.scalatest.junit.JUnitRunner
+import org.slf4j.LoggerFactory
+import parallelai.spyglass.hbase.HBaseConstants.SourceMode
+import parallelai.spyglass.hbase.HBaseSource
+import scala._
+
+@RunWith(classOf[JUnitRunner])
+class DumpUnGrobidedJobTest extends FunSpec with TupleConversions {
+
+  val output = "/tmp/testOutput"
+  val (testTable, testHost) = ("test-table", "dummy-host:2181")
+
+  val log = LoggerFactory.getLogger(this.getClass.getName)
+
+  val statusCode: Long = 200
+  val statusBytes = Bytes.toBytes(statusCode)
+
+  val sampleDataGrobid : List[List[Array[Byte]]] = List(
+    ("sha1:SDKUVHC3YNNEGH5WAG5ZAAXWAEBNX4WT", statusBytes),
+    ("sha1:35985C3YNNEGH5WAG5ZAAXWAEBNXJW56", statusBytes),
+    ("sha1:885C3YNNEGH5WAG5ZAAXWA8BNXJWT6CZ", statusBytes),
+    ("sha1:00904C3YNNEGH5WAG5ZA9XWAEBNXJWT6", statusBytes),
+    ("sha1:249C3YNNEGH5WAG5ZAAXWAEBNXJWT6CZ", statusBytes),
+    ("sha1:095893C3YNNEGH5WAG5ZAAXWAEBNXJWT", statusBytes))
+      .map(pair => List(Bytes.toBytes(pair._1), pair._2))
+
+  val sampleDataFile : List[List[Array[Byte]]] = List(
+    ("sha1:K2DKSSVTXWPRMFDTWSTCQW3RVWRIOV3Q", """{c-json-data}""", "application/pdf", """{cdx-json-data}"""),
+    ("sha1:C3YNNEGH5WAG5ZAAXWAEBNXJWT6CZ3WU", """{c-json-data}""", "application/pdf", """{cdx-json-data}"""),
+    ("sha1:SDKUVHC3YNNEGH5WAG5ZAAXWAEBNX4WT", """{c-json-data}""", "application/pdf", """{cdx-json-data}"""),
+    ("sha1:35985C3YNNEGH5WAG5ZAAXWAEBNXJW56", """{c-json-data}""", "application/pdf", """{cdx-json-data}"""),
+    ("sha1:885C3YNNEGH5WAG5ZAAXWA8BNXJWT6CZ", """{c-json-data}""", "application/pdf", """{cdx-json-data}"""),
+    ("sha1:00904C3YNNEGH5WAG5ZA9XWAEBNXJWT6", """{c-json-data}""", "application/pdf", """{cdx-json-data}"""),
+    ("sha1:249C3YNNEGH5WAG5ZAAXWAEBNXJWT6CZ", """{c-json-data}""", "application/pdf", """{cdx-json-data}"""),
+    ("sha1:095893C3YNNEGH5WAG5ZAAXWAEBNXJWT", """{c-json-data}""", "application/pdf", """{cdx-json-data}"""))
+      .map(pair => List(Bytes.toBytes(pair._1),
+                        Bytes.toBytes(pair._2),
+                        Bytes.toBytes(pair._3),
+                        Bytes.toBytes(pair._4)))
+
+  JobTest("sandcrawler.DumpUnGrobidedJob")
+    .arg("test", "")
+    .arg("app.conf.path", "app.conf")
+    .arg("output", output)
+    .arg("hbase-table", testTable)
+    .arg("zookeeper-hosts", testHost)
+    .arg("debug", "true")
+    .source[Tuple](DumpUnGrobidedJob.getHBaseColSource(testTable, testHost),
+      sampleDataGrobid.map(l => new Tuple(l.map(s => {new ImmutableBytesWritable(s)}):_*)))
+    .source[Tuple](DumpUnGrobidedJob.getHBaseKeySource(testTable, testHost),
+      sampleDataFile.map(l => new Tuple(l.map(s => {new ImmutableBytesWritable(s)}):_*)))
+    .sink[Tuple](TypedTsv[(String,String,String,String)](output)) {
+      outputBuffer =>
+      it("should return correct-length list.") {
+        assert(outputBuffer.size === 2)
+      }
+    }
+    .run
+    .finish
+}