aboutsummaryrefslogtreecommitdiffstats
path: root/scalding/src/main/scala
diff options
context:
space:
mode:
Diffstat (limited to 'scalding/src/main/scala')
-rw-r--r--scalding/src/main/scala/sandcrawler/HBaseBuilder.scala46
-rw-r--r--scalding/src/main/scala/sandcrawler/HBaseRowCountJob.scala18
2 files changed, 55 insertions, 9 deletions
diff --git a/scalding/src/main/scala/sandcrawler/HBaseBuilder.scala b/scalding/src/main/scala/sandcrawler/HBaseBuilder.scala
new file mode 100644
index 0000000..87cc0cb
--- /dev/null
+++ b/scalding/src/main/scala/sandcrawler/HBaseBuilder.scala
@@ -0,0 +1,46 @@
+package sandcrawler
+
+import cascading.tuple.Fields
+import parallelai.spyglass.hbase.HBaseConstants.SourceMode
+import parallelai.spyglass.hbase.HBaseSource
+import scala._
+
+object HBaseBuilder {
+ // map from column families to column names
+ val schema = Map("f" -> List("c"),
+ "file" -> List("size", "mime", "cdx"),
+ "grobid0" -> List("status_code", "quality", "status", "tei_xml", "tei_json", "metadata"),
+ "match0" -> List("status", "doi", "info"))
+ val inverseSchema = {for ((k,vs) <- schema; v <-vs) yield (k + ":" + v)}.toList
+
+ // The argument should be a comma-separated list of family:column, such as "f:c, file:size".
+ @throws(classOf[IllegalArgumentException])
+ def parseColSpec(colSpecs: List[String]) : (List[String], List[Fields]) = {
+ // Verify that all column specifiers are legal.
+ for (colSpec <- colSpecs) {
+ if (!(inverseSchema contains colSpec)) {
+ throw new IllegalArgumentException("No such column: " + colSpec)
+ }
+ val pair = colSpec split(":")
+ if (colSpec.split(":").length != 2) {
+ throw new IllegalArgumentException("Bad column specifier " + colSpec +
+ " (specifiers should be family:name)")
+ }
+ }
+
+ // Produce and return a tuple containing:
+ // 1. A list of column families.
+ // 2. A corresponding list of Fields, each containing column names.
+ val groupMap: Map[String, List[String]] = colSpecs.groupBy(c => (c split ":")(0))
+ val families = groupMap.keys.toList
+ val groupedColNames : List[List[String]] = families map {fam => {
+ val cols = {groupMap(fam).map(v => v.split(":")(1))}
+ cols}}
+ (families, groupedColNames.map({fields => new Fields(fields : _*)}))
+ }
+
+ def build(table: String, server: String, colSpec: List[String], sourceMode: SourceMode, keyList: List[String] = List("key")) = {
+ val (families, fields) = parseColSpec(colSpec)
+ new HBaseSource(table, server, new Fields("key"), families, fields, sourceMode = sourceMode, keyList = keyList)
+ }
+}
diff --git a/scalding/src/main/scala/sandcrawler/HBaseRowCountJob.scala b/scalding/src/main/scala/sandcrawler/HBaseRowCountJob.scala
index d47fe60..9d074ab 100644
--- a/scalding/src/main/scala/sandcrawler/HBaseRowCountJob.scala
+++ b/scalding/src/main/scala/sandcrawler/HBaseRowCountJob.scala
@@ -16,17 +16,17 @@ class HBaseRowCountJob(args: Args) extends JobBase(args) with HBasePipeConversio
val output = args("output")
- val hbs = new HBaseSource(
- //"table_name",
- //"quorum_name:2181",
- "wbgrp-journal-extract-0-qa", // HBase Table Name
- "mtrcs-zk1.us.archive.org:2181", // HBase Zookeeper server (to get runtime config info; can be array?)
- new Fields("key"),
- List("file"),
- List(new Fields("size", "mimetype")),
- sourceMode = SourceMode.SCAN_ALL)
+ HBaseRowCountJob.getHBaseSource
.read
.debug
.groupAll { _.size('count) }
.write(Tsv(output))
}
+
+object HBaseRowCountJob {
+ def getHBaseSource = HBaseBuilder.build(
+ "wbgrp-journal-extract-0-qa", // HBase Table Name
+ "mtrcs-zk1.us.archive.org:2181", // HBase Zookeeper server (to get runtime config info; can be array?)
+ List("file:size", "file:mime"),
+ SourceMode.SCAN_ALL)
+}