aboutsummaryrefslogtreecommitdiffstats
path: root/scalding
diff options
context:
space:
mode:
authorBryan Newbold <bnewbold@archive.org>2018-07-17 18:53:37 -0700
committerEllen Spertus <ellen.spertus@gmail.com>2018-07-19 15:50:26 -0700
commitf88cfd9dc54ffa3d2e8067ebc368ca59f75a40c8 (patch)
tree77d319ac8c0b006980706422bb4033e678445137 /scalding
parentd2928196f994dfe3bac79c84c8b9ca54cc3fe81e (diff)
downloadsandcrawler-f88cfd9dc54ffa3d2e8067ebc368ca59f75a40c8.tar.gz
sandcrawler-f88cfd9dc54ffa3d2e8067ebc368ca59f75a40c8.zip
more complete CdxBackfillJob (untested)
Diffstat (limited to 'scalding')
-rw-r--r--scalding/src/test/scala/sandcrawler/CdxBackfillJob.scala70
1 files changed, 68 insertions, 2 deletions
diff --git a/scalding/src/test/scala/sandcrawler/CdxBackfillJob.scala b/scalding/src/test/scala/sandcrawler/CdxBackfillJob.scala
index a442d79..de94494 100644
--- a/scalding/src/test/scala/sandcrawler/CdxBackfillJob.scala
+++ b/scalding/src/test/scala/sandcrawler/CdxBackfillJob.scala
@@ -1,8 +1,17 @@
package sandcrawler
-import cascading.tuple.Fields
import org.scalatest._
+import cascading.tuple.{Tuple, Fields}
+import com.twitter.scalding.{JobTest, Tsv, TypedTsv, TupleConversions, TextLine}
+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.HBaseSource
+import parallelai.spyglass.hbase.HBaseConstants.SourceMode
class CdxBackfillTest extends FlatSpec with Matchers {
@@ -37,7 +46,7 @@ class CdxBackfillTest extends FlatSpec with Matchers {
"#edu,upenn,ldc)/sites/www.ldc.upenn.edu/files/medar2009-large-arabic-broadcast-collection.pdf 20170828233154 https://www.ldc.upenn.edu/sites/www.ldc.upenn.edu/files/medar2009-large-arabic-broadcast-collection.pdf application/pdf 200 WL3FEA62TEU4F52Y5DOVQ62VET4QJW7G - - 210251 931661233 SEMSCHOLAR-PDF-CRAWL-2017-08-04-20170828231135742-00000-00009-wbgrp-svc284/SEMSCHOLAR-PDF-CRAWL-2017-08-04-20170828232253025-00005-3480~wbgrp-svc284.us.archive.org~8443.warc.gz"))
// missing two fields
assert(false === isCdxLine(
- "#edu,upenn,ldc)/sites/www.ldc.upenn.edu/files/medar2009-large-arabic-broadcast-collection.pdf 20170828233154 https://www.ldc.upenn.edu/sites/www.ldc.upenn.edu/files/medar2009-large-arabic-broadcast-collection.pdf application/pdf 200 WL3FEA62TEU4F52Y5DOVQ62VET4QJW7G 210251 931661233 SEMSCHOLAR-PDF-CRAWL-2017-08-04-20170828231135742-00000-00009-wbgrp-svc284/SEMSCHOLAR-PDF-CRAWL-2017-08-04-20170828232253025-00005-3480~wbgrp-svc284.us.archive.org~8443.warc.gz"))
+ "edu,upenn,ldc)/sites/www.ldc.upenn.edu/files/medar2009-large-arabic-broadcast-collection.pdf 20170828233154 https://www.ldc.upenn.edu/sites/www.ldc.upenn.edu/files/medar2009-large-arabic-broadcast-collection.pdf application/pdf 200 WL3FEA62TEU4F52Y5DOVQ62VET4QJW7G 210251 931661233 SEMSCHOLAR-PDF-CRAWL-2017-08-04-20170828231135742-00000-00009-wbgrp-svc284/SEMSCHOLAR-PDF-CRAWL-2017-08-04-20170828232253025-00005-3480~wbgrp-svc284.us.archive.org~8443.warc.gz"))
// extra field
assert(false === isCdxLine(
"edu,upenn,ldc)/sites/www.ldc.upenn.edu/files/medar2009-large-arabic-broadcast-collection.pdf 20170828233154 https://www.ldc.upenn.edu/sites/www.ldc.upenn.edu/files/medar2009-large-arabic-broadcast-collection.pdf application/pdf 200 WL3FEA62TEU4F52Y5DOVQ62VET4QJW7G - - 210251 931661233 SEMSCHOLAR-PDF-CRAWL-2017-08-04-20170828231135742-00000-00009-wbgrp-svc284/SEMSCHOLAR-PDF-CRAWL-2017-08-04-20170828232253025-00005-3480~wbgrp-svc284.us.archive.org~8443.warc.gz -"))
@@ -49,3 +58,60 @@ class CdxBackfillTest extends FlatSpec with Matchers {
}
}
+
+@RunWith(classOf[JUnitRunner])
+class CdxBackfillJobTest extends FunSpec with TupleConversions {
+
+ val output = "/tmp/testOutput"
+ val (testTable, testHost, testCdxFile) = ("test-table", "dummy-host:2181", "test_file.cdx")
+
+ val log = LoggerFactory.getLogger(this.getClass.getName)
+
+ val dummySizeBytes = Bytes.toBytes(100)
+
+ val sampleData = List(
+ List(Bytes.toBytes("sha1:K2DKSSVTXWPRMFDTWSTCQW3RVWRIOV3Q"), dummySizeBytes),
+ List(Bytes.toBytes("sha1:C3YNNEGH5WAG5ZAAXWAEBNXJWT6CZ3WU"), dummySizeBytes),
+ List(Bytes.toBytes("sha1:SDKUVHC3YNNEGH5WAG5ZAAXWAEBNX4WT"), dummySizeBytes),
+ List(Bytes.toBytes("sha1:095893C3YNNEGH5WAG5ZAAXWAEBNXJWT"), dummySizeBytes)
+ )
+ val sampleCdxLines = List(
+ // clean line
+ "edu,upenn,ldc)/sites/www.ldc.upenn.edu/files/medar2009-large-arabic-broadcast-collection.pdf 20170828233154 https://www.ldc.upenn.edu/sites/www.ldc.upenn.edu/files/medar2009-large-arabic-broadcast-collection.pdf application/pdf 200 WL3FEA62TEU4F52Y5DOVQ62VET4QJW7G - - 210251 931661233 SEMSCHOLAR-PDF-CRAWL-2017-08-04-20170828231135742-00000-00009-wbgrp-svc284/SEMSCHOLAR-PDF-CRAWL-2017-08-04-20170828232253025-00005-3480~wbgrp-svc284.us.archive.org~8443.warc.gz",
+ // has existing SHA1
+ "edu,upenn,ldc)/sites/www.ldc.upenn.edu/files/medar2009-large-arabic-broadcast-collection.pdf 20170828233154 https://www.ldc.upenn.edu/sites/www.ldc.upenn.edu/files/medar2009-large-arabic-broadcast-collection.pdf application/pdf 200 C3YNNEGH5WAG5ZAAXWAEBNXJWT6CZ3WU - - 210251 931661233 SEMSCHOLAR-PDF-CRAWL-2017-08-04-20170828231135742-00000-00009-wbgrp-svc284/SEMSCHOLAR-PDF-CRAWL-2017-08-04-20170828232253025-00005-3480~wbgrp-svc284.us.archive.org~8443.warc.gz",
+ // HTTP status code
+ "edu,upenn,ldc)/sites/www.ldc.upenn.edu/files/medar2009-large-arabic-broadcast-collection.pdf 20170828233154 https://www.ldc.upenn.edu/sites/www.ldc.upenn.edu/files/medar2009-large-arabic-broadcast-collection.pdf application/pdf 301 WL3FEA62TEU4F52Y5DOVQ62VET4QJW7G - - 210251 931661233 SEMSCHOLAR-PDF-CRAWL-2017-08-04-20170828231135742-00000-00009-wbgrp-svc284/SEMSCHOLAR-PDF-CRAWL-2017-08-04-20170828232253025-00005-3480~wbgrp-svc284.us.archive.org~8443.warc.gz",
+ // not CDX (prefixed with hash)
+ "#edu,upenn,ldc)/sites/www.ldc.upenn.edu/files/medar2009-large-arabic-broadcast-collection.pdf 20170828233154 https://www.ldc.upenn.edu/sites/www.ldc.upenn.edu/files/medar2009-large-arabic-broadcast-collection.pdf application/pdf 200 WL3FEA62TEU4F52Y5DOVQ62VET4QJW7G - - 210251 931661233 SEMSCHOLAR-PDF-CRAWL-2017-08-04-20170828231135742-00000-00009-wbgrp-svc284/SEMSCHOLAR-PDF-CRAWL-2017-08-04-20170828232253025-00005-3480~wbgrp-svc284.us.archive.org~8443.warc.gz"
+ )
+
+ JobTest("sandcrawler.CdxBackfillJob")
+ .arg("test", "")
+ .arg("app.conf.path", "app.conf")
+ .arg("output", output)
+ .arg("hbase-table", testTable)
+ .arg("zookeeper-hosts", testHost)
+ .arg("cdx-input-path", testCdxFile)
+ .arg("debug", "true")
+ .source[Tuple](CdxBackfillJob.getHBaseSource(testTable, testHost),
+ sampleData.map(l => new Tuple(l.map(s => {new ImmutableBytesWritable(s)}):_*)))
+ .source[String](TextLine(testCdxFile), sampleCdxLines)
+ .sink[Tuple](CdxBackfillJob.getHBaseSink(testTable, testHost)) {
+ outputBuffer =>
+
+ it("should return a 1-element list (after join).") {
+ // XXX:
+ assert(outputBuffer.size === 1)
+ }
+
+ // Convert List[Tuple] to Map[Long, Long].
+ val counts = outputBuffer.map(t => (t.getLong(0), t.getLong(1))).toMap
+ it("should have the appropriate number of each status type") {
+ // XXX:
+ assert(counts(1) == 3)
+ }
+ }
+ .run
+ .finish
+}