diff options
author | Ellen Spertus <ellen.spertus@gmail.com> | 2018-07-25 20:33:38 -0700 |
---|---|---|
committer | Ellen Spertus <ellen.spertus@gmail.com> | 2018-07-25 20:33:38 -0700 |
commit | 0f0152189cf6df0f4b56d92149a60e902eb20be6 (patch) | |
tree | 8d22242ed5ae38caa086be16dee3a07a7fc44989 /scalding/src/test | |
parent | 4b63570522e5ebbc73980356372c39ce7547ba68 (diff) | |
download | sandcrawler-0f0152189cf6df0f4b56d92149a60e902eb20be6.tar.gz sandcrawler-0f0152189cf6df0f4b56d92149a60e902eb20be6.zip |
Fixed bug with reading from TextLine. (Thanks, Bryan\!) Still had to comment out some tests.
Diffstat (limited to 'scalding/src/test')
-rw-r--r-- | scalding/src/test/scala/sandcrawler/HBaseCrossrefScoreTest.scala | 9 |
1 files changed, 4 insertions, 5 deletions
diff --git a/scalding/src/test/scala/sandcrawler/HBaseCrossrefScoreTest.scala b/scalding/src/test/scala/sandcrawler/HBaseCrossrefScoreTest.scala index dc96003..96c7770 100644 --- a/scalding/src/test/scala/sandcrawler/HBaseCrossrefScoreTest.scala +++ b/scalding/src/test/scala/sandcrawler/HBaseCrossrefScoreTest.scala @@ -178,18 +178,17 @@ class HBaseCrossrefScoreTest extends FlatSpec with Matchers { .arg("debug", "true") .source[Tuple](HBaseCrossrefScore.getHBaseSource(testTable, testHost), grobidSampleData.map(l => new Tuple(l.map(s => {new ImmutableBytesWritable(s)}):_*))) - .source(TextLine(input), List(( + .source(TextLine(input), List( CrossrefString.replace("<<TITLE>>", "Title 1: TNG").replace("<<DOI>>", "DOI-0"), CrossrefString.replace("<<TITLE>>", "Title 1: TNG").replace("<<DOI>>", "DOI-0.5"), CrossrefString.replace("<<TITLE>>", "Title 1: TNG").replace("<<DOI>>", "DOI-0.75"), - CrossrefString.replace("<<TITLE>>", "Title 2: Rebooted").replace("<<DOI>>", "DOI-1")))) - .sink[(String, String, String, String, String, - String)](TypedTsv[(String, String, String, String, String, String)](output)) { + CrossrefString.replace("<<TITLE>>", "Title 2: Rebooted").replace("<<DOI>>", "DOI-1"))) + .sink[(String, String, String)](TypedTsv[(String, String, String)](output)) { outputBuffer => - /* it should "return a 3-element list" in { outputBuffer should have length 3 } + /* it should "return the right first entry" in { val (slug, slug0, slug1, sha1, grobidJson, crossrefJson) = outputBuffer(0) slug shouldBe "title1" |