aboutsummaryrefslogtreecommitdiffstats
path: root/scalding
diff options
context:
space:
mode:
authorBryan Newbold <bnewbold@archive.org>2018-08-22 12:59:17 -0700
committerBryan Newbold <bnewbold@archive.org>2018-08-22 12:59:17 -0700
commit7087760028cb55b8290783fc8c28958108f00f36 (patch)
treec51a9f2a00f5d2da80ca64c05677e96a9c901436 /scalding
parentb628b7026ab8e7abf4beeaaad99d831b49578483 (diff)
downloadsandcrawler-7087760028cb55b8290783fc8c28958108f00f36.tar.gz
sandcrawler-7087760028cb55b8290783fc8c28958108f00f36.zip
add more punctuation characters to slug filter
Diffstat (limited to 'scalding')
-rw-r--r--scalding/src/main/scala/sandcrawler/StringUtilities.scala2
-rw-r--r--scalding/src/test/scala/sandcrawler/ScorableFeaturesTest.scala2
2 files changed, 2 insertions, 2 deletions
diff --git a/scalding/src/main/scala/sandcrawler/StringUtilities.scala b/scalding/src/main/scala/sandcrawler/StringUtilities.scala
index 2745875..e03b60d 100644
--- a/scalding/src/main/scala/sandcrawler/StringUtilities.scala
+++ b/scalding/src/main/scala/sandcrawler/StringUtilities.scala
@@ -36,7 +36,7 @@ object StringUtilities {
// Source: https://stackoverflow.com/a/30076541/631051
def removePunctuation(s: String) : String = {
- s.replaceAll("""[\p{Punct}]""", "")
+ s.replaceAll("""[\p{Punct}’·“”‘’“”«»「」]""", "")
}
// Adapted from: https://stackoverflow.com/a/16018452/631051
diff --git a/scalding/src/test/scala/sandcrawler/ScorableFeaturesTest.scala b/scalding/src/test/scala/sandcrawler/ScorableFeaturesTest.scala
index 9459749..d742384 100644
--- a/scalding/src/test/scala/sandcrawler/ScorableFeaturesTest.scala
+++ b/scalding/src/test/scala/sandcrawler/ScorableFeaturesTest.scala
@@ -49,7 +49,7 @@ class ScorableFeaturesTest extends FlatSpec with Matchers {
}
it should "strip special characters" in {
- titleToSlug(":;!',|\"\'`.#?!-@*/\\=+~%$^{}()[]<>-_") shouldBe Scorable.NoSlug
+ titleToSlug(":;!',|\"\'`.#?!-@*/\\=+~%$^{}()[]<>-_’·“”‘’“”«»「」") shouldBe Scorable.NoSlug
// TODO: titleToSlug("©™₨№…") shouldBe Scorable.NoSlug
// TODO: titleToSlug("πµΣσ") shouldBe Scorable.NoSlug
}