From f8a0c99b270ebcd6e239c6e26190cf7200389ced Mon Sep 17 00:00:00 2001 From: Bryan Newbold Date: Mon, 27 Aug 2018 15:04:04 -0700 Subject: more special characters to strip --- scalding/src/main/scala/sandcrawler/StringUtilities.scala | 2 +- scalding/src/test/scala/sandcrawler/ScorableFeaturesTest.scala | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) (limited to 'scalding') diff --git a/scalding/src/main/scala/sandcrawler/StringUtilities.scala b/scalding/src/main/scala/sandcrawler/StringUtilities.scala index e03b60d..9150ced 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}’·“”‘’“”«»「」¿–±§ʿ]""", "") } // 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 474f69a..450c169 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(":;!',|\"\'`.#?!-@*/\\=+~%$^{}()[]<>-_’·“”‘’“”«»「」") shouldBe Scorable.NoSlug + titleToSlug(":;!',|\"\'`.#?!-@*/\\=+~%$^{}()[]<>-_’·“”‘’“”«»「」¿–±§ʿ") shouldBe Scorable.NoSlug // TODO: titleToSlug("©™₨№…") shouldBe Scorable.NoSlug // TODO: titleToSlug("πµΣσ") shouldBe Scorable.NoSlug } -- cgit v1.2.3