added smsifier with smsspec

build.sbt

@@ -2,7 +2,7 @@ organization := "hr.element.etb"
 
 name         := "etb-slug"
 
-version      := "0.0.1"
+version      := "0.0.2"
 
 // ### Build settings ###
 
@@ -33,4 +33,4 @@ publishArtifact in (Compile, packageDoc) := false
 
 // ### Misc ###
 
-initialCommands := "import hr.element.doit.slug._"
+//initialCommands := "import hr.element.doit.slug._"

src/main/scala/hr/element/etb/slug/SMSify.scala

@@ -0,0 +1,49 @@
+package hr.element.etb
+package slug
+
+import com.ibm.icu.text.Transliterator
+import java.util.regex.Pattern
+
+
+
+object SMSifier {
+  protected val SMSUnsafeChars = """[^!-/0-9:-@A-Za-z£¥\n\s]+"""
+
+  private val default =
+    new SMSifier(TransliterateRules.latinToASCII, "-")
+
+  def apply(text: String) = default(text)
+}
+
+
+case class SMSifier (transRules: String, replacement: String) {
+  private val Trans = Transliterator.getInstance(transRules)
+
+  private val SMSUnsafeReplacePattern =
+    "(%s|%s)+" format(SMSifier.SMSUnsafeChars, Pattern.quote(replacement)) r
+
+  private val SMSUnsafeTrimPattern =
+    "^%s|%1$s$" format(SMSifier.SMSUnsafeChars) r
+
+  private val SMSWhiteSpace =
+    """\s+"""r
+
+  // convert latin letters to ASCII (ex. đ->d)
+  protected val transliterate =
+    Trans.transliterate(_: String)
+
+  // trim all SMS non-safe characters from the beginning and the end
+  protected val trim =
+    SMSUnsafeTrimPattern.replaceAllIn(_: String, "")
+
+  protected val trimWhiteSpaceToOne =
+    SMSWhiteSpace.replaceAllIn(_: String, " ")
+
+  // replace all SMS non-safe characters
+  protected val sanitize =
+    SMSUnsafeReplacePattern.replaceAllIn(_: String, replacement)
+
+  def apply(text: String) = {
+    (transliterate andThen trim andThen trimWhiteSpaceToOne andThen sanitize)(text)
+  }
+}

src/test/scala/hr/element/etb/slug/SMSSpec.scala

@@ -0,0 +1,42 @@
+package hr.element.etb.slug
+package test
+
+import org.scalatest._
+import org.scalatest.matchers._
+
+
+
+class SMSSpec extends FeatureSpec with GivenWhenThen with MustMatchers{
+
+  feature("SMS URL sanitization"){
+    info("SMS must convert every evil character combination into a pretty URL-safe string")
+
+    scenario("SMS unsafe character removal"){
+      val in = "aZ{}[]! a";
+      given ("a string containing: %s" format in)
+      val res = "aZ-! a"
+      then ("it should return a string: %s" format res)
+      val out = SMSifier(in)
+      out must equal (res)
+    }
+
+    scenario("SMS unsafe character trimming"){
+      val in = "!aćsčš ";
+      given ("a string containing: %s" format in)
+      val res = "!acscs "
+      then ("it should return a string: %s" format res)
+      val out = SMSifier(in)
+      out must equal (res)
+    }
+
+    scenario("SMS whitespace trimming to one character"){
+      val in = "Helooooooooooooo   oooo";
+      given ("a string containing: %s" format in)
+      val res = "Helooooooooooooo oooo";
+      then ("it should return a string: %s" format res)
+      val out = SMSifier(in)
+      out must equal (res)
+    }
+  }
+
+}

src/test/scala/hr/element/etb/slug/SluggifySpec.scala

@@ -34,5 +34,14 @@ class SluggifySpec extends FeatureSpec with GivenWhenThen with MustMatchers {
       val out = Sluggifier(in)
       out must equal (res)
     }
+
+    scenario("URL unsafe character trimming2"){
+      val in = """~ $Hajduk ! Is better!!!Slayer rules$ ~""" mkString;
+      given ("an URL unsafe string: %s" format in)
+      val res = "hajduk-is-better-slayer-rules"
+      then ("the resulting slug must be trimmed: %s" format res)
+      val out = Sluggifier(in)
+      out must equal (res)
+    }
   }
 }