Merge pull request #12 from indix/mpn

Basic MPN standardization

Author: Rajesh Muppalla

util-core/src/main/scala/com/indix/utils/core/MPN.scala

@@ -0,0 +1,44 @@
+package com.indix.utils.core
+
+import org.apache.commons.lang3.StringUtils
+import org.apache.commons.lang3.text.WordUtils
+
+object MPN {
+  // Some domain specific keywords known to be invalid
+  val BlackListedMpns = Set("unknown", "none", "does not apply", "non applicable", "na", "n/a", "various")
+
+  val StopChars = Set(' ', '-', '_', '.', '/')
+  val TerminateChars = Set(',', '"', '*', '%')
+
+  val MaxLen = 50
+  val MinLen = 3
+
+  def isValidIdentifier(input: String): Boolean = {
+    input match {
+      case _ if StringUtils.isBlank(input) || input.length > MaxLen || input.length < MinLen => false
+      case _ if input.count(c => TerminateChars.contains(c)) > 1 => false
+      case _ if BlackListedMpns.contains(input.toLowerCase) => false
+      case _ if isTitleCase(input) => false
+      // Unicode strings yet to be handled
+      case _ => true
+    }
+  }
+
+  // Does not consider numbers or one word strings as title-case phrase
+  def isTitleCase(str: String): Boolean = {
+    val words = str.split(' ').filter(_.nonEmpty)
+    if (words.length < 2) false
+    else words.forall(w => w == WordUtils.capitalizeFully(w) && !StringUtils.isNumeric(w))
+  }
+
+  def standardizeMPN(input: String): Option[String] = {
+    if (isValidIdentifier(input)) {
+      Some(input)
+    } else if (StringUtils.isBlank(input)) {
+      None
+    } else if (input.indexWhere(c => TerminateChars.contains(c)) > 0) {
+      Some(input.substring(0, input.indexWhere(c => TerminateChars.contains(c))))
+    }
+    else None
+  }
+}

util-core/src/test/scala/com/indix/utils/core/MPNSpec.scala

@@ -0,0 +1,43 @@
+package com.indix.utils.core
+
+import org.scalatest.{FlatSpec, Matchers}
+
+class MPNSpec extends FlatSpec with Matchers {
+
+  behavior of "MPN"
+
+  it should "check title case" in {
+    MPN.isTitleCase("Key Shell") should be(true)
+    MPN.isTitleCase("Samsung Galaxy A8") should be(true)
+    MPN.isTitleCase("Tempered Glass") should be(true)
+
+    MPN.isTitleCase("1442820G1") should be(false)
+    MPN.isTitleCase("Macbook") should be(false)
+    MPN.isTitleCase("CE 7200") should be(false)
+  }
+
+  it should "validate identifier" in {
+    MPN.isValidIdentifier(null) should be (false)
+    MPN.isValidIdentifier("") should be (false)
+    MPN.isValidIdentifier("51") should be (false)
+    MPN.isValidIdentifier("NA") should be (false)
+    MPN.isValidIdentifier("Does not apply") should be (false)
+
+
+    MPN.isValidIdentifier("DT.VFGAA.003") should be (true)
+    MPN.isValidIdentifier("A55BM-A/USB3") should be (true)
+    MPN.isValidIdentifier("cASSP1598345-10") should be (true)
+    MPN.isValidIdentifier("016393B119058-Regular-18x30-BE-BK") should be (true)
+    MPN.isValidIdentifier("PJS2V") should be (true)
+  }
+
+  it should "standardize MPN" in {
+    MPN.standardizeMPN(null) should be (None)
+    MPN.standardizeMPN("Does not apply") should be (None)
+    MPN.standardizeMPN("PJS2V") should be (Some("PJS2V"))
+
+    MPN.standardizeMPN("105200010437-07-70% All Windows") should be (Some("105200010437-07-70"))
+    MPN.standardizeMPN("30634190, 30753839, 31253006") should be (Some("30634190"))
+  }
+
+}