Bugfix/date relaxed matcher subword month

src/main/scala/com/johnsnowlabs/nlp/annotators/DateMatcher.scala

@@ -16,6 +16,8 @@
 
 package com.johnsnowlabs.nlp.annotators
 
+import com.johnsnowlabs.nlp.util.regex.RuleFactory
+import com.johnsnowlabs.nlp.util.regex.RuleFactory.RuleMatch
 import com.johnsnowlabs.nlp.{Annotation, AnnotatorModel, HasSimpleAnnotate}
 import org.apache.spark.ml.util.{DefaultParamsReadable, Identifiable}
 
@@ -151,8 +153,21 @@ class DateMatcher(override val uid: String) extends AnnotatorModel[DateMatcher]
     }
   }
 
+  private def isNotMonthSubwordMatch(text: String, d: RuleMatch): Boolean = {
+    val words = text.replaceAll("""([?.!:]|\b\p{IsLetter}{1,2}\b)\s*""", "").split(" ")
+    val notSubWordMatches = words
+      .map(_.toLowerCase)
+      .filter( w => w.contains(d.content.matched.toLowerCase) && w.length <= d.content.matched.length)
+
+    notSubWordMatches.length match {
+      case 1 => true
+      case _ => false
+    }
+  }
+
   private def extractRelaxedDate(text: String): Option[MatchedDateTime] = {
-    val possibleDates = relaxedFactory.findMatch(text)
+    val possibleDates: Seq[RuleFactory.RuleMatch] = relaxedFactory.findMatch(text)
+
     if (possibleDates.length > 1) {
       var dayMatch = $(defaultDayWhenMissing)
       var monthMatch = defaultMonthWhenMissing
@@ -163,7 +178,10 @@ class DateMatcher(override val uid: String) extends AnnotatorModel[DateMatcher]
         dayMatch = dayCandidate.get.content.matched.filter(_.isDigit).toInt
       }
 
-      val monthCandidate = possibleDates.find(_.identifier == "relaxed months exclusive")
+      val monthCandidate = possibleDates
+        .find(_.identifier == "relaxed months exclusive")
+        .filter(d => isNotMonthSubwordMatch(text, d))
+
       if (monthCandidate.isDefined && monthCandidate.get.content.matched.length > 2) {
         val month = monthCandidate.get.content.matched.toLowerCase().take(3)
         if (shortMonths.contains(month))

src/main/scala/com/johnsnowlabs/nlp/annotators/DateMatcherUtils.scala

@@ -39,7 +39,7 @@ trait DateMatcherUtils extends Params {
   private val formalDateAlt2 = new Regex("\\b(\\d{2,4})[-/](0?[1-9]|1[012])[-/]([0-2]?[1-9]|[1-3][0-1])\\b", "year", "month", "day")
   private val formalDateShort = new Regex("\\b(0?[1-9]|1[012])[-/](\\d{2,4})\\b", "month", "year")
 
-  private val months = Seq("january", "february", "march", "april", "may", "june", "july", "august", "september", "october", "november", "december")
+  protected val months = Seq("january", "february", "march", "april", "may", "june", "july", "august", "september", "october", "november", "december")
   protected val shortMonths = Seq("jan", "feb", "mar", "apr", "may", "jun", "jul", "aug", "sep", "oct", "nov", "dec")
 
   /** Relaxed dates, e.g. March 2nd */

src/test/scala/com/johnsnowlabs/nlp/annotators/DateMatcherTestSpec.scala

@@ -201,4 +201,27 @@ class DateMatcherTestSpec extends AnyFlatSpec with DateMatcherBehaviors {
     assert(dateMatcherRead.getFormat == dateMatcher.getFormat)
   }
 
+  "a DateMatcher" should "correctly disambiguate the monthly sub-words in text" taggedAs FastTest in {
+    val data: Dataset[Row] = DataBuilder.multipleDataBuild(
+      Array("right over-the-needle catheter system 18 gauge;1 1/2 in length")
+    )
+
+    val date = new DateMatcher()
+      .setInputCols("document")
+      .setOutputCol("date")
+      .setAnchorDateYear(1900)
+      .setFormat("yyyy/MM/dd")
+      .transform(data)
+
+    val results = Annotation.collect(date, "date").flatten.toSeq
+
+    val defaultYearWhenMissing = Calendar.getInstance.get(Calendar.YEAR)
+    val defaultMonthWhenMissing = "01"
+
+    val expectedDates = Seq(
+      Annotation(DATE, 38, 44, s"$defaultYearWhenMissing/$defaultMonthWhenMissing/18", Map("sentence" -> "0")))
+
+    assert(results == expectedDates)
+  }
+
 }