Merge pull request #814 from clulab/kwalcock/debugNumbers

Fail better on percentages

main/src/main/resources/org/clulab/numeric/number.yml

@@ -27,7 +27,7 @@ rules:
       # numbers 10-19
       (ten | eleven | twelve | dozen | thirteen | fourteen | fifteen | sixteen | seventeen | eighteen | nineteen |
       # or numbers 20-99
-      twenty @WordDigit? | thirty @WordDigit? | fourty @WordDigit? | fifty @WordDigit? |
+      twenty @WordDigit? | thirty @WordDigit? | forty @WordDigit? | fifty @WordDigit? |
       sixty @WordDigit? | seventy @WordDigit? | eighty @WordDigit? | ninety @WordDigit? |
       "twenty-one" | "twenty-two" | "twenty-three" | "twenty-four" | "twenty-five" | "twenty-six" | "twenty-seven" | "twenty-eight" | "twenty-nine" |
       "thirty-one" | "thirty-two" | "thirty-three" | "thirty-four" | "thirty-five" | "thirty-six" | "thirty-seven" | "thirty-eight" | "thirty-nine" |
@@ -91,7 +91,7 @@ rules:
       # numbers 10-19
       ten | eleven | twelve | dozen | thirteen | fourteen | fifteen | sixteen | seventeen | eighteen | nineteen |
       # or numbers 20-99
-      twenty @WordDigit? | thirty @WordDigit? | fourty @WordDigit? | fifty @WordDigit? |
+      twenty @WordDigit? | thirty @WordDigit? | forty @WordDigit? | fifty @WordDigit? |
       sixty @WordDigit? | seventy @WordDigit? | eighty @WordDigit? | ninety @WordDigit? |
       "twenty-one" | "twenty-two" | "twenty-three" | "twenty-four" | "twenty-five" | "twenty-six" | "twenty-seven" | "twenty-eight" | "twenty-nine" |
       "thirty-one" | "thirty-two" | "thirty-three" | "thirty-four" | "thirty-five" | "thirty-six" | "thirty-seven" | "thirty-eight" | "thirty-nine" |

main/src/main/scala/org/clulab/numeric/actions/NumericActions.scala

@@ -16,9 +16,9 @@ class NumericActions(seasonNormalizer: SeasonNormalizer, unitNormalizer: UnitNor
   /** Converts a sequence of mentions to new types given the converter function */
   private def convert(mentions: Seq[Mention], converter: Mention => Mention, converterName: String): Seq[Mention] = {
     val convertedMentions = new ArrayBuffer[Mention]()
-    for(m <- mentions) {
+    for (m <- mentions) {
       try {
-        convertedMentions += converter(m )
+        convertedMentions += converter(m)
       } catch {
         case e: Exception =>
           // sometimes these conversions fail, mainly on broken texts

main/src/main/scala/org/clulab/numeric/mentions/MeasurementMention.scala

@@ -5,34 +5,38 @@ import org.clulab.odin.{Attachment, TextBoundMention}
 import org.clulab.processors.Document
 import org.clulab.struct.Interval
 
-class MeasurementMention ( labels: Seq[String],
-                           tokenInterval: Interval,
-                           sentence: Int,
-                           document: Document,
-                           keep: Boolean,
-                           foundBy: String,
-                           attachments: Set[Attachment],
-                           val value: Option[Seq[String]],
-                           val unit: Option[Seq[String]],
-                           val fromRange: Boolean,
-                           unitNormalizer: BaseUnitNormalizer = NullUnitNormalizer)
-  extends TextBoundMention(labels, tokenInterval, sentence, document, keep, foundBy, attachments) with Norm {
-
-  override def neNorm: String = {
+class MeasurementMention(
+  labels: Seq[String],
+  tokenInterval: Interval,
+  sentence: Int,
+  document: Document,
+  keep: Boolean,
+  foundBy: String,
+  attachments: Set[Attachment],
+  val value: Option[Seq[String]],
+  val unit: Option[Seq[String]],
+  val fromRange: Boolean,
+  unitNormalizer: BaseUnitNormalizer = NullUnitNormalizer
+) extends TextBoundMention(labels, tokenInterval, sentence, document, keep, foundBy, attachments) with Norm {
+  val cachedNeNorm: String = calcNeNorm()
+
+  override def neNorm: String = cachedNeNorm
+
+  def calcNeNorm(): String = {
     assert(value.nonEmpty)
     assert(unit.nonEmpty)
 
     val numValueOpt =
-      if(fromRange) {
-        Some(value.get.head)
-      } else {
-        NumberParser.parse(value.get)
-      }
-    if(numValueOpt.isEmpty)
-      throw new RuntimeException(s"ERROR: could not parse the number [${value.mkString(" ")}] in the measurement ${raw.mkString(" ")}!")
-    val unitNorm = unitNormalizer.norm(unit.get)
-
-    s"${numValueOpt.get} $unitNorm"
+        if (fromRange) Some(value.get.head)
+        else NumberParser.parse(value.get)
+
+    numValueOpt
+        .map { numValue =>
+          val unitNorm = unitNormalizer.norm(unit.get)
+
+          s"$numValue $unitNorm"
+        }
+        .getOrElse(throw new RuntimeException(s"ERROR: could not parse the number [${value.mkString(" ")}] in the measurement ${raw.mkString(" ")}!"))
   }
 
   override def neLabel: String = {

main/src/main/scala/org/clulab/numeric/mentions/PercentageMention.scala

@@ -5,24 +5,30 @@ import org.clulab.odin.{Attachment, TextBoundMention}
 import org.clulab.processors.Document
 import org.clulab.struct.Interval
 
-class PercentageMention ( labels: Seq[String],
-                          tokenInterval: Interval,
-                          sentence: Int,
-                          document: Document,
-                          keep: Boolean,
-                          foundBy: String,
-                          attachments: Set[Attachment],
-                          val value: Option[Seq[String]])
-  extends TextBoundMention(labels, tokenInterval, sentence, document, keep, foundBy, attachments) with Norm {
+class PercentageMention(
+  labels: Seq[String],
+  tokenInterval: Interval,
+  sentence: Int,
+  document: Document,
+  keep: Boolean,
+  foundBy: String,
+  attachments: Set[Attachment],
+  val value: Option[Seq[String]]
+) extends TextBoundMention(labels, tokenInterval, sentence, document, keep, foundBy, attachments) with Norm {
+  val cachedNeNorm: String = calcNeNorm()
 
-  override def neNorm: String = {
+  override def neNorm: String = cachedNeNorm
+
+  // This isn't a valid PercentageMention unless the neNorm can be calculated.
+  // See if it can be in the constructor and cache the calculated value for reuse.
+  def calcNeNorm(): String = {
     assert(value.nonEmpty)
 
     val numValueOpt = NumberParser.parse(value.get)
-    if(numValueOpt.isEmpty)
-      throw new RuntimeException(s"ERROR: could not parse the number [${value.mkString(" ")}] in the percentage ${raw.mkString(" ")}!")
 
-    s"${numValueOpt.get} %"
+    numValueOpt
+        .map(numValue => s"$numValue %")
+        .getOrElse(throw new RuntimeException(s"ERROR: could not parse the number [${value.mkString(" ")}] in the percentage ${raw.mkString(" ")}!"))
   }
 
   override def neLabel: String = {

main/src/test/scala/org/clulab/numeric/TestNumericMentions.scala

@@ -0,0 +1,26 @@
+package org.clulab.numeric
+
+import org.clulab.processors.clu.CluProcessor
+import org.clulab.utils.Test
+
+class TestNumericMentions extends Test {
+  val processor = new CluProcessor()
+
+  behavior of "MeasurementMention"
+
+  it should "not throw an exception" in {
+    val text = "(c) a retention licence, fourty five days; and  ."
+    val document = processor.annotate(text)
+
+    document
+  }
+
+  behavior of "PercentageMention"
+
+  it should "not throw an exception" in {
+    val text = "pdf 75 Some of the major sectoral plans include the national energy policy, transport policy, forest and wildlife policy, national environmental sanitation strategy etc. 76 https://www.thegasconsortium.com/documents/GMP-Final-Jun16.pdf 77 http://www.energycom.gov.gh/files/Renewable-Energy-Masterplan-February-2019.pdf 78 http://energycom.gov.gh/files/SE4ALL-GHANA%20ACTION%20PLAN.pdf 79 http://www.energycom.gov.gh/files/Ghana%20Integrated%20Power%20System%20Master%20Plan%20_Volume%202.pdf Page | 80 In 2017, the Ministry of Energy started a comprehensive review of the National Energy Policy, which is still ongoing."
+    val document = processor.annotate(text)
+
+    document
+  }
+}

version.sbt

@@ -1 +1 @@
-version in ThisBuild := "8.5.4"
+version in ThisBuild := "8.5.5-SNAPSHOT"