Fix another two corner cases

lfoppiano · Mar 28, 2024 · 10edbf0 · 10edbf0
1 parent f23e4d3
commit 10edbf0
Show file tree

Hide file tree

Showing 3 changed files with 29 additions and 2 deletions.
diff --git a/resources/dataset/values/corpus/trainingdata1.tei.xml b/resources/dataset/values/corpus/trainingdata1.tei.xml
@@ -11,5 +11,6 @@
     <value><base>10</base><pow>-1</pow></value>
 <!--    <value>e<exp>-1</exp></value>-->
     <value><number>10</number></value>
+    <value><alpha>Six</alpha>, <number>12</number></value>
 </values>
 
diff --git a/src/main/java/org/grobid/core/utilities/WordsToNumber.java b/src/main/java/org/grobid/core/utilities/WordsToNumber.java
@@ -33,9 +33,9 @@ public class WordsToNumber {
 
     private final String VALUES_PATH = "lexicon/en/values.json";
 
-    private final Pattern NUMERIC_PATTERN = Pattern.compile("[0-9.,]+", Pattern.CASE_INSENSITIVE);
+    private final Pattern NUMERIC_PATTERN = Pattern.compile("\\b(?:\\d+(?:[.,]\\d+)*|\\d+[.,]?\\d*\\b)\\b", Pattern.CASE_INSENSITIVE);
     private final Pattern OUT_OF_PATTERN_NUMBERS = Pattern.compile("([0-9.,]+)( out)? of (the )?([0-9.,]+)", Pattern.CASE_INSENSITIVE);
-    private final Pattern OUT_OF_PATTERN_ALPHABETIC = Pattern.compile("([A-Za-z ]+) out of (the )?([A-Za-z]+)", Pattern.CASE_INSENSITIVE);
+    private final Pattern OUT_OF_PATTERN_ALPHABETIC = Pattern.compile("([A-Za-z ]+) out of ([a-z]+ )?([A-Za-z]+)", Pattern.CASE_INSENSITIVE);
 
     private static List<String> bases = null;
     private static List<String> tens = null;

diff --git a/src/test/kotlin/org/grobid/core/utilities/WordsToNumberTest.kt b/src/test/kotlin/org/grobid/core/utilities/WordsToNumberTest.kt
@@ -164,6 +164,22 @@ class WordsToNumberTest {
         MatcherAssert.assertThat(number, Is.`is`(BigDecimal("0.75")))
     }
 
+    @Test
+    @Throws(Exception::class)
+    fun testConvertFractions6_1() {
+        val input = "three out of these four"
+        val number = target.normalize(input, Locale.ENGLISH)
+        MatcherAssert.assertThat(number, Is.`is`(BigDecimal("0.75")))
+    }
+
+    @Test
+    @Throws(Exception::class)
+    fun testConvertFractions6_2() {
+        val input = "three out of that four"
+        val number = target.normalize(input, Locale.ENGLISH)
+        MatcherAssert.assertThat(number, Is.`is`(BigDecimal("0.75")))
+    }
+
     @Test
     @Throws(Exception::class)
     fun testConvertFractions4Numeric() {
@@ -186,4 +202,14 @@ class WordsToNumberTest {
         val input = "a temperature of 20"
         target.normalize(input, Locale.ENGLISH)
     }
+
+
+    @Throws(Exception::class)
+    fun testErrorCase_1() {
+        val input = "six, 12"
+        val output = target.normalize(input, Locale.ENGLISH)
+
+
+    }
+
 }