Small improvement to Semantic Type NAME.FIRST_LAST to reject more company names (Issue #151) + lib upgrades

Author: tsegall

ChangeLog.md

@@ -1,6 +1,11 @@
 
 ## Changes ##
 
+
+### 17.1.10
+ - INT: Bump logback-classic to 1.5.23, SpotBugs to 6.4.8, commons-text:1.15.0, google phonenumber to 9.0.21,
+ - ENH: Small improvement to Semantic Type NAME.FIRST_LAST to reject more company names (Issue #151)
+
 ### 17.1.9
  - INT: Bump logback-classic to 1.5.21, SpotBugs to 6.4.7, commons-validator to 1.10.1, gradle to 9.2.1, google phonenumber to 9.0.19
  - INT: Switch examples to use non-deprecated method for registering plugins

build.gradle

@@ -1,6 +1,6 @@
 plugins {
 	id "com.github.ben-manes.versions"
-	id "com.github.spotbugs" version "6.4.7" apply false
+	id "com.github.spotbugs" version "6.4.8" apply false
 }
 
 wrapper {

settings.gradle

@@ -40,7 +40,7 @@ includeBuild 'examples/core/speed'
 dependencyResolutionManagement {
 	versionCatalogs {
 		libs {
-			version('fta', '17.1.9')
+			version('fta', '17.1.10')
 			version('jacoco', '0.8.12')
 
 // https://mvnrepository.com/artifact/de.siegmar/fastcsv
@@ -58,9 +58,9 @@ dependencyResolutionManagement {
 // https://mvnrepository.com/artifact/com.fasterxml.jackson.datatype/jackson-datatype-jsr310
 			library('jacksonDataType', 'com.fasterxml.jackson.datatype:jackson-datatype-jsr310:2.20.1')
 // https://mvnrepository.com/artifact/com.googlecode.libphonenumber/libphonenumber
-			library('googlePhoneNumber', 'com.googlecode.libphonenumber:libphonenumber:9.0.20')
+			library('googlePhoneNumber', 'com.googlecode.libphonenumber:libphonenumber:9.0.21')
 // https://mvnrepository.com/artifact/org.apache.commons/commons-text
-			library('commonsText', 'org.apache.commons:commons-text:1.14.0')
+			library('commonsText', 'org.apache.commons:commons-text:1.15.0')
 // https://mvnrepository.com/artifact/com.github.krraghavan/xeger
 			library('xeger', 'com.github.krraghavan:xeger:1.0.0-RELEASE')
 // https://mvnrepository.com/artifact/dk.brics.automaton/automaton
@@ -70,7 +70,7 @@ dependencyResolutionManagement {
 // https://mvnrepository.com/artifact/com.google.guava/guava
 			library('guava', 'com.google.guava:guava:33.5.0-jre')
 // https://mvnrepository.com/artifact/ch.qos.logback/logback-classic
-			library('logbackClassic', 'ch.qos.logback:logback-classic:1.5.21')
+			library('logbackClassic', 'ch.qos.logback:logback-classic:1.5.23')
 // https://mvnrepository.com/artifact/com.datadoghq/sketches-java
 			library('sketches', 'com.datadoghq:sketches-java:0.8.3')
 		}

types/src/main/java/com/cobber/fta/plugins/NameFirstLast.java

@@ -50,9 +50,16 @@ public class NameFirstLast extends LogicalTypeInfinite {
 
 	static {
 		excludes = new HashSet<>();
+		excludes.add("COMPANY");
+		excludes.add("CORP");
+		excludes.add("CORP.");
+		excludes.add("CORPORATION");
 		excludes.add("INC");
 		excludes.add("INC.");
+		excludes.add("INCORPORATED");
+		excludes.add("LIMITED");
 		excludes.add("LLC");
+		excludes.add("SERVICES");
 	}
 
 	/**
@@ -169,7 +176,7 @@ public boolean isValid(final String input, final boolean detectMode, final long
 		final String firstName = trimmed.substring(0, firstSpace);
 		final String lastName = trimmed.substring(lastSpace + 1);
 
-		if (excludes.contains(lastName))
+		if (excludes.contains(lastName.toUpperCase(locale)))
 			return false;
 
 		firstNames.mergeIfSpace(firstName, count, Long::sum);