Series import: extract country name from from "Minerals-Maldives" string.

Author: php-coder

src/main/java/ru/mystamps/web/service/SeriesInfoExtractorServiceImpl.java

@@ -24,6 +24,7 @@
 import java.util.regex.Matcher;
 import java.util.regex.Pattern;
 import java.util.stream.Collectors;
+import java.util.stream.Stream;
 
 import org.apache.commons.lang3.StringUtils;
 
@@ -145,8 +146,18 @@ protected List<Integer> extractCountry(String fragment) {
 
 		log.debug("Determining country from a fragment: '{}'", fragment);
 
-		String[] names = StringUtils.split(fragment, "\n\t ,.");
-		List<String> candidates = Arrays.stream(names)
+		String[] words = StringUtils.split(fragment, "\n\t ,.");
+		
+		Stream<String> names = Arrays.stream(words);
+		
+		// Generate more candidates by split their names by a hyphen.
+		// For example: "Minerals-Maldives" becomes [ "Minerals", "Maldives" ]
+		Stream<String> additionalNames = Arrays.stream(words)
+			.filter(el -> el.contains("-"))
+			.map(el -> StringUtils.split(el, '-'))
+			.flatMap(Arrays::stream);
+		
+		List<String> candidates = Stream.concat(names, additionalNames)
 			.filter(SeriesInfoExtractorServiceImpl::validCountryName)
 			.distinct()
 			.limit(MAX_CANDIDATES_FOR_LOOKUP)

src/test/groovy/ru/mystamps/web/service/SeriesInfoExtractorServiceImplTest.groovy

@@ -149,6 +149,16 @@ class SeriesInfoExtractorServiceImplTest extends Specification {
 			1 * countryService.findIdsByNames(expectedCandidates) >> Random.listOfIntegers()
 	}
 
+	def 'extractCountry() should generate additional candidates by split words by a hyphen'() {
+		given:
+			String fragment = 'foo-bar'
+			List<String> expectedCandidates = [ 'foo-bar', 'foo', 'bar' ]
+		when:
+			service.extractCountry(fragment)
+		then:
+			1 * countryService.findIdsByNames(expectedCandidates) >> Random.listOfIntegers()
+	}
+	
 	def 'extractCountry() should try to search country names with candidate as a prefix'() {
 		given:
 			List<Integer> expectedResult = Random.listOfIntegers()