Merge pull request #1172 from adomasven/fix/clean-authors

Get a better last name if current guess starts with weird symbols

chrome/content/zotero/xpcom/utilities.js

@@ -179,9 +179,13 @@ Zotero.Utilities = {
 				var lastName = author;
 			}
 		} else {
-			var spaceIndex = author.lastIndexOf(" ");
-			var lastName = author.substring(spaceIndex+1);
-			var firstName = author.substring(0, spaceIndex);
+			// Don't parse "Firstname Lastname [Country]" as "[Country], Firstname Lastname"
+			var spaceIndex = author.length;
+			do {
+				spaceIndex = author.lastIndexOf(" ", spaceIndex-1);
+				var lastName = author.substring(spaceIndex + 1);
+				var firstName = author.substring(0, spaceIndex);
+			} while (!Zotero.Utilities.XRegExp('\\pL').test(lastName[0]) && spaceIndex > 0)
 		}
 
 		if(firstName && allCapsRe.test(firstName) &&

test/tests/utilitiesTest.js

@@ -16,6 +16,21 @@ describe("Zotero.Utilities", function() {
 				}
 			}
 		});
+		
+		it('should not parse words starting with symbols as last name', function() {
+			let author = Zotero.Utilities.cleanAuthor('First Middle Last [CountryName]', false);
+			assert.equal(author.firstName, 'First Middle');
+			// Brackets at the beginning and end of a string get removed for strings
+			// such as [First Last] -> Last, First.
+			// The current output is not ideal, but better than "[CountryName, First Middle Last"
+			assert.equal(author.lastName, 'Last [CountryName');
+		});
+		
+		it('should parse names starting with unicode characters correctly', function() {
+			let author = Zotero.Utilities.cleanAuthor('Ąžuolas Žolynas', false);
+			assert.equal(author.firstName, 'Ąžuolas');
+			assert.equal(author.lastName, 'Žolynas');
+		})
 	});
 	describe("cleanISBN", function() {
 		let cleanISBN = Zotero.Utilities.cleanISBN;