Improve author parsing (#4931)

Fixes #4864. The author class added an additional space when converting "A O" to the dotted version "A. O.".

CHANGELOG.md

@@ -82,6 +82,7 @@ We refer to [GitHub issues](https://github.com/JabRef/jabref/issues) by using `#
 - We fixed an issue where a non-existing aux file in a group made it impossible to open the library. [#4735](https://github.com/JabRef/jabref/issues/4735)
 - We fixed an issue where some journal names were wrongly marked as abbreviated. [#4115](https://github.com/JabRef/jabref/issues/4115)
 - We fixed an issue where the custom file column were sorted incorrectly. https://github.com/JabRef/jabref/issues/3119
+- We improved the parsing of author names whose infix is abbreviated without a dot. [#4864](https://github.com/JabRef/jabref/issues/4864)
 - We fixed an issues where the entry losses focus when a field is edited and at the same time used for sorting. https://github.com/JabRef/jabref/issues/3373
 - We fixed an issue where the menu on Mac OS was not displayed in the usual Mac-specific way. https://github.com/JabRef/jabref/issues/3146
 - We improved the integrity check for page numbers. [#4113](https://github.com/JabRef/jabref/issues/4113) and [feature request in the forum](http://discourse.jabref.org/t/pages-field-allow-use-of-en-dash/1199)

src/main/java/org/jabref/model/entry/Author.java

@@ -56,7 +56,6 @@ public Author(String first, String firstabbr, String von, String last, String jr
     }
 
     public static String addDotIfAbbreviation(String name) {
-        // Avoid arrayindexoutof.... :
         if ((name == null) || name.isEmpty()) {
             return name;
         }
@@ -114,23 +113,26 @@ public static String addDotIfAbbreviation(String name) {
             // AA -> A. A.
             // Only append ". " if the rest of the 'word' is uppercase
             boolean nextWordIsUppercase = true;
+            char furtherChar = Character.MIN_VALUE;
             for (int j = i + 1; j < name.length(); j++) {
-                char furtherChar = name.charAt(j);
-                if (Character.isWhitespace(furtherChar) || (furtherChar == '-') || (furtherChar == '~')
-                        || (furtherChar == '.')) {
+                furtherChar = name.charAt(j);
+                if (Character.isWhitespace(furtherChar) || (furtherChar == '-') || (furtherChar == '~') || (furtherChar == '.')) {
                     // end of word
                     break;
                 }
 
-                boolean furtherIsUppercaseLetter = Character.isLetter(furtherChar)
-                        && Character.isUpperCase(furtherChar);
+                boolean furtherIsUppercaseLetter = Character.isLetter(furtherChar) && Character.isUpperCase(furtherChar);
                 if (!furtherIsUppercaseLetter) {
                     nextWordIsUppercase = false;
                     break;
                 }
             }
             if (nextWordIsUppercase) {
-                sb.append(". ");
+                if (Character.isWhitespace(furtherChar)) {
+                    sb.append(".");
+                } else {
+                    sb.append(". ");
+                }
             }
         }
 

src/main/java/org/jabref/model/entry/AuthorListParser.java

@@ -80,7 +80,6 @@ public class AuthorListParser {
      * @return a parsed list of persons
      */
     public AuthorList parse(String listOfNames) {
-
         Objects.requireNonNull(listOfNames);
 
         // initialization of parser
@@ -103,7 +102,6 @@ public AuthorList parse(String listOfNames) {
      * empty.
      */
     private Optional<Author> getAuthor() {
-
         List<Object> tokens = new ArrayList<>(); // initialization
         int vonStart = -1;
         int lastStart = -1;
@@ -154,7 +152,7 @@ private Optional<Author> getAuthor() {
                         vonStart = tokens.size() - TOKEN_GROUP_LENGTH;
                         break;
                     }
-                } else if ((lastStart < 0) && tokenCase) {
+                } else if (tokenCase) {
                     lastStart = tokens.size() - TOKEN_GROUP_LENGTH;
                     break;
                 }
@@ -255,13 +253,10 @@ private Optional<Author> getAuthor() {
         }
 
         // Third step: do actual splitting, construct Author object
-        String firstPart = firstPartStart < 0 ? null : concatTokens(tokens, firstPartStart, firstPartEnd, OFFSET_TOKEN,
-                false);
-        String firstAbbr = firstPartStart < 0 ? null : concatTokens(tokens, firstPartStart, firstPartEnd,
-                OFFSET_TOKEN_ABBR, true);
+        String firstPart = firstPartStart < 0 ? null : concatTokens(tokens, firstPartStart, firstPartEnd, OFFSET_TOKEN, false);
+        String firstAbbr = firstPartStart < 0 ? null : concatTokens(tokens, firstPartStart, firstPartEnd, OFFSET_TOKEN_ABBR, true);
         String vonPart = vonPartStart < 0 ? null : concatTokens(tokens, vonPartStart, vonPartEnd, OFFSET_TOKEN, false);
-        String lastPart = lastPartStart < 0 ? null : concatTokens(tokens, lastPartStart, lastPartEnd, OFFSET_TOKEN,
-                false);
+        String lastPart = lastPartStart < 0 ? null : concatTokens(tokens, lastPartStart, lastPartEnd, OFFSET_TOKEN, false);
         String jrPart = jrPartStart < 0 ? null : concatTokens(tokens, jrPartStart, jrPartEnd, OFFSET_TOKEN, false);
 
         if ((firstPart != null) && (lastPart != null) && lastPart.equals(lastPart.toUpperCase(Locale.ROOT)) && (lastPart.length() < 5)

src/test/java/org/jabref/model/entry/AuthorListParameterTest.java

@@ -1,6 +1,5 @@
 package org.jabref.model.entry;
 
-import java.util.Arrays;
 import java.util.stream.Stream;
 
 import org.junit.jupiter.params.ParameterizedTest;
@@ -9,32 +8,26 @@
 
 import static org.junit.jupiter.api.Assertions.assertEquals;
 
-public class AuthorListParameterTest {
+class AuthorListParameterTest {
 
     private static Stream<Arguments> data() {
-
         return Stream.of(
-                Arguments.of("王, 军", authorList(new Author("军", "军.", null, "王", null))),
-                Arguments.of("Doe, John", authorList(new Author("John", "J.", null, "Doe", null))),
-                Arguments.of("von Berlichingen zu Hornberg, Johann Gottfried",
-                        authorList(new Author("Johann Gottfried", "J. G.", "von", "Berlichingen zu Hornberg", null))),
-                //Arguments.of("Robert and Sons, Inc.", authorList(new Author(null, null, null, "Robert and Sons, Inc.", null))),
-                //Arguments.of("al-Ṣāliḥ, Abdallāh", authorList(new Author("Abdallāh", "A.", null, "al-Ṣāliḥ", null))),
-                Arguments.of("de la Vallée Poussin, Jean Charles Gabriel",
-                        authorList(new Author("Jean Charles Gabriel", "J. C. G.", "de la", "Vallée Poussin", null))),
-                Arguments.of("de la Vallée Poussin, J. C. G.",
-                        authorList(new Author("J. C. G.", "J. C. G.", "de la", "Vallée Poussin", null))),
-                Arguments.of("{K}ent-{B}oswell, E. S.", authorList(new Author("E. S.", "E. S.", null, "{K}ent-{B}oswell", null))));
-    }
-
-    private static AuthorList authorList(Author author) {
-        return new AuthorList(Arrays.asList(author));
+                Arguments.of("王, 军", new Author("军", "军.", null, "王", null)),
+                Arguments.of("Doe, John", new Author("John", "J.", null, "Doe", null)),
+                Arguments.of("von Berlichingen zu Hornberg, Johann Gottfried", new Author("Johann Gottfried", "J. G.", "von", "Berlichingen zu Hornberg", null)),
+                //Arguments.of("Robert and Sons, Inc.", new Author(null, null, null, "Robert and Sons, Inc.", null))),
+                //Arguments.of("al-Ṣāliḥ, Abdallāh", new Author("Abdallāh", "A.", null, "al-Ṣāliḥ", null))),
+                Arguments.of("de la Vallée Poussin, Jean Charles Gabriel", new Author("Jean Charles Gabriel", "J. C. G.", "de la", "Vallée Poussin", null)),
+                Arguments.of("de la Vallée Poussin, J. C. G.", new Author("J. C. G.", "J. C. G.", "de la", "Vallée Poussin", null)),
+                Arguments.of("{K}ent-{B}oswell, E. S.", new Author("E. S.", "E. S.", null, "{K}ent-{B}oswell", null)),
+                Arguments.of("Uhlenhaut, N Henriette", new Author("N Henriette", "N. H.", null, "Uhlenhaut", null))
+        );
     }
 
     @ParameterizedTest
     @MethodSource("data")
-    void parseCorrectly(String authorsString, AuthorList authorsParsed) {
+    void parseCorrectly(String authorsString, Author authorsParsed) {
         AuthorListParser parser = new AuthorListParser();
-        assertEquals(authorsParsed, parser.parse(authorsString));
+        assertEquals(new AuthorList(authorsParsed), parser.parse(authorsString));
     }
 }

src/test/java/org/jabref/model/entry/AuthorTest.java

@@ -4,10 +4,10 @@
 
 import static org.junit.jupiter.api.Assertions.assertEquals;
 
-public class AuthorTest {
+class AuthorTest {
 
     @Test
-    public void addDotIfAbbreviationAddDot() {
+    void addDotIfAbbreviationAddDot() {
         assertEquals("O.", Author.addDotIfAbbreviation("O"));
         assertEquals("A. O.", Author.addDotIfAbbreviation("AO"));
         assertEquals("A. O.", Author.addDotIfAbbreviation("AO."));
@@ -16,7 +16,12 @@ public void addDotIfAbbreviationAddDot() {
     }
 
     @Test
-    public void addDotIfAbbreviationDoNotAddDot() {
+    void addDotIfAbbreviationDoesNotAddMultipleSpaces() {
+        assertEquals("A. O.", Author.addDotIfAbbreviation("A O"));
+    }
+
+    @Test
+    void addDotIfAbbreviationDoNotAddDot() {
         assertEquals("O.", Author.addDotIfAbbreviation("O."));
         assertEquals("A. O.", Author.addDotIfAbbreviation("A. O."));
         assertEquals("A.-O.", Author.addDotIfAbbreviation("A.-O."));
@@ -32,7 +37,6 @@ public void addDotIfAbbreviationDoNotAddDot() {
         assertEquals("{\\'{E}}douard", Author.addDotIfAbbreviation("{\\'{E}}douard"));
         assertEquals("J{\\\"o}rg", Author.addDotIfAbbreviation("J{\\\"o}rg"));
         assertEquals("Moore, O. and O. Moore", Author.addDotIfAbbreviation("Moore, O. and O. Moore"));
-        assertEquals("Moore, O. and O. Moore and Moore, O. O.",
-                Author.addDotIfAbbreviation("Moore, O. and O. Moore and Moore, O. O."));
+        assertEquals("Moore, O. and O. Moore and Moore, O. O.", Author.addDotIfAbbreviation("Moore, O. and O. Moore and Moore, O. O."));
     }
 }