Pagenumbers CleanupAction

src/main/java/net/sf/jabref/gui/actions/CleanUpAction.java

@@ -41,6 +41,8 @@
 
 import com.jgoodies.forms.builder.FormBuilder;
 import com.jgoodies.forms.layout.FormLayout;
+import net.sf.jabref.logic.cleanup.PageNumbersCleanup;
+import net.sf.jabref.logic.formatter.FieldFormatters;
 import net.sf.jabref.logic.l10n.Localization;
 import net.sf.jabref.model.entry.BibtexEntry;
 import net.sf.jabref.logic.util.DOI;
@@ -445,12 +447,12 @@ private static void doCleanUpMonth(BibtexEntry entry, NamedCompound ce) {
 
     private static void doCleanUpPageNumbers(BibtexEntry entry, NamedCompound ce) {
         String oldValue = entry.getField("pages");
-        if (oldValue == null) {
-            return;
-        }
-        String newValue = oldValue.replaceAll(" *(\\d+) *- *(\\d+) *", "$1--$2");
-        if (!oldValue.equals(newValue)) {
-            entry.setField("pages", newValue);
+        // run formatter
+        new PageNumbersCleanup(entry).cleanup();
+        // new value
+        String newValue = entry.getField("pages");
+        // undo action
+        if(!oldValue.equals(newValue)) {
             ce.addEdit(new UndoableFieldChange(entry, "pages", oldValue, newValue));
         }
     }

src/main/java/net/sf/jabref/logic/cleanup/PageNumbersCleanup.java

@@ -0,0 +1,30 @@
+package net.sf.jabref.logic.cleanup;
+
+import net.sf.jabref.logic.formatter.FieldFormatters;
+import net.sf.jabref.logic.formatter.PageNumbersFormatter;
+import net.sf.jabref.model.entry.BibtexEntry;
+
+/**
+ * This class includes sensible defaults for consistent formatting of BibTex page numbers.
+ */
+public class PageNumbersCleanup {
+    private BibtexEntry entry;
+
+    public PageNumbersCleanup(BibtexEntry entry) {
+        this.entry = entry;
+    }
+
+    /**
+     * Format page numbers, separated either by commas or double-hyphens.
+     * Converts the range number format of the <code>pages</code> field to page_number--page_number.
+     *
+     * @see{PageNumbersFormatter}
+     */
+    public void cleanup() {
+        final String field = "pages";
+
+        String value = entry.getField(field);
+        String newValue = FieldFormatters.PAGE_NUMBERS.format(value);
+        entry.setField(field, newValue);
+    }
+}

src/main/java/net/sf/jabref/logic/formatter/FieldFormatters.java

@@ -0,0 +1,10 @@
+package net.sf.jabref.logic.formatter;
+
+import java.util.Arrays;
+import java.util.List;
+
+public class FieldFormatters {
+    public static final PageNumbersFormatter PAGE_NUMBERS = new PageNumbersFormatter();
+
+    public static final List<Formatter> ALL = Arrays.asList(PAGE_NUMBERS);
+}

src/main/java/net/sf/jabref/logic/formatter/Formatter.java

@@ -0,0 +1,21 @@
+package net.sf.jabref.logic.formatter;
+
+/**
+ * Formatter Interface
+ */
+public interface Formatter {
+    /**
+     * Returns a human readable name of the formatter usable for e.g. in the GUI
+     *
+     * @return the name of the formatter
+     */
+    String getName();
+
+    /**
+     * Formats a field value by with a particular formatter transformation.
+     *
+     * @param value the input String
+     * @return the formatted output String
+     */
+    String format(String value);
+}

src/main/java/net/sf/jabref/logic/formatter/PageNumbersFormatter.java

@@ -0,0 +1,51 @@
+package net.sf.jabref.logic.formatter;
+
+import java.util.regex.Matcher;
+import java.util.regex.Pattern;
+
+/**
+ * This class includes sensible defaults for consistent formatting of BibTex page numbers.
+ */
+public class PageNumbersFormatter implements Formatter {
+    @Override
+    public String getName() {
+        return "Page numbers";
+    }
+
+    /**
+     * Format page numbers, separated either by commas or double-hyphens.
+     * Converts the range number format of the <code>pages</code> field to page_number--page_number.
+     * Removes all literals except [0-9,-].
+     * Keeps the existing String if the resulting field does not match the expected Regex.
+     *
+     * <example>
+     *     1-2 -> 1--2
+     *     1,2,3 -> 1,2,3
+     *     {1}-{2} -> 1--2
+     *     Invalid -> Invalid
+     * </example>
+     */
+    public String format(String value) {
+        final String rejectLiterals = "[^0-9,-]";
+        final Pattern pagesPattern = Pattern.compile("\\A(\\d+)-{1,2}(\\d+)\\Z");
+        final String replace = "$1--$2";
+
+        // nothing to do
+        if (value == null || value.isEmpty()) {
+            return value;
+        }
+
+        // remove unwanted literals incl. whitespace
+        String cleanValue = value.replaceAll(rejectLiterals, "");
+        // try to find pages pattern
+        Matcher matcher = pagesPattern.matcher(cleanValue);
+        // replace
+        String newValue = matcher.replaceFirst(replace);
+        // replacement?
+        if(!newValue.equals(cleanValue)) {
+            // write field
+            return newValue;
+        }
+        return value;
+    }
+}

src/test/java/net/sf/jabref/logic/cleanup/PageNumbersCleanupTest.java

@@ -0,0 +1,39 @@
+package net.sf.jabref.logic.cleanup;
+
+import junit.framework.Assert;
+import net.sf.jabref.model.entry.BibtexEntry;
+import org.junit.After;
+import org.junit.Before;
+import org.junit.Test;
+
+import static org.junit.Assert.*;
+
+public class PageNumbersCleanupTest {
+    private BibtexEntry entry;
+
+    @Before
+    public void setUp() {
+        entry = new BibtexEntry();
+    }
+
+    @After
+    public void teardown() {
+        entry = null;
+    }
+
+    @Test
+    public void formatPageNumbers() {
+        entry.setField("pages", "1-2");
+        new PageNumbersCleanup(entry).cleanup();
+
+        Assert.assertEquals("1--2", entry.getField("pages"));
+    }
+
+    @Test
+    public void onlyFormatPageNumbersField() {
+        entry.setField("otherfield", "1-2");
+        new PageNumbersCleanup(entry).cleanup();
+
+        Assert.assertEquals("1-2", entry.getField("otherfield"));
+    }
+}

src/test/java/net/sf/jabref/logic/formatter/PageNumbersFormatterTest.java

@@ -0,0 +1,62 @@
+package net.sf.jabref.logic.formatter;
+
+import junit.framework.Assert;
+import org.junit.After;
+import org.junit.Before;
+import org.junit.Test;
+
+import static org.junit.Assert.*;
+
+public class PageNumbersFormatterTest {
+    private PageNumbersFormatter formatter;
+
+    @Before
+    public void setUp() {
+        formatter = new PageNumbersFormatter();
+    }
+
+    @After
+    public void teardown() {
+        formatter = null;
+    }
+
+    @Test
+    public void formatPageNumbers() {
+        expectCorrect("1-2", "1--2");
+    }
+
+    @Test
+    public void formatPageNumbersCommaSeparated() {
+        expectCorrect("1,2,3", "1,2,3");
+    }
+
+    @Test
+    public void ignoreWhitespaceInPageNumbers() {
+        expectCorrect("   1  - 2 ", "1--2");
+    }
+
+    @Test
+    public void keepCorrectlyFormattedPageNumbers() {
+        expectCorrect("1--2", "1--2");
+    }
+
+    @Test
+    public void formatPageNumbersEmptyFields() {
+        expectCorrect("", "");
+        expectCorrect(null, null);
+    }
+
+    @Test
+    public void formatPageNumbersRemoveUnexpectedLiterals() {
+        expectCorrect("{1}-{2}", "1--2");
+    }
+
+    @Test
+    public void formatPageNumbersRegexNotMatching() {
+        expectCorrect("12", "12");
+    }
+
+    private void expectCorrect(String input, String expected) {
+        Assert.assertEquals(expected, formatter.format(input));
+    }
+}