WIP: Add swedish short words filter

src/org/daisy/dotify/translator/impl/sv_SE/SwedishBrailleFilter.java

@@ -24,13 +24,25 @@ public SwedishBrailleFilter(String locale) {
 		this(locale, false);
 	}
 
+	/**
+	 * Creates a new Swedish braille filter.
+	 * @param locale the locale
+	 * @param strict if true the result is braille only, if false the result
+	 * 		contains break point characters such as space, dash and soft hyphen.
+	 */
+	public SwedishBrailleFilter(String locale, boolean strict) {
+		this(locale, strict, false);
+	}
+
 	/**
 	 * Creates a new Swedish braille filter with the specified mode.
 	 * @param locale the locale
 	 * @param strict if true the result is braille only, if false the result 
 	 * 			contains break point characters such as space, dash and soft hyphen.
+	 * @param useContractedBraille Whether or not the string should be translated to contracted braille.
+	 *
 	 */
-	public SwedishBrailleFilter(String locale, boolean strict) {
+	public SwedishBrailleFilter(String locale, boolean strict, boolean useContractedBraille) {
 		filters = new CombinationFilter();
 		// Remove zero width space
 		filters.add(new RegexFilter("\\u200B", ""));
@@ -42,6 +54,12 @@ public SwedishBrailleFilter(String locale, boolean strict) {
 		filters.add(new CapitalizationMarkers());
 
 		Locale l = FilterLocale.parse(locale).toLocale();
+
+		if (useContractedBraille) {
+			// Text to braille, shorthand format
+			filters.add(new SwedishContractedBrailleFilter());
+		}
+
 		// Text to braille, Pas 1
 		filters.add(new UCharFilter(getResource("sv_SE-pas1.xml"), l));
 		// Text to braille, Pas 2

src/org/daisy/dotify/translator/impl/sv_SE/SwedishBrailleFilterFactory.java

@@ -35,7 +35,16 @@ public BrailleFilter newFilter(String locale, String mode) throws TranslatorConf
 				throw new SwedishFilterConfigurationException(e);
 			}
 			return new DefaultBrailleFilter(new SwedishBrailleFilter(loc.get(), true), loc.get(), sap, hyphenatorService);
-		} 
+		} else if (loc.isPresent() && mode.equals(TranslatorType.CONTRACTED.toString())) {
+
+			DefaultMarkerProcessor sap;
+			try {
+				sap = new SwedishMarkerProcessorFactory().newMarkerProcessor(loc.get(), mode);
+			} catch (SwedishMarkerProcessorConfigurationException e) {
+				throw new SwedishFilterConfigurationException(e);
+			}
+			return new DefaultBrailleFilter(new SwedishBrailleFilter(loc.get(), true, true), loc.get(), sap, hyphenatorService);
+		}
 		throw new SwedishFilterConfigurationException("Factory does not support " + locale + "/" + mode);
 	}
 

src/org/daisy/dotify/translator/impl/sv_SE/SwedishBrailleFilterFactoryService.java

@@ -36,6 +36,8 @@ public SwedishBrailleFilterFactoryService() {
 		this.specs = new ArrayList<>();
 		String displayName = RESOURCE_BUNDLE.getString("uncontracted-6-dot");
 		String desc = RESOURCE_BUNDLE.getString("uncontracted-description");
+		String contractedDisplayName = RESOURCE_BUNDLE.getString("contracted-6-dot");
+		String contractedDesc = RESOURCE_BUNDLE.getString("contracted-description");
 		specs.add(new TranslatorSpecification("sv", TranslatorMode.Builder
 				.withType(TranslatorType.UNCONTRACTED)
 				.displayName(displayName)
@@ -46,11 +48,27 @@ public SwedishBrailleFilterFactoryService() {
 				.displayName(displayName)
 				.description(desc)
 				.build()));
-	}
-
-	@Override
-	public boolean supportsSpecification(String locale, String mode) {
-		return ("sv".equalsIgnoreCase(locale) || "sv-SE".equalsIgnoreCase(locale)) && mode.equals(TranslatorType.UNCONTRACTED.toString());
+		specs.add(new TranslatorSpecification("sv", TranslatorMode.Builder
+				.withType(TranslatorType.CONTRACTED)
+				.displayName(contractedDisplayName)
+				.description(contractedDesc)
+				.build()));
+		specs.add(new TranslatorSpecification("sv-SE", TranslatorMode.Builder
+				.withType(TranslatorType.CONTRACTED)
+				.displayName(contractedDisplayName)
+				.description(contractedDesc)
+				.build()));
+    }
+
+    @Override
+    public boolean supportsSpecification(String locale, String mode) {
+        if (("sv".equalsIgnoreCase(locale) || "sv-SE".equalsIgnoreCase(locale)) && mode.equals(TranslatorType.UNCONTRACTED.toString())) {
+            return true;
+        } else if (("sv".equalsIgnoreCase(locale) || "sv-SE".equalsIgnoreCase(locale)) && mode.equals(TranslatorType.CONTRACTED.toString())) {
+            return true;
+        } else {
+            return false;
+        }
 	}
 
 	@Override
@@ -74,7 +92,7 @@ public void setHyphenator(HyphenatorFactoryMakerService hyphenator) {
 	public void unsetHyphenator(HyphenatorFactoryMakerService hyphenator) {
 		this.hyphenator = null;
 	}
-	
+
 	@Override
 	public Collection<TranslatorSpecification> listSpecifications() {
 		return specs;

src/org/daisy/dotify/translator/impl/sv_SE/SwedishBrailleTranslatorFactory.java

@@ -41,6 +41,18 @@ public BrailleTranslator newTranslator(String locale, String mode) throws Transl
 			return new SimpleBrailleTranslator(
 					new DefaultBrailleFilter(new SwedishBrailleFilter(loc.get()), loc.get(), sap, hyphenatorService),
 					new DefaultBrailleFinalizer(), mode);
+		} else if (loc.isPresent() && mode.equals(TranslatorType.CONTRACTED.toString())) {
+
+			DefaultMarkerProcessor sap;
+			try {
+				sap = new SwedishMarkerProcessorFactory().newMarkerProcessor(loc.get(), mode);
+			} catch (SwedishMarkerProcessorConfigurationException e) {
+				throw new SwedishTranslatorConfigurationException(e);
+			}
+
+			return new SimpleBrailleTranslator(
+					new DefaultBrailleFilter(new SwedishBrailleFilter(loc.get(), false, true), loc.get(), sap, hyphenatorService),
+					new DefaultBrailleFinalizer(), mode);
 		} else if (loc.isPresent() && mode.equals(TranslatorType.PRE_TRANSLATED.toString())) {
 			return new SimpleBrailleTranslator(
 					new PreTranslatedBrailleFilter(),

src/org/daisy/dotify/translator/impl/sv_SE/SwedishBrailleTranslatorFactoryService.java

@@ -35,14 +35,21 @@ public class SwedishBrailleTranslatorFactoryService implements
 	public SwedishBrailleTranslatorFactoryService() {
 		this.specs = new ArrayList<>();
 		String uncontracted = RESOURCE_BUNDLE.getString("uncontracted-6-dot");
+		String contracted = RESOURCE_BUNDLE.getString("contracted-6-dot");
 		String preTranslated = RESOURCE_BUNDLE.getString("pre-translated");
 		String descUncontracted = RESOURCE_BUNDLE.getString("uncontracted-description");
+		String descContracted = RESOURCE_BUNDLE.getString("contracted-description");
 		String descPreTranslated = RESOURCE_BUNDLE.getString("pre-translated-description");
 		specs.add(new TranslatorSpecification("sv", TranslatorMode.Builder
 				.withType(TranslatorType.UNCONTRACTED)
 				.displayName(uncontracted)
 				.description(descUncontracted)
 				.build()));
+		specs.add(new TranslatorSpecification("sv", TranslatorMode.Builder
+				.withType(TranslatorType.CONTRACTED)
+				.displayName(contracted)
+				.description(descContracted)
+				.build()));
 		specs.add(new TranslatorSpecification("sv", TranslatorMode.Builder
 				.withType(TranslatorType.PRE_TRANSLATED)
 				.displayName(preTranslated)
@@ -53,6 +60,11 @@ public SwedishBrailleTranslatorFactoryService() {
 				.displayName(uncontracted)
 				.description(descUncontracted)
 				.build()));
+		specs.add(new TranslatorSpecification("sv-SE", TranslatorMode.Builder
+				.withType(TranslatorType.CONTRACTED)
+				.displayName(contracted)
+				.description(descContracted)
+				.build()));
 		specs.add(new TranslatorSpecification("sv-SE", TranslatorMode.Builder
 				.withType(TranslatorType.PRE_TRANSLATED)
 				.displayName(preTranslated)

src/org/daisy/dotify/translator/impl/sv_SE/SwedishContractedBrailleFilter.java

@@ -0,0 +1,89 @@
+package org.daisy.dotify.translator.impl.sv_SE;
+
+import org.daisy.dotify.common.text.StringFilter;
+
+import java.io.IOException;
+import java.net.URL;
+
+import java.util.HashMap;
+import java.util.Iterator;
+import java.util.Properties;
+import java.util.Set;
+import java.util.regex.Matcher;
+import java.util.regex.Pattern;
+
+/**
+ *
+ */
+public class SwedishContractedBrailleFilter implements StringFilter {
+
+    private HashMap<String, String> contractedBrailleMap;
+    private static final String CONTRACTED_BRAILLE_TABLE_PATH = "sv_SE-single-character-contracted-words.xml";
+
+    public static final String CAPITAL_CHAR_MARKER = "\u2820";
+    public static final String SOFT_HYPHEN = "\u00AD";
+
+    /**
+     * Todo: add support for different grades of contraction.
+     *      For now only single character contracted words is supported
+     */
+    public SwedishContractedBrailleFilter() {
+        this.contractedBrailleMap = new HashMap<>();
+        this.loadTable();
+    }
+
+    /**
+     * Search and replace with contracted braille.
+     * First split the string on space character and go through all words and look for a match in the contractin table.
+     *
+     * @param str - The string that should be filterd
+     * @return the filtered string
+     */
+    @Override
+    public String filter(String str) {
+
+        String[] words = str.split("\\s");
+        // Handle Edge case with no words.
+        if (words.length == 0) {
+            return str;
+        }
+        // Strip string from
+        Pattern pattern = Pattern.compile(CAPITAL_CHAR_MARKER + "*([\\p{javaUpperCase}\\p{javaLowerCase}"+SOFT_HYPHEN+"]+)");
+        StringBuilder sb = new StringBuilder();
+        String key, replace;
+
+        for (String word: words) {
+            Matcher matcher = pattern.matcher(word);
+            if (matcher.find()) {
+                key = matcher.group(1).toLowerCase().replace(SOFT_HYPHEN, "");
+                if (this.contractedBrailleMap.containsKey(key)) {
+                    replace = this.contractedBrailleMap.get(key);
+                    word = word.substring(0, matcher.start(1)) + replace + word.substring(matcher.end(1));
+                }
+            }
+            sb.append(word);
+            sb.append(" ");
+        }
+        str = sb.toString().trim();
+        return str;
+    }
+
+    /**
+     * Loads a table using the Properties class.
+     */
+    private void loadTable() {
+        URL tableURL = this.getClass().getResource(CONTRACTED_BRAILLE_TABLE_PATH);
+        Properties props = new Properties();
+        try {
+            props.loadFromXML(tableURL.openStream());
+        } catch (IOException e) {
+            e.printStackTrace();
+            return;
+        }
+        Set<?> keys = props.keySet();
+        for (Iterator<?> it = keys.iterator(); it.hasNext(); ) {
+            String key = (String) it.next();
+            contractedBrailleMap.put(key, props.getProperty(key));
+        }
+    }
+}

src/org/daisy/dotify/translator/impl/sv_SE/SwedishMarkerProcessorFactory.java

@@ -18,7 +18,7 @@ class SwedishMarkerProcessorFactory {
 
 	public DefaultMarkerProcessor newMarkerProcessor(String locale, String mode) throws SwedishMarkerProcessorConfigurationException {
 		if (FilterLocale.parse(locale).equals(sv)||FilterLocale.parse(locale).equals(sv_SE)) {
-			if (mode.equals(TranslatorType.UNCONTRACTED.toString())) {
+			if (mode.equals(TranslatorType.UNCONTRACTED.toString()) || mode.equals(TranslatorType.CONTRACTED.toString())) {
 
 				// Svenska skrivregler för punktskrift 2009, page 34
 				RegexMarkerDictionary strong = new RegexMarkerDictionary.Builder().

src/org/daisy/dotify/translator/impl/sv_SE/messages.properties

@@ -1,4 +1,6 @@
-uncontracted-6-dot=Uncontracted (6-dot) 
+uncontracted-6-dot=Uncontracted (6-dot)
 uncontracted-description=Uncontracted Swedish braille
+contracted-6-dot=Contracted (6-dot)
+contracted-description=Contracted Swedish braille
 pre-translated=Pre-translated
 pre-translated-description=Pre-translated braille

src/org/daisy/dotify/translator/impl/sv_SE/sv_SE-single-character-contracted-words.xml

@@ -0,0 +1,42 @@
+<?xml version="1.0" encoding="utf-8"?>
+<!DOCTYPE properties SYSTEM "http://java.sun.com/dtd/properties.dtd">
+<properties>
+    <comment>Braille translation table for Swedish short script words.</comment>
+
+    <entry key="att">a</entry>
+    <entry key="bli">b</entry>
+    <entry key="och">c</entry>
+    <entry key="där">d</entry>
+    <entry key="eller">e</entry>
+    <entry key="från">f</entry>
+    <entry key="genom">g</entry>
+    <entry key="han">h</entry>
+    <entry key="jag">j</entry>
+    <entry key="kan">k</entry>
+    <entry key="lika">l</entry>
+    <entry key="man">m</entry>
+    <entry key="när">n</entry>
+    <entry key="på">p</entry>
+    <entry key="under">q</entry>
+    <entry key="har">r</entry>
+    <entry key="som">s</entry>
+    <entry key="till">t</entry>
+    <entry key="utan">u</entry>
+    <entry key="vid">v</entry>
+    <entry key="vad">w</entry>
+    <entry key="över">x</entry>
+    <entry key="mycket">y</entry>
+    <entry key="efter">z</entry>
+    <entry key="är">ä</entry>
+
+    <entry key="en">ê</entry>
+    <entry key="med">î</entry>
+    <entry key="er">û</entry>
+    <entry key="ett">§</entry>
+    <entry key="för">ë</entry>
+    <entry key="inte">ü</entry>
+    <entry key="de">ô</entry>
+    <entry key="det">è</entry>
+    <entry key="den">\</entry>
+    <entry key="var">à</entry>
+</properties>

test/org/daisy/dotify/translator/impl/sv_SE/SwedishBrailleFilterTest.java

@@ -237,4 +237,44 @@ public void testFractions_001() {
 	public void testCapitalIWithDot() {
 		assertEquals("⠠⠈⠊", filter.filter("İ"));
 	}
+
+	@Test
+	public void testFilterAppliesContractedBrailleFilterCorrectly(){
+		SwedishBrailleFilter filter_1 = new SwedishBrailleFilter("sv-SE", false, true);
+		String test_string = "jag vet att";
+		String filtered_string = filter_1.filter(test_string);
+		assertEquals("⠚ ⠧⠑⠞ ⠁", filtered_string);
+	}
+
+	@Test
+	public void testFilterAppliesContractedBrailleFilterCorrectlyWithCharCapitalization() {
+		SwedishBrailleFilter filter_1 = new SwedishBrailleFilter("sv-SE", false, true);
+		String test_string = "Hon hade många husdjur";
+		String filtered_string = filter_1.filter(test_string);
+		assertEquals("⠠⠓⠕⠝ ⠓⠁⠙⠑ ⠍⠡⠝⠛⠁ ⠓⠥⠎⠙⠚⠥⠗", filtered_string);
+	}
+
+	@Test
+	public void testFilterDoesNotCrashWhensContractedBrailleFilterIsAppliedForEmptyStrings() {
+		SwedishBrailleFilter filter_1 = new SwedishBrailleFilter("sv-SE", false, true);
+		String test_string = "";
+		String filtered_string = filter_1.filter(test_string);
+		assertEquals("", filtered_string);
+	}
+
+	@Test
+	public void testFilterAppliesContractedBrailleFilterCorrectlyWithWordCapitalization(){
+		SwedishBrailleFilter filter_1 = new SwedishBrailleFilter("sv-SE", false, true);
+		String test_string = "HEY jag heter java";
+		String filtered_string = filter_1.filter(test_string);
+		assertEquals("⠠⠠⠓⠑⠽ ⠚ ⠓⠑⠞⠑⠗ ⠚⠁⠧⠁", filtered_string);
+	}
+
+	@Test
+	public void testFilterAppliesContractedBrailleFilterCorrectlyWithContractionWithSequentialCaptalLetters(){
+		SwedishBrailleFilter filter_1 = new SwedishBrailleFilter("sv-SE", false, true);
+		String test_string = "JAG SAKNAR HONOM MYCKET";
+		String filtered_string = filter_1.filter(test_string);
+		assertEquals("⠠⠠⠠⠚ ⠎⠁⠅⠝⠁⠗ ⠓⠕⠝⠕⠍ ⠽⠱", filtered_string);
+	}
 }