From 83c78559bd3edd90c6f52461754d1265af824b5d Mon Sep 17 00:00:00 2001
From: aStereoID <astrid.schmiedel@serbski-institut.de>
Date: Wed, 10 Apr 2024 12:20:17 +0200
Subject: [PATCH] Add support for abbreviations expansion

---
 .../java/marytts/language/dsb/Preprocess.java | 44 ++++++++++++++++++-
 1 file changed, 43 insertions(+), 1 deletion(-)

diff --git a/src/main/java/marytts/language/dsb/Preprocess.java b/src/main/java/marytts/language/dsb/Preprocess.java
index c9a4de9..17d66ff 100644
--- a/src/main/java/marytts/language/dsb/Preprocess.java
+++ b/src/main/java/marytts/language/dsb/Preprocess.java
@@ -31,6 +31,7 @@
 public class Preprocess extends InternalModule {
 
     static final ULocale locale = new ULocale.Builder().setLanguage("dsb").build();
+    private Map<String, String> abbreviations;
     private Map<String, String> symbols;
     private RuleBasedNumberFormat ruleBasedNumberFormat;
     private NumberFormat numberFormat;
@@ -39,6 +40,26 @@ public Preprocess() throws MaryConfigurationException {
         super("Preprocess", MaryDataType.TOKENS, MaryDataType.WORDS, locale.toLocale());
         initNumberExpansion("formatRules.txt");
         initSymbolExpansion("symbols.csv");
+        initAbbreviationExpansion("abbreviations.csv");
+    }
+
+	private void initAbbreviationExpansion(String resourceName) throws MaryConfigurationException {
+        try {
+            abbreviations = new HashMap<>();
+            InputStream abbreviationsStream = this.getClass().getResourceAsStream(resourceName);
+            InputStreamReader abbreviationsReader = new InputStreamReader(abbreviationsStream, Charsets.UTF_8);
+            CSVParser csv = CSVFormat.Builder.create(CSVFormat.DEFAULT)
+                    .setHeader("abbreviation", "expansion")
+                    .build()
+                    .parse(abbreviationsReader);
+            for (CSVRecord record : csv) {
+                String abbreviation = record.get("abbreviation");
+                String expansion = record.get("expansion");
+                abbreviations.put(abbreviation, expansion);
+            }
+        } catch (Exception exception) {
+            throw new MaryConfigurationException(String.format("Could not load abbreviations from %s.%s", this.getClass().getCanonicalName(), resourceName), exception);
+        }
     }
 
     private void initSymbolExpansion(String resourceName) throws MaryConfigurationException {
@@ -73,6 +94,7 @@ private void initNumberExpansion(String resourceName) throws MaryConfigurationEx
 
     public MaryData process(MaryData d) {
         Document doc = d.getDocument();
+        expandAllAbbreviations(doc);
         expandAllSymbols(doc);
         expandAllNumbers(doc);
         MaryData result = new MaryData(getOutputType(), d.getLocale());
@@ -80,6 +102,26 @@ public MaryData process(MaryData d) {
         return result;
     }
 
+	private void expandAllAbbreviations(Document document) {
+        TreeWalker treeWalker = ((DocumentTraversal) document).createTreeWalker(document, NodeFilter.SHOW_ELEMENT,
+                new NameNodeFilter(MaryXML.TOKEN), false);
+        Element token;
+        while ((token = (Element) treeWalker.nextNode()) != null) {
+            String tokenText = MaryDomUtils.tokenText(token);
+            String expandedAbbreviation = expandAbbreviation(tokenText);
+            if (expandedAbbreviation != tokenText) {
+                MaryDomUtils.setTokenText(token, expandedAbbreviation);
+            }
+        }
+    }
+
+    protected String expandAbbreviation(String abbreviation) {
+        if (symbols.containsKey(abbreviation))
+            return abbreviations.get(abbreviation);
+        else
+            return abbreviation;
+    }
+
     private void expandAllSymbols(Document document) {
         TreeWalker treeWalker = ((DocumentTraversal) document).createTreeWalker(document, NodeFilter.SHOW_ELEMENT,
                 new NameNodeFilter(MaryXML.TOKEN), false);
@@ -126,4 +168,4 @@ protected String spelloutNumber(Number number) {
         return ruleBasedNumberFormat.format(number);
     }
 
-}
+}
\ No newline at end of file