From 2fdf5cd298a8767a8bf8e875fd90c92a05257462 Mon Sep 17 00:00:00 2001
From: =?utf8?q?Reimar=20D=C3=B6ffinger?= <Reimar.Doeffinger@gmx.de>
Date: Wed, 16 Dec 2015 23:35:44 +0100
Subject: [PATCH] Fix for splitting Mandarin/Cantonese/...

---
 .../dictionary/engine/WiktionarySplitter.java     | 15 +++++++++++++++
 .../parser/wiktionary/WiktionaryLangs.java        |  2 +-
 2 files changed, 16 insertions(+), 1 deletion(-)

diff --git a/src/com/hughes/android/dictionary/engine/WiktionarySplitter.java b/src/com/hughes/android/dictionary/engine/WiktionarySplitter.java
index a5dfebc..12b0c52 100644
--- a/src/com/hughes/android/dictionary/engine/WiktionarySplitter.java
+++ b/src/com/hughes/android/dictionary/engine/WiktionarySplitter.java
@@ -170,6 +170,7 @@ public class WiktionarySplitter extends org.xml.sax.helpers.DefaultHandler {
     }
     
     String text = textBuilder.toString();
+    String translingual = "";
     
     while (text.length() > 0) {
       // Find start.
@@ -181,6 +182,19 @@ public class WiktionarySplitter extends org.xml.sax.helpers.DefaultHandler {
       
       final String heading = startMatcher.group();
       for (final Selector selector : currentSelectors) {
+        if (heading.indexOf("Translingual") != -1) {
+          // Find end.
+          final int depth = startMatcher.group(1).length();
+          final Pattern endPattern = Pattern.compile(String.format("^={1,%d}[^=].*$", depth), Pattern.MULTILINE);
+
+          final Matcher endMatcher = endPattern.matcher(text);
+          if (endMatcher.find()) {
+            int end = endMatcher.start();
+            translingual = text.substring(0, endMatcher.start());
+            text = text.substring(end);
+            break;
+          }
+        }
         if (selector.pattern.matcher(heading).find()) {
           
           // Find end.
@@ -205,6 +219,7 @@ public class WiktionarySplitter extends org.xml.sax.helpers.DefaultHandler {
                      sectionText.charAt(dummy_end + 1) == '\n') ++dummy_end;
               sectionText = sectionText.substring(dummy_end);
           }
+          if (heading.indexOf("Japanese") == -1) sectionText += translingual;
           final Section section = new Section(title, heading, sectionText);
           
           try {
diff --git a/src/com/hughes/android/dictionary/parser/wiktionary/WiktionaryLangs.java b/src/com/hughes/android/dictionary/parser/wiktionary/WiktionaryLangs.java
index 5c039e8..813c69b 100644
--- a/src/com/hughes/android/dictionary/parser/wiktionary/WiktionaryLangs.java
+++ b/src/com/hughes/android/dictionary/parser/wiktionary/WiktionaryLangs.java
@@ -35,7 +35,6 @@ public class WiktionaryLangs {
     isoCodeToEnWikiName.put("BG", "Bulgarian");
     isoCodeToEnWikiName.put("CA", "Catalan");
     isoCodeToEnWikiName.put("SH", "Serbo-Croatian");
-    isoCodeToEnWikiName.put("HR", "Croatian");
     isoCodeToEnWikiName.put("CS", "Czech");
     isoCodeToEnWikiName.put("ZH", "Chinese");
     isoCodeToEnWikiName.put("cmn", "Mandarin");
@@ -113,6 +112,7 @@ public class WiktionaryLangs {
     // No longer exists in EN:
     // isoCodeToEnWikiName.put("BS", "Bosnian");
     // isoCodeToEnWikiName.put("SR", "Serbian");
+    // isoCodeToEnWikiName.put("HR", "Croatian");
     
     // Font doesn't work:
     //isoCodeToEnWikiName.put("MY", "Burmese");
-- 
2.43.0