Italian verb conjugations!

[DictionaryPC.git] / src / com / hughes / android / dictionary / parser / wiktionary / WholeSectionToHtmlParser.java

diff --git a/src/com/hughes/android/dictionary/parser/wiktionary/WholeSectionToHtmlParser.java b/src/com/hughes/android/dictionary/parser/wiktionary/WholeSectionToHtmlParser.java
index bba139b6bd45c15c7af3812a3294f943c81c5dd1..57f4d587010c6d66d722f26acd8afb640daf293c 100644 (file)
--- a/src/com/hughes/android/dictionary/parser/wiktionary/WholeSectionToHtmlParser.java
+++ b/src/com/hughes/android/dictionary/parser/wiktionary/WholeSectionToHtmlParser.java
@@ -6,8 +6,10 @@ import com.hughes.android.dictionary.engine.IndexBuilder;
 import com.hughes.android.dictionary.engine.IndexBuilder.TokenData;
 import com.hughes.android.dictionary.engine.IndexedEntry;
 import com.hughes.android.dictionary.parser.WikiTokenizer;
+import com.hughes.util.StringUtil;
 
 import org.apache.commons.lang3.StringEscapeUtils;
+import org.apache.commons.lang3.StringUtils;
 
 import java.util.ArrayList;
 import java.util.LinkedHashMap;
@@ -23,6 +25,8 @@ public class WholeSectionToHtmlParser extends AbstractWiktionaryParser {
         boolean skipSection(final String name);
         boolean skipWikiLink(final WikiTokenizer wikiTokenizer);
         String adjustWikiLink(String wikiLinkDest);
+        void addFunctionCallbacks(
+                Map<String, FunctionCallback<WholeSectionToHtmlParser>> functionCallbacks);
     }
     static final Map<String,LangConfig> isoToLangConfig = new LinkedHashMap<String,LangConfig>();
     static {
@@ -47,6 +51,12 @@ public class WholeSectionToHtmlParser extends AbstractWiktionaryParser {
                     return null;
                 }
                 return wikiLinkDest;
+            }
+
+            @Override
+            public void addFunctionCallbacks(
+                    Map<String, FunctionCallback<WholeSectionToHtmlParser>> functionCallbacks) {
+                EnFunctionCallbacks.addGenericCallbacks(functionCallbacks);
             }});
     }
 
@@ -62,12 +72,13 @@ public class WholeSectionToHtmlParser extends AbstractWiktionaryParser {
     }
 
     @Override
-    void parseSection(String heading, String text) {
+    public void parseSection(String heading, String text) {
         HtmlEntry htmlEntry = new HtmlEntry(entrySource, StringEscapeUtils.escapeHtml3(title));
         IndexedEntry indexedEntry = new IndexedEntry(htmlEntry);
 
         final AppendAndIndexWikiCallback<WholeSectionToHtmlParser> callback = new AppendCallback(
                 this);
+        langConfig.addFunctionCallbacks(callback.functionCallbacks);
 
         callback.builder = new StringBuilder();
         callback.indexedEntry = indexedEntry;
@@ -87,6 +98,8 @@ public class WholeSectionToHtmlParser extends AbstractWiktionaryParser {
     @Override
     void removeUselessArgs(Map<String, String> namedArgs) {
     }
+    
+    static final Pattern ALL_ASCII = Pattern.compile("[\\p{ASCII}]*");
 
     class AppendCallback extends AppendAndIndexWikiCallback<WholeSectionToHtmlParser> {
         public AppendCallback(WholeSectionToHtmlParser parser) {
@@ -95,7 +108,12 @@ public class WholeSectionToHtmlParser extends AbstractWiktionaryParser {
 
         @Override
         public void onPlainText(String plainText) {
-            super.onPlainText(StringEscapeUtils.escapeHtml3(plainText));
+            final String htmlEscaped = StringEscapeUtils.escapeHtml3(plainText);
+            if (ALL_ASCII.matcher(htmlEscaped).matches()) {
+                super.onPlainText(htmlEscaped);
+            } else { 
+                super.onPlainText(StringUtil.escapeToPureHtmlUnicode(plainText));
+            }
         }
 
         @Override