Encode URLs as ASCII, avoid UTF-8.

[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 8a4bb216c9b75967fc713512ef9f0f32bc63fe34..502e92a9b542558b8dc653f57a026071eb2e9bc4 100644 (file)
--- a/src/com/hughes/android/dictionary/parser/wiktionary/WholeSectionToHtmlParser.java
+++ b/src/com/hughes/android/dictionary/parser/wiktionary/WholeSectionToHtmlParser.java
@@ -132,7 +132,6 @@ public class WholeSectionToHtmlParser extends AbstractWiktionaryParser {
             public void addFunctionCallbacks(
                     Map<String, FunctionCallback<WholeSectionToHtmlParser>> functionCallbacks) {
                 // TODO: need Spanish variant
-                EnFunctionCallbacks.addGenericCallbacks(functionCallbacks);
             }
         });
 
@@ -157,7 +156,7 @@ public class WholeSectionToHtmlParser extends AbstractWiktionaryParser {
             @Override
             public boolean skipWikiLink(WikiTokenizer wikiTokenizer) {
                 final String wikiText = wikiTokenizer.wikiLinkText();
-                if (wikiText.startsWith("???Category:")) {
+                if (wikiText.startsWith("Kategorie:")) {
                     return true;
                 }
                 return false;
@@ -205,7 +204,7 @@ public class WholeSectionToHtmlParser extends AbstractWiktionaryParser {
             @Override
             public boolean skipWikiLink(WikiTokenizer wikiTokenizer) {
                 final String wikiText = wikiTokenizer.wikiLinkText();
-                if (wikiText.startsWith("???Category:")) {
+                if (wikiText.startsWith("Categoria:")) {
                     return true;
                 }
                 return false;
@@ -233,7 +232,7 @@ public class WholeSectionToHtmlParser extends AbstractWiktionaryParser {
         });
 
 
-        final Pattern frSkipSections = Pattern.compile(".*(Traductions).*");
+        final Pattern frSkipSections = Pattern.compile(".*(Traductions|[Aa]nagrammes).*");
         isoToLangConfig.put("FR", new LangConfig() {
             @Override
             public boolean skipSection(String headingText) {
@@ -245,11 +244,18 @@ public class WholeSectionToHtmlParser extends AbstractWiktionaryParser {
                 if (sectionName.equalsIgnoreCase("Synonymes")) {
                     return EntryTypeName.SYNONYM_MULTI;
                 }
+                if (sectionName.equalsIgnoreCase("Antonymes")) {
+                    return EntryTypeName.ANTONYM_MULTI;
+                }
                 return null;
             }
             
             @Override
             public boolean skipWikiLink(WikiTokenizer wikiTokenizer) {
+                final String wikiText = wikiTokenizer.wikiLinkText();
+                if (wikiText.startsWith("Catégorie:")) {
+                    return true;
+                }
                 return false;
             }
             @Override
@@ -312,7 +318,7 @@ public class WholeSectionToHtmlParser extends AbstractWiktionaryParser {
             final String webUrl = String.format(webUrlTemplate, title);
            // URI.create can raise an exception e.g. if webUrl contains %, just ignore those cases.
            try {
-            callback.builder.append(String.format("<p> <a href=\"%s\">%s</a>", URI.create(webUrl).toString(), escapeHtmlLiteral(webUrl)));
+            callback.builder.append(String.format("<p> <a href=\"%s\">%s</a>", URI.create(webUrl).toASCIIString(), escapeHtmlLiteral(webUrl)));
            } catch (Exception e)
            {}
         }