X-Git-Url: http://gitweb.fperrin.net/?a=blobdiff_plain;f=src%2Fcom%2Fhughes%2Fandroid%2Fdictionary%2Fparser%2Fwiktionary%2FWholeSectionToHtmlParser.java;h=0f7ae2d1cef8d94876a6b97ad0627a7633ee4562;hb=1b515f031d39e758e8e6339c03e124f1548579cc;hp=70a02c37f826433b9cda1a6626339ebf0dffce61;hpb=cd86f5f7abf5e545daf89c03d9dbbfd4e5bc65a4;p=DictionaryPC.git

diff --git a/src/com/hughes/android/dictionary/parser/wiktionary/WholeSectionToHtmlParser.java b/src/com/hughes/android/dictionary/parser/wiktionary/WholeSectionToHtmlParser.java
index 70a02c3..0f7ae2d 100644
--- a/src/com/hughes/android/dictionary/parser/wiktionary/WholeSectionToHtmlParser.java
+++ b/src/com/hughes/android/dictionary/parser/wiktionary/WholeSectionToHtmlParser.java
@@ -1,6 +1,7 @@
 
 package com.hughes.android.dictionary.parser.wiktionary;
 
+import com.hughes.android.dictionary.HtmlDisplayActivity;
 import com.hughes.android.dictionary.engine.EntryTypeName;
 import com.hughes.android.dictionary.engine.HtmlEntry;
 import com.hughes.android.dictionary.engine.IndexBuilder;
@@ -23,8 +24,9 @@ public class WholeSectionToHtmlParser extends AbstractWiktionaryParser {
 
     interface LangConfig {
         boolean skipSection(final String name);
+        EntryTypeName sectionNameToEntryType(String sectionName);
         boolean skipWikiLink(final WikiTokenizer wikiTokenizer);
-        String adjustWikiLink(String wikiLinkDest);
+        String adjustWikiLink(String wikiLinkDest, final String wikiLinkText);
         void addFunctionCallbacks(
                 Map<String, FunctionCallback<WholeSectionToHtmlParser>> functionCallbacks);
     }
@@ -36,7 +38,25 @@ public class WholeSectionToHtmlParser extends AbstractWiktionaryParser {
             public boolean skipSection(String headingText) {
                 return enSkipSections.matcher(headingText).matches();
             }
-
+            
+            @Override
+            public EntryTypeName sectionNameToEntryType(String sectionName) {
+                if (sectionName.equalsIgnoreCase("Synonyms")) {
+                    return EntryTypeName.SYNONYM_MULTI;
+                }
+                if (sectionName.equalsIgnoreCase("Antonyms")) {
+                    return EntryTypeName.ANTONYM_MULTI;
+                }
+                if (EnParser.partOfSpeechHeader.matcher(sectionName).matches()) {
+                    // We need to put it in the other index, too.
+                    return null;
+                }
+                if (sectionName.equalsIgnoreCase("Derived Terms")) {
+                    return null;
+                }
+                return null;
+            }
+            
             @Override
             public boolean skipWikiLink(WikiTokenizer wikiTokenizer) {
                 final String wikiText = wikiTokenizer.wikiLinkText();
@@ -46,10 +66,17 @@ public class WholeSectionToHtmlParser extends AbstractWiktionaryParser {
                 return false;
             }
             @Override
-            public String adjustWikiLink(String wikiLinkDest) {
+            public String adjustWikiLink(String wikiLinkDest, String wikiLinkText) {
                 if (wikiLinkDest.startsWith("w:") || wikiLinkDest.startsWith("Image:")) {
                     return null;
                 }
+                final int hashPos = wikiLinkDest.indexOf("#");
+                if (hashPos != -1) {
+                    wikiLinkDest = wikiLinkDest.substring(0, hashPos);
+                    if (wikiLinkDest.isEmpty()) {
+                        wikiLinkDest = wikiLinkText;
+                    }
+                }
                 return wikiLinkDest;
             }
 
@@ -57,14 +84,18 @@ public class WholeSectionToHtmlParser extends AbstractWiktionaryParser {
             public void addFunctionCallbacks(
                     Map<String, FunctionCallback<WholeSectionToHtmlParser>> functionCallbacks) {
                 EnFunctionCallbacks.addGenericCallbacks(functionCallbacks);
-            }});
+            }
+        });
         
         final LangConfig basicLangConfig = new LangConfig() {
             @Override
             public boolean skipSection(String headingText) {
                 return false;
             }
-
+            @Override
+            public EntryTypeName sectionNameToEntryType(String sectionName) {
+                return EntryTypeName.WIKTIONARY_MENTIONED;
+            }
             @Override
             public boolean skipWikiLink(WikiTokenizer wikiTokenizer) {
                 final String wikiText = wikiTokenizer.wikiLinkText();
@@ -74,7 +105,7 @@ public class WholeSectionToHtmlParser extends AbstractWiktionaryParser {
                 return false;
             }
             @Override
-            public String adjustWikiLink(String wikiLinkDest) {
+            public String adjustWikiLink(String wikiLinkDest, final String wikiLinkText) {
                 return wikiLinkDest;
             }
 
@@ -89,11 +120,14 @@ public class WholeSectionToHtmlParser extends AbstractWiktionaryParser {
     }
 
     final IndexBuilder titleIndexBuilder;
+    final IndexBuilder defIndexBuilder;
     final String skipLangIso;
     final LangConfig langConfig;
+    
 
-    public WholeSectionToHtmlParser(final IndexBuilder titleIndexBuilder, final String wiktionaryIso, final String skipLangIso) {
+    public WholeSectionToHtmlParser(final IndexBuilder titleIndexBuilder, final IndexBuilder defIndexBuilder, final String wiktionaryIso, final String skipLangIso) {
         this.titleIndexBuilder = titleIndexBuilder;
+        this.defIndexBuilder = defIndexBuilder;
         assert isoToLangConfig.containsKey(wiktionaryIso): wiktionaryIso;
         this.langConfig = isoToLangConfig.get(wiktionaryIso);
         this.skipLangIso = skipLangIso;
@@ -104,7 +138,7 @@ public class WholeSectionToHtmlParser extends AbstractWiktionaryParser {
     @Override
     public void parseSection(String heading, String text) {
         assert entrySource != null;
-        final HtmlEntry htmlEntry = new HtmlEntry(entrySource, StringEscapeUtils.escapeHtml3(title));
+        final HtmlEntry htmlEntry = new HtmlEntry(entrySource, title);
         indexedEntry = new IndexedEntry(htmlEntry);
 
         final AppendAndIndexWikiCallback<WholeSectionToHtmlParser> callback = new AppendCallback(
@@ -139,8 +173,6 @@ public class WholeSectionToHtmlParser extends AbstractWiktionaryParser {
 
 
 
-    static final Pattern ALL_ASCII = Pattern.compile("[\\p{ASCII}]*");
-
     class AppendCallback extends AppendAndIndexWikiCallback<WholeSectionToHtmlParser> {
         public AppendCallback(WholeSectionToHtmlParser parser) {
             super(parser);
@@ -149,7 +181,7 @@ public class WholeSectionToHtmlParser extends AbstractWiktionaryParser {
         @Override
         public void onPlainText(String plainText) {
             final String htmlEscaped = StringEscapeUtils.escapeHtml3(plainText);
-            if (ALL_ASCII.matcher(htmlEscaped).matches()) {
+            if (StringUtil.isAscii(htmlEscaped)) {
                 super.onPlainText(htmlEscaped);
             } else { 
                 super.onPlainText(StringUtil.escapeToPureHtmlUnicode(plainText));
@@ -167,12 +199,16 @@ public class WholeSectionToHtmlParser extends AbstractWiktionaryParser {
             }
             String linkDest;
             if (wikiTokenizer.wikiLinkDest() != null) {
-                linkDest = langConfig.adjustWikiLink(wikiTokenizer.wikiLinkDest());
+                linkDest = langConfig.adjustWikiLink(wikiTokenizer.wikiLinkDest(), wikiTokenizer.wikiLinkText());
             } else {
                 linkDest = wikiTokenizer.wikiLinkText();
             }
+            if (sectionEntryTypeName != null) {
+                // TODO: inside a definition, this could be the wrong language.
+                titleIndexBuilder.addEntryWithString(indexedEntry, wikiTokenizer.wikiLinkText(), sectionEntryTypeName);
+            }
             if (linkDest != null) {
-                builder.append(String.format("<a href=\"%s\">", linkDest));
+                builder.append(String.format("<a href=\"%s\">", HtmlEntry.formatQuickdicUrl("", linkDest)));
                 super.onWikiLink(wikiTokenizer);
                 builder.append(String.format("</a>"));
             } else {
@@ -197,10 +233,14 @@ public class WholeSectionToHtmlParser extends AbstractWiktionaryParser {
         @Override
         public void onNewline(WikiTokenizer wikiTokenizer) {
         }
+        
+        EntryTypeName sectionEntryTypeName;
+        IndexBuilder currentIndexBuilder;
 
         @Override
         public void onHeading(WikiTokenizer wikiTokenizer) {
             final String headingText = wikiTokenizer.headingWikiText();
+            sectionEntryTypeName = langConfig.sectionNameToEntryType(headingText);
             final int depth = wikiTokenizer.headingDepth();
             if (langConfig.skipSection(headingText)) {
                 while ((wikiTokenizer = wikiTokenizer.nextToken()) != null) {