Fix flags of Scotland. Fix bug with URL encoding HTMLEntry titles.

[Dictionary.git] / src / com / hughes / android / dictionary / engine / HtmlEntry.java

diff --git a/src/com/hughes/android/dictionary/engine/HtmlEntry.java b/src/com/hughes/android/dictionary/engine/HtmlEntry.java
index 7e24a441f49b307ed885a9bddcb0690a81680018..612c6c1b7040ed1acd4fab4050768033a58ffe55 100644 (file)
--- a/src/com/hughes/android/dictionary/engine/HtmlEntry.java
+++ b/src/com/hughes/android/dictionary/engine/HtmlEntry.java
@@ -1,5 +1,10 @@
 package com.hughes.android.dictionary.engine;
 
+import android.content.Intent;
+import android.util.Log;
+
+import com.hughes.android.dictionary.C;
+import com.hughes.util.StringUtil;
 import com.hughes.util.raf.RAFListSerializer;
 import com.hughes.util.raf.RAFSerializable;
 import com.ibm.icu.text.Transliterator;
@@ -7,30 +12,44 @@ import com.ibm.icu.text.Transliterator;
 import java.io.IOException;
 import java.io.PrintStream;
 import java.io.RandomAccessFile;
+import java.lang.ref.SoftReference;
 import java.util.List;
 import java.util.regex.Pattern;
 
 public class HtmlEntry extends AbstractEntry implements RAFSerializable<HtmlEntry>, Comparable<HtmlEntry> {
   
-  // Both are HTML escaped already.
+  // Title is not HTML escaped.
   public final String title;
+  public final LazyHtmlLoader lazyHtmlLoader;
   public String html;
   
   public HtmlEntry(final EntrySource entrySource, String title) {
     super(entrySource);
     this.title = title;
+    lazyHtmlLoader = null;
   }
   
   public HtmlEntry(Dictionary dictionary, RandomAccessFile raf, final int index) throws IOException {
     super(dictionary, raf, index);
     title = raf.readUTF();
-    html = raf.readUTF();
+    lazyHtmlLoader = new LazyHtmlLoader(raf);
+    html = null;
   }
+  
   @Override
   public void write(RandomAccessFile raf) throws IOException {
     super.write(raf);
     raf.writeUTF(title);
-    raf.writeUTF(html);
+
+    final byte[] bytes = getHtml().getBytes("UTF-8");
+    final byte[] zipBytes = StringUtil.zipBytes(bytes);
+    raf.writeInt(bytes.length);
+    raf.writeInt(zipBytes.length);
+    raf.write(zipBytes);
+  }
+  
+  String getHtml() {
+      return html != null ? html : lazyHtmlLoader.getHtml();
   }
 
   @Override
@@ -45,7 +64,6 @@ public class HtmlEntry extends AbstractEntry implements RAFSerializable<HtmlEntr
     return new Row(this.index, rowIndex, dictionaryIndex);
   }
 
-  
   static final class Serializer implements RAFListSerializer<HtmlEntry> {
     
     final Dictionary dictionary;
@@ -66,7 +84,7 @@ public class HtmlEntry extends AbstractEntry implements RAFSerializable<HtmlEntr
   };
 
   public String getRawText(final boolean compact) {
-    return title + ":\n" + html;
+    return title + ":\n" + getHtml();
   }
 
   
@@ -75,7 +93,7 @@ public class HtmlEntry extends AbstractEntry implements RAFSerializable<HtmlEntr
     if (title.compareTo(another.title) != 0) {
       return title.compareTo(another.title);
     }
-    return html.compareTo(another.html);
+    return getHtml().compareTo(another.getHtml());
   }
   
   @Override
@@ -112,7 +130,7 @@ public class HtmlEntry extends AbstractEntry implements RAFSerializable<HtmlEntr
     @Override
     public void print(PrintStream out) {
       final HtmlEntry entry = getEntry();
-      out.println("HtmlEntry (shortened): " + entry.title);
+      out.println("See also HtmlEntry:" + entry.title);
     }
 
     @Override
@@ -135,17 +153,81 @@ public class HtmlEntry extends AbstractEntry implements RAFSerializable<HtmlEntr
       }
       return RowMatchType.BAG_OF_WORDS_MATCH;
     }
-    
   }
 
-    public static String htmlBody(final List<HtmlEntry> htmlEntries) {
+    public static String htmlBody(final List<HtmlEntry> htmlEntries, final String indexShortName) {
         final StringBuilder result = new StringBuilder();
         for (final HtmlEntry htmlEntry : htmlEntries) {
-            result.append(String.format("<h1><a href=\"%s\">%s</a></h1>\n(%s)\n<p>%s\n", 
-                    htmlEntry.title, htmlEntry.title, htmlEntry.entrySource.name,
-                    htmlEntry.html));
+            final String titleEscaped = StringUtil.escapeUnicodeToPureHtml(htmlEntry.title);
+            result.append(String.format("<h1><a href=\"%s\">%s</a></h1>\n<p>%s\n", 
+                    formatQuickdicUrl(indexShortName, htmlEntry.title), titleEscaped,
+                    htmlEntry.getHtml()));
         }
         return result.toString();
     }
+    
+    public static String formatQuickdicUrl(final String indexShortName, final String text) {
+        assert !indexShortName.contains(":");
+        assert text.length() > 0;
+        return String.format("q://d?%s&%s", indexShortName, StringUtil.encodeForUrl(text));
+    }
+    
+    public static boolean isQuickdicUrl(String url) {
+        return url.startsWith("q://d?");
+    }
+    
+    public static void quickdicUrlToIntent(final String url, final Intent intent) {
+        int firstColon = url.indexOf("?");
+        if (firstColon == -1) return;
+        int secondColon = url.indexOf("&", firstColon + 1);
+        if (secondColon == -1) return;
+        intent.putExtra(C.SEARCH_TOKEN, StringUtil.decodeFromUrl(url.substring(secondColon + 1)));
+    }
+    
+    // --------------------------------------------------------------------
+    
+    public static final class LazyHtmlLoader {
+        final RandomAccessFile raf;
+        final long offset;
+        final int numBytes;
+        final int numZipBytes;
+        
+        // Not sure this volatile is right, but oh well.
+        volatile SoftReference<String> htmlRef = new SoftReference<String>(null);
+        
+        private LazyHtmlLoader(final RandomAccessFile raf) throws IOException {
+            this.raf = raf;
+            numBytes = raf.readInt();
+            numZipBytes = raf.readInt();
+            offset = raf.getFilePointer();
+            raf.skipBytes(numZipBytes);
+        }
+        
+        public String getHtml() {
+            String html = htmlRef.get();
+            if (html != null) {
+                return html;
+            }
+            System.out.println("Loading Html: numBytes=" + numBytes + ", numZipBytes=" + numZipBytes);
+            final byte[] bytes = new byte[numBytes];
+            final byte[] zipBytes = new byte[numZipBytes];
+            synchronized (raf) {
+                try {
+                    raf.seek(offset);
+                    raf.read(zipBytes);
+                } catch (IOException e) {
+                    throw new RuntimeException(e);
+                }
+            }
+            try {
+                StringUtil.unzipFully(zipBytes, bytes);
+                html = new String(bytes, "UTF-8");
+            } catch (IOException e) {
+                throw new RuntimeException(e);
+            }
+            htmlRef = new SoftReference<String>(html);
+            return html;
+        }
+    }
 
 }