From 467b2d17b3d3fb0635022fc00a09216a9efca58e Mon Sep 17 00:00:00 2001
From: Xiao YiFang <yifang.xiao@gmail.com>
Date: Mon, 23 May 2022 00:13:24 +0800
Subject: [PATCH 1/2] fix: duplicated entries in epwing dictionary

---
 epwing_book.cc | 42 +++++++++++++++++++++++++++++++++---------
 epwing_book.hh |  2 +-
 2 files changed, 34 insertions(+), 10 deletions(-)

diff --git a/epwing_book.cc b/epwing_book.cc
index bec5995d..111f031a 100644
--- a/epwing_book.cc
+++ b/epwing_book.cc
@@ -850,7 +850,7 @@ void EpwingBook::getFirstHeadword( EpwingHeadword & head )
   fixHeadword( head.headword );
 
   EWPos epos( pos.page, pos.offset );
-  allHeadwordPositions[ head.headword ] = epos;
+  allHeadwordPositions[ head.headword ] << epos;
 }
 
 bool EpwingBook::getNextHeadword( EpwingHeadword & head )
@@ -881,13 +881,25 @@ bool EpwingBook::getNextHeadword( EpwingHeadword & head )
 
       if( allHeadwordPositions.contains( head.headword ) )
       {
-        EWPos epos = allHeadwordPositions[ head.headword ];
-        if( pos.page != epos.first || abs( pos.offset - epos.second ) > 4 )
+        // existed position
+        bool existed = false;
+        foreach( EWPos epos, allHeadwordPositions[ head.headword ] )
+        {
+          if( pos.page == epos.first && abs( pos.offset - epos.second ) <= 4 )
+          {
+            existed = true;
+            break;
+          }
+        }
+        if( !existed )
+        {
+          allHeadwordPositions[ head.headword ] << EWPos( pos.page, pos.offset );
           return true;
+        }
       }
       else
       {
-        allHeadwordPositions[ head.headword ] = EWPos( pos.page, pos.offset );
+        allHeadwordPositions[ head.headword ]<<EWPos( pos.page, pos.offset );
         return true;
       }
     }
@@ -943,14 +955,26 @@ bool EpwingBook::getNextHeadword( EpwingHeadword & head )
 
     if( allHeadwordPositions.contains( head.headword ) )
     {
-      EWPos epos = allHeadwordPositions[ head.headword ];
-      if( pos.page != epos.first || abs( pos.offset - epos.second ) > 4 )
-        break;
+      // existed position
+      bool existed = false;
+      foreach( EWPos epos, allHeadwordPositions[ head.headword ] )
+      {
+        if( pos.page == epos.first && abs( pos.offset - epos.second ) <= 4 )
+        {
+          existed = true;
+          break;
+        }
+      }
+      if( !existed )
+      {
+        allHeadwordPositions[ head.headword ] << EWPos( pos.page, pos.offset );
+        return true;
+      }
     }
     else
     {
-      allHeadwordPositions[ head.headword ] = EWPos( pos.page, pos.offset );
-      break;
+      allHeadwordPositions[ head.headword ]<<EWPos( pos.page, pos.offset );
+      return true;
     }
   }
 
diff --git a/epwing_book.hh b/epwing_book.hh
index 5e2b99b5..49e5f9ad 100644
--- a/epwing_book.hh
+++ b/epwing_book.hh
@@ -78,7 +78,7 @@ class EpwingBook
   QStringList imageCacheList, soundsCacheList, moviesCacheList, fontsCacheList;
   QMap< QString, QString > baseFontsMap, customFontsMap;
   QVector< int > refPages, refOffsets;
-  QMap< QString, EWPos > allHeadwordPositions;
+  QMap< QString, QList<EWPos> > allHeadwordPositions;
   QVector< EWPos > LinksQueue;
   int refOpenCount, refCloseCount;
   static Mutex libMutex;

From b41bef9a314c73458eaded8ae3517acf9e7fa1eb Mon Sep 17 00:00:00 2001
From: Xiao YiFang <yifang.xiao@gmail.com>
Date: Mon, 23 May 2022 00:26:12 +0800
Subject: [PATCH 2/2] fix: skip too long headword in epwing dictionary

---
 epwing.cc | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/epwing.cc b/epwing.cc
index cd5eec5c..9d67cd2d 100644
--- a/epwing.cc
+++ b/epwing.cc
@@ -1054,7 +1054,8 @@ vector< sptr< Dictionary::Class > > makeDictionaries(
 
             for( ; ; )
             {
-              if( !head.headword.isEmpty() )
+              //skip too long headword
+              if( !head.headword.isEmpty() && head.headword.size() < 30 )
               {
                 uint32_t offset = chunks.startNewBlock();
                 chunks.addToBlock( &head.page, sizeof( head.page ) );