fix segmentation on CJK characters when updating selection

Makefile

@@ -347,7 +347,8 @@ fetchthirdparty:
 		git apply ../lvfntman-RegisterExternalFont.patch && \
 		git apply ../lvtinydom-registerEmbeddedFonts.patch && \
 		git apply ../epubfmt-EmbeddedFontStyleParser.patch && \
-		git apply ../lvtextfm-punctuation.patch
+		git apply ../lvtextfm-punctuation.patch && \
+		git apply ../lvtinydom-CJK-word.patch
 	# CREngine patch: disable fontconfig
 	grep USE_FONTCONFIG $(CRENGINE_DIR)/crengine/include/crsetup.h \
 		&& grep -v USE_FONTCONFIG \

kpvcrlib/lvtextfm-punctuation.patch

@@ -11,20 +11,85 @@ index 49125a7..2fd9f6a 100644
          int maxw = 0;
          for ( int i=0; chars[i]; i++ ) {
              int w = getCharWidth( chars[i] );
+diff --git a/crengine/include/lvstring.h b/crengine/include/lvstring.h
+index 6985cd6..a38c4fa 100644
+--- a/crengine/include/lvstring.h
++++ b/crengine/include/lvstring.h
+@@ -25,6 +25,14 @@
+ #define UNICODE_NO_BREAK_SPACE   0x00a0
+ #define UNICODE_HYPHEN   0x2010
+ #define UNICODE_NB_HYPHEN   0x2011
++#define UNICODE_CJK_IDEOGRAPHS_BEGIN 0x4e00
++#define UNICODE_CJK_IDEOGRAPHS_END 0x9FFF
++#define UNICODE_CJK_PUNCTUATION_BEGIN 0x3000
++#define UNICODE_CJK_PUNCTUATION_END 0x303F
++#define UNICODE_GENERAL_PUNCTUATION_BEGIN 0x2000
++#define UNICODE_GENERAL_PUNCTUATION_END 0x206F
++#define UNICODE_CJK_PUNCTUATION_HALF_AND_FULL_WIDTH_BEGIN 0xFF01
++#define UNICODE_CJK_PUNCTUATION_HALF_AND_FULL_WIDTH_END 0xFFEE
+
+
+
+@@ -91,6 +99,7 @@ int decodeDecimal( const lChar16 * str, int len );
+ #define CH_PROP_SIGN        0x0100 ///< sign character flag
+ #define CH_PROP_ALPHA_SIGN  0x0200 ///< alpha sign character flag
+ #define CH_PROP_DASH        0x0400 ///< minus, emdash, endash, ... (- signs)
++#define CH_PROP_CJK         0x0800 ///< CJK ideographs
+
+ /// retrieve character properties mask array for wide c-string
+ void lStr_getCharProps( const lChar16 * str, int sz, lUInt16 * props );
+diff --git a/crengine/src/lvstring.cpp b/crengine/src/lvstring.cpp
+index 27adddd..21611a0 100644
+--- a/crengine/src/lvstring.cpp
++++ b/crengine/src/lvstring.cpp
+@@ -4063,6 +4063,12 @@ inline lUInt16 getCharProp(lChar16 ch) {
+         return char_props_1f00[ch & 255];
+     else if (ch>=0x2012 && ch<=0x2015)
+         return CH_PROP_DASH|CH_PROP_SIGN;
++    else if (ch>=UNICODE_CJK_IDEOGRAPHS_BEGIN && ch<=UNICODE_CJK_IDEOGRAPHS_END)
++        return CH_PROP_CJK;
++    else if ((ch>=UNICODE_CJK_PUNCTUATION_BEGIN && ch<=UNICODE_CJK_PUNCTUATION_END) ||
++             (ch>=UNICODE_GENERAL_PUNCTUATION_BEGIN && ch<=UNICODE_GENERAL_PUNCTUATION_END) ||
++             (ch>=UNICODE_CJK_PUNCTUATION_HALF_AND_FULL_WIDTH_BEGIN && ch<=UNICODE_CJK_PUNCTUATION_HALF_AND_FULL_WIDTH_END))
++        return CH_PROP_PUNCT;
+     return 0;
+ }
+
 diff --git a/crengine/src/lvtextfm.cpp b/crengine/src/lvtextfm.cpp
-index 57bbf3b..e6f5913 100755
+index 57bbf3b..5239f9a 100755
 --- a/crengine/src/lvtextfm.cpp
 +++ b/crengine/src/lvtextfm.cpp
 @@ -712,7 +712,7 @@ public:
              } else {
                  lastWord = true;
              }
 -            if ( i>wstart && (newSrc!=lastSrc || space || lastWord) ) {
-+            if ( i>wstart && (newSrc!=lastSrc || space || lastWord || isCJKIdeograph(m_flags[i])) ) {
++            if ( i>wstart && (newSrc!=lastSrc || space || lastWord || isCJKIdeograph(m_text[i])) ) {
                  // create and add new word
                  formatted_word_t * word = lvtextAddFormattedWord(frmline);
                  int b;
-@@ -811,7 +811,9 @@ public:
+@@ -794,7 +794,10 @@ public:
+ //                            CRLog::trace("Double nbsp text[-1]=%04x", m_text[wstart-1]);
+ //                        else
+                         frmline->words[frmline->word_count-2].flags |= LTEXT_WORD_CAN_ADD_SPACE_AFTER;
+-                    } if ( m_flags[i-1] & LCHAR_ALLOW_WRAP_AFTER )
++                    } else if (frmline->word_count>1 && isCJKIdeograph(m_text[i])) {
++                        frmline->words[frmline->word_count-2].flags |= LTEXT_WORD_CAN_ADD_SPACE_AFTER;
++                    }
++                    if ( m_flags[i-1] & LCHAR_ALLOW_WRAP_AFTER )
+                         word->flags |= LTEXT_WORD_CAN_BREAK_LINE_AFTER;
+                     if ( word->t.start==0 && srcline->flags & LTEXT_IS_LINK )
+                         word->flags |= LTEXT_WORD_IS_LINK_START;
+@@ -803,7 +806,7 @@ public:
+                         int endp = i-1;
+                         int lastc = m_text[endp];
+                         int wAlign = font->getVisualAligmentWidth();
+-                        word->width += wAlign;
++                        word->width += wAlign/2;
+                         while ( (lastc==' ') && endp>0 ) { // || lastc=='\r' || lastc=='\n'
+                             word->width -= m_widths[endp] - m_widths[endp-1];
+                             endp--;
+@@ -811,7 +814,9 @@ public:
                          }
                          if ( word->flags & LTEXT_WORD_CAN_HYPH_BREAK_LINE_AFTER ) {
                              word->width -= font->getHyphenWidth(); // TODO: strange fix - need some other solution
@@ -35,19 +100,10 @@ index 57bbf3b..e6f5913 100755
                              int w = font->getCharWidth(lastc);
                              TR("floating: %c w=%d", lastc, w);
                              word->width -= w;
-@@ -869,6 +871,29 @@ public:
+@@ -869,6 +874,20 @@ public:
          return 0;
      }
 
-+#define UNICODE_CJK_IDEOGRAPHS_BEGIN 0x4e00
-+#define UNICODE_CJK_IDEOGRAPHS_END 0x9FFF
-+#define UNICODE_CJK_PUNCTUATION_BEGIN 0x3000
-+#define UNICODE_CJK_PUNCTUATION_END 0x303F
-+#define UNICODE_GENERAL_PUNCTUATION_BEGIN 0x2000
-+#define UNICODE_GENERAL_PUNCTUATION_END 0x206F
-+#define UNICODE_CJK_PUNCTUATION_HALF_AND_FULL_WIDTH_BEGIN 0xFF01
-+#define UNICODE_CJK_PUNCTUATION_HALF_AND_FULL_WIDTH_END 0xFFEE
-+
 +    bool isCJKIdeograph(lChar16 c) {
 +       return c >= UNICODE_CJK_IDEOGRAPHS_BEGIN && c <= UNICODE_CJK_IDEOGRAPHS_END;
 +    }
@@ -65,15 +121,15 @@ index 57bbf3b..e6f5913 100755
      /// Split paragraph into lines
      void processParagraph( int start, int end )
      {
-@@ -927,7 +952,6 @@ public:
+@@ -927,7 +946,6 @@ public:
                      }
                  }
              }
 -            maxWidth -= visialAlignmentWidth;
          }
  #endif
 
-@@ -944,6 +968,8 @@ public:
+@@ -944,6 +962,8 @@ public:
              int lastMandatoryWrap = -1;
              int spaceReduceWidth = 0; // max total line width which can be reduced by narrowing of spaces
              int firstCharMargin = getAdditionalCharWidthOnLeft(pos); // for first italic char with elements below baseline
@@ -82,7 +138,7 @@ index 57bbf3b..e6f5913 100755
              for ( i=pos; i<m_length; i++ ) {
                  if ( x + m_widths[i]-w0 > maxWidth + spaceReduceWidth - firstCharMargin)
                      break;
-@@ -952,7 +978,7 @@ public:
+@@ -952,7 +972,7 @@ public:
                      lastMandatoryWrap = i;
                      break;
                  }
@@ -91,7 +147,7 @@ index 57bbf3b..e6f5913 100755
                      lastNormalWrap = i;
                  else if ( flags & LCHAR_DEPRECATED_WRAP_AFTER )
                      lastDeprecatedWrap = i;
-@@ -1029,6 +1055,32 @@ public:
+@@ -1029,6 +1049,32 @@ public:
              }
              bool needReduceSpace = true; // todo: calculate whether space reducing required
              int endp = wrapPos+(lastMandatoryWrap<0 ? 1 : 0);

kpvcrlib/lvtinydom-CJK-word.patch

@@ -0,0 +1,15 @@
+diff --git a/crengine/src/lvtinydom.cpp b/crengine/src/lvtinydom.cpp
+index fe2b879..df50386 100755
+--- a/crengine/src/lvtinydom.cpp
++++ b/crengine/src/lvtinydom.cpp
+@@ -6874,6 +6874,10 @@ public:
+                 _list.add( ldomWord( node, beginOfWord, i ) );
+                 beginOfWord = -1;
+             }
++            if (lGetCharProps(text[i]) == CH_PROP_CJK) {
++                _list.add( ldomWord( node, i, i+1 ) );
++                beginOfWord = -1;
++            }
+         }
+     }
+     /// called for each found node in range