From 35a897b248ca48583149dd2b26de63d68b9bfbbb Mon Sep 17 00:00:00 2001
From: Andy McFadden <thefadden@gmail.com>
Date: Sat, 11 Nov 2017 14:53:28 -0800
Subject: [PATCH] Add support for AW5 inverse and MouseText

AppleWorks 5 added inverse text and MouseText.  We can handle inverse
text with RTF features, and convert MouseText to something vaguely
similar in the Unicode symbol set.
---
 reformat/AppleWorks.cpp   | 89 ++++++++++++++++++++++++++++++++-------
 reformat/ReformatBase.cpp | 77 +++++++++++++++++++++++++++++++--
 reformat/ReformatBase.h   | 11 ++++-
 3 files changed, 156 insertions(+), 21 deletions(-)

diff --git a/reformat/AppleWorks.cpp b/reformat/AppleWorks.cpp
index 756ccde..b45fdde 100644
--- a/reformat/AppleWorks.cpp
+++ b/reformat/AppleWorks.cpp
@@ -24,6 +24,20 @@
  *  array of line records
  *  $ff $ff
  *  optional tags
+ *
+ * AppleWorks 5.0 introduced inverse and MouseText characters.
+ * These just use previously-unused byte ranges.  The full set
+ * of values is thus:
+ *  00-1f special
+ *  20-7f plain ASCII
+ *  80-9f inverse upper (map to 40-5f)
+ *  a0-bf inverse symbols/numbers (map to 20-3f)
+ *  c0-df MouseText
+ *  e0-ff inverse lower (map to 60-7f)
+ *
+ * We can output MouseText as Unicode symbols.  Inverse text can use the Rich
+ * Text "highlight" feature; the "background color" feature doesn't seem to
+ * have any effect.
  */
 
 /*
@@ -308,6 +322,7 @@ int ReformatAWP::HandleTextRecord(uint8_t lineRecData,
     uint8_t byteCountPlusCR;
     int byteCount = lineRecData;
     bool noOutput = false;
+    bool inverse = false;
     int ic;
 
     tabFlags = Read8(pSrcPtr, pLength);
@@ -372,40 +387,40 @@ int ReformatAWP::HandleTextRecord(uint8_t lineRecData,
                 break;
             case kSpecialCharEnterKeyboard:
                 if (fShowEmbeds) {
-                    RTFSetColor(kColorBlue);
+                    TextColor oldColor = RTFSetColor(kColorBlue);
                     BufPrintf("<kdb-entry>");
-                    RTFSetColor(kColorNone);
+                    RTFSetColor(oldColor);
                 }
                 break;
             case kSpecialCharPrintPageNumber:
                 if (fShowEmbeds) {
-                    RTFSetColor(kColorBlue);
+                    TextColor oldColor = RTFSetColor(kColorBlue);
                     BufPrintf("<page#>");
-                    RTFSetColor(kColorNone);
+                    RTFSetColor(oldColor);
                 }
                 break;
             case kSpecialCharStickySpace:
                 /* MSWord uses "\~", but RichEdit ignores that */
-                BufPrintf(" ");
+                BufPrintf("\u00a0");    // Unicode NO-BREAK SPACE
                 break;
             case kSpecialCharMailMerge:
                 if (fShowEmbeds) {
-                    RTFSetColor(kColorBlue);
+                    TextColor oldColor = RTFSetColor(kColorBlue);
                     BufPrintf("<mail-merge>");
-                    RTFSetColor(kColorNone);
+                    RTFSetColor(oldColor);
                 }
             case kSpecialCharPrintDate:
                 if (fShowEmbeds) {
-                    RTFSetColor(kColorBlue);
+                    TextColor oldColor = RTFSetColor(kColorBlue);
                     BufPrintf("<date>");
-                    RTFSetColor(kColorNone);
+                    RTFSetColor(oldColor);
                 }
                 break;
             case kSpecialCharPrintTime:
                 if (fShowEmbeds) {
-                    RTFSetColor(kColorBlue);
+                    TextColor oldColor = RTFSetColor(kColorBlue);
                     BufPrintf("<time>");
-                    RTFSetColor(kColorNone);
+                    RTFSetColor(oldColor);
                 }
                 break;
             case kSpecialCharTab:
@@ -421,19 +436,61 @@ int ReformatAWP::HandleTextRecord(uint8_t lineRecData,
             default:
                 LOGI(" AWP unhandled special char 0x%02x", ic);
                 if (fShowEmbeds) {
-                    RTFSetColor(kColorBlue);
+                    TextColor oldColor = RTFSetColor(kColorBlue);
                     BufPrintf("^");
-                    RTFSetColor(kColorNone);
+                    RTFSetColor(oldColor);
                 }
             }
         } else {
-            if (fUseRTF)
-                RTFPrintChar(ic);
-            else
+            // Character.
+            bool wantInverse = false;
+            uint16_t mtLow = 0, mtHigh = 0;
+
+            if (ic >= 0x80 && ic <= 0x9f) {
+                // inverse upper; map 100x xxxx --> 010x xxxx
+                ic ^= 0xc0;
+                wantInverse = true;
+            } else if (ic >= 0xa0 && ic <= 0xbf || ic >= 0xe0 && ic <= 0xff) {
+                // inverse symbols; map 101x xxxx --> 001x xxxx
+                // inverse lower; map 111x xxxx --> 011x xxxx
+                ic ^= 0x80;
+                wantInverse = true;
+            } else if (ic >= 0xc0 && ic <= 0xdf) {
+                // MouseText characters
+                MouseTextToUTF16(ic & 0x1f, &mtLow, &mtHigh);
+                ic = '?';
+            } else {
+                // plain ASCII
+            }
+
+            if (wantInverse && !inverse) {
+                inverse = true;
+                RTFInverseOn();
+            } else if (!wantInverse && inverse) {
+                inverse = false;
+                RTFInverseOff();
+            }
+
+            if (fUseRTF) {
+                if (mtLow != 0) {
+                    if (mtHigh != 0) {
+                        RTFPrintUTF16Char(mtHigh);
+                    }
+                    RTFPrintUTF16Char(mtLow);
+                } else {
+                    RTFPrintChar(ic);
+                }
+            } else {
+                // Plain text output.
                 BufPrintf("%c", PrintableChar(ic));
+            }
         }
     }
 
+    if (inverse) {
+        RTFInverseOff();
+    }
+
     /* if there's a carriage return at the end of the line, output it now */
     if (byteCountPlusCR & kCRatEOL) {
         RTFNewPara();
diff --git a/reformat/ReformatBase.cpp b/reformat/ReformatBase.cpp
index 9dff6bb..0208661 100644
--- a/reformat/ReformatBase.cpp
+++ b/reformat/ReformatBase.cpp
@@ -241,6 +241,28 @@ void ReformatText::RTFUnderlineOff(void)
     }
 }
 
+void ReformatText::RTFInverseOn(void)
+{
+    if (fInverseEnabled)
+        return;
+    if (fUseRTF) {
+        RTFSetColor(TextColor::kColorWhite);
+        BufPrintf("\\highlight1 ");     // black
+        fInverseEnabled = true;
+    }
+}
+
+void ReformatText::RTFInverseOff(void)
+{
+    if (!fInverseEnabled)
+        return;
+    if (fUseRTF) {
+        RTFSetColor(TextColor::kColorNone);
+        BufPrintf("\\highlight0 ");
+        fInverseEnabled = false;
+    }
+}
+
 void ReformatText::RTFOutlineOn(void)
 {
     if (fOutlineEnabled)
@@ -321,14 +343,14 @@ void ReformatText::RTFSuperscriptOff(void)
     }
 }
 
-void ReformatText::RTFSetColor(TextColor color)
+ReformatText::TextColor ReformatText::RTFSetColor(TextColor color)
 {
-    if (color == fTextColor)
-        return;
-    if (fUseRTF) {
+    TextColor oldColor = fTextColor;
+    if (color != fTextColor && fUseRTF) {
         BufPrintf("\\cf%d ", color);
         fTextColor = color;
     }
+    return oldColor;
 }
 
 /*
@@ -721,6 +743,53 @@ void ReformatText::BufHexDump(const uint8_t* srcBuf, long srcLen)
     }
 }
 
+// Thanks to: http://hoop-la.ca/apple2/docs/mousetext/unicode.html
+static const uint32_t gMouseTextConv[32] = {
+    0xd83cdf4e,     // 00 U+1f34e RED APPLE
+    0xd83cdf4f,     // 01 U+1f34f GREEN APPLE
+    //0x00002316,     // 02 U+2316 POSITION INDICATOR
+    0x000025c4,     // 02 U+25c4 BLACK LEFT-POINTING POINTER
+    0x000023f3,     // 03 U+23f3 HOURGLASS WITH FLOWING SAND
+    0x00002713,     // 04 U+2713 CHECK MARK
+    0x00002705,     // 05 U+2705 WHITE HEAVY CHECK MARK
+    0x000023ce,     // 06 U+23ce RETURN SYMBOL
+    //0xd83cdf54,     // 07 U+1f354 HAMBURGER
+    0x00002630,     // 07 U+2630 TRIGRAM FOR HEAVEN (actually want 4 lines, not 3)
+    0x00002190,     // 08 U+2190 LEFTWARDS ARROW
+    0x00002026,     // 09 U+2026 HORIZONTAL ELLIPSIS
+    0x00002193,     // 0a U+2193 DOWNWARDS ARROW
+    0x00002191,     // 0b U+2191 UPWARDS ARROW
+    0x00002594,     // 0c U+2594 UPPER ONE EIGHTH BLOCK
+    0x000021b5,     // 0d U+21b5 DOWNWARDS ARROW WITH CORNER LEFTWARDS
+    0x00002589,     // 0e U+2589 LEFT SEVEN EIGHTHS BLOCK
+    0x000021e4,     // 0f U+21e4 LEFTWARDS ARROW TO BAR
+    0x000021e5,     // 10 U+21e5 RIGHTWARDS ARROW TO BAR
+    0x00002913,     // 11 U+2913 DOWNWARDS ARROW TO BAR
+    0x00002912,     // 12 U+2912 UPWARDS ARROW TO BAR
+    0x00002500,     // 13 U+2500 BOX DRAWINGS LIGHT HORIZONTAL
+    0x0000231e,     // 14 U+231e BOTTOM LEFT CORNER
+    0x00002192,     // 15 U+2192 UPWARDS ARROW TO BAR
+    //0xd83dde7e,     // 16 U+1f67e CHECKER BOARD
+    0x00002591,     // 16 U+2591 LIGHT SHADE
+    //0xd83dde7f,     // 17 U+1f67f REVERSE CHECKER BOARD
+    0x00002592,     // 17 U+2592 MEDIUM SHADE
+    0xd83ddcc1,     // 18 U+1f4c1 FILE FOLDER
+    0xd83ddcc2,     // 19 U+1f4c2 OPEN FILE FOLDER
+    0x00002595,     // 1a U+2595 RIGHT ONE EIGHTH BLOCK
+    0x00002666,     // 1b U+2666 BLACK DIAMOND SUIT
+    0x0000203e,     // 1c U+203e OVERLINE -- wrong, want top/bottom
+    0x0000256c,     // 1d U+256c BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL
+    //0xd83ddcbe,     // 1e U+1f4be FLOPPY DISK
+    0x000022a1,     // 1e U+22a1 SQUARED DOT OPERATOR (seems better than 25a3)
+    0x0000258f,     // 1f U+258f LEFT ONE EIGHTH BLOCK
+};
+void ReformatText::MouseTextToUTF16(uint8_t mtVal, uint16_t* pLow, uint16_t* pHigh) {
+    ASSERT(mtVal < 32);
+
+    *pLow = gMouseTextConv[mtVal] & 0xffff;
+    *pHigh = gMouseTextConv[mtVal] >> 16;
+}
+
 
 /*
  * ==========================================================================
diff --git a/reformat/ReformatBase.h b/reformat/ReformatBase.h
index 884d459..c7a4e4d 100644
--- a/reformat/ReformatBase.h
+++ b/reformat/ReformatBase.h
@@ -198,6 +198,7 @@ public:
         fBoldEnabled(false),
         fItalicEnabled(false),
         fUnderlineEnabled(false),
+        fInverseEnabled(false),
         fOutlineEnabled(false),
         fShadowEnabled(false),
         fSuperscriptEnabled(false),
@@ -305,6 +306,8 @@ protected:
     void RTFItalicOff(void);
     void RTFUnderlineOn(void);
     void RTFUnderlineOff(void);
+    void RTFInverseOn(void);
+    void RTFInverseOff(void);
     void RTFParaLeft(void);
     void RTFParaRight(void);
     void RTFParaCenter(void);
@@ -317,7 +320,7 @@ protected:
     void RTFOutlineOff(void);
     void RTFShadowOn(void);
     void RTFShadowOff(void);
-    void RTFSetColor(TextColor color);
+    TextColor RTFSetColor(TextColor color);
     void RTFSetFont(RTFFont font);
     void RTFSetFontSize(int points);
     void RTFLeftMargin(int margin);
@@ -382,6 +385,11 @@ protected:
             fExpBuf.Printf("%c", ch);
     }
 
+    // Converts a MouseText value (0-31) to a 16-bit UTF-16 value.  If
+    // the character is in the BMP then only low is used.  If it requires
+    // encoding as a surrogate pair, high will be nonzero.
+    void MouseTextToUTF16(uint8_t mtVal, uint16_t* pLow, uint16_t* pHigh);
+
 private:
     DECLARE_COPY_AND_OPEQ(ReformatText)
     int CreateWorkBuf(void);
@@ -394,6 +402,7 @@ private:
     bool    fBoldEnabled;
     bool    fItalicEnabled;
     bool    fUnderlineEnabled;
+    bool    fInverseEnabled;
     bool    fOutlineEnabled;
     bool    fShadowEnabled;
     bool    fSuperscriptEnabled;