Misc bug fixes from Karl Koehler.

Mon May  1 11:58:55 2000  Owen Taylor  <otaylor@redhat.com>

	* modules/arabic/*: Misc bug fixes from Karl Koehler.

5 files changed, 74 insertions, 71 deletions

diff --git a/modules/arabic/arabic-x.c b/modules/arabic/arabic-x.c
index 098799ce..ec598b17 100644
--- a/modules/arabic/arabic-x.c
+++ b/modules/arabic/arabic-x.c
@@ -17,7 +17,7 @@
 #include "mulefont.h"
 #include "langboxfont.h"
 
-/*  #define DEBUG   */
+/*  #define DEBUG    */
 #ifdef DEBUG
 #include <stdio.h>
 #endif
@@ -161,9 +161,10 @@ find_unic_font (PangoFont *font,char* charsets[],PangoXSubfont* rfonts)
 }
 
 static char *default_charset[] = {
-      "iso10646-1", 
-      "iso8859-6.8x",
-      "mulearabic-2" 
+    "iso10646-1",  
+    "iso8859-6.8x",
+    /*          "iso8859-6.8x" */
+    "mulearabic-2"    
 };
   
 
@@ -171,7 +172,7 @@ static char *default_charset[] = {
 static void
 set_glyph (PangoGlyphString *glyphs, 
            PangoFont *font, PangoXSubfont subfont,
-           int i, int cluster_start, int glyph)
+           int i, int cluster_start, int glyph, int is_vowel)
 {
     PangoRectangle logical_rect;
 #ifdef DEBUG
@@ -193,7 +194,7 @@ set_glyph (PangoGlyphString *glyphs,
 
     pango_font_get_glyph_extents (font, glyphs->glyphs[i].glyph, NULL, &logical_rect);
     glyphs->log_clusters[i] = cluster_start;
-    if (arabic_isvowel(glyph))
+    if (is_vowel)
         {
             glyphs->glyphs[i].geometry.width = 0;
         }
@@ -248,7 +249,7 @@ arabic_engine_shape (PangoFont        *font,
                 {
                     set_glyph (glyphs, font, 
                                PANGO_X_GLYPH_SUBFONT (unknown_glyph), i, 
-                               p - text, PANGO_X_GLYPH_INDEX (unknown_glyph));
+                               p - text, PANGO_X_GLYPH_INDEX (unknown_glyph),0);
                     p = unicode_next_utf8 (p);
                 }
             return;
@@ -282,6 +283,9 @@ arabic_engine_shape (PangoFont        *font,
     p    = text;
     pold = p;
     i    = n_chars-1;
+#ifdef DEBUG
+    fprintf(stderr,"[ar]: after shaping : %i glyphs ",n_glyph);
+#endif
     while(i >= 0)
         {
             if (wc[i] == 0)
@@ -290,9 +294,11 @@ arabic_engine_shape (PangoFont        *font,
                     i--;
                 }
             else 
-                {
-                    int cluster_start = arabic_isvowel (wc[i]) 
-                        ? pold - text : p - text;
+                {   
+                    int cluster_start ;
+                    int is_vowel      = arabic_isvowel(wc[i]);
+                    cluster_start     = is_vowel ? pold - text : p - text;
+
                     if ( lvl ==  1 )
                         {
 #ifdef DEBUG
@@ -303,22 +309,10 @@ arabic_engine_shape (PangoFont        *font,
                         }
                     else if ( lvl == 2 )
                         {
-#ifdef DEBUG
-                            fprintf(stderr,"[ar] lbox-recoding char %x",
-                                    wc[i]);
-#endif
                             if (( i > 0 )&&(wc[i-1] == 0))
                                 {
                                     arabic_lbox_recode(&subfont,&(wc[i]),
                                                        &(wc[i-1]), arfonts);
-#ifdef DEBUG
-                                    if (wc[i-1] != 0)
-                                        {
-                                            fprintf(stderr,"[ar] lbox-recoded"
-                                                    " special char %x",
-                                                    wc[i-1]);
-                                        }
-#endif
                                 }
                             else 
                                 arabic_lbox_recode(&subfont,&(wc[i]),NULL,
@@ -326,7 +320,7 @@ arabic_engine_shape (PangoFont        *font,
                         }
                     
                     set_glyph(glyphs, font, subfont, n_glyph - 1,
-                              cluster_start, wc[i]);
+                              cluster_start, wc[i], is_vowel);
       
                     pold = p;
                     p    = unicode_next_utf8 (p);
diff --git a/modules/arabic/arabic.c b/modules/arabic/arabic.c
index 098799ce..ec598b17 100644
--- a/modules/arabic/arabic.c
+++ b/modules/arabic/arabic.c
@@ -17,7 +17,7 @@
 #include "mulefont.h"
 #include "langboxfont.h"
 
-/*  #define DEBUG   */
+/*  #define DEBUG    */
 #ifdef DEBUG
 #include <stdio.h>
 #endif
@@ -161,9 +161,10 @@ find_unic_font (PangoFont *font,char* charsets[],PangoXSubfont* rfonts)
 }
 
 static char *default_charset[] = {
-      "iso10646-1", 
-      "iso8859-6.8x",
-      "mulearabic-2" 
+    "iso10646-1",  
+    "iso8859-6.8x",
+    /*          "iso8859-6.8x" */
+    "mulearabic-2"    
 };
   
 
@@ -171,7 +172,7 @@ static char *default_charset[] = {
 static void
 set_glyph (PangoGlyphString *glyphs, 
            PangoFont *font, PangoXSubfont subfont,
-           int i, int cluster_start, int glyph)
+           int i, int cluster_start, int glyph, int is_vowel)
 {
     PangoRectangle logical_rect;
 #ifdef DEBUG
@@ -193,7 +194,7 @@ set_glyph (PangoGlyphString *glyphs,
 
     pango_font_get_glyph_extents (font, glyphs->glyphs[i].glyph, NULL, &logical_rect);
     glyphs->log_clusters[i] = cluster_start;
-    if (arabic_isvowel(glyph))
+    if (is_vowel)
         {
             glyphs->glyphs[i].geometry.width = 0;
         }
@@ -248,7 +249,7 @@ arabic_engine_shape (PangoFont        *font,
                 {
                     set_glyph (glyphs, font, 
                                PANGO_X_GLYPH_SUBFONT (unknown_glyph), i, 
-                               p - text, PANGO_X_GLYPH_INDEX (unknown_glyph));
+                               p - text, PANGO_X_GLYPH_INDEX (unknown_glyph),0);
                     p = unicode_next_utf8 (p);
                 }
             return;
@@ -282,6 +283,9 @@ arabic_engine_shape (PangoFont        *font,
     p    = text;
     pold = p;
     i    = n_chars-1;
+#ifdef DEBUG
+    fprintf(stderr,"[ar]: after shaping : %i glyphs ",n_glyph);
+#endif
     while(i >= 0)
         {
             if (wc[i] == 0)
@@ -290,9 +294,11 @@ arabic_engine_shape (PangoFont        *font,
                     i--;
                 }
             else 
-                {
-                    int cluster_start = arabic_isvowel (wc[i]) 
-                        ? pold - text : p - text;
+                {   
+                    int cluster_start ;
+                    int is_vowel      = arabic_isvowel(wc[i]);
+                    cluster_start     = is_vowel ? pold - text : p - text;
+
                     if ( lvl ==  1 )
                         {
 #ifdef DEBUG
@@ -303,22 +309,10 @@ arabic_engine_shape (PangoFont        *font,
                         }
                     else if ( lvl == 2 )
                         {
-#ifdef DEBUG
-                            fprintf(stderr,"[ar] lbox-recoding char %x",
-                                    wc[i]);
-#endif
                             if (( i > 0 )&&(wc[i-1] == 0))
                                 {
                                     arabic_lbox_recode(&subfont,&(wc[i]),
                                                        &(wc[i-1]), arfonts);
-#ifdef DEBUG
-                                    if (wc[i-1] != 0)
-                                        {
-                                            fprintf(stderr,"[ar] lbox-recoded"
-                                                    " special char %x",
-                                                    wc[i-1]);
-                                        }
-#endif
                                 }
                             else 
                                 arabic_lbox_recode(&subfont,&(wc[i]),NULL,
@@ -326,7 +320,7 @@ arabic_engine_shape (PangoFont        *font,
                         }
                     
                     set_glyph(glyphs, font, subfont, n_glyph - 1,
-                              cluster_start, wc[i]);
+                              cluster_start, wc[i], is_vowel);
       
                     pold = p;
                     p    = unicode_next_utf8 (p);
diff --git a/modules/arabic/arconv.c b/modules/arabic/arconv.c
index 668378fc..94da5a4c 100644
--- a/modules/arabic/arconv.c
+++ b/modules/arabic/arconv.c
@@ -67,7 +67,7 @@ static shapestruct chartable [] =
     {0x63D,      0x0000,0}, /* simple       */
     {0x63E,      0x0000,0}, /* indexing     */
     {0x63F,      0x0000,0}, /*  :           */
-    {0x640,      0x0640,1}, /*              */
+    {0x640,      0x0640,4}, /* tatweel      */
     {0x641,      0xFED1,4}, /* FA           */
     {0x642,      0xFED5,4}, /* QAF          */
     {0x643,      0xFED9,4}, /* KAF          */
@@ -89,24 +89,24 @@ static shapestruct chartable [] =
     {0x686,      0xFB7A,4}, /* Persian: Tcheh                    */
     {0x687,      0xFB7E,4}, /* Sindhi:                           */
     {0x68C,      0xFB84,2}, /* Sindhi: DAHAL                     */
-    {0x68D,      0xFB82,2}, /* Sindhi                            */             
+    {0x68D,      0xFB82,2}, /* Sindhi                            */ 
     {0x68E,      0xFB86,2}, /*                                   */
-    {0x691,      0xFB8C,2}, /* Urdu                              */             
+    {0x691,      0xFB8C,2}, /* Urdu                              */
     {0x698,      0xFB8A,2}, /* Persian: JEH                      */
-    {0x6A4,      0xFB6A,4}, /* VEH: latin compatibility          */             
-    {0x6A6,      0xFB6E,4}, /* Sindhi                            */             
+    {0x6A4,      0xFB6A,4}, /* VEH: latin compatibility          */
+    {0x6A6,      0xFB6E,4}, /* Sindhi                            */
     {0x6A9,      0xFB8E,4}, /* Persan K                          */
-    {0x6AA,      0xFB8E,4}, /* extrawide KAF-> Persian KAF       */             
-    {0x6AF,      0xFB92,4}, /* Persian: GAF                      */             
+    {0x6AA,      0xFB8E,4}, /* extrawide KAF-> Persian KAF       */
+    {0x6AF,      0xFB92,4}, /* Persian: GAF                      */
     {0x6B1,      0xFB9A,4}, /* Sindhi:                           */
     {0x6B3,      0xFB97,4}, /*                                   */
     {0x6BA,      0xFB9E,2}, /* Urdu:NUN GHUNNA                   */
-    {0x6BB,      0xFBA0,4}, /* Sindhi:                           */             
+    {0x6BB,      0xFBA0,4}, /* Sindhi:                           */
     {0x6BE,      0xFBAA,4}, /* HA special                        */
-    {0x6C0,      0xFBA4,2}, /* izafet: HA HAMZA                  */             
+    {0x6C0,      0xFBA4,2}, /* izafet: HA HAMZA                  */
     {0x6C1,      0xFBA6,4}, /* Urdu:                             */
     {0x6D2,      0xFBAE,2}, /* YA barree                         */
-    {0x6D3,      0xFBB0,2}, /* YA BARREE HAMZA                   */             
+    {0x6D3,      0xFBB0,2}, /* YA BARREE HAMZA                   */
 
     {0xFEF5,     0xFEF5,2}, /* Lam-Alif Madda  */
     {0xFEF7,     0xFEF7,2}, /* Lam-Alif Hamza  */
@@ -119,8 +119,9 @@ static shapestruct chartable [] =
 #define ALIFMADDA  0x622
 #define LAM        0x644
 #define HAMZA      0x621
+#define TATWEEL    0x640
 
-/* Hmaza below ( saves Kasra and special cases ), Hamza above ( always joins ).
+/* Hamza below ( saves Kasra and special cases ), Hamza above ( always joins ).
  * As I don't know what sHAMZA is good for I don't handle it.
  */
 #define iHAMZA      0x654
@@ -221,11 +222,12 @@ static GUChar4
 charshape(GUChar4 s,short which)
 { /* which 0=alone 1=end 2=start 3=middle */
     int j = 0;
-    if ((s >= chartable[1].basechar)  && (s <= 0x64A)) 
+    if ((s >= chartable[1].basechar)  && (s <= 0x64A) && ( s != TATWEEL)) 
         {   /* basic character */ 
             return chartable[s-chartable[0].basechar].charstart+which;
         } 
-    else if ( (s >= chartable[1].basechar)  && ( s <= 0xFEFB ))
+    else if ( (s >= chartable[1].basechar)  && ( s <= 0xFEFB ) 
+              && (s != TATWEEL))
         {   /* special char or  Lam-Alif */
             while ( chartable[j].basechar < s) j++;
             return chartable[j].charstart+which;
@@ -274,6 +276,9 @@ ligature(GUChar4* string,int si,int len,charstruct* oldchar)
     if (arabic_isvowel(newchar))
         {
             retval = 1;
+            if ((oldchar->vowel != 0)&&(newchar != SHADDA)){
+                retval = 2; /* we eliminate the old vowel .. */
+            }
             switch(newchar)
                 {
                 case SHADDA: oldchar->mark1 = newchar; break;
@@ -436,8 +441,11 @@ shape(int olen,int* len,GUChar4* string,int level)
                     curchar.lignum++;
                 }
             else if ( ( join == 2 )
-		      ||((join == 3)&&(level != 2))  ) 
+                      ||((join == 3)&&(level != 2))  
+                      ||((join == 1)&&(level == 1)
+                         && arabic_isvowel(string[si])) ) 
                 { /*  Lam-Alif in Langbox-font is no ligature */
+                    /*  No vowels in Langbox-font */
                     (*len)--;
                 }
             si--;
diff --git a/modules/arabic/langboxfont.c b/modules/arabic/langboxfont.c
index cdfac60c..3073b539 100644
--- a/modules/arabic/langboxfont.c
+++ b/modules/arabic/langboxfont.c
@@ -153,7 +153,7 @@ static fontentry charmap [] =
     { 0xFEDD,0xE4  }, /* LAM      */
     { 0xFEDE,0xE4  }, 
     { 0xFEDF,0xFA  }, 
-    { 0xFEE0,0xA6  }, 
+    { 0xFEE0,0xFA  }, 
     { 0xFEE1,0xE5  }, /* MIM      */
     { 0xFEE2,0xE5  }, 
     { 0xFEE3,0xFB  }, 
@@ -181,9 +181,10 @@ arabic_lbox_recode(PangoXSubfont* subfont,int* glyph,int* glyph2,
                    PangoXSubfont* lboxfonts)
 {
     int letter=*glyph;
+    
+    *subfont = lboxfonts[0];
     if ((letter >= 0x660)&&(letter <= 0x669)) /* indic numeral */
         {
-            *subfont = lboxfonts[0];
             *glyph   = letter - 0x660 + 0xB0;
         }
     else if ((letter >= 0xFE80)&&(letter <= 0xFEF4))
@@ -196,17 +197,14 @@ arabic_lbox_recode(PangoXSubfont* subfont,int* glyph,int* glyph2,
                             letter,charmap[letter-0xFE80].unicodechar);
                 }
 #endif
-            *subfont = lboxfonts[0];
             *glyph   = charmap[letter-0xFE80].charindex;
         }
     else if ((letter >= 0x64B)&&(letter <= 0x652))
         { /* a vowel */
-            *subfont = lboxfonts[0];
             *glyph   = letter - 0x64B + 0xA8;
         }
     else if ((letter >= 0xFEF5)&&(letter <= 0xFEFC))
         { /* Lam-Alif. Langbox solved the problem in their own way ... */
-            *subfont = lboxfonts[0];
 #ifdef DEBUG
             fprintf(stderr,"[ar] lbox-recoding chars %x",
                     letter);
@@ -233,17 +231,22 @@ arabic_lbox_recode(PangoXSubfont* subfont,int* glyph,int* glyph2,
         }
     else if (letter == 0x621)
         {
-            *subfont = lboxfonts[0];
             *glyph   = charmap[0].charindex;
         }
+    else if (letter == 0x640)
+        { /* tatweel */
+            *glyph   = 0xE0;
+        }
+    else if (letter == 0x61F)
+        { /* question mark */
+            *glyph   = 0xBF; 
+        }
     else if (letter < 0xB0 )
         {
-            *subfont = lboxfonts[0];
             *glyph   = letter;
         }
     else
         {
-            *subfont = lboxfonts[0];
             *glyph   = 0x20; /* we don't have this thing -- use a space */
             /* This has to be something that does not print anything !! */
         }
diff --git a/modules/arabic/mulefont.c b/modules/arabic/mulefont.c
index edb440a8..e087ce0d 100644
--- a/modules/arabic/mulefont.c
+++ b/modules/arabic/mulefont.c
@@ -247,11 +247,15 @@ arabic_mule_recode(PangoXSubfont* subfont,int* glyph,PangoXSubfont* mulefonts)
             *subfont = mulefonts[charmap[0].fontindex];
             *glyph   = charmap[0].charindex;
         }
+    else if (letter == 0x61F)
+        { /* question mark */
+            *subfont = mulefonts[1];
+            *glyph   = 0x29; 
+        }
     else
         {
-            *subfont = mulefonts[charmap[1].fontindex];
-            *glyph   = 0x20; /* we don't have this thing -- use a space */
-            /* This has to be something that does not print anything !! */
+            *subfont = mulefonts[1];
+            *glyph   = 0x26; /* we don't have this thing -- use a dot */
         }
 }