added glyph pairing to align zone detector

diff --git a/lib/devices/swf.c b/lib/devices/swf.c
index 3277938..a3f57e9 100644 (file)
--- a/lib/devices/swf.c
+++ b/lib/devices/swf.c
@@ -623,6 +623,7 @@ static void chararray_writetotag(chararray_t*_chardata, TAG*tag)
     int lastx;
     int lasty;
     int lastsize;
+    int lastchar;
     int charids[128];
     int charadvance[128];
     int charstorepos;
@@ -646,6 +647,7 @@ static void chararray_writetotag(chararray_t*_chardata, TAG*tag)
         lastx = CHARMIDX;
         lasty = CHARMIDY;
         lastsize = -1;
+       lastchar = -1;
 
         if(pass==1)
         {
@@ -670,9 +672,9 @@ static void chararray_writetotag(chararray_t*_chardata, TAG*tag)
 
                charatposition_t*chr = &chardata->chr[t];
 
-               if(lastfont != chardata->chr[t].font || 
-                       lastx!=chardata->chr[t].x ||
-                       lasty!=chardata->chr[t].y ||
+               if(lastfont != chr->font || 
+                       lastx!=chr->x ||
+                       lasty!=chr->y ||
                        !colorcompare(&color, &chardata->chr[t].color) ||
                        charstorepos==127 ||
                        lastsize != chardata->chr[t].size ||
@@ -736,6 +738,12 @@ static void chararray_writetotag(chararray_t*_chardata, TAG*tag)
                    lastx = chr->x;
                    lasty = chr->y;
                    lastsize = chr->size;
+               } else {
+                   assert(lastchar>=0);
+                   if(pass==1 && lastchar!=chr->charid) {
+                       swf_FontUsePair(chr->font, lastchar, chr->charid);
+                       swf_FontUsePair(chr->font, chr->charid, lastchar);
+                   }
                }
 
                if(islast)
@@ -745,7 +753,7 @@ static void chararray_writetotag(chararray_t*_chardata, TAG*tag)
                if(t<chardata->pos-1) nextx = chardata->chr[t+1].x;
                if(t==chardata->pos-1 && chardata->next) nextx = chardata->next->chr[0].x;
                int dx = nextx-chr->x;
-
+               
                int advance;
                if(dx>=0 && (dx<(1<<(advancebits-1)) || pass==0)) {
                   advance = dx;
@@ -754,8 +762,10 @@ static void chararray_writetotag(chararray_t*_chardata, TAG*tag)
                   advance = 0;
                   lastx=chr->x;
                }
+
                charids[charstorepos] = chr->charid;
                charadvance[charstorepos] = advance;
+               lastchar = chr->charid;
                charstorepos ++;
            }
            chardata = chardata->next;

diff --git a/lib/modules/swfalignzones.c b/lib/modules/swfalignzones.c
index 1961903..f05cad8 100644 (file)
--- a/lib/modules/swfalignzones.c
+++ b/lib/modules/swfalignzones.c
@@ -40,7 +40,7 @@ static void find_best(float*_row, int width, int*_x1, int*_x2, int min_size, int
     float max1=-1e20,max2=-1e20;
     int t;
     float*row = malloc(sizeof(float)*(width+1));
-    int filter_size = 20;
+    int filter_size = 10;
     float* filter = malloc(sizeof(float)*(filter_size*2+1));
     double var = filter_size/3;
     for(t=-filter_size;t<=filter_size;t++) {
@@ -169,7 +169,7 @@ static ALIGNZONE detect_for_char(SWFFONT * f, int nr, float*row, float*column, S
     /* find two best x values */
     int x1=-1,y1=-1,x2=-1,y2=-1;
 
-    int nr_x = 1;
+    int nr_x = 0;
     find_best(row, width, &x1, &x2, f->use->smallest_size, 
                char_bbox.xmin - font_bbox.xmin,
                char_bbox.xmax - font_bbox.xmin, nr_x,
@@ -210,9 +210,8 @@ void swf_FontCreateAlignZones(SWFFONT * f)
            f->alignzones[t].dy = 0xffff;//floatToF16(460.80 / 1024.0);
        }
     } else {
-       int t;
        SRECT bounds = {0,0,0,0};
-       
+       int t;
        for(t=0;t<f->numchars;t++) {
            SRECT b = f->layout->bounds[t];
            negate_y(&b);
@@ -231,7 +230,25 @@ void swf_FontCreateAlignZones(SWFFONT * f)
        for(t=0;t<=height;t++) {column_global[t]/=f->numchars/2;}
 
        for(t=0;t<f->numchars;t++) {
-           memcpy(column, column_global, sizeof(float)*(height+1));
+           //memcpy(column, column_global, sizeof(float)*(height+1));
+
+           memset(column, 0, sizeof(float)*(height+1));
+           int s;
+           int drawn = 0;
+           printf("[font %d] pairing %c with ", f->id, f->glyph2ascii[t]);
+           for(s=0;s<f->use->num_neighbors;s++) {
+               if(f->use->neighbors[s].char2 == t) {
+                   printf("%c ", f->glyph2ascii[f->use->neighbors[s].char1]);
+                   draw_char(f, f->use->neighbors[s].char1, row, column, bounds);
+                   drawn++;
+               }
+           }
+           printf("\n");
+
+           for(s=0;s<=height;s++) {
+               column[t] /= drawn*2;
+           }
+
            memset(row, 0, sizeof(float)*(width+1));
            draw_char(f, t, row, column, bounds);
            

diff --git a/lib/modules/swftext.c b/lib/modules/swftext.c
index 194ec72..2ce669a 100644 (file)
--- a/lib/modules/swftext.c
+++ b/lib/modules/swftext.c
@@ -625,6 +625,12 @@ static void font_freeusage(SWFFONT*f)
        if(f->use->chars) {
            rfx_free(f->use->chars);f->use->chars = 0;
        }
+       if(f->use->neighbors) {
+           rfx_free(f->use->neighbors);f->use->neighbors = 0;
+       }
+       if(f->use->neighbors_hash) {
+           rfx_free(f->use->neighbors_hash);f->use->neighbors_hash = 0;
+       }
        rfx_free(f->use); f->use = 0;
     }
 }
@@ -851,12 +857,9 @@ int swf_FontInitUsage(SWFFONT * f)
        fprintf(stderr, "Usage initialized twice");
        return -1;
     }
-    f->use = (FONTUSAGE*)rfx_alloc(sizeof(FONTUSAGE));
-    f->use->is_reduced = 0;
+    f->use = (FONTUSAGE*)rfx_calloc(sizeof(FONTUSAGE));
     f->use->smallest_size = 0xffff;
-    f->use->used_glyphs = 0;
     f->use->chars = (int*)rfx_calloc(sizeof(f->use->chars[0]) * f->numchars);
-    f->use->glyphs_specified = 0;
     return 0;
 }
 
@@ -906,6 +909,78 @@ int swf_FontUseAll(SWFFONT* f)
     return 0;
 }
 
+static unsigned hash2(int char1, int char2)
+{
+    unsigned hash = char1^(char2<<8);
+    hash += (hash << 3);
+    hash ^= (hash >> 11);
+    hash += (hash << 15);
+    return hash;
+}
+static void hashadd(FONTUSAGE*u, int char1, int char2, int nr)
+{
+    unsigned hash = hash2(char1, char2);
+    while(1) {
+       hash = hash%u->neighbors_hash_size;
+       if(!u->neighbors_hash[hash]) {
+          u->neighbors_hash[hash] = nr+1;
+          return;
+       }
+       hash++;
+    }
+}
+int swf_FontUseGetPair(SWFFONT * f, int char1, int char2)
+{
+    FONTUSAGE*u = f->use;
+    if(!u || !u->neighbors_hash_size) 
+       return 0;
+    unsigned hash = hash2(char1, char2);
+    while(1) {
+       hash = hash%u->neighbors_hash_size;
+       int pos = u->neighbors_hash[hash];
+       if(!pos)
+           return 0;
+       if(pos && 
+          u->neighbors[pos-1].char1 == char1 &&
+          u->neighbors[pos-1].char2 == char2) {
+           return pos;
+       }
+       hash++;
+    }
+
+}
+void swf_FontUsePair(SWFFONT * f, int char1, int char2)
+{
+    if (!f->use)
+       swf_FontInitUsage(f);
+    FONTUSAGE*u = f->use;
+
+    if(u->num_neighbors*3 >= u->neighbors_hash_size*2) {
+       if(u->neighbors_hash) {
+           free(u->neighbors_hash);
+       }
+       u->neighbors_hash_size = u->neighbors_hash_size?u->neighbors_hash_size*2:1024;
+       u->neighbors_hash = rfx_calloc(u->neighbors_hash_size*sizeof(int));
+       int t;
+       for(t=0;t<u->num_neighbors;t++) {
+           hashadd(u, u->neighbors[t].char1, u->neighbors[t].char2, t);
+       }
+    }
+
+    if(!swf_FontUseGetPair(f, char1, char2)) {
+       if(u->num_neighbors == u->neighbors_size) {
+           u->neighbors_size += 4096;
+           u->neighbors = rfx_realloc(u->neighbors, sizeof(SWFGLYPHPAIR)*u->neighbors_size);
+       }
+       u->neighbors[u->num_neighbors].char1 = char1;
+       u->neighbors[u->num_neighbors].char2 = char2;
+       hashadd(u, char1, char2, u->num_neighbors);
+       u->num_neighbors++;
+    } else {
+       // increase?
+    }
+}
+
 int swf_FontUseGlyph(SWFFONT * f, int glyph, U16 size)
 {
     if (!f->use)

diff --git a/lib/rfxswf.h b/lib/rfxswf.h
index 0baf61e..4ca38a2 100644 (file)
--- a/lib/rfxswf.h
+++ b/lib/rfxswf.h
@@ -530,12 +530,24 @@ typedef struct
   SHAPE *     shape;
 } SWFGLYPH;
 
+typedef struct _SWFGLYPHPAIR
+{
+    U16 char1;
+    U16 char2;
+} SWFGLYPHPAIR;
+
 typedef struct _FONTUSAGE
 { int* chars;
   char is_reduced;
   int used_glyphs;
   int glyphs_specified;
   U16 smallest_size;
+
+  SWFGLYPHPAIR* neighbors;
+  int num_neighbors;
+  int neighbors_size;
+  int* neighbors_hash;
+  int neighbors_hash_size;
 } FONTUSAGE;
 
 #define FONT_STYLE_BOLD 1
@@ -631,6 +643,8 @@ int swf_FontReduce_swfc(SWFFONT * f);
 
 int swf_FontInitUsage(SWFFONT * f);
 int swf_FontUseGlyph(SWFFONT * f, int glyph, U16 size);
+void swf_FontUsePair(SWFFONT * f, int char1, int char2);
+int swf_FontUseGetPair(SWFFONT * f, int char1, int char2);
 int swf_FontUseAll(SWFFONT* f);
 int swf_FontUseUTF8(SWFFONT * f, U8 * s, U16 size);
 int swf_FontUse(SWFFONT* f,U8 * s);

diff --git a/src/Makefile.in b/src/Makefile.in
index 463b8d9..0c1381d 100644 (file)
--- a/src/Makefile.in
+++ b/src/Makefile.in
@@ -56,6 +56,15 @@ parser.$(O): parser.yy.c parser.h ../lib/q.h
 ../lib/librfxswf$(A): ../lib/modules/swfrender.c ../lib/modules/swfshape.c ../lib/modules/swftext.c ../lib/modules/swffont.c ../lib/modules/swfbits.c ../lib/rfxswf.c ../lib/devices/swf.c ../lib/modules/swfalignzones.c
        cd ../lib;$(MAKE) librfxswf$(A);cd -
 
+../lib/libpdf$(A): ../lib/pdf/GFXOutputDev.cc
+       cd ../lib;$(MAKE) libpdf$(A);cd -
+
+../lib/libgfx$(A): ../lib/devices/*.c
+       cd ../lib;$(MAKE) libgfx$(A);cd -
+
+../lib/libgfxswf$(A): ../lib/devices/swf.c
+       cd ../lib;$(MAKE) libgfxswf$(A);cd -
+
 # TODO: include the following rule only if lex is available
 parser.yy.c: parser.lex
        flex -B -s -oparser.yy.c parser.lex