finetuned space char detection

[swftools.git] / lib / pdf / InfoOutputDev.cc

diff --git a/lib/pdf/InfoOutputDev.cc b/lib/pdf/InfoOutputDev.cc
index 8a00d98..a142c77 100644 (file)
--- a/lib/pdf/InfoOutputDev.cc
+++ b/lib/pdf/InfoOutputDev.cc
@@ -15,9 +15,11 @@
 #endif
 #include "GfxState.h"
 #include "../log.h"
+#include "../q.h"
 #include <math.h>
+#include <assert.h>
 
-/* there's not yet a way to set this */
+int config_addspace = 1;
 int config_fontquality = 10;
 int config_bigchar = 0;
 
@@ -70,6 +72,7 @@ FontInfo::FontInfo(char*id)
     this->lastx = 0;
     this->lasty = 0;
     this->gfxfont = 0;
+    this->space_char = -1;
 }
 FontInfo::~FontInfo()
 {
@@ -88,9 +91,46 @@ FontInfo::~FontInfo()
        }
     }
     free(glyphs);glyphs=0;
+    if(this->gfxfont)
+        gfxfont_free(this->gfxfont);
 }
 
-gfxfont_t* createGfxFont(FontInfo*src)
+static int findSpace(gfxfont_t*font)
+{
+    int first_space = -1;
+    int t;
+    for(t=0;t<font->num_glyphs;t++) {
+       gfxglyph_t*g = &font->glyphs[t];
+       if(GLYPH_IS_SPACE(g)) {
+           if(g->unicode == 32) return t;
+           if(first_space<0)
+               first_space = t;
+       }
+    }
+    if(font->num_glyphs>32 && GLYPH_IS_SPACE(&font->glyphs[32])) {
+       return 32;
+    }
+    return first_space;
+}
+
+static int addSpace(gfxfont_t*font)
+{
+    font->num_glyphs++;
+    font->glyphs = (gfxglyph_t*)realloc(font->glyphs, sizeof(gfxglyph_t)*font->num_glyphs);
+    gfxglyph_t*g = &font->glyphs[font->num_glyphs-1];
+    memset(g, 0, sizeof(*g));
+    g->unicode = 32;
+    //g->advance = font->ascent;
+    g->advance = fabs(font->ascent - font->descent)*2 / 3;
+    if(font->max_unicode > 32)
+       font->unicode2glyph[32] = font->num_glyphs-1;
+#if 0
+    g->line = gfxline_makerectangle(0, -font->ascent, g->advance, font->descent);
+#endif
+    return font->num_glyphs-1;
+}
+
+static gfxfont_t* createGfxFont(FontInfo*src)
 {
     gfxfont_t*font = (gfxfont_t*)malloc(sizeof(gfxfont_t));
     memset(font, 0, sizeof(gfxfont_t));
@@ -177,12 +217,37 @@ gfxfont_t* createGfxFont(FontInfo*src)
     return font;
 }
 
+static float find_average_glyph_advance(gfxfont_t*f)
+{
+    if(!f->num_glyphs)
+       return 0.0;
+
+    float*values = (float*)malloc(sizeof(float)*f->num_glyphs);
+    int t;
+    for(t=0;t<f->num_glyphs;t++) {
+       values[t] = f->glyphs[t].advance;
+    }
+    float m = medianf(values, f->num_glyphs);
+    free(values);
+    return m;
+}
 
 gfxfont_t* FontInfo::getGfxFont()
 {
     if(!this->gfxfont) {
         this->gfxfont = createGfxFont(this);
         this->gfxfont->id = strdup(this->id);
+       this->space_char = findSpace(this->gfxfont);
+       this->average_advance = find_average_glyph_advance(this->gfxfont);
+
+       if(this->space_char>=0) {
+           msg("<debug> Font %s has space char %d (unicode=%d)", 
+                   this->id, this->space_char, 
+                   this->gfxfont->glyphs[this->space_char].unicode);
+       } else if(config_addspace) {
+           this->space_char = addSpace(this->gfxfont);
+           msg("<debug> Appending space char to font %s, position %d", this->gfxfont->id, this->space_char);
+       }
     }
     return this->gfxfont;
 }
@@ -338,7 +403,7 @@ void InfoOutputDev::drawChar(GfxState *state, double x, double y,
        g->advance = currentfont->splash_font->last_advance;
        g->unicode = 0;
     }
-    if(uLen && (u[0]>=32 && u[0]<g->unicode || !g->unicode)) {
+    if(uLen && ((u[0]>=32 && u[0]<g->unicode) || !g->unicode)) {
        g->unicode = u[0];
     }
     if(currentfont->lastchar>=0 && currentfont->lasty == y) {