add a font preprocessor

CMakeLists.txt

@@ -66,6 +66,8 @@ add_executable(pdf2htmlEX
     src/ff/ff.c
     src/BackgroundRenderer.h 
     src/BackgroundRenderer.cc 
+    src/FontPreprocessor.h
+    src/FontPreprocessor.cc
     src/Consts.h 
     src/Consts.cc 
     src/util.h

TODO

@@ -1,3 +1,5 @@
+disable annotation
+
 valgrind
 
 re-encoded only used glyphs

src/BackgroundRenderer.h

@@ -24,7 +24,7 @@ public:
             allowAntialiasA) { }
   virtual ~BackgroundRenderer() { }
   
-  void drawChar(GfxState *state, double x, double y,
+  virtual void drawChar(GfxState *state, double x, double y,
       double dx, double dy,
       double originX, double originY,
       CharCode code, int nBytes, Unicode *u, int uLen);

src/HTMLRenderer.h

@@ -30,7 +30,7 @@
 
 #include "Param.h"
 #include "util.h"
-
+#include "FontPreprocessor.h"
 
 /*
  * Naming Convention
@@ -124,6 +124,7 @@ class HTMLRenderer : public OutputDev
         ////////////////////////////////////////////////////
         // misc
         ////////////////////////////////////////////////////
+        
         void add_tmp_file (const std::string & fn);
         void clean_tmp_files ();
         boost::filesystem::path dump_embedded_font (GfxFont * font, long long fn_id);
@@ -333,6 +334,7 @@ class HTMLRenderer : public OutputDev
         // for font reencoding
         int32_t * cur_mapping;
         char ** cur_mapping2;
+        FontPreprocessor font_preprocessor;
 
         ////////////////////////////////////////////////////
         // styles & resources

src/HTMLRenderer/general.cc

@@ -51,12 +51,20 @@ HTMLRenderer::~HTMLRenderer()
 
 void HTMLRenderer::process(PDFDoc *doc)
 {
-    cerr << "Working: ";
-
     xref = doc->getXRef();
 
-    BackgroundRenderer * bg_renderer = nullptr;
+    cerr << "Preprocessing: ";
+    for(int i = param->first_page; i <= param->last_page ; ++i) 
+    {
+        doc->displayPage(&font_preprocessor, i, param->h_dpi, param->v_dpi,
+                0, true, false, false,
+                nullptr, nullptr, nullptr, nullptr);
+        cerr << "." << flush;
+    }
+    cerr << endl;
 
+    cerr << "Working: ";
+    BackgroundRenderer * bg_renderer = nullptr;
     if(param->process_nontext)
     {
         // Render non-text objects as image

src/HTMLRenderer/install.cc

@@ -25,7 +25,7 @@ const FontInfo * HTMLRenderer::install_font(GfxFont * font)
 {
     assert(sizeof(long long) == 2*sizeof(int));
                 
-    long long fn_id = (font == nullptr) ? 0 : *reinterpret_cast<long long*>(font->getID());
+    long long fn_id = (font == nullptr) ? 0 : hash_ref(font->getID());
 
     auto iter = font_name_map.find(fn_id);
     if(iter != font_name_map.end())

src/HTMLRenderer/text.cc

@@ -217,8 +217,12 @@ void HTMLRenderer::embed_font(const path & filepath, GfxFont * font, FontInfo &
 
                 memset(cur_mapping2, 0, 256 * sizeof(char*));
 
+                const char * used_map = font_preprocessor.get_code_map(hash_ref(font->getID()));
+
                 for(int i = 0; i < 256; ++i)
                 {
+                    if(!used_map[i]) continue;
+
                     auto cn = font_8bit->getCharName(i);
                     if(cn == nullptr)
                     {

src/util.h

@@ -21,6 +21,7 @@
 #include <CharTypes.h>
 #include <UTF8.h>
 #include <GlobalParams.h>
+#include <Object.h>
 
 #include "Consts.h"
 
@@ -49,6 +50,11 @@ static inline bool _tm_equal(const double * tm1, const double * tm2, int size =
     return true;
 }
 
+static inline long long hash_ref(const Ref * id)
+{
+    return (((long long)(id->num)) << (sizeof(id->gen)*8)) | (id->gen);
+}
+
 /*
  * http://en.wikipedia.org/wiki/HTML_decimal_character_rendering
  */