seems better font re-encoding

CMakeLists.txt

@@ -14,8 +14,8 @@ set(PDF2HTMLEX_VERSION "0.1")
 
 set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Wall -Wunused-function")
 set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=c++0x")
-set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -O2")
-#set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -ggdb")
+#set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -O2")
+set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -ggdb")
 
 configure_file (src/config.h.in src/config.h)
 

share/all.css

@@ -19,7 +19,6 @@
 }
 #pdf-main .l {
   position:absolute;
-  height:0;
   white-space:pre;
   font-size:1px;
   transform-origin:0% 100%;

src/BackgroundRenderer.cc

@@ -18,5 +18,6 @@ void BackgroundRenderer::drawChar(GfxState *state, double x, double y,
   double originX, double originY,
   CharCode code, int nBytes, Unicode *u, int uLen)
 {
+    SplashOutputDev::drawChar(state,x,y,dx,dy,originX,originY,code, nBytes, u, uLen);
 }
 

src/HTMLRenderer/install.cc

@@ -42,6 +42,7 @@ long long HTMLRenderer::install_font(GfxFont * font)
     if(param->debug)
     {
         cerr << "Install font: (" << (font->getID()->num) << ' ' << (font->getID()->gen) << ") -> " << format("f%|1$x|")%new_fn_id << endl;
+        cerr << "Ascent: " << (font->getAscent()) << " Descent: " << (font->getDescent()) << endl;
     }
 
     if(font->getType() == fontType3) {
@@ -133,19 +134,6 @@ void HTMLRenderer::install_embedded_font(GfxFont * font, const string & suffix,
     int code2GID_len = 0;
     int maxcode = 0;
 
-    // if we cannot map to unicode through ctu, map the char to private Unicode values
-    auto map_to_unicode = [&ctu](int c)->Unicode
-    {
-        Unicode *u;
-        int n = 0;
-        if(ctu)
-        {
-            n = ctu->mapToUnicode(c, &u);
-        }
-
-        return check_unicode(u, n, c);
-    };
-
     if(!font->isCIDFont())
     {
         maxcode = 0xff;
@@ -201,8 +189,21 @@ void HTMLRenderer::install_embedded_font(GfxFont * font, const string & suffix,
     for(int i = 0; i <= maxcode; ++i)
     {
         map_fout << format("0x%|1$X|") % ((code2GID && (i < code2GID_len))? code2GID[i] : i);
-        map_fout << format(" 0x%|1$X|") % map_to_unicode(i);
-        map_fout << format(" # 0x%|1$X|") % i << endl;
+
+        Unicode u, *pu;
+        int n = 0;
+        if(ctu)
+            n = ctu->mapToUnicode(i, &pu);
+
+        u = check_unicode(pu, n, i, font);
+
+        map_fout << format(" 0x%|1$X|") % u;
+        map_fout << format(" # 0x%|1$X|") % i;
+
+        for(int j = 0; j < n; ++j)
+            map_fout << format(" 0x%|1$X|") % pu[j];
+
+        map_fout << endl;
     }
 
     script_fout << format("LoadEncodingFile(%1%, \"%2%\")") % (tmp_dir / (fn+".encoding")) % fn << endl;

src/HTMLRenderer/state.cc

@@ -352,8 +352,8 @@ void HTMLRenderer::prepare_line(GfxState * state)
             double x,y; // in user space
             state->transform(state->getCurX(), state->getCurY(), &x, &y);
 
-            html_fout << format("<div style=\"bottom:%1%px;left:%2%px;\" class=\"l t%|3$x|\"><em></em>") 
-                % y % x % cur_tm_id;
+            html_fout << format("<div style=\"left:%1%px;bottom:%2%px;\" class=\"l t%|3$x|\"><em></em>") 
+                % x % y % cur_tm_id;
 
             //resync position
             draw_ty = cur_ty;

src/HTMLRenderer/text.cc

@@ -196,7 +196,7 @@ void HTMLRenderer::drawString(GfxState * state, GooString * s)
             ++nSpaces;
         }
         
-        Unicode uu = check_unicode(u, uLen, code);
+        Unicode uu = check_unicode(u, uLen, code, font);
         outputUnicodes(html_fout, &uu, 1);
 
         dx += dx1;

src/util.h

@@ -15,8 +15,10 @@
 #include <ostream>
 
 #include <GfxState.h>
+#include <GfxFont.h>
 #include <CharTypes.h>
 #include <UTF8.h>
+#include <GlobalParams.h>
 
 #include "Consts.h"
 
@@ -65,12 +67,38 @@ static inline bool isLegalUnicode(Unicode u)
     return true;
 }
 
-static inline Unicode check_unicode(Unicode * u, int len, CharCode code)
+/*
+ * We have to use a single Unicode value to reencode fonts
+ * if we got multi-unicode values, it might be expanded ligature, try to restore it
+ * if we cannot figure it out at the end, use a private mapping
+ */
+static inline Unicode check_unicode(Unicode * u, int len, CharCode code, GfxFont * font)
 {
-    if((len == 0) || (len > 1) || (!isLegalUnicode(*u)))
-        return (Unicode)(code + 0xE000);
-    else
-        return *u;
+    Unicode private_mapping = (Unicode)(code + 0xE000);
+
+    if(len == 0)
+        return private_mapping;
+
+    if(len == 1)
+    {
+        if(isLegalUnicode(*u))
+            return *u;
+    }
+
+    if(!font->isCIDFont())
+    {
+        char * cname = dynamic_cast<Gfx8BitFont*>(font)->getCharName(code);
+        // may be untranslated ligature
+        if(cname)
+        {
+            Unicode ou = globalParams->mapNameToUnicode(cname);
+
+            if(isLegalUnicode(ou))
+                return ou;
+        }
+    }
+
+    return private_mapping;
 }
 
 static inline void outputUnicodes(std::ostream & out, const Unicode * u, int uLen)