remove scripts, link to fontforge directly

CMakeLists.txt

@@ -9,6 +9,9 @@ find_package(Boost REQUIRED COMPONENTS program_options filesystem system)
 include_directories(${Boost_INCLUDE_DIRS})
 link_directories ( ${Boost_LIBRARY_DIRS} )
 include_directories(${CMAKE_SOURCE_DIR}/src)
+pkg_check_modules(FONTFORGE REQUIRED fontforge)
+include_directories(${FONTFORGE_INCLUDE_DIRS})
+link_directories ( ${FONTFORGE_LIBRARY_DIRS} )
 
 set(PDF2HTMLEX_VERSION "0.2")
 set(ARCHIVE_NAME pdf2htmlex-${PDF2HTMLEX_VERSION})
@@ -17,10 +20,14 @@ add_custom_target(dist
         | bzip2 > ${CMAKE_BINARY_DIR}/${ARCHIVE_NAME}.tar.bz2
     WORKING_DIRECTORY ${CMAKE_SOURCE_DIR})
 
+set(CMAKE_C_FLAGS "${CMAKE_CXX_FLAGS} -Wall -Wunused-function")
+#set(CMAKE_C_FLAGS "${CMAKE_CXX_FLAGS} -O2")
+set(CMAKE_C_FLAGS "${CMAKE_CXX_FLAGS} -ggdb")
+
 set(CMAKE_CXX_FLAGS "${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 "${CMAKE_CXX_FLAGS} -O2")
-#set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -ggdb")
+set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -ggdb")
 
 configure_file (${CMAKE_SOURCE_DIR}/src/config.h.in ${CMAKE_SOURCE_DIR}/src/config.h)
 
@@ -34,6 +41,8 @@ add_executable(pdf2htmlEX
     src/HTMLRenderer/text.cc
     src/HTMLRenderer/image.cc
     src/HTMLRenderer/namespace.h
+    src/ff/ff.h
+    src/ff/ff.c
     src/BackgroundRenderer.h 
     src/BackgroundRenderer.cc 
     src/Consts.h 
@@ -41,7 +50,7 @@ add_executable(pdf2htmlEX
     src/util.h
     src/config.h)
 
-target_link_libraries(pdf2htmlEX poppler boost_program_options boost_filesystem boost_system)
+target_link_libraries(pdf2htmlEX poppler boost_program_options boost_filesystem boost_system fontforge python2.7)
 
 install (TARGETS pdf2htmlEX DESTINATION bin)
 file (GLOB datafiles share/*)

TODO

@@ -1,3 +1,5 @@
+cmake - pkgconfig - fontforge - python
+
 link fontforge instead of script
 
 option to break ligatures

share/unify.pe

@@ -1,31 +0,0 @@
-# script piece to unify the metrics
-# by WangLu
-# 2012.08.31
-wa = GetOS2Value("WinAscent")
-wd = GetOS2Value("WinDescent")
-ta = GetOS2Value("TypoAscent")
-td = GetOS2Value("TypoDescent")
-ha = GetOS2Value("HHeadAscent")
-hd = GetOS2Value("HHeadDescent")
-a = wa
-if (ta > a)
-    a = ta
-endif
-if (ha > a)
-    a = ha
-endif
-d = wd
-if (-td > d)
-    d = -td
-endif
-if (-hd > d)
-    d = -hd
-endif
-SetOS2Value("WinAscent", a)
-SetOS2Value("WinDescent", d)
-SetOS2Value("HHeadAscent", a)
-SetOS2Value("HHeadDescent", -d)
-Print(ta-td)
-Print(a)
-Print(d)
-# script end

src/HTMLRenderer.h

@@ -297,9 +297,6 @@ class HTMLRenderer : public OutputDev
         static const std::string NECK_HTML_FILENAME;
         static const std::string TAIL_HTML_FILENAME;
         static const std::string CSS_FILENAME;
-        static const std::string UNIFY_SCRIPT_FILENAME;
-        // for cross-platform purpose, use a "null" file instead of /dev/null
-        static const std::string NULL_FILENAME;
 };
 
 #endif /* HTMLRENDERER_H_ */

src/HTMLRenderer/general.cc

@@ -13,6 +13,7 @@
 #include "BackgroundRenderer.h"
 #include "config.h"
 #include "namespace.h"
+#include "ff/ff.h"
 
 using std::fixed;
 using std::flush;
@@ -26,6 +27,7 @@ HTMLRenderer::HTMLRenderer(const Param * param)
     ,dest_dir(param->dest_dir)
     ,tmp_dir(param->tmp_dir)
 {
+    ff_init();
 }
 
 HTMLRenderer::~HTMLRenderer()
@@ -242,5 +244,3 @@ const std::string HTMLRenderer::HEAD_HTML_FILENAME = "head.html";
 const std::string HTMLRenderer::NECK_HTML_FILENAME = "neck.html";
 const std::string HTMLRenderer::TAIL_HTML_FILENAME = "tail.html";
 const std::string HTMLRenderer::CSS_FILENAME = "all.css";
-const std::string HTMLRenderer::UNIFY_SCRIPT_FILENAME = "unify.pe";
-const std::string HTMLRenderer::NULL_FILENAME = "null";

src/HTMLRenderer/text.cc

@@ -1,5 +1,5 @@
 /*
- * text.ccc
+ * text.cc
  *
  * Handling text & font, and relative stuffs
  *
@@ -18,6 +18,7 @@
 #include <CharCodeToUnicode.h>
 #include <fofi/FoFiTrueType.h>
 
+#include "ff/ff.h"
 #include "HTMLRenderer.h"
 #include "namespace.h"
 #include "config.h"
@@ -162,11 +163,7 @@ void HTMLRenderer::embed_font(const path & filepath, GfxFont * font, FontInfo &
 
     string fn = (format("f%|1$x|") % info.id).str();
 
-    path script_path = tmp_dir / (fn + ".pe");
+    ff_load_font(filepath.c_str());
-    ofstream script_fout(script_path, ofstream::binary);
-    add_tmp_file(fn+".pe");
-
-    script_fout << format("Open(%1%, 1)") % filepath << endl;
 
     int * code2GID = nullptr;
     int code2GID_len = 0;
@@ -200,7 +197,7 @@ void HTMLRenderer::embed_font(const path & filepath, GfxFont * font, FontInfo &
             maxcode = 0xff;
             if((suffix == ".ttf") || (suffix == ".ttc") || (suffix == ".otf"))
             {
-                script_fout << "Reencode(\"original\")" << endl;
+                ff_reencode("original", 0);
                 FoFiTrueType *fftt = nullptr;
                 if((fftt = FoFiTrueType::load((char*)filepath.c_str())) != nullptr)
                 {
@@ -246,8 +243,8 @@ void HTMLRenderer::embed_font(const path & filepath, GfxFont * font, FontInfo &
                 }
                 out << "] def" << endl;
 
-                script_fout << format("LoadEncodingFile(%1%)") % (tmp_dir / (fn+"_.encoding")) << endl;
+                ff_load_encoding((tmp_dir / (fn+"_.encoding")).c_str(), nullptr);
-                script_fout << format("Reencode(\"%1%\")") % fn << endl;
+                ff_reencode(fn.c_str(), 0);
             }
         }
         else
@@ -256,7 +253,7 @@ void HTMLRenderer::embed_font(const path & filepath, GfxFont * font, FontInfo &
 
             if(suffix == ".ttf")
             {
-                script_fout << "Reencode(\"original\")" << endl;
+                ff_reencode("original", 0);
 
                 GfxCIDFont * _font = dynamic_cast<GfxCIDFont*>(font);
 
@@ -266,7 +263,7 @@ void HTMLRenderer::embed_font(const path & filepath, GfxFont * font, FontInfo &
             }
             else
             {
-                script_fout << "CIDFlatten()" << endl;
+                ff_cidflatten();
             }
         }
 
@@ -280,46 +277,49 @@ void HTMLRenderer::embed_font(const path & filepath, GfxFont * font, FontInfo &
          * - For 8bit nonTruetype fonts:
          *   Try to calculate the correct Unicode value from the glyph names, unless param->always_apply_tounicode is set
          * 
+         * TODO: build Encoding directly, without read/write files
          */
 
         auto ctu = font->getToUnicode();
-
-        ofstream map_fout(tmp_dir / (fn + ".encoding"));
-        add_tmp_file(fn+".encoding");
-
         int cnt = 0;
-        for(int i = 0; i <= maxcode; ++i)
+
         {
-            if((suffix != ".ttf") && (font_8bit != nullptr) && (font_8bit->getCharName(i) == nullptr))
+            ofstream map_fout(tmp_dir / (fn + ".encoding"));
-                continue;
+            add_tmp_file(fn+".encoding");
 
-            ++ cnt;
+            for(int i = 0; i <= maxcode; ++i)
-            map_fout << format("0x%|1$X|") % ((code2GID && (i < code2GID_len))? code2GID[i] : i);
-
-            Unicode u, *pu=&u;
-
-            if(info.use_tounicode)
             {
-                int n = 0;
+                if((suffix != ".ttf") && (font_8bit != nullptr) && (font_8bit->getCharName(i) == nullptr))
-                if(ctu)
+                    continue;
-                    n = ctu->mapToUnicode(i, &pu);
-                u = check_unicode(pu, n, i, font);
-            }
-            else
-            {
-                u = unicode_from_font(i, font);
-            }
 
-            map_fout << format(" 0x%|1$X|") % u;
+                ++ cnt;
-            map_fout << format(" # 0x%|1$X|") % i;
+                map_fout << format("0x%|1$X|") % ((code2GID && (i < code2GID_len))? code2GID[i] : i);
 
-            map_fout << endl;
+                Unicode u, *pu=&u;
+
+                if(info.use_tounicode)
+                {
+                    int n = 0;
+                    if(ctu)
+                        n = ctu->mapToUnicode(i, &pu);
+                    u = check_unicode(pu, n, i, font);
+                }
+                else
+                {
+                    u = unicode_from_font(i, font);
+                }
+
+                map_fout << format(" 0x%|1$X|") % u;
+                map_fout << format(" # 0x%|1$X|") % i;
+
+                map_fout << endl;
+            }
         }
 
         if(cnt > 0)
         {
-            script_fout << format("LoadEncodingFile(%1%, \"%2%\")") % (tmp_dir / (fn+".encoding")) % fn << endl;
+            ff_load_encoding((tmp_dir / (fn+".encoding")).c_str(), fn.c_str());
-            script_fout << format("Reencode(\"%1%\", 1)") % fn << endl;
+            ff_reencode(fn.c_str(), 1);
         }
 
         if(ctu)
@@ -336,42 +336,37 @@ void HTMLRenderer::embed_font(const path & filepath, GfxFont * font, FontInfo &
      * Trying to unify them 
      */
     // Generate an intermediate ttf font in order to retrieve the metrics
-    add_tmp_file(fn + "_.ttf");
+    // TODO: see if we can get the values without save/load
-    script_fout << format("Generate(%1%)") % (tmp_dir / (fn + "_.ttf")) << endl;
+    
-    script_fout << "Close()" << endl;
+    string tmp_fn = fn+"_.ttf";
-    script_fout << format("Open(%1%, 1)") % (tmp_dir / (fn + "_.ttf")) << endl;
+    add_tmp_file(tmp_fn);
-    script_fout << ifstream(PDF2HTMLEX_DATA_PATH / UNIFY_SCRIPT_FILENAME).rdbuf();
+    auto tmp_path = tmp_dir / tmp_fn;
-    script_fout << format("Generate(%1%)") % dest << endl;
+    ff_save(tmp_path.c_str());
-    script_fout.close();
+    ff_close();
+    ff_load_font(tmp_path.c_str());
 
-    if(system((boost::format("fontforge -script %1% 1>%2% 2>%3%") % script_path % (tmp_dir / (fn+".info")) % (tmp_dir / NULL_FILENAME)).str().c_str()) != 0)
+    int em = ff_get_em_size();
-        cerr << "Warning: fontforge failed." << endl;
+    int ascent = ff_get_max_ascent();
+    int descent = ff_get_max_descent();
 
-    add_tmp_file(fn+".info");
+    ff_set_ascent(ascent);
-    add_tmp_file(NULL_FILENAME);
+    ff_set_descent(descent);
+    ff_save(dest.c_str());
+    ff_close();
 
-    // read metric
+    if(em != 0)
-    int em, ascent, descent;
-    if(ifstream(tmp_dir / (fn+".info")) >> em >> ascent >> descent)
     {
-        if(em != 0)
+        info.ascent = ((double)ascent) / em;
-        {
+        info.descent = -((double)descent) / em;
-            info.ascent = ((double)ascent) / em;
-            info.descent = -((double)descent) / em;
-        }
-        else
-        {
-            info.ascent = 0;
-            info.descent = 0;
-        }
     }
     else
     {
-        cerr << "Warning: cannot read font info for " << fn << endl;
+        info.ascent = 0;
-        info.ascent = font->getAscent();
+        info.descent = 0;
-        info.descent = font->getDescent();
     }
 
+    // read metric
+
     if(param->debug)
     {
         cerr << "Ascent: " << info.ascent << " Descent: " << info.descent << endl;

src/ff/ff.c

@@ -0,0 +1,156 @@
+/*
+ * ff.c
+ *
+ * Processing fonts using Fontforge
+ *
+ * by WangLu
+ * 2012.09.03
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <stdarg.h>
+
+#include <fontforge.h>
+#include <baseviews.h>
+
+#include "ff.h"
+
+SplineFont * cur_font = NULL;
+
+static void err(const char * format, ...)
+{
+    va_list al;
+    va_start(al, format);
+    vfprintf(stderr, format, al);
+    va_end(al);
+    exit(-1);
+}
+static char * strcopy(const char * str)
+{
+    if(str == NULL) return NULL;
+
+    char * _ = strdup(str);
+    if(!_)
+        err("Not enough memory");
+    return _;
+}
+
+static int max(int a, int b)
+{
+    return (a>b) ? a : b;
+}
+
+void ff_init(void)
+{
+    InitSimpleStuff();
+    if ( default_encoding==NULL )
+        default_encoding=FindOrMakeEncoding("ISO8859-1");
+    if ( default_encoding==NULL )
+        default_encoding=&custom;	/* In case iconv is broken */
+}
+void ff_load_font(const char * filename)
+{
+    char * _filename = strcopy(filename);
+    cur_font = LoadSplineFont(_filename, 1);
+    free(_filename);
+
+    if(!cur_font)
+        err("Cannot load font %s\n", filename);
+
+    if(!cur_font->fv)
+        FVAppend(_FontViewCreate(cur_font));
+}
+
+void ff_load_encoding(const char * filename, const char * encname)
+{
+    char * _filename = strcopy(filename);
+    char * _encname = strcopy(encname);
+    ParseEncodingFile(_filename, _encname);
+    free(_encname);
+    free(_filename);
+}
+
+void ff_reencode(const char * encname, int force)
+{
+    Encoding * enc = FindOrMakeEncoding(encname);
+    if(!enc)
+        err("Unknown encoding %s\n", encname);
+
+    if(force)
+    {
+        SFForceEncoding(cur_font, cur_font->fv->map, enc);
+    }
+    else
+    {
+        EncMapFree(cur_font->fv->map);
+        cur_font->fv->map= EncMapFromEncoding(cur_font, enc);
+    }
+
+    SFReplaceEncodingBDFProps(cur_font, cur_font->fv->map);
+}
+
+void ff_cidflatten(void)
+{
+    printf("cid flatten\n");
+
+    if(!cur_font->cidmaster)
+        err("Cannot flatten a non-CID font");
+    SFFlatten(cur_font->cidmaster);
+}
+
+void ff_save(const char * filename)
+{
+    char * _filename = strcopy(filename);
+    char * _ = strcopy("");
+
+    int r = GenerateScript(cur_font, _filename
+            , _, -1, -1, NULL, NULL, cur_font->fv->map, NULL, ly_fore);
+    
+    free(_);
+    free(_filename);
+
+    if(!r)
+        err("Cannot save font to %s\n", filename);
+}
+
+void ff_close(void)
+{
+    FontViewClose(cur_font->fv);
+    cur_font = NULL;
+}
+
+int ff_get_em_size(void)
+{
+    return (cur_font->pfminfo.os2_typoascent - cur_font->pfminfo.os2_typodescent);
+}
+
+int ff_get_max_ascent(void)
+{
+    return max(cur_font->pfminfo.os2_winascent,
+            max(cur_font->pfminfo.os2_typoascent,
+                cur_font->pfminfo.hhead_ascent));
+}
+
+int ff_get_max_descent(void)
+{
+    return max(cur_font->pfminfo.os2_windescent,
+            max(-cur_font->pfminfo.os2_typodescent,
+                -cur_font->pfminfo.hhead_descent));
+}
+
+void ff_set_ascent(int a)
+{
+    cur_font->pfminfo.os2_winascent = a;
+    cur_font->pfminfo.os2_typoascent = a;
+    cur_font->pfminfo.hhead_ascent = a;
+}
+
+void ff_set_descent(int d)
+{
+    cur_font->pfminfo.os2_windescent = d;
+    cur_font->pfminfo.os2_typodescent = -d;
+    cur_font->pfminfo.hhead_descent = -d;
+}
+

src/ff/ff.h

@@ -0,0 +1,32 @@
+/*
+ * ff.h
+ *
+ * Processing fonts using Fontforge
+ *
+ * fontforge.h cannot be included in C++
+ * So this wrapper in C publishes several functions we need
+ *
+ * by WangLu
+ * 2012.09.03
+ */
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+void ff_init(void);
+void ff_load_font(const char * filename);
+void ff_load_encoding(const char * filename, const char * encname);
+void ff_reencode(const char * encname, int force);
+void ff_cidflatten(void);
+void ff_save(const char * filename);
+void ff_close(void);
+int ff_get_em_size(void);
+int ff_get_max_ascent(void);
+int ff_get_max_descent(void);
+void ff_set_ascent(int a);
+void ff_set_descent(int d);
+
+#ifdef __cplusplus
+}
+#endif