experimental: fix with of glphs accoring to the info in PDF

src/HTMLRenderer/general.cc

@@ -45,6 +45,7 @@ HTMLRenderer::HTMLRenderer(const Param * param)
     ffw_init(param->debug);
     cur_mapping = new int32_t [0x10000];
     cur_mapping2 = new char* [0x100];
+    width_list = new int [0x10000];
 }
 
 HTMLRenderer::~HTMLRenderer()
@@ -53,6 +54,7 @@ HTMLRenderer::~HTMLRenderer()
     clean_tmp_files();
     delete [] cur_mapping;
     delete [] cur_mapping2;
+    delete [] width_list;
 }
 
 static GBool annot_cb(Annot *, void *) {

src/HTMLRenderer/text.cc

@@ -11,6 +11,7 @@
 #include <algorithm>
 #include <unordered_set>
 #include <cctype>
+#include <cmath>
 
 #include <CharCodeToUnicode.h>
 #include <fofi/FoFiTrueType.h>
@@ -24,6 +25,7 @@ namespace pdf2htmlEX {
 using std::unordered_set;
 using std::min;
 using std::all_of;
+using std::round;
 
 string HTMLRenderer::dump_embedded_font (GfxFont * font, long long fn_id)
 {
@@ -165,6 +167,7 @@ void HTMLRenderer::embed_font(const string & filepath, GfxFont * font, FontInfo
     int maxcode = 0;
 
     Gfx8BitFont * font_8bit = nullptr;
+    GfxCIDFont * font_cid = nullptr;
 
     string suffix = get_suffix(filepath);
     for(auto iter = suffix.begin(); iter != suffix.end(); ++iter)
@@ -175,6 +178,13 @@ void HTMLRenderer::embed_font(const string & filepath, GfxFont * font, FontInfo
 
     const char * used_map = nullptr;
 
+    ffw_metric(&info.ascent, &info.descent, &info.em_size);
+
+    if(param->debug)
+    {
+        cerr << "Ascent: " << info.ascent << " Descent: " << info.descent << endl;
+    }
+
     if(!get_metric_only)
     {
         used_map = font_preprocessor.get_code_map(hash_ref(font->getID()));
@@ -250,6 +260,7 @@ void HTMLRenderer::embed_font(const string & filepath, GfxFont * font, FontInfo
         }
         else
         {
+            font_cid = dynamic_cast<GfxCIDFont*>(font);
             maxcode = 0xffff;
 
             if(is_truetype_suffix(suffix))
@@ -278,6 +289,8 @@ void HTMLRenderer::embed_font(const string & 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
          * 
+         *
+         * Also fill in the width_list, and set widths accordingly
          */
 
 
@@ -286,18 +299,27 @@ void HTMLRenderer::embed_font(const string & filepath, GfxFont * font, FontInfo
             bool name_conflict_warned = false;
 
             auto ctu = font->getToUnicode();
-            memset(cur_mapping, 0, 0x10000 * sizeof(int32_t));
+            memset(cur_mapping, -1, 0x10000 * sizeof(*cur_mapping));
+            memset(width_list, -1, 0x1000 * sizeof(*width_list));
 
             if(code2GID)
                 maxcode = min(maxcode, code2GID_len - 1);
 
+            bool is_truetype = is_truetype_suffix(suffix);
             int max_key = maxcode;
+            /*
+             * Traverse all possible codes
+             */
             for(int i = 0; i <= maxcode; ++i)
             {
                 if(!used_map[i])
                     continue;
 
-                if(is_truetype_suffix(suffix) && (font_8bit != nullptr) && (font_8bit->getCharName(i) == nullptr))
+                /*
+                 * Skip glyphs without names (only for non-ttf fonts)
+                 */
+                if(!is_truetype && (font_8bit != nullptr) 
+                        && (font_8bit->getCharName(i) == nullptr))
                 {
                     continue;
                 }
@@ -338,9 +360,22 @@ void HTMLRenderer::embed_font(const string & filepath, GfxFont * font, FontInfo
                         cerr << "Warning: encoding confliction detected in font: " << hex << info.id << dec << endl;
                     }
                 }
+
+                if(font_8bit)
+                {
+                    width_list[k] = (int)round(font_8bit->getWidth(i) * info.em_size);
+                }
+                else
+                {
+                    char buf[2];  
+                    buf[0] = (i >> 8) & 0xff;
+                    buf[1] = (i & 0xff);
+                    width_list[k] = (int)round(font_cid->getWidth(buf, 2) * info.em_size);
+                }
             }
 
             ffw_reencode_raw(cur_mapping, max_key + 1, 1);
+            ffw_set_widths(width_list, max_key + 1);
 
             if(ctu)
                 ctu->decRefCnt();
@@ -357,21 +392,15 @@ void HTMLRenderer::embed_font(const string & filepath, GfxFont * font, FontInfo
         // TODO: see if we can get the values without save/load
 
 
+        /*
         auto fn = str_fmt("%s/f%llx_.ttf", param->tmp_dir.c_str(), info.id);
         add_tmp_file((char*)fn);
 
         ffw_save((char*)fn);
         ffw_close();
         ffw_load_font((char*)fn);
+        */
     }
-
-    ffw_metric(&info.ascent, &info.descent);
-
-    if(param->debug)
-    {
-        cerr << "Ascent: " << info.ascent << " Descent: " << info.descent << endl;
-    }
-
     {
         auto fn = str_fmt("%s/f%llx%s", 
                 (param->single_html ? param->tmp_dir : param->dest_dir).c_str(),

src/ffw.c

@@ -18,6 +18,11 @@
 
 #include "ffw.h"
 
+static inline int min(int a, int b)
+{
+    return (a<b)?a:b;
+}
+
 static FontViewBase * cur_fv = NULL;
 static Encoding * original_enc = NULL;
 static Encoding * enc_head = NULL;
@@ -209,13 +214,17 @@ void ffw_close(void)
     cur_fv = NULL;
 }
 
-void ffw_metric(double * ascent, double * descent)
+void ffw_metric(double * ascent, double * descent, int * em_size)
 {
-    DBounds bb;
     SplineFont * sf = cur_fv->sf;
+
+    DBounds bb;
     SplineFontFindBounds(sf, &bb);
+
     struct pfminfo * info = &sf->pfminfo;
 
+    *em_size = sf->ascent + sf->descent;
+
     /*
     //debug
     printf("bb %lf %lf\n", bb.maxy, bb.miny);
@@ -267,3 +276,29 @@ void ffw_metric(double * ascent, double * descent)
     sf->changed = true;
 }
 
+/*
+ * TODO:bitmap, reference have not been considered in this function
+ */
+void ffw_set_widths(int * width_list, int mapping_len)
+{
+    SplineFont * sf = cur_fv->sf;
+
+    EncMap * map = cur_fv->map;
+    int i;
+    int imax = min(mapping_len, map->enccount);
+    for(i = 0; i < imax; ++i)
+    {
+        // TODO why need this
+        // when width_list[i] == -1, the code itself should be unused.
+        // but might be reference within ttf etc
+        if(width_list[i] == -1) continue;
+
+        int j = map->map[i];
+        if(j == -1) continue;
+
+        SplineChar * sc = sf->glyphs[j];
+        if(sc == NULL) continue;
+
+        sc->width = width_list[i];
+    }
+}

src/include/HTMLRenderer.h

@@ -357,6 +357,7 @@ class HTMLRenderer : public OutputDev
         // for font reencoding
         int32_t * cur_mapping;
         char ** cur_mapping2;
+        int * width_list;
         FontPreprocessor font_preprocessor;
 
         // for string formatting

src/include/ffw.h

@@ -31,7 +31,9 @@ void ffw_save(const char * filename);
 void ffw_close(void);
 
 // fix metrics and get them
-void ffw_metric(double * ascent, double * descent);
+void ffw_metric(double * ascent, double * descent, int * em_size);
+
+void ffw_set_widths(int * width_list, int mapping_len);
 
 #ifdef __cplusplus
 }

src/include/util.h

@@ -109,6 +109,7 @@ class FontInfo
 public:
     long long id;
     bool use_tounicode;
+    int em_size;
     double ascent, descent;
     bool has_space; // whether space is included in the font
 };