2012-08-14 06:35:55 +00:00
|
|
|
/*
|
2013-02-05 14:00:23 +00:00
|
|
|
* font.cc
|
2012-08-14 06:35:55 +00:00
|
|
|
*
|
2013-02-05 14:00:23 +00:00
|
|
|
* Font processing
|
2012-08-14 06:35:55 +00:00
|
|
|
*
|
2013-02-05 06:36:36 +00:00
|
|
|
* Copyright (C) 2012,2013 Lu Wang <coolwanglu@gmail.com>
|
2012-08-14 06:35:55 +00:00
|
|
|
*/
|
|
|
|
|
|
|
|
#include <iostream>
|
2012-08-30 00:39:49 +00:00
|
|
|
#include <cmath>
|
2012-08-30 17:08:31 +00:00
|
|
|
#include <algorithm>
|
2013-02-05 14:00:23 +00:00
|
|
|
#include <sstream>
|
|
|
|
#include <cctype>
|
2012-08-14 06:35:55 +00:00
|
|
|
|
2012-09-27 09:17:08 +00:00
|
|
|
#include <GlobalParams.h>
|
2012-08-31 09:06:19 +00:00
|
|
|
|
2012-09-27 09:17:08 +00:00
|
|
|
#include "Param.h"
|
2012-08-14 06:35:55 +00:00
|
|
|
#include "HTMLRenderer.h"
|
2012-11-29 09:28:05 +00:00
|
|
|
#include "util/namespace.h"
|
2012-11-29 10:16:05 +00:00
|
|
|
#include "util/math.h"
|
2012-11-29 10:28:07 +00:00
|
|
|
#include "util/misc.h"
|
2013-02-05 14:00:23 +00:00
|
|
|
#include "util/base64stream.h"
|
2012-08-26 15:56:38 +00:00
|
|
|
|
2012-09-12 15:26:14 +00:00
|
|
|
namespace pdf2htmlEX {
|
|
|
|
|
2012-08-31 15:14:05 +00:00
|
|
|
using std::abs;
|
2012-11-29 10:28:07 +00:00
|
|
|
using std::cerr;
|
|
|
|
using std::endl;
|
2012-08-31 15:14:05 +00:00
|
|
|
|
2012-09-04 15:33:15 +00:00
|
|
|
const FontInfo * HTMLRenderer::install_font(GfxFont * font)
|
2012-08-14 06:35:55 +00:00
|
|
|
{
|
|
|
|
assert(sizeof(long long) == 2*sizeof(int));
|
|
|
|
|
2012-09-06 16:58:23 +00:00
|
|
|
long long fn_id = (font == nullptr) ? 0 : hash_ref(font->getID());
|
2012-08-14 06:35:55 +00:00
|
|
|
|
|
|
|
auto iter = font_name_map.find(fn_id);
|
|
|
|
if(iter != font_name_map.end())
|
2012-09-04 15:33:15 +00:00
|
|
|
return &(iter->second);
|
2012-08-14 06:35:55 +00:00
|
|
|
|
|
|
|
long long new_fn_id = font_name_map.size();
|
|
|
|
|
2013-01-19 12:13:31 +00:00
|
|
|
auto cur_info_iter = font_name_map.insert(make_pair(fn_id, FontInfo())).first;
|
|
|
|
|
|
|
|
FontInfo & new_font_info = cur_info_iter->second;
|
|
|
|
new_font_info.id = new_fn_id;
|
|
|
|
new_font_info.use_tounicode = true;
|
2012-08-14 06:35:55 +00:00
|
|
|
|
|
|
|
if(font == nullptr)
|
|
|
|
{
|
2013-01-19 12:13:31 +00:00
|
|
|
new_font_info.ascent = 0;
|
|
|
|
new_font_info.descent = 0;
|
|
|
|
new_font_info.is_type3 = false;
|
|
|
|
|
2012-08-14 06:35:55 +00:00
|
|
|
export_remote_default_font(new_fn_id);
|
2013-01-19 12:13:31 +00:00
|
|
|
|
|
|
|
return &(new_font_info);
|
2012-08-14 06:35:55 +00:00
|
|
|
}
|
|
|
|
|
2013-01-19 12:13:31 +00:00
|
|
|
new_font_info.ascent = font->getAscent();
|
|
|
|
new_font_info.descent = font->getDescent();
|
|
|
|
new_font_info.is_type3 = (font->getType() == fontType3);
|
2012-08-30 16:25:05 +00:00
|
|
|
|
2012-08-14 06:35:55 +00:00
|
|
|
if(param->debug)
|
|
|
|
{
|
2012-09-07 16:38:41 +00:00
|
|
|
cerr << "Install font: (" << (font->getID()->num) << ' ' << (font->getID()->gen) << ") -> " << "f" << hex << new_fn_id << dec << endl;
|
2012-08-14 06:35:55 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
if(font->getType() == fontType3) {
|
2012-08-14 09:13:29 +00:00
|
|
|
cerr << "Type 3 fonts are unsupported and will be rendered as Image" << endl;
|
2012-08-14 06:35:55 +00:00
|
|
|
export_remote_default_font(new_fn_id);
|
2012-09-04 15:33:15 +00:00
|
|
|
return &(cur_info_iter->second);
|
2012-08-14 06:35:55 +00:00
|
|
|
}
|
|
|
|
if(font->getWMode()) {
|
2012-08-14 09:13:29 +00:00
|
|
|
cerr << "Writing mode is unsupported and will be rendered as Image" << endl;
|
2012-08-14 06:35:55 +00:00
|
|
|
export_remote_default_font(new_fn_id);
|
2012-09-04 15:33:15 +00:00
|
|
|
return &(cur_info_iter->second);
|
2012-08-14 06:35:55 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
auto * font_loc = font->locateFont(xref, gTrue);
|
|
|
|
if(font_loc != nullptr)
|
|
|
|
{
|
|
|
|
switch(font_loc -> locType)
|
|
|
|
{
|
|
|
|
case gfxFontLocEmbedded:
|
2012-08-31 07:27:17 +00:00
|
|
|
install_embedded_font(font, cur_info_iter->second);
|
2012-08-14 06:35:55 +00:00
|
|
|
break;
|
|
|
|
case gfxFontLocExternal:
|
2012-08-31 07:27:17 +00:00
|
|
|
install_external_font(font, cur_info_iter->second);
|
2012-08-14 06:35:55 +00:00
|
|
|
break;
|
|
|
|
case gfxFontLocResident:
|
2012-08-31 07:27:17 +00:00
|
|
|
install_base_font(font, font_loc, cur_info_iter->second);
|
2012-08-14 06:35:55 +00:00
|
|
|
break;
|
|
|
|
default:
|
2012-08-14 09:13:29 +00:00
|
|
|
cerr << "TODO: other font loc" << endl;
|
2012-08-14 06:35:55 +00:00
|
|
|
export_remote_default_font(new_fn_id);
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
delete font_loc;
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
export_remote_default_font(new_fn_id);
|
|
|
|
}
|
|
|
|
|
2012-09-04 15:33:15 +00:00
|
|
|
return &(cur_info_iter->second);
|
2012-08-14 06:35:55 +00:00
|
|
|
}
|
|
|
|
|
2012-08-31 07:27:17 +00:00
|
|
|
void HTMLRenderer::install_embedded_font(GfxFont * font, FontInfo & info)
|
2012-08-14 06:35:55 +00:00
|
|
|
{
|
2012-08-31 07:27:17 +00:00
|
|
|
auto path = dump_embedded_font(font, info.id);
|
2012-09-09 17:27:32 +00:00
|
|
|
|
2012-08-31 07:27:17 +00:00
|
|
|
if(path != "")
|
2012-08-26 15:56:38 +00:00
|
|
|
{
|
2012-08-31 07:27:17 +00:00
|
|
|
embed_font(path, font, info);
|
2013-01-29 12:07:51 +00:00
|
|
|
export_remote_font(info, param->font_suffix, font);
|
2012-08-26 15:56:38 +00:00
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
2012-08-31 07:27:17 +00:00
|
|
|
export_remote_default_font(info.id);
|
2012-08-27 16:14:11 +00:00
|
|
|
}
|
2012-08-31 07:27:17 +00:00
|
|
|
}
|
2012-08-27 16:14:11 +00:00
|
|
|
|
2012-08-31 07:27:17 +00:00
|
|
|
void HTMLRenderer::install_base_font(GfxFont * font, GfxFontLoc * font_loc, FontInfo & info)
|
|
|
|
{
|
2012-08-31 09:47:37 +00:00
|
|
|
string psname(font_loc->path->getCString());
|
|
|
|
string basename = psname.substr(0, psname.find('-'));
|
|
|
|
|
2012-09-30 07:43:23 +00:00
|
|
|
GfxFontLoc * localfontloc = font->locateFont(xref, gFalse);
|
2012-08-31 07:27:17 +00:00
|
|
|
if(param->embed_base_font)
|
2012-08-30 15:36:30 +00:00
|
|
|
{
|
2012-08-31 07:50:14 +00:00
|
|
|
if(localfontloc != nullptr)
|
2012-08-30 16:25:05 +00:00
|
|
|
{
|
2012-09-30 07:43:23 +00:00
|
|
|
embed_font(localfontloc->path->getCString(), font, info);
|
2013-01-29 12:07:51 +00:00
|
|
|
export_remote_font(info, param->font_suffix, font);
|
2012-09-30 07:43:23 +00:00
|
|
|
delete localfontloc;
|
|
|
|
return;
|
2012-08-30 16:25:05 +00:00
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
2012-09-07 16:38:41 +00:00
|
|
|
cerr << "Cannot embed base font: f" << hex << info.id << dec << ' ' << psname << endl;
|
2012-08-31 09:47:37 +00:00
|
|
|
// fallback to exporting by name
|
2012-08-30 16:25:05 +00:00
|
|
|
}
|
|
|
|
|
2012-08-30 15:36:30 +00:00
|
|
|
}
|
|
|
|
|
2012-08-14 06:35:55 +00:00
|
|
|
string cssfont;
|
|
|
|
auto iter = BASE_14_FONT_CSS_FONT_MAP.find(basename);
|
|
|
|
if(iter == BASE_14_FONT_CSS_FONT_MAP.end())
|
|
|
|
{
|
2012-08-14 09:13:29 +00:00
|
|
|
cerr << "PS Font: " << basename << " not found in the base 14 font map" << endl;
|
2012-08-14 06:35:55 +00:00
|
|
|
cssfont = "";
|
|
|
|
}
|
|
|
|
else
|
|
|
|
cssfont = iter->second;
|
|
|
|
|
2012-08-31 07:50:14 +00:00
|
|
|
// still try to get an idea of read ascent/descent
|
|
|
|
if(localfontloc != nullptr)
|
|
|
|
{
|
|
|
|
// fill in ascent/descent only, do not embed
|
2012-09-09 16:21:46 +00:00
|
|
|
embed_font(string(localfontloc->path->getCString()), font, info, true);
|
2012-08-31 09:47:37 +00:00
|
|
|
delete localfontloc;
|
2012-08-31 07:50:14 +00:00
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
info.ascent = font->getAscent();
|
|
|
|
info.descent = font->getDescent();
|
|
|
|
}
|
2012-08-31 07:27:17 +00:00
|
|
|
|
2012-08-31 07:50:14 +00:00
|
|
|
export_local_font(info, font, psname, cssfont);
|
2012-08-14 06:35:55 +00:00
|
|
|
}
|
|
|
|
|
2012-08-31 07:50:14 +00:00
|
|
|
void HTMLRenderer::install_external_font(GfxFont * font, FontInfo & info)
|
2012-08-14 06:35:55 +00:00
|
|
|
{
|
2012-08-14 09:13:29 +00:00
|
|
|
string fontname(font->getName()->getCString());
|
2012-08-14 06:35:55 +00:00
|
|
|
|
|
|
|
// resolve bad encodings in GB
|
|
|
|
auto iter = GB_ENCODED_FONT_NAME_MAP.find(fontname);
|
|
|
|
if(iter != GB_ENCODED_FONT_NAME_MAP.end())
|
|
|
|
{
|
|
|
|
fontname = iter->second;
|
2012-08-14 09:13:29 +00:00
|
|
|
cerr << "Warning: workaround for font names in bad encodings." << endl;
|
2012-08-14 06:35:55 +00:00
|
|
|
}
|
2012-08-31 09:06:19 +00:00
|
|
|
|
2012-08-31 09:47:37 +00:00
|
|
|
GfxFontLoc * localfontloc = font->locateFont(xref, gFalse);
|
2012-08-31 09:06:19 +00:00
|
|
|
|
2012-08-31 09:47:37 +00:00
|
|
|
if(param->embed_external_font)
|
2012-08-31 09:06:19 +00:00
|
|
|
{
|
2012-08-31 09:47:37 +00:00
|
|
|
if(localfontloc != nullptr)
|
|
|
|
{
|
2012-09-09 16:21:46 +00:00
|
|
|
embed_font(string(localfontloc->path->getCString()), font, info);
|
2013-01-29 12:07:51 +00:00
|
|
|
export_remote_font(info, param->font_suffix, font);
|
2012-08-31 09:47:37 +00:00
|
|
|
delete localfontloc;
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
2012-09-07 16:38:41 +00:00
|
|
|
cerr << "Cannot embed external font: f" << hex << info.id << dec << ' ' << fontname << endl;
|
2012-08-31 09:47:37 +00:00
|
|
|
// fallback to exporting by name
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
// still try to get an idea of read ascent/descent
|
|
|
|
if(localfontloc != nullptr)
|
|
|
|
{
|
|
|
|
// fill in ascent/descent only, do not embed
|
2012-09-09 16:21:46 +00:00
|
|
|
embed_font(string(localfontloc->path->getCString()), font, info, true);
|
2012-08-31 09:47:37 +00:00
|
|
|
delete localfontloc;
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
info.ascent = font->getAscent();
|
|
|
|
info.descent = font->getDescent();
|
2012-08-31 09:06:19 +00:00
|
|
|
}
|
2012-08-14 06:35:55 +00:00
|
|
|
|
2012-08-31 07:50:14 +00:00
|
|
|
export_local_font(info, font, fontname, "");
|
2012-08-14 06:35:55 +00:00
|
|
|
}
|
2013-02-05 14:00:23 +00:00
|
|
|
|
|
|
|
void HTMLRenderer::export_remote_font(const FontInfo & info, const string & suffix, GfxFont * font)
|
|
|
|
{
|
|
|
|
string mime_type, format;
|
|
|
|
if(suffix == ".ttf")
|
|
|
|
{
|
|
|
|
format = "truetype";
|
|
|
|
mime_type = "application/x-font-ttf";
|
|
|
|
}
|
|
|
|
else if(suffix == ".otf")
|
|
|
|
{
|
|
|
|
format = "opentype";
|
|
|
|
mime_type = "application/x-font-otf";
|
|
|
|
}
|
|
|
|
else if(suffix == ".woff")
|
|
|
|
{
|
|
|
|
format = "woff";
|
|
|
|
mime_type = "application/font-woff";
|
|
|
|
}
|
|
|
|
else if(suffix == ".eot")
|
|
|
|
{
|
|
|
|
format = "embedded-opentype";
|
|
|
|
mime_type = "application/vnd.ms-fontobject";
|
|
|
|
}
|
|
|
|
else if(suffix == ".svg")
|
|
|
|
{
|
|
|
|
format = "svg";
|
|
|
|
mime_type = "image/svg+xml";
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
cerr << "Warning: unknown font suffix: " << suffix << endl;
|
|
|
|
}
|
|
|
|
|
|
|
|
f_css.fs << "@font-face{"
|
|
|
|
<< "font-family:f" << info.id << ";"
|
|
|
|
<< "src:url(";
|
|
|
|
|
|
|
|
{
|
|
|
|
auto fn = str_fmt("f%llx%s", info.id, suffix.c_str());
|
|
|
|
if(param->single_html)
|
|
|
|
{
|
|
|
|
auto path = param->tmp_dir + "/" + (char*)fn;
|
|
|
|
ifstream fin(path, ifstream::binary);
|
|
|
|
if(!fin)
|
|
|
|
throw "Cannot locate font file: " + path;
|
|
|
|
f_css.fs << "'data:font/" + mime_type + ";base64," << base64stream(fin) << "'";
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
f_css.fs << (char*)fn;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
f_css.fs << ")"
|
|
|
|
<< "format(\"" << format << "\");"
|
|
|
|
<< "}" // end of @font-face
|
|
|
|
<< ".f" << info.id << "{"
|
|
|
|
<< "font-family:f" << info.id << ";"
|
|
|
|
<< "line-height:" << round(info.ascent - info.descent) << ";"
|
|
|
|
<< "font-style:normal;"
|
|
|
|
<< "font-weight:normal;"
|
|
|
|
<< "visibility:visible;"
|
|
|
|
<< "}" // end of .f
|
|
|
|
<< endl;
|
|
|
|
}
|
|
|
|
|
|
|
|
static string general_font_family(GfxFont * font)
|
|
|
|
{
|
|
|
|
if(font->isFixedWidth())
|
|
|
|
return "monospace";
|
|
|
|
else if (font->isSerif())
|
|
|
|
return "serif";
|
|
|
|
else
|
|
|
|
return "sans-serif";
|
|
|
|
}
|
|
|
|
|
|
|
|
// TODO: this function is called when some font is unable to process, may use the name there as a hint
|
|
|
|
void HTMLRenderer::export_remote_default_font(long long fn_id)
|
|
|
|
{
|
|
|
|
f_css.fs << ".f" << fn_id << "{font-family:sans-serif;visibility:hidden;}" << endl;
|
|
|
|
}
|
|
|
|
|
|
|
|
void HTMLRenderer::export_local_font(const FontInfo & info, GfxFont * font, const string & original_font_name, const string & cssfont)
|
|
|
|
{
|
|
|
|
f_css.fs << ".f" << info.id << "{";
|
|
|
|
f_css.fs << "font-family:" << ((cssfont == "") ? (original_font_name + "," + general_font_family(font)) : cssfont) << ";";
|
|
|
|
|
|
|
|
string fn = original_font_name;
|
|
|
|
for(auto iter = fn.begin(); iter != fn.end(); ++iter)
|
|
|
|
*iter = tolower(*iter);
|
|
|
|
|
|
|
|
if(font->isBold() || (fn.find("bold") != string::npos))
|
|
|
|
f_css.fs << "font-weight:bold;";
|
|
|
|
else
|
|
|
|
f_css.fs << "font-weight:normal;";
|
|
|
|
|
|
|
|
if(fn.find("oblique") != string::npos)
|
|
|
|
f_css.fs << "font-style:oblique;";
|
|
|
|
else if(font->isItalic() || (fn.find("italic") != string::npos))
|
|
|
|
f_css.fs << "font-style:italic;";
|
|
|
|
else
|
|
|
|
f_css.fs << "font-style:normal;";
|
|
|
|
|
|
|
|
f_css.fs << "line-height:" << round(info.ascent - info.descent) << ";";
|
|
|
|
|
|
|
|
f_css.fs << "visibility:visible;";
|
|
|
|
|
|
|
|
f_css.fs << "}" << endl;
|
|
|
|
}
|
|
|
|
|
|
|
|
} //namespace pdf2hmlEX
|