DenysMedvid/pdf2htmlEX
1
0
Fork 0
mirror of https://github.com/pdf2htmlEX/pdf2htmlEX.git synced 2024-12-22 13:00:08 +00:00
Code

add manifest & split-pages

Browse Source
This commit is contained in:
Lu Wang 2012-09-12 23:26:14 +08:00
parent 78bf7fcaa5
commit 61cdce7e7d
20 changed files with 292 additions and 110 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
pdf2htmlEX.1
View File

@ -58,6 +58,11 @@ Whether to embed everything into one HTML file.
If switched out, there will be several files generated along with the HTML file including files for fonts, css, images. If switched out, there will be several files generated along with the HTML file including files for fonts, css, images.
.TP .TP
.B --split-pages <0|1> (Default: 0)
If turned on, each page is saved in a separated files, also the generated css file will be store separatedly as if single-html=0
The output files will be named as <output-filename>0.page, <output-file>1.page, ...
.TP
.B --embed-base-font <0|1> (Default: 1) .B --embed-base-font <0|1> (Default: 1)
Whether to embed base 14 fonts. Whether to embed base 14 fonts.
@ -100,6 +105,11 @@ Treat space characters as offsets, which may increase the size of the output.
Turn it on if space characters are not displayed correctly, or you want to remove positional spaces. Turn it on if space characters are not displayed correctly, or you want to remove positional spaces.
.TP .TP
.B --css-filename <filename> (Default: "")
Specify the filename of the generated css file, if not embedded.
If it's empty, the file name will be determined automatically.
.TP
.B --font-suffix <suffix> (Default: ".ttf"), --font-format <format> (Default: "truetype") .B --font-suffix <suffix> (Default: ".ttf"), --font-format <format> (Default: "truetype")
Specify the suffix and format of fonts extracted from the PDF file. They should be consistent. Specify the suffix and format of fonts extracted from the PDF file. They should be consistent.
.TP .TP

0
share/all.css → share/base.css
View File

7
share/head.html
View File

@ -1,7 +0,0 @@
<!-- Created by pdf2htmlEX (http://github.com/coolwanglu/pdf2htmlEX) -->
<!-- head.html by WangLu 2012.08.14 -->
<!DOCTYPE html>
<html>
<head>
<meta charset="utf-8">
<!-- head.html END -->

2
share/jquery.js vendored Normal file
View File

File diff suppressed because one or more lines are too long

48
share/manifest Normal file
View File

@ -0,0 +1,48 @@
# manifest
# by WangLu
# 2012.09.12
#
# Syntax
# The first char of each line is the command
# Empty lines are ignored
#
# # - comment
# @ - include file
# $ - special use for pdf2htmlEX
#
# Special
# If a line contains """ only, all text until next """ will be included
"""
<!-- Created by pdf2htmlEX (http://github.com/coolwanglu/pdf2htmlEX) -->
<!DOCTYPE html>
<html>
<head>
<meta charset="utf-8">
"""
@base.css
$css
@jquery.js
"""
<title></title>
</head>
<body>
<div id="pdf-main">
"""
$pages
"""
</div>
"""
@scroll.js
"""
</body>
</html>
"""
# MANIFEST END

7
share/neck.html
View File

@ -1,7 +0,0 @@
<!-- neck.html by WangLu 2012.08.15 -->
<title>pdf2htmlEX</title>
<script src="http://ajax.googleapis.com/ajax/libs/jquery/1.8.1/jquery.min.js"></script>
</head>
<body>
<div id="pdf-main">
<!-- neck.html END -->

8
share/tail.html → share/scroll.js
View File

@ -1,10 +1,6 @@
<!-- tail.html by Hongliang Tian 2012.09.11 -->
</div>
<script type="text/javascript">
$(function() { $(function() {
var $pages = $(".p"), var $pages = $(".p"),
$pageWrappers = $(".b"), $pageWrappers = $(".b"),
$pageWrappers = $(".pw"),
$main = $("#pdf-main"), $main = $("#pdf-main"),
l = $pages.length; l = $pages.length;
@ -63,7 +59,3 @@ $(function() {
// Trigger the event // Trigger the event
$("#pdf-main").scroll(); $("#pdf-main").scroll();
}); });
</script>
</body>
</html>
<!-- tail.html END -->

3
src/HTMLRenderer/LineBuffer.cc
View File

@ -12,6 +12,8 @@
#include "HTMLRenderer.h" #include "HTMLRenderer.h"
#include "namespace.h" #include "namespace.h"
namespace pdf2htmlEX {
using std::min; using std::min;
using std::max; using std::max;
using std::vector; using std::vector;
@ -272,3 +274,4 @@ int HTMLRenderer::LineBuffer::State::diff(const State & s) const
} }
const char * HTMLRenderer::LineBuffer::State::format_str = "fsclwr"; const char * HTMLRenderer::LineBuffer::State::format_str = "fsclwr";
} //namespace pdf2htmlEX

49
src/HTMLRenderer/export.cc
View File

@ -13,23 +13,25 @@
#include "HTMLRenderer.h" #include "HTMLRenderer.h"
#include "namespace.h" #include "namespace.h"
namespace pdf2htmlEX {
void HTMLRenderer::export_remote_font(const FontInfo & info, const string & suffix, const string & fontfileformat, GfxFont * font) void HTMLRenderer::export_remote_font(const FontInfo & info, const string & suffix, const string & fontfileformat, GfxFont * font)
{ {
allcss_fout << "@font-face{font-family:f" << info.id << ";src:url("; css_fout << "@font-face{font-family:f" << info.id << ";src:url(";
{ {
auto fn = str_fmt("f%llx%s", info.id, suffix.c_str()); auto fn = str_fmt("f%llx%s", info.id, suffix.c_str());
if(param->single_html) if(param->single_html)
{ {
allcss_fout << "'data:font/opentype;base64," << base64stream(ifstream(tmp_dir + "/" + (char*)fn, ifstream::binary)) << "'"; css_fout << "'data:font/opentype;base64," << base64stream(ifstream(tmp_dir + "/" + (char*)fn, ifstream::binary)) << "'";
} }
else else
{ {
allcss_fout << (char*)fn; css_fout << (char*)fn;
} }
} }
allcss_fout << ")format(\"" << fontfileformat << "\");}.f" << info.id << "{font-family:f" << info.id << ";line-height:" << (info.ascent - info.descent) << ";}" << endl; css_fout << ")format(\"" << fontfileformat << "\");}.f" << info.id << "{font-family:f" << info.id << ";line-height:" << (info.ascent - info.descent) << ";}" << endl;
} }
static string general_font_family(GfxFont * font) static string general_font_family(GfxFont * font)
@ -45,39 +47,39 @@ static string general_font_family(GfxFont * font)
// TODO: this function is called when some font is unable to process, may use the name there as a hint // 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) void HTMLRenderer::export_remote_default_font(long long fn_id)
{ {
allcss_fout << ".f" << fn_id << "{font-family:sans-serif;color:transparent;visibility:hidden;}" << endl; css_fout << ".f" << fn_id << "{font-family:sans-serif;color:transparent;visibility:hidden;}" << endl;
} }
void HTMLRenderer::export_local_font(const FontInfo & info, GfxFont * font, const string & original_font_name, const string & cssfont) void HTMLRenderer::export_local_font(const FontInfo & info, GfxFont * font, const string & original_font_name, const string & cssfont)
{ {
allcss_fout << ".f" << info.id << "{"; css_fout << ".f" << info.id << "{";
allcss_fout << "font-family:" << ((cssfont == "") ? (original_font_name + "," + general_font_family(font)) : cssfont) << ";"; css_fout << "font-family:" << ((cssfont == "") ? (original_font_name + "," + general_font_family(font)) : cssfont) << ";";
string fn = original_font_name; string fn = original_font_name;
for(auto iter = fn.begin(); iter != fn.end(); ++iter) for(auto iter = fn.begin(); iter != fn.end(); ++iter)
*iter = tolower(*iter); *iter = tolower(*iter);
if(font->isBold() || (fn.find("bold") != string::npos)) if(font->isBold() || (fn.find("bold") != string::npos))
allcss_fout << "font-weight:bold;"; css_fout << "font-weight:bold;";
if(fn.find("oblique") != string::npos) if(fn.find("oblique") != string::npos)
allcss_fout << "font-style:oblique;"; css_fout << "font-style:oblique;";
else if(font->isItalic() || (fn.find("italic") != string::npos)) else if(font->isItalic() || (fn.find("italic") != string::npos))
allcss_fout << "font-style:italic;"; css_fout << "font-style:italic;";
allcss_fout << "line-height:" << (info.ascent - info.descent) << ";"; css_fout << "line-height:" << (info.ascent - info.descent) << ";";
allcss_fout << "}" << endl; css_fout << "}" << endl;
} }
void HTMLRenderer::export_font_size (long long fs_id, double font_size) void HTMLRenderer::export_font_size (long long fs_id, double font_size)
{ {
allcss_fout << ".s" << fs_id << "{font-size:" << font_size << "px;}" << endl; css_fout << ".s" << fs_id << "{font-size:" << font_size << "px;}" << endl;
} }
void HTMLRenderer::export_transform_matrix (long long tm_id, const double * tm) void HTMLRenderer::export_transform_matrix (long long tm_id, const double * tm)
{ {
allcss_fout << ".t" << tm_id << "{"; css_fout << ".t" << tm_id << "{";
// always ignore tm[4] and tm[5] because // always ignore tm[4] and tm[5] because
// we have already shifted the origin // we have already shifted the origin
@ -88,30 +90,30 @@ void HTMLRenderer::export_transform_matrix (long long tm_id, const double * tm)
{ {
const auto & prefix = *iter; const auto & prefix = *iter;
// PDF use a different coordinate system from Web // PDF use a different coordinate system from Web
allcss_fout << prefix << "transform:matrix(" css_fout << prefix << "transform:matrix("
<< tm[0] << ',' << tm[0] << ','
<< -tm[1] << ',' << -tm[1] << ','
<< -tm[2] << ',' << -tm[2] << ','
<< tm[3] << ','; << tm[3] << ',';
allcss_fout << "0,0);"; css_fout << "0,0);";
} }
allcss_fout << "}" << endl; css_fout << "}" << endl;
} }
void HTMLRenderer::export_letter_space (long long ls_id, double letter_space) void HTMLRenderer::export_letter_space (long long ls_id, double letter_space)
{ {
allcss_fout << ".l" << ls_id << "{letter-spacing:" << letter_space << "px;}" << endl; css_fout << ".l" << ls_id << "{letter-spacing:" << letter_space << "px;}" << endl;
} }
void HTMLRenderer::export_word_space (long long ws_id, double word_space) void HTMLRenderer::export_word_space (long long ws_id, double word_space)
{ {
allcss_fout << ".w" << ws_id << "{word-spacing:" << word_space << "px;}" << endl; css_fout << ".w" << ws_id << "{word-spacing:" << word_space << "px;}" << endl;
} }
void HTMLRenderer::export_color (long long color_id, const GfxRGB * rgb) void HTMLRenderer::export_color (long long color_id, const GfxRGB * rgb)
{ {
allcss_fout << ".c" << color_id << "{color:rgb(" css_fout << ".c" << color_id << "{color:rgb("
<< dec << (int)colToByte(rgb->r) << "," << (int)colToByte(rgb->g) << "," << (int)colToByte(rgb->b) << ");}" << hex << dec << (int)colToByte(rgb->r) << "," << (int)colToByte(rgb->g) << "," << (int)colToByte(rgb->b) << ");}" << hex
<< endl; << endl;
} }
@ -119,13 +121,14 @@ void HTMLRenderer::export_color (long long color_id, const GfxRGB * rgb)
void HTMLRenderer::export_whitespace (long long ws_id, double ws_width) void HTMLRenderer::export_whitespace (long long ws_id, double ws_width)
{ {
if(ws_width > 0) if(ws_width > 0)
allcss_fout << "._" << ws_id << "{display:inline-block;width:" << ws_width << "px;}" << endl; css_fout << "._" << ws_id << "{display:inline-block;width:" << ws_width << "px;}" << endl;
else else
allcss_fout << "._" << ws_id << "{display:inline;margin-left:" << ws_width << "px;}" << endl; css_fout << "._" << ws_id << "{display:inline;margin-left:" << ws_width << "px;}" << endl;
} }
void HTMLRenderer::export_rise (long long rise_id, double rise) void HTMLRenderer::export_rise (long long rise_id, double rise)
{ {
allcss_fout << ".r" << rise_id << "{top:" << (-rise) << "px;}" << endl; css_fout << ".r" << rise_id << "{top:" << (-rise) << "px;}" << endl;
} }
}

199
src/HTMLRenderer/general.cc
View File

@ -8,6 +8,7 @@
*/ */
#include <cstdio> #include <cstdio>
#include <ostream>
#include <splash/SplashBitmap.h> #include <splash/SplashBitmap.h>
@ -17,8 +18,11 @@
#include "ff.h" #include "ff.h"
#include "pdf2htmlEX-config.h" #include "pdf2htmlEX-config.h"
namespace pdf2htmlEX {
using std::fixed; using std::fixed;
using std::flush; using std::flush;
using std::ostream;
static void dummy(void *, enum ErrorCategory, int pos, char *) static void dummy(void *, enum ErrorCategory, int pos, char *)
{ {
@ -83,6 +87,14 @@ void HTMLRenderer::process(PDFDoc *doc)
for(int i = param->first_page; i <= param->last_page ; ++i) for(int i = param->first_page; i <= param->last_page ; ++i)
{ {
if(param->split_pages)
{
auto page_fn = str_fmt("%s/__pages%x", tmp_dir.c_str(), i);
html_fout.open((char*)page_fn, ofstream::binary);
add_tmp_file((char*)page_fn);
fix_stream(html_fout);
}
if(param->process_nontext) if(param->process_nontext)
{ {
doc->displayPage(bg_renderer, i, param->h_dpi, param->v_dpi, doc->displayPage(bg_renderer, i, param->h_dpi, param->v_dpi,
@ -104,8 +116,14 @@ void HTMLRenderer::process(PDFDoc *doc)
0, true, false, false, 0, true, false, false,
nullptr, nullptr, nullptr, nullptr); nullptr, nullptr, nullptr, nullptr);
if(param->split_pages)
{
html_fout.close();
}
cerr << "." << flush; cerr << "." << flush;
} }
post_process(); post_process();
if(bg_renderer) if(bg_renderer)
@ -116,53 +134,116 @@ void HTMLRenderer::process(PDFDoc *doc)
void HTMLRenderer::pre_process() void HTMLRenderer::pre_process()
{ {
// we may output utf8 characters, so use binary // we may output utf8 characters, so always use binary
if(param->single_html)
{ {
{ /*
auto fn = str_fmt("%s/%s", tmp_dir.c_str(), CSS_FILENAME.c_str()); * If single-html && !split-pages
allcss_fout.open((char*)fn, ofstream::binary); * we have to keep the generated css file into a temporary place
* and embed it into the main html later
*
*
* If single-html && split-page
* as there's no place to embed the css file, just leave it alone (into dest_dir)
*
* If !single-html
* leave it in dest_dir
*/
auto fn = (param->single_html && (!param->split_pages))
? str_fmt("%s/__css", tmp_dir.c_str())
: str_fmt("%s/%s", dest_dir.c_str(), param->css_filename.c_str());
if(param->single_html)
add_tmp_file((char*)fn); add_tmp_file((char*)fn);
}
{ css_path = (char*)fn,
// don't use output_file directly css_fout.open(css_path, ofstream::binary);
// otherwise it'll be a disaster when tmp_dir == dest_dir fix_stream(css_fout);
auto tmp_output_fn = str_fmt("%s/%s.part", tmp_dir.c_str(), param->output_filename.c_str());
add_tmp_file((char*)tmp_output_fn);
html_fout.open((char*)tmp_output_fn, ofstream::binary);
}
} }
else
// if split-pages is specified, open & close the file in the process loop
// if not, open the file here:
if(!param->split_pages);
{ {
html_fout.open(str_fmt("%s/%s", dest_dir.c_str(), param->output_filename.c_str()), ofstream::binary); /*
allcss_fout.open(str_fmt("%s/%s", dest_dir.c_str(), CSS_FILENAME.c_str()), ofstream::binary); * If single-html
* we have to keep the html file (for page) into a temporary place
* because we'll have to embed css before it
*
* Otherwise just generate it
*/
auto fn = (param->single_html)
? str_fmt("%s/__pages", tmp_dir.c_str())
: str_fmt("%s/%s", dest_dir.c_str(), param->output_filename.c_str());
html_fout << ifstream(str_fmt("%s/%s", PDF2HTMLEX_DATA_PATH.c_str(), HEAD_HTML_FILENAME.c_str()), ifstream::binary).rdbuf(); if(param->single_html)
html_fout << "<link rel=\"stylesheet\" type=\"text/css\" href=\"" << CSS_FILENAME << "\"/>" << endl; add_tmp_file((char*)fn);
html_fout << ifstream(str_fmt("%s/%s", PDF2HTMLEX_DATA_PATH.c_str(), NECK_HTML_FILENAME.c_str()), ifstream::binary).rdbuf();
html_path = (char*)fn;
html_fout.open(html_path, ofstream::binary);
fix_stream(html_fout);
} }
fix_stream(html_fout);
fix_stream(allcss_fout);
allcss_fout << ifstream(str_fmt("%s/%s", PDF2HTMLEX_DATA_PATH.c_str(), CSS_FILENAME.c_str()), ifstream::binary).rdbuf();
} }
void HTMLRenderer::post_process() void HTMLRenderer::post_process()
{ {
if(!param->single_html) // close files
{
html_fout << ifstream(str_fmt("%s/%s", PDF2HTMLEX_DATA_PATH.c_str(), TAIL_HTML_FILENAME.c_str()), ifstream::binary).rdbuf();
}
html_fout.close(); html_fout.close();
allcss_fout.close(); css_fout.close();
if(param->single_html) //only when !split-page, do we have some work left to do
if(!param->split_pages)
return;
ofstream output((char*)str_fmt("%s/%s", dest_dir.c_str(), param->output_filename.c_str()));
fix_stream(output);
// apply manifest
ifstream manifest_fin((char*)str_fmt("%s/%s", PDF2HTMLEX_DATA_PATH.c_str(), MANIFEST_FILENAME.c_str()));
bool embed_string = false;
string line;
while(getline(manifest_fin, line))
{ {
process_single_html(); if(embed_string)
{
output << line << endl;
continue;
}
if(line.empty() || line[0] == '#')
continue;
if(line == "\"\"\"")
{
embed_string = !embed_string;
continue;
}
if(line[0] == '@')
{
embed_file(output, PDF2HTMLEX_DATA_PATH + "/" + line.substr(1), true);
continue;
}
if(line[0] == '$')
{
if(line == "$css")
{
embed_file(output, css_path, false);
}
else if (line == "$pages")
{
output << ifstream(html_path, ifstream::binary).rdbuf();
}
else
{
cerr << "Warning: unknown line in manifest: " << line << endl;
}
continue;
}
cerr << "Warning: unknown line in manifest: " << line << endl;
} }
} }
@ -182,7 +263,7 @@ void HTMLRenderer::startPage(int pageNum, GfxState *state)
{ {
if(param->single_html) if(param->single_html)
{ {
html_fout << "'data:image/png;base64," << base64stream(ifstream(str_fmt("%s/p%llx.png", tmp_dir.c_str(), pageNum) , ifstream::binary)) << "'"; html_fout << "'data:image/png;base64," << base64stream(ifstream((char*)str_fmt("%s/p%llx.png", tmp_dir.c_str(), pageNum) , ifstream::binary)) << "'";
} }
else else
{ {
@ -225,23 +306,6 @@ void HTMLRenderer::endPage() {
html_fout << "</div></div>" << endl; html_fout << "</div></div>" << endl;
} }
void HTMLRenderer::process_single_html()
{
ofstream out (dest_dir + "/" + param->output_filename, ofstream::binary);
out << ifstream(PDF2HTMLEX_DATA_PATH + "/" + HEAD_HTML_FILENAME , ifstream::binary).rdbuf();
out << "<style type=\"text/css\">" << endl;
out << ifstream(tmp_dir + "/" + CSS_FILENAME, ifstream::binary).rdbuf();
out << "</style>" << endl;
out << ifstream(PDF2HTMLEX_DATA_PATH + "/" + NECK_HTML_FILENAME, ifstream::binary).rdbuf();
out << ifstream(tmp_dir + "/" + (param->output_filename + ".part"), ifstream::binary).rdbuf();
out << ifstream(PDF2HTMLEX_DATA_PATH + "/" + TAIL_HTML_FILENAME, ifstream::binary).rdbuf();
}
void HTMLRenderer::fix_stream (std::ostream & out) void HTMLRenderer::fix_stream (std::ostream & out)
{ {
out << fixed << hex; out << fixed << hex;
@ -274,7 +338,32 @@ void HTMLRenderer::clean_tmp_files()
cerr << "Remove temporary directory: " << tmp_dir << endl; cerr << "Remove temporary directory: " << tmp_dir << endl;
} }
const std::string HTMLRenderer::HEAD_HTML_FILENAME = "head.html"; void HTMLRenderer::embed_file(ostream & out, const string & path, bool copy)
const std::string HTMLRenderer::NECK_HTML_FILENAME = "neck.html"; {
const std::string HTMLRenderer::TAIL_HTML_FILENAME = "tail.html"; string fn = get_filename(path);
const std::string HTMLRenderer::CSS_FILENAME = "all.css"; string suffix = get_suffix(fn);
auto iter = EMBED_STRING_MAP.find(make_pair(suffix, param->single_html));
if(iter == EMBED_STRING_MAP.end())
{
cerr << "Warning: unknown suffix in manifest: " << suffix << endl;
return;
}
if(param->single_html)
{
cerr << iter->second.first << endl
<< ifstream(path, ifstream::binary).rdbuf()
<< iter->second.second << endl;
}
else
{
cerr << iter->second.first
<< fn
<< iter->second.second;
}
}
const std::string HTMLRenderer::MANIFEST_FILENAME = "manifest";
}// namespace pdf2htmlEX

4
src/HTMLRenderer/image.cc
View File

@ -10,6 +10,8 @@
#include "HTMLRenderer.h" #include "HTMLRenderer.h"
#include "namespace.h" #include "namespace.h"
namespace pdf2htmlEX {
void HTMLRenderer::drawImage(GfxState * state, Object * ref, Stream * str, int width, int height, GfxImageColorMap * colorMap, GBool interpolate, int *maskColors, GBool inlineImg) void HTMLRenderer::drawImage(GfxState * state, Object * ref, Stream * str, int width, int height, GfxImageColorMap * colorMap, GBool interpolate, int *maskColors, GBool inlineImg)
{ {
return OutputDev::drawImage(state,ref,str,width,height,colorMap,interpolate,maskColors,inlineImg); return OutputDev::drawImage(state,ref,str,width,height,colorMap,interpolate,maskColors,inlineImg);
@ -59,3 +61,5 @@ void HTMLRenderer::drawImage(GfxState * state, Object * ref, Stream * str, int w
++ image_count; ++ image_count;
#endif #endif
} }
} // namespace pdf2htmlEX

4
src/HTMLRenderer/install.cc
View File

@ -17,6 +17,8 @@
#include "namespace.h" #include "namespace.h"
#include "util.h" #include "util.h"
namespace pdf2htmlEX {
using std::abs; using std::abs;
const FontInfo * HTMLRenderer::install_font(GfxFont * font) const FontInfo * HTMLRenderer::install_font(GfxFont * font)
@ -290,3 +292,5 @@ long long HTMLRenderer::install_rise(double rise)
export_rise(new_rise_id, rise); export_rise(new_rise_id, rise);
return new_rise_id; return new_rise_id;
} }
} // namespace pdf2htmlEX

5
src/HTMLRenderer/state.cc
View File

@ -17,6 +17,9 @@
#include "HTMLRenderer.h" #include "HTMLRenderer.h"
#include "namespace.h" #include "namespace.h"
namespace pdf2htmlEX {
using std::max; using std::max;
using std::abs; using std::abs;
@ -358,3 +361,5 @@ void HTMLRenderer::close_line()
line_buf.flush(); line_buf.flush();
} }
} }
} //namespace pdf2htmlEX

4
src/HTMLRenderer/text.cc
View File

@ -19,6 +19,8 @@
#include "HTMLRenderer.h" #include "HTMLRenderer.h"
#include "namespace.h" #include "namespace.h"
namespace pdf2htmlEX {
using std::unordered_set; using std::unordered_set;
using std::min; using std::min;
using std::all_of; using std::all_of;
@ -486,3 +488,5 @@ void HTMLRenderer::drawString(GfxState * state, GooString * s)
draw_tx += dx + dxerr * cur_font_size * hs; draw_tx += dx + dxerr * cur_font_size * hs;
draw_ty += dy; draw_ty += dy;
} }
} // namespace pdf2htmlEX

14
src/include/HTMLRenderer.h
View File

@ -84,7 +84,6 @@ class HTMLRenderer : public OutputDev
virtual void pre_process(); virtual void pre_process();
virtual void post_process(); virtual void post_process();
virtual void process_single_html();
// Start a page. // Start a page.
virtual void startPage(int pageNum, GfxState *state); virtual void startPage(int pageNum, GfxState *state);
@ -169,6 +168,11 @@ class HTMLRenderer : public OutputDev
void export_whitespace(long long ws_id, double ws_width); void export_whitespace(long long ws_id, double ws_width);
void export_rise(long long rise_id, double rise); void export_rise(long long rise_id, double rise);
// depending on single-html, to embed the content or add a link to it
// "copy": indicates whether to copy the file into dest_dir, if not embedded
void embed_file(std::ostream & out, const std::string & path, bool copy);
//////////////////////////////////////////////////// ////////////////////////////////////////////////////
// state tracking // state tracking
//////////////////////////////////////////////////// ////////////////////////////////////////////////////
@ -360,13 +364,11 @@ class HTMLRenderer : public OutputDev
const Param * param; const Param * param;
std::string dest_dir, tmp_dir; std::string dest_dir, tmp_dir;
std::ofstream html_fout, allcss_fout; std::ofstream html_fout, css_fout;
std::string html_path, css_path;
std::set<std::string> tmp_files; std::set<std::string> tmp_files;
static const std::string HEAD_HTML_FILENAME; static const std::string MANIFEST_FILENAME;
static const std::string NECK_HTML_FILENAME;
static const std::string TAIL_HTML_FILENAME;
static const std::string CSS_FILENAME;
}; };
} //namespace pdf2htmlEX } //namespace pdf2htmlEX

2
src/include/Param.h
View File

@ -30,6 +30,7 @@ struct Param
int process_nontext; int process_nontext;
int single_html; int single_html;
int split_pages;
int embed_base_font; int embed_base_font;
int embed_external_font; int embed_external_font;
int decompose_ligature; int decompose_ligature;
@ -41,6 +42,7 @@ struct Param
int tounicode; int tounicode;
int space_as_offset; int space_as_offset;
std::string css_filename;
std::string font_suffix, font_format; std::string font_suffix, font_format;
int debug; int debug;

2
src/include/namespace.h
View File

@ -19,7 +19,5 @@ using std::make_pair;
using std::ifstream; using std::ifstream;
using std::ofstream; using std::ofstream;
using namespace pdf2htmlEX;
#endif // NAMESPACE_H__ #endif // NAMESPACE_H__

4
src/include/util.h
View File

@ -33,6 +33,10 @@ static const double DEFAULT_DPI = 72.0;
extern const std::map<std::string, std::string> BASE_14_FONT_CSS_FONT_MAP; extern const std::map<std::string, std::string> BASE_14_FONT_CSS_FONT_MAP;
extern const std::map<std::string, std::string> GB_ENCODED_FONT_NAME_MAP; extern const std::map<std::string, std::string> GB_ENCODED_FONT_NAME_MAP;
// map to embed files into html
// key: (suffix, if_embed_content)
// value: (prefix string, suffix string)
extern const std::map<std::pair<std::string, bool>, std::pair<std::string, std::string> > EMBED_STRING_MAP;
// mute gcc warning of unused function // mute gcc warning of unused function
namespace namespace

23
src/pdf2htmlEX.cc
View File

@ -64,6 +64,7 @@ void parse_options (int argc, char **argv)
.add("process-nontext", &param.process_nontext, 1, "process nontext objects") .add("process-nontext", &param.process_nontext, 1, "process nontext objects")
.add("single-html", &param.single_html, 1, "combine everything into one single HTML file") .add("single-html", &param.single_html, 1, "combine everything into one single HTML file")
.add("split-pages", &param.split_pages, 0, "split pages into separated files")
.add("embed-base-font", &param.embed_base_font, 0, "embed local matched font for base 14 fonts in the PDF file") .add("embed-base-font", &param.embed_base_font, 0, "embed local matched font for base 14 fonts in the PDF file")
.add("embed-external-font", &param.embed_external_font, 0, "embed local matched font for external fonts in the PDF file") .add("embed-external-font", &param.embed_external_font, 0, "embed local matched font for external fonts in the PDF file")
.add("decompose-ligature", &param.decompose_ligature, 0, "decompose ligatures, for example 'fi' -> 'f''i'") .add("decompose-ligature", &param.decompose_ligature, 0, "decompose ligatures, for example 'fi' -> 'f''i'")
@ -75,6 +76,7 @@ void parse_options (int argc, char **argv)
.add("tounicode", &param.tounicode, 0, "Specify how to deal with ToUnicode map, 0 for auto, 1 for forced, -1 for disabled") .add("tounicode", &param.tounicode, 0, "Specify how to deal with ToUnicode map, 0 for auto, 1 for forced, -1 for disabled")
.add("space-as-offset", &param.space_as_offset, 0, "treat space characters as offsets") .add("space-as-offset", &param.space_as_offset, 0, "treat space characters as offsets")
.add("css-filename", &param.css_filename, "", "Specify the file name of the generated css file")
.add("font-suffix", &param.font_suffix, ".ttf", "suffix for extracted font files") .add("font-suffix", &param.font_suffix, ".ttf", "suffix for extracted font files")
.add("font-format", &param.font_format, "opentype", "format for extracted font files") .add("font-format", &param.font_format, "opentype", "format for extracted font files")
@ -180,11 +182,28 @@ int main(int argc, char **argv)
if(get_suffix(param.input_filename) == ".pdf") if(get_suffix(param.input_filename) == ".pdf")
{ {
param.output_filename = s.substr(0, s.size() - 4) + ".html"; param.output_filename = s.substr(0, s.size() - 4);
if(!param.split_pages)
param.output_filename = s.substr(0, s.size() - 4) + ".html";
} }
else else
{ {
param.output_filename = s + ".html"; if(!param.split_pages)
param.output_filename = s + ".html";
}
}
if(param.css_filename == "")
{
const string s = get_filename(param.input_filename);
if(get_suffix(param.input_filename) == ".pdf")
{
param.output_filename = s.substr(0, s.size() - 4) + ".css";
}
else
{
if(!param.split_pages)
param.output_filename = s + ".css";
} }
} }

7
src/util.cc
View File

@ -46,6 +46,13 @@ const map<string, string> GB_ENCODED_FONT_NAME_MAP({
{"\xC1\xA5\xCA\xE9", "SimLi"}, {"\xC1\xA5\xCA\xE9", "SimLi"},
}); });
const std::map<std::pair<std::string, bool>, std::pair<std::string, std::string> > EMBED_STRING_MAP({
{{".css", 0}, {"<link rel=\"stylesheet\" type=\"text/css\" href=\"", "\"/>"}},
{{".css", 1}, {"<style type=\"text/css\">", "</style>"}},
{{".js", 0}, {"<script type=\"text/javascript\" src=\"", "\"></script>"}},
{{".js", 1}, {"<script type=\"text/javascript\">", "</script>"}}
});
bool isLegalUnicode(Unicode u) bool isLegalUnicode(Unicode u)
{ {
/* /*