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

clean code

Browse Source
This commit is contained in:
Lu Wang 2012-08-14 17:50:16 +08:00
parent e050500328
commit 0182ea303b
9 changed files with 57 additions and 47 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

13
src/HTMLRenderer.h
View File

@ -13,6 +13,7 @@
#include <map> #include <map>
#include <vector> #include <vector>
#include <boost/format.hpp>
#include <boost/filesystem/fstream.hpp> #include <boost/filesystem/fstream.hpp>
#include <OutputDev.h> #include <OutputDev.h>
@ -103,7 +104,7 @@ class HTMLRenderer : public OutputDev
* remote font: to be retrieved from the web server * remote font: to be retrieved from the web server
* local font: to be substituted with a local (client side) font * local font: to be substituted with a local (client side) font
*/ */
void export_remote_font(long long fn_id, const std::string & suffix, const std::string & format, GfxFont * font); void export_remote_font(long long fn_id, const std::string & suffix, const std::string & fontfileformat, GfxFont * font);
void export_remote_default_font(long long fn_id); void export_remote_default_font(long long fn_id);
void export_local_font(long long fn_id, GfxFont * font, const std::string & original_font_name, const std::string & cssfont); void export_local_font(long long fn_id, GfxFont * font, const std::string & original_font_name, const std::string & cssfont);
void export_font_size(long long fs_id, double font_size); void export_font_size(long long fs_id, double font_size);
@ -126,7 +127,7 @@ class HTMLRenderer : public OutputDev
XRef * xref; XRef * xref;
// page info // page info
int pageNum ; int pageNum;
double pageWidth ; double pageWidth ;
double pageHeight ; double pageHeight ;
@ -134,11 +135,11 @@ class HTMLRenderer : public OutputDev
//////////////////////////////////////////////////// ////////////////////////////////////////////////////
// states // states
//////////////////////////////////////////////////// ////////////////////////////////////////////////////
bool all_changed;
// if we have a pending opened line // if we have a pending opened line
bool line_opened; bool line_opened;
// any state changed
bool all_changed;
// current position // current position
double cur_tx, cur_ty; // real text position, in text coords double cur_tx, cur_ty; // real text position, in text coords
bool text_pos_changed; bool text_pos_changed;
@ -177,8 +178,6 @@ class HTMLRenderer : public OutputDev
// styles & resources // styles & resources
//////////////////////////////////////////////////// ////////////////////////////////////////////////////
boost::filesystem::ofstream html_fout, allcss_fout, fontscript_fout;
std::unordered_map<long long, FontInfo> font_name_map; std::unordered_map<long long, FontInfo> font_name_map;
std::map<double, long long> font_size_map; std::map<double, long long> font_size_map;
std::map<double, long long> whitespace_map; std::map<double, long long> whitespace_map;
@ -188,6 +187,8 @@ class HTMLRenderer : public OutputDev
int image_count; int image_count;
const Param * param; const Param * param;
boost::filesystem::path dest_dir, tmp_dir;
boost::filesystem::ofstream html_fout, allcss_fout, fontscript_fout;
}; };
#endif /* HTMLRENDERER_H_ */ #endif /* HTMLRENDERER_H_ */

30
src/HTMLRenderer/export.cc
View File

@ -7,13 +7,13 @@
* 2012.08.14 * 2012.08.14
*/ */
#include "HTMLRenderer.h"
#include <boost/algorithm/string.hpp> #include <boost/algorithm/string.hpp>
#include <boost/format.hpp> #include <boost/format.hpp>
using std::string; #include "HTMLRenderer.h"
using std::endl; #include "namespace.h"
using boost::algorithm::ifind_first;
/* /*
* CSS classes * CSS classes
@ -34,9 +34,9 @@ using std::endl;
* *
*/ */
void HTMLRenderer::export_remote_font(long long fn_id, const string & suffix, const string & format, GfxFont * font) void HTMLRenderer::export_remote_font(long long fn_id, const string & suffix, const string & fontfileformat, GfxFont * font)
{ {
allcss_fout << boost::format("@font-face{font-family:f%|1$x|;src:url(f%|1$x|%2%)format(\"%3%\");}.f%|1$x|{font-family:f%|1$x|;") % fn_id % suffix % format; allcss_fout << format("@font-face{font-family:f%|1$x|;src:url(f%|1$x|%2%)format(\"%3%\");}.f%|1$x|{font-family:f%|1$x|;") % fn_id % suffix % fontfileformat;
double a = font->getAscent(); double a = font->getAscent();
double d = font->getDescent(); double d = font->getDescent();
@ -65,18 +65,18 @@ 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 << boost::format(".f%|1$x|{font-family:sans-serif;color:transparent;visibility:hidden;}")%fn_id << endl; allcss_fout << format(".f%|1$x|{font-family:sans-serif;color:transparent;visibility:hidden;}")%fn_id << endl;
} }
void HTMLRenderer::export_local_font(long long fn_id, GfxFont * font, const string & original_font_name, const string & cssfont) void HTMLRenderer::export_local_font(long long fn_id, GfxFont * font, const string & original_font_name, const string & cssfont)
{ {
allcss_fout << boost::format(".f%|1$x|{") % fn_id; allcss_fout << format(".f%|1$x|{") % fn_id;
allcss_fout << "font-family:" << ((cssfont == "") ? (original_font_name + "," + general_font_family(font)) : cssfont) << ";"; allcss_fout << "font-family:" << ((cssfont == "") ? (original_font_name + "," + general_font_family(font)) : cssfont) << ";";
if(font->isBold()) if(font->isBold())
allcss_fout << "font-weight:bold;"; allcss_fout << "font-weight:bold;";
if(boost::algorithm::ifind_first(original_font_name, "oblique")) if(ifind_first(original_font_name, "oblique"))
allcss_fout << "font-style:oblique;"; allcss_fout << "font-style:oblique;";
else if(font->isItalic()) else if(font->isItalic())
allcss_fout << "font-style:italic;"; allcss_fout << "font-style:italic;";
@ -98,17 +98,17 @@ void HTMLRenderer::export_local_font(long long fn_id, GfxFont * font, const stri
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 << boost::format(".s%|1$x|{font-size:%2%px;}") % fs_id % font_size << endl; allcss_fout << format(".s%|1$x|{font-size:%2%px;}") % fs_id % font_size << endl;
} }
void HTMLRenderer::export_whitespace (long long ws_id, double ws_width) void HTMLRenderer::export_whitespace (long long ws_id, double ws_width)
{ {
allcss_fout << boost::format(".w%|1$x|{width:%2%px;}") % ws_id % ws_width << endl; allcss_fout << format(".w%|1$x|{width:%2%px;}") % ws_id % ws_width << 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 << boost::format(".t%|1$x|{") % tm_id; allcss_fout << format(".t%|1$x|{") % tm_id;
// TODO: recognize common matices // TODO: recognize common matices
if(_tm_equal(tm, id_matrix)) if(_tm_equal(tm, id_matrix))
@ -127,9 +127,9 @@ void HTMLRenderer::export_transform_matrix (long long tm_id, const double * tm)
<< tm[3] << ','; << tm[3] << ',';
if(prefix == "-moz-") if(prefix == "-moz-")
allcss_fout << boost::format("%1%px,%2%px);") % tm[4] % -tm[5]; allcss_fout << format("%1%px,%2%px);") % tm[4] % -tm[5];
else else
allcss_fout << boost::format("%1%,%2%);") % tm[4] % -tm[5]; allcss_fout << format("%1%,%2%);") % tm[4] % -tm[5];
} }
} }
allcss_fout << "}" << endl; allcss_fout << "}" << endl;
@ -137,7 +137,7 @@ void HTMLRenderer::export_transform_matrix (long long tm_id, const double * tm)
void HTMLRenderer::export_color(long long color_id, const GfxRGB * rgb) void HTMLRenderer::export_color(long long color_id, const GfxRGB * rgb)
{ {
allcss_fout << boost::format(".c%|1$x|{color:rgb(%2%,%3%,%4%);}") allcss_fout << format(".c%|1$x|{color:rgb(%2%,%3%,%4%);}")
% color_id % (int)colToByte(rgb->r) % (int)colToByte(rgb->g) % (int)colToByte(rgb->b) % color_id % (int)colToByte(rgb->r) % (int)colToByte(rgb->g) % (int)colToByte(rgb->b)
<< endl; << endl;
} }

14
src/HTMLRenderer/general.cc
View File

@ -23,11 +23,13 @@
HTMLRenderer::HTMLRenderer(const Param * param) HTMLRenderer::HTMLRenderer(const Param * param)
:line_opened(false) :line_opened(false)
,html_fout(param->output_filename.c_str(), ofstream::binary)
,allcss_fout("all.css")
,fontscript_fout(TMP_DIR+"/convert.pe")
,image_count(0) ,image_count(0)
,param(param) ,param(param)
,dest_dir(param->dest_dir)
,tmp_dir(TMP_DIR)
,html_fout(dest_dir / param->output_filename, ofstream::binary) // we may output utf8 characters, so use binary
,allcss_fout(dest_dir / "all.css", ofstream::binary)
,fontscript_fout(tmp_dir / "convert.pe", ofstream::binary)
{ {
// install default font & size // install default font & size
install_font(nullptr); install_font(nullptr);
@ -76,7 +78,7 @@ void HTMLRenderer::process(PDFDoc *doc)
doc->displayPage(bg_renderer, i, param->h_dpi2, param->v_dpi2, doc->displayPage(bg_renderer, i, param->h_dpi2, param->v_dpi2,
0, true, false, false, 0, true, false, false,
nullptr, nullptr, nullptr, nullptr); nullptr, nullptr, nullptr, nullptr);
bg_renderer->getBitmap()->writeImgFile(splashFormatPng, (char*)(boost::format("p%|1$x|.png")%i).str().c_str(), param->h_dpi2, param->v_dpi2); bg_renderer->getBitmap()->writeImgFile(splashFormatPng, (char*)(format("p%|1$x|.png")%i).str().c_str(), param->h_dpi2, param->v_dpi2);
cerr << "."; cerr << ".";
cerr.flush(); cerr.flush();
@ -104,9 +106,9 @@ void HTMLRenderer::startPage(int pageNum, GfxState *state)
assert(!line_opened); assert(!line_opened);
html_fout << boost::format("<div id=\"page-%3%\" class=\"p\" style=\"width:%1%px;height:%2%px;") % pageWidth % pageHeight % pageNum; html_fout << format("<div id=\"page-%3%\" class=\"p\" style=\"width:%1%px;height:%2%px;") % pageWidth % pageHeight % pageNum;
html_fout << boost::format("background-image:url(p%|3$x|.png);background-position:0 0;background-size:%1%px %2%px;background-repeat:no-repeat;") % pageWidth % pageHeight % pageNum; html_fout << format("background-image:url(p%|3$x|.png);background-position:0 0;background-size:%1%px %2%px;background-repeat:no-repeat;") % pageWidth % pageHeight % pageNum;
html_fout << "\">" << endl; html_fout << "\">" << endl;

10
src/HTMLRenderer/image.cc
View File

@ -16,12 +16,14 @@ const int *int_p_NULL = nullptr;
#include "HTMLRenderer.h" #include "HTMLRenderer.h"
#include "namespace.h" #include "namespace.h"
using namespace boost::gil;
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)
{ {
if(maskColors) if(maskColors)
return; return;
boost::gil::rgb8_image_t img(width, height); rgb8_image_t img(width, height);
auto imgview = view(img); auto imgview = view(img);
auto loc = imgview.xy_at(0,0); auto loc = imgview.xy_at(0,0);
@ -36,7 +38,7 @@ void HTMLRenderer::drawImage(GfxState * state, Object * ref, Stream * str, int w
GfxRGB rgb; GfxRGB rgb;
colorMap->getRGB(p, &rgb); colorMap->getRGB(p, &rgb);
*loc = boost::gil::rgb8_pixel_t(colToByte(rgb.r), colToByte(rgb.g), colToByte(rgb.b)); *loc = rgb8_pixel_t(colToByte(rgb.r), colToByte(rgb.g), colToByte(rgb.b));
p += colorMap->getNumPixelComps(); p += colorMap->getNumPixelComps();
@ -46,7 +48,7 @@ void HTMLRenderer::drawImage(GfxState * state, Object * ref, Stream * str, int w
loc = imgview.xy_at(0, i+1); loc = imgview.xy_at(0, i+1);
} }
boost::gil::png_write_view((boost::format("i%|1$x|.png")%image_count).str(), imgview); png_write_view((format("i%|1$x|.png")%image_count).str(), imgview);
img_stream->close(); img_stream->close();
delete img_stream; delete img_stream;
@ -55,7 +57,7 @@ void HTMLRenderer::drawImage(GfxState * state, Object * ref, Stream * str, int w
double * ctm = state->getCTM(); double * ctm = state->getCTM();
ctm[4] = ctm[5] = 0.0; ctm[4] = ctm[5] = 0.0;
html_fout << boost::format("<img class=\"i t%2%\" style=\"left:%3%px;bottom:%4%px;width:%5%px;height:%6%px;\" src=\"i%|1$x|.png\" />") % image_count % install_transform_matrix(ctm) % state->getCurX() % state->getCurY() % width % height << endl; html_fout << format("<img class=\"i t%2%\" style=\"left:%3%px;bottom:%4%px;width:%5%px;height:%6%px;\" src=\"i%|1$x|.png\" />") % image_count % install_transform_matrix(ctm) % state->getCurX() % state->getCurY() % width % height << endl;
++ image_count; ++ image_count;

22
src/HTMLRenderer/install.cc
View File

@ -38,7 +38,7 @@ long long HTMLRenderer::install_font(GfxFont * font)
if(param->debug) if(param->debug)
{ {
cerr << "Install font: (" << (font->getID()->num) << ' ' << (font->getID()->gen) << ") -> " << boost::format("f%|1$x|")%new_fn_id << endl; cerr << "Install font: (" << (font->getID()->num) << ' ' << (font->getID()->gen) << ") -> " << format("f%|1$x|")%new_fn_id << endl;
} }
if(font->getType() == fontType3) { if(font->getType() == fontType3) {
@ -117,9 +117,9 @@ void HTMLRenderer::install_embedded_font(GfxFont * font, const string & suffix,
* generate an encoding file and let fontforge handle it. * generate an encoding file and let fontforge handle it.
*/ */
string fn = (boost::format("f%|1$x|") % fn_id).str(); string fn = (format("f%|1$x|") % fn_id).str();
fontscript_fout << boost::format("Open(\"%1%/%2%%3%\",1)") % TMP_DIR % fn % suffix << endl; fontscript_fout << format("Open(\"%1%\", 1)") % (tmp_dir / (fn + suffix)) << endl;
auto ctu = font->getToUnicode(); auto ctu = font->getToUnicode();
int * code2GID = nullptr; int * code2GID = nullptr;
@ -141,7 +141,7 @@ void HTMLRenderer::install_embedded_font(GfxFont * font, const string & suffix,
} }
else else
{ {
fontscript_fout << boost::format("Reencode(\"original\")") << endl; fontscript_fout << format("Reencode(\"original\")") << endl;
int len; int len;
// code2GID has been stored for embedded CID fonts // code2GID has been stored for embedded CID fonts
code2GID = dynamic_cast<GfxCIDFont*>(font)->getCodeToGIDMap(nullptr, &len); code2GID = dynamic_cast<GfxCIDFont*>(font)->getCodeToGIDMap(nullptr, &len);
@ -150,7 +150,7 @@ void HTMLRenderer::install_embedded_font(GfxFont * font, const string & suffix,
if(maxcode > 0) if(maxcode > 0)
{ {
ofstream map_fout((boost::format("%1%/%2%.encoding") % TMP_DIR % fn).str().c_str()); ofstream map_fout(tmp_dir / (fn + ".encoding"));
int cnt = 0; int cnt = 0;
for(int i = 0; i <= maxcode; ++i) for(int i = 0; i <= maxcode; ++i)
{ {
@ -160,24 +160,24 @@ void HTMLRenderer::install_embedded_font(GfxFont * font, const string & suffix,
if(n > 0) if(n > 0)
{ {
++cnt; ++cnt;
map_fout << boost::format("0x%|1$X|") % (code2GID ? code2GID[i] : i); map_fout << format("0x%|1$X|") % (code2GID ? code2GID[i] : i);
for(int j = 0; j < n; ++j) for(int j = 0; j < n; ++j)
map_fout << boost::format(" 0x%|1$X|") % u[j]; map_fout << format(" 0x%|1$X|") % u[j];
map_fout << boost::format(" # 0x%|1$X|") % i << endl; map_fout << format(" # 0x%|1$X|") % i << endl;
} }
} }
if(cnt > 0) if(cnt > 0)
{ {
fontscript_fout << boost::format("LoadEncodingFile(\"%1%/%2%.encoding\", \"%2%\")") % TMP_DIR % fn << endl; fontscript_fout << format("LoadEncodingFile(\"%1%\", \"%2%\")") % (tmp_dir / (fn+".encoding")) % fn << endl;
fontscript_fout << boost::format("Reencode(\"%1%\", 1)") % fn << endl; fontscript_fout << format("Reencode(\"%1%\", 1)") % fn << endl;
} }
} }
ctu->decRefCnt(); ctu->decRefCnt();
} }
fontscript_fout << boost::format("Generate(\"%1%.ttf\")") % fn << endl; fontscript_fout << format("Generate(\"%1%.ttf\")") % (dest_dir / fn) << endl;
export_remote_font(fn_id, ".ttf", "truetype", font); export_remote_font(fn_id, ".ttf", "truetype", font);
} }

1
src/HTMLRenderer/namespace.h
View File

@ -18,6 +18,7 @@ using std::make_pair;
using boost::filesystem::ifstream; using boost::filesystem::ifstream;
using boost::filesystem::ofstream; using boost::filesystem::ofstream;
using boost::filesystem::path; using boost::filesystem::path;
using boost::format;
#endif // NAMESPACE_H__ #endif // NAMESPACE_H__

10
src/HTMLRenderer/text.cc
View File

@ -113,7 +113,7 @@ string HTMLRenderer::dump_embedded_font (GfxFont * font, long long fn_id)
} }
obj.streamReset(); obj.streamReset();
outf.open((boost::format("%1%/f%|2$x|%3%")%TMP_DIR%fn_id%suffix).str().c_str(), ofstream::binary); outf.open(tmp_dir / (format("f%|1$x|%2%")%fn_id%suffix).str(), ofstream::binary);
while((len = obj.streamGetChars(1024, (Guchar*)buf)) > 0) while((len = obj.streamGetChars(1024, (Guchar*)buf)) > 0)
{ {
outf.write(buf, len); outf.write(buf, len);
@ -162,7 +162,7 @@ void HTMLRenderer::drawString(GfxState * state, GooString * s)
{ {
double w; double w;
auto wid = install_whitespace(target, w); auto wid = install_whitespace(target, w);
html_fout << boost::format("<span class=\"w w%|1$x|\"> </span>") % wid; html_fout << format("<span class=\"w w%|1$x|\"> </span>") % wid;
draw_tx += w / draw_scale; draw_tx += w / draw_scale;
} }
} }
@ -179,11 +179,11 @@ void HTMLRenderer::drawString(GfxState * state, GooString * s)
// classes // classes
html_fout << "<div class=\"l " html_fout << "<div class=\"l "
<< boost::format("f%|1$x| s%|2$x| c%|3$x|") % cur_fn_id % cur_fs_id % cur_color_id; << format("f%|1$x| s%|2$x| c%|3$x|") % cur_fn_id % cur_fs_id % cur_color_id;
// "t0" is the id_matrix // "t0" is the id_matrix
if(cur_tm_id != 0) if(cur_tm_id != 0)
html_fout << boost::format(" t%|1$x|") % cur_tm_id; html_fout << format(" t%|1$x|") % cur_tm_id;
{ {
double x,y; // in user space double x,y; // in user space
@ -211,7 +211,7 @@ void HTMLRenderer::drawString(GfxState * state, GooString * s)
html_fout << "\""; html_fout << "\"";
double x,y; double x,y;
state->transform(state->getCurX(), state->getCurY(), &x, &y); state->transform(state->getCurX(), state->getCurY(), &x, &y);
html_fout << boost::format("data-lx=\"%5%\" data-ly=\"%6%\" data-drawscale=\"%4%\" data-x=\"%1%\" data-y=\"%2%\" data-hs=\"%3%") html_fout << format("data-lx=\"%5%\" data-ly=\"%6%\" data-drawscale=\"%4%\" data-x=\"%1%\" data-y=\"%2%\" data-hs=\"%3%")
%x%y%(state->getHorizScaling())%draw_scale%state->getLineX()%state->getLineY(); %x%y%(state->getHorizScaling())%draw_scale%state->getLineX()%state->getLineY();
#endif #endif
} }

3
src/Param.h
View File

@ -15,6 +15,9 @@ struct Param
{ {
std::string owner_password, user_password; std::string owner_password, user_password;
std::string input_filename, output_filename; std::string input_filename, output_filename;
std::string dest_dir;
int first_page, last_page; int first_page, last_page;
double h_dpi, v_dpi; double h_dpi, v_dpi;

1
src/pdf2htmlEX.cc
View File

@ -121,6 +121,7 @@ po::variables_map parse_options (int argc, char **argv)
("metadata,m", "show the document meta data in JSON") ("metadata,m", "show the document meta data in JSON")
("owner-password,o", po::value<string>(&param.owner_password)->default_value(""), "owner password (for encrypted files)") ("owner-password,o", po::value<string>(&param.owner_password)->default_value(""), "owner password (for encrypted files)")
("user-password,u", po::value<string>(&param.user_password)->default_value(""), "user password (for encrypted files)") ("user-password,u", po::value<string>(&param.user_password)->default_value(""), "user password (for encrypted files)")
("dest-dir", po::value<string>(&param.dest_dir)->default_value("."), "destination directory")
("hdpi", po::value<double>(&param.h_dpi)->default_value(72.0), "horizontal DPI for text") ("hdpi", po::value<double>(&param.h_dpi)->default_value(72.0), "horizontal DPI for text")
("vdpi", po::value<double>(&param.v_dpi)->default_value(72.0), "vertical DPI for text") ("vdpi", po::value<double>(&param.v_dpi)->default_value(72.0), "vertical DPI for text")
("hdpi2", po::value<double>(&param.h_dpi2)->default_value(144.0), "horizontal DPI for non-text") ("hdpi2", po::value<double>(&param.h_dpi2)->default_value(144.0), "horizontal DPI for non-text")