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

Merge branch 'devv'

Browse Source
This commit is contained in:
Lu Wang 2012-08-13 19:13:31 +08:00
commit c3c80260cb
3 changed files with 63 additions and 43 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
README.md
View File

@ -31,7 +31,7 @@ Dependency
---------------------------- ----------------------------
* libpoppler with xpdf header >= 0.20.2 * libpoppler with xpdf header >= 0.20.2
* boost c++ library (format, program options, gil) * boost c++ library (format, program options, gil)
* fontforge **Please use [my modified version](http://github.com/coolwanglu/fontforge) before the bugs are fixed** * fontforge **Please use [the lastest version](https://github.com/fontforge/fontforge)**
HOW TO COMPILE HOW TO COMPILE
---------------------------- ----------------------------

15
src/Consts.cc
View File

@ -30,8 +30,8 @@ const std::string HTML_HEAD = "<!DOCTYPE html>\n\
overflow:hidden;\ overflow:hidden;\
display:none;\ display:none;\
}\ }\
.p > .l {\ .p > .l {\
position:absolute; \ position:absolute;\
white-space:pre;\ white-space:pre;\
}\ }\
.l > .w {\ .l > .w {\
@ -44,6 +44,9 @@ const std::string HTML_HEAD = "<!DOCTYPE html>\n\
::-moz-selection{\ ::-moz-selection{\
background: rgba(168,209,255,0.5);\ background: rgba(168,209,255,0.5);\
}\ }\
.p > .i {\
position:absolute;\
}\
</style><link rel=\"stylesheet\" type=\"text/css\" href=\"all.css\" />\ </style><link rel=\"stylesheet\" type=\"text/css\" href=\"all.css\" />\
<script type=\"text/javascript\">\ <script type=\"text/javascript\">\
function show_pages()\ function show_pages()\
@ -51,10 +54,10 @@ function show_pages()\
var pages = document.getElementById('pdf-main').childNodes;\ var pages = document.getElementById('pdf-main').childNodes;\
var idx = 0;\ var idx = 0;\
var f = function(){\ var f = function(){\
if (idx < pages.length) {\ if (idx < pages.length) {\
try{\ try{\
pages[idx].style.display='block';\ pages[idx].style.display='block';\
}catch(e){}\ }catch(e){}\
++idx;\ ++idx;\
setTimeout(f,100);\ setTimeout(f,100);\
}\ }\

89
src/HTMLRenderer.cc
View File

@ -39,6 +39,7 @@ const int *int_p_NULL = nullptr;
* p - Page * p - Page
* l - Line * l - Line
* w - White space * w - White space
* i - Image
* *
* *
* Reusable CSS classes * Reusable CSS classes
@ -69,14 +70,12 @@ HTMLRenderer::HTMLRenderer(const Param * param)
black.r = black.g = black.b = 0; black.r = black.g = black.b = 0;
install_color(&black); install_color(&black);
html_fout << HTML_HEAD; html_fout << HTML_HEAD << endl;
html_fout << endl;
} }
HTMLRenderer::~HTMLRenderer() HTMLRenderer::~HTMLRenderer()
{ {
html_fout << HTML_TAIL; html_fout << HTML_TAIL << endl;
html_fout << endl;
} }
void HTMLRenderer::process(PDFDoc *doc) void HTMLRenderer::process(PDFDoc *doc)
@ -133,8 +132,7 @@ void HTMLRenderer::startPage(int pageNum, GfxState *state)
html_fout << boost::format("background-image:url(p%|3$x|.png);background-position:0 0;background-size:%1%px %2%px;background-repeat:no-repeat;") % pageWidth % pageHeight % pageNum; html_fout << boost::format("background-image:url(p%|3$x|.png);background-position:0 0;background-size:%1%px %2%px;background-repeat:no-repeat;") % pageWidth % pageHeight % pageNum;
html_fout << "\">"; html_fout << "\">" << endl;
html_fout << endl;
cur_fn_id = cur_fs_id = cur_tm_id = cur_color_id = 0; cur_fn_id = cur_fs_id = cur_tm_id = cur_color_id = 0;
cur_tx = cur_ty = 0; cur_tx = cur_ty = 0;
@ -154,17 +152,14 @@ void HTMLRenderer::startPage(int pageNum, GfxState *state)
void HTMLRenderer::endPage() { void HTMLRenderer::endPage() {
close_cur_line(); close_cur_line();
// close page // close page
html_fout << "</div>"; html_fout << "</div>" << endl;
html_fout << endl;
} }
void HTMLRenderer::close_cur_line() void HTMLRenderer::close_cur_line()
{ {
if(line_opened) if(line_opened)
{ {
html_fout << "</div>"; html_fout << "</div>" << endl;
html_fout << endl;
line_opened = false; line_opened = false;
} }
} }
@ -376,7 +371,10 @@ void HTMLRenderer::drawString(GfxState * state, GooString * s)
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)
{ {
boost::gil::rgb16_image_t img(width, height); if(maskColors)
return;
boost::gil::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);
@ -391,7 +389,7 @@ 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::rgb16_pixel_t(rgb.r, rgb.g, rgb.b); *loc = boost::gil::rgb8_pixel_t(colToByte(rgb.r), colToByte(rgb.g), colToByte(rgb.b));
p += colorMap->getNumPixelComps(); p += colorMap->getNumPixelComps();
@ -402,10 +400,17 @@ void HTMLRenderer::drawImage(GfxState * state, Object * ref, Stream * str, int w
} }
boost::gil::png_write_view((boost::format("i%|1$x|.png")%image_count).str(), imgview); boost::gil::png_write_view((boost::format("i%|1$x|.png")%image_count).str(), imgview);
img_stream->close(); img_stream->close();
delete img_stream; delete img_stream;
close_cur_line();
double * ctm = state->getCTM();
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$x|.png\" />") % image_count % install_transform_matrix(ctm) % state->getCurX() % state->getCurY() % width % height << endl;
++ image_count; ++ image_count;
} }
@ -608,6 +613,25 @@ err:
void HTMLRenderer::install_embedded_font(GfxFont * font, const std::string & suffix, long long fn_id) void HTMLRenderer::install_embedded_font(GfxFont * font, const std::string & suffix, long long fn_id)
{ {
// TODO Should use standard way to handle CID fonts // TODO Should use standard way to handle CID fonts
/*
* How it works:
*
* 1.dump the font file directly from the font descriptor and put the glyphs into the correct slots
*
* for nonCID
* nothing need to do
*
* for CID + nonTrueType
* Flatten the font
*
* for CID Truetype
* Just use glyph order, and later we'll map GID (instead of char code) to Unicode
*
*
* 2. map charcode (or GID for CID truetype) to Unicode
*
* generate an encoding file and let fontforge handle it.
*/
std::string fn = (boost::format("f%|1$x|") % fn_id).str(); std::string fn = (boost::format("f%|1$x|") % fn_id).str();
@ -618,13 +642,11 @@ void HTMLRenderer::install_embedded_font(GfxFont * font, const std::string & suf
if(ctu) if(ctu)
{ {
// TODO: ctu could be CID2Unicode for CID fonts // TODO: ctu could be CID2Unicode for CID fonts
int maxcode = 0; int maxcode = 0;
if(!font->isCIDFont()) if(!font->isCIDFont())
{ {
maxcode = 0xff; maxcode = 0xff;
//TODO read code2GID for TrueType
} }
else else
{ {
@ -777,16 +799,13 @@ void HTMLRenderer::export_remote_font(long long fn_id, const string & suffix, co
allcss_fout << "line-height:" << (a-d) << ";"; allcss_fout << "line-height:" << (a-d) << ";";
allcss_fout << "}"; allcss_fout << "}" << endl;
allcss_fout << endl;
} }
// TODO: this function is called when some font is unable to process, may use the name there as a hint // TODO: this function is called when some font is unable to process, may use the name there as a hint
void HTMLRenderer::export_remote_default_font(long long fn_id) void HTMLRenderer::export_remote_default_font(long long fn_id)
{ {
allcss_fout << boost::format(".f%|1$x|{font-family:sans-serif;color:transparent;visibility:hidden;}")%fn_id; allcss_fout << boost::format(".f%|1$x|{font-family:sans-serif;color:transparent;visibility:hidden;}")%fn_id << endl;
allcss_fout << endl;
} }
void HTMLRenderer::export_local_font(long long fn_id, GfxFont * font, const string & original_font_name, const string & cssfont) void HTMLRenderer::export_local_font(long long fn_id, GfxFont * font, const string & original_font_name, const string & cssfont)
@ -813,9 +832,7 @@ void HTMLRenderer::export_local_font(long long fn_id, GfxFont * font, const stri
allcss_fout << "line-height:" << (a-d) << ";"; allcss_fout << "line-height:" << (a-d) << ";";
allcss_fout << "}"; allcss_fout << "}" << endl;
allcss_fout << endl;
} }
std::string HTMLRenderer::general_font_family(GfxFont * font) std::string HTMLRenderer::general_font_family(GfxFont * font)
@ -830,14 +847,12 @@ std::string HTMLRenderer::general_font_family(GfxFont * font)
void HTMLRenderer::export_font_size (long long fs_id, double font_size) void HTMLRenderer::export_font_size (long long fs_id, double font_size)
{ {
allcss_fout << boost::format(".s%|1$x|{font-size:%2%px;}") % fs_id % font_size; allcss_fout << boost::format(".s%|1$x|{font-size:%2%px;}") % fs_id % font_size << endl;
allcss_fout << endl;
} }
void HTMLRenderer::export_whitespace (long long ws_id, double ws_width) void HTMLRenderer::export_whitespace (long long ws_id, double ws_width)
{ {
allcss_fout << boost::format(".w%|1$x|{width:%2%px;}") % ws_id % ws_width; allcss_fout << boost::format(".w%|1$x|{width:%2%px;}") % ws_id % ws_width << endl;
allcss_fout << endl;
} }
void HTMLRenderer::export_transform_matrix (long long tm_id, const double * tm) void HTMLRenderer::export_transform_matrix (long long tm_id, const double * tm)
@ -866,15 +881,14 @@ void HTMLRenderer::export_transform_matrix (long long tm_id, const double * tm)
allcss_fout << boost::format("%1%,%2%);") % tm[4] % -tm[5]; allcss_fout << boost::format("%1%,%2%);") % tm[4] % -tm[5];
} }
} }
allcss_fout << "}"; allcss_fout << "}" << endl;
allcss_fout << endl;
} }
void HTMLRenderer::export_color(long long color_id, const GfxRGB * rgb) void HTMLRenderer::export_color(long long color_id, const GfxRGB * rgb)
{ {
allcss_fout << boost::format(".c%|1$x|{color:rgb(%2%,%3%,%4%);}") allcss_fout << boost::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)
allcss_fout << endl; << endl;
} }
void HTMLRenderer::check_state_change(GfxState * state) void HTMLRenderer::check_state_change(GfxState * state)
@ -944,14 +958,17 @@ void HTMLRenderer::check_state_change(GfxState * state)
if(need_rescale_font) if(need_rescale_font)
{ {
draw_scale = std::sqrt(cur_ctm[2] * cur_ctm[2] + cur_ctm[3] * cur_ctm[3]); double new_draw_ctm[6];
memcpy(new_draw_ctm, cur_ctm, sizeof(new_draw_ctm));
draw_scale = std::sqrt(new_draw_ctm[2] * new_draw_ctm[2] + new_draw_ctm[3] * new_draw_ctm[3]);
double new_draw_font_size = cur_font_size; double new_draw_font_size = cur_font_size;
if(_is_positive(draw_scale)) if(_is_positive(draw_scale))
{ {
new_draw_font_size *= draw_scale; new_draw_font_size *= draw_scale;
for(int i = 0; i < 4; ++i) for(int i = 0; i < 4; ++i)
cur_ctm[i] /= draw_scale; new_draw_ctm[i] /= draw_scale;
} }
else else
{ {
@ -964,9 +981,9 @@ void HTMLRenderer::check_state_change(GfxState * state)
cur_fs_id = install_font_size(draw_font_size); cur_fs_id = install_font_size(draw_font_size);
close_line = true; close_line = true;
} }
if(!(_tm_equal(cur_ctm, draw_ctm))) if(!(_tm_equal(new_draw_ctm, draw_ctm)))
{ {
memcpy(draw_ctm, cur_ctm, sizeof(draw_ctm)); memcpy(draw_ctm, new_draw_ctm, sizeof(draw_ctm));
cur_tm_id = install_transform_matrix(draw_ctm); cur_tm_id = install_transform_matrix(draw_ctm);
close_line = true; close_line = true;
} }