DenysMedvid/pdf2htmlEX
1
0
Fork 0
mirror of https://github.com/pdf2htmlEX/pdf2htmlEX.git synced 2024-12-22 04:50:09 +00:00
Code

[experimental], styling change within a line

Browse Source
This commit is contained in:
Lu Wang 2012-08-16 18:11:22 +08:00
parent ad6d6a31a0
commit dc0fd477d8
5 changed files with 139 additions and 113 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

16
src/HTMLRenderer.h
View File

@ -8,7 +8,6 @@
#ifndef HTMLRENDERER_H_
#define HTMLRENDERER_H_
#include <algorithm>
#include <unordered_map>
#include <map>
#include <vector>
@ -164,7 +163,8 @@ class HTMLRenderer : public OutputDev
////////////////////////////////////////////////////
void check_state_change(GfxState * state);
void reset_state_track();
void close_cur_line();
void prepare_line(); // close current span or div if necessary, according to new_line_status
void close_line();
////////////////////////////////////////////////////
@ -182,13 +182,15 @@ class HTMLRenderer : public OutputDev
////////////////////////////////////////////////////
// states
////////////////////////////////////////////////////
// if we have a pending opened line
//line status
//indicating the status for current line & next line
//see comments: meaning for current line || meaning for next line
enum class LineStatus
{
CLOSED,
SPAN,
DIV
} line_status;
NONE, // no line is opened (last <div> is closed) || stay with the same style
SPAN, // there's a pending opening <span> (within a pending opening <div>) || open a new <span> if possible, otherwise a new <div>
DIV // there's a pending opening <div> (but no <span>) || has to open a new <div>
} line_status, new_line_status;
// The order is according to the appearance in check_state_change
// any state changed

6
src/HTMLRenderer/general.cc
View File

@ -19,7 +19,7 @@ using boost::filesystem::remove;
using boost::filesystem::filesystem_error;
HTMLRenderer::HTMLRenderer(const Param * param)
:line_status(LineStatus::CLOSED)
:line_status(LineStatus::NONE)
,image_count(0)
,param(param)
,dest_dir(param->dest_dir)
@ -142,7 +142,7 @@ void HTMLRenderer::startPage(int pageNum, GfxState *state)
this->pageWidth = state->getPageWidth();
this->pageHeight = state->getPageHeight();
assert(line_status == LineStatus::CLOSED);
assert(line_status == LineStatus::NONE);
html_fout << format("<div id=\"p%|1$x|\" class=\"p\" style=\"width:%2%px;height:%3%px;") % pageNum % pageWidth % pageHeight;
@ -186,7 +186,7 @@ void HTMLRenderer::startPage(int pageNum, GfxState *state)
}
void HTMLRenderer::endPage() {
close_cur_line();
close_line();
// close page
html_fout << "</div>" << endl;
}

2
src/HTMLRenderer/image.cc
View File

@ -53,7 +53,7 @@ void HTMLRenderer::drawImage(GfxState * state, Object * ref, Stream * str, int w
img_stream->close();
delete img_stream;
close_cur_line();
close_line();
double * ctm = state->getCTM();
ctm[4] = ctm[5] = 0.0;

154
src/HTMLRenderer/state.cc
View File

@ -7,10 +7,13 @@
* 2012.08.14
*/
#include <algorithm>
#include "HTMLRenderer.h"
#include "namespace.h"
using std::max;
void HTMLRenderer::updateAll(GfxState * state)
{
all_changed = true;
@ -63,10 +66,13 @@ void HTMLRenderer::updateFillColor(GfxState * state)
}
void HTMLRenderer::check_state_change(GfxState * state)
{
//TODO:
// close <span> but not <div>, to use the first style of the line
// DEPENDENCY WARNING
// don't adjust the order of state checking
bool close_line = false;
new_line_status = LineStatus::NONE;
bool need_recheck_position = false;
bool need_rescale_font = false;
@ -97,7 +103,7 @@ void HTMLRenderer::check_state_change(GfxState * state)
// should draw_scale be updated, we'll close the line anyway
if(!(abs((cur_ty + cur_rise) - draw_ty) * draw_scale < param->v_eps))
{
close_line = true;
new_line_status = max(new_line_status, LineStatus::DIV);
}
}
@ -108,7 +114,7 @@ void HTMLRenderer::check_state_change(GfxState * state)
if(!(new_fn_id == cur_fn_id))
{
close_line = true;
new_line_status = max(new_line_status, LineStatus::SPAN);
cur_fn_id = new_fn_id;
}
@ -167,13 +173,13 @@ void HTMLRenderer::check_state_change(GfxState * state)
if(!(_equal(new_draw_font_size, draw_font_size)))
{
close_line = true;
new_line_status = max(new_line_status, LineStatus::SPAN);
draw_font_size = new_draw_font_size;
cur_fs_id = install_font_size(draw_font_size);
}
if(!(_tm_equal(new_draw_ctm, draw_ctm)))
{
close_line = true;
new_line_status = max(new_line_status, LineStatus::DIV);
memcpy(draw_ctm, new_draw_ctm, sizeof(draw_ctm));
cur_tm_id = install_transform_matrix(draw_ctm);
}
@ -186,7 +192,7 @@ void HTMLRenderer::check_state_change(GfxState * state)
double new_letter_space = state->getCharSpace();
if(!_equal(cur_letter_space, new_letter_space))
{
close_line = true;
new_line_status = max(new_line_status, LineStatus::SPAN);
cur_letter_space = new_letter_space;
cur_ls_id = install_letter_space(cur_letter_space * draw_scale);
}
@ -199,14 +205,12 @@ void HTMLRenderer::check_state_change(GfxState * state)
double new_word_space = state->getWordSpace();
if(!_equal(cur_word_space, new_word_space))
{
close_line = true;
new_line_status = max(new_line_status, LineStatus::SPAN);
cur_word_space = new_word_space;
cur_ws_id = install_word_space(cur_word_space * draw_scale);
}
}
// TODO, we may use nested span if only color is changed
// color
if(all_changed || color_changed)
{
@ -214,17 +218,103 @@ void HTMLRenderer::check_state_change(GfxState * state)
state->getFillRGB(&new_color);
if(!((new_color.r == cur_color.r) && (new_color.g == cur_color.g) && (new_color.b == cur_color.b)))
{
close_line = true;
new_line_status = max(new_line_status, LineStatus::SPAN);
cur_color = new_color;
cur_color_id = install_color(&new_color);
}
}
prepare_line();
// TODO: move the following to prepare_line ??
if(line_status == LineStatus::NONE)
{
new_line_status = LineStatus::DIV;
//resync position
draw_ty = cur_ty + cur_rise;
draw_tx = cur_tx;
}
else
{
assert(new_line_status != LineStatus::DIV);
// horizontal position
// try to merge with the last line if possible
double target = (cur_tx - draw_tx) * draw_scale;
if(abs(target) < param->h_eps)
{
// ignore it
}
else
{
// don't close a pending span here, keep the styling
if(target > param->h_eps)
{
double w;
auto wid = install_whitespace(target, w);
html_fout << format("<span class=\"_ _%|1$x|\"> </span>") % wid;
draw_tx += w / draw_scale;
}
else
{
// shift left
// TODO, create a class for this
html_fout << format("<span style=\"margin-left:%1%px\"></span>") % target;
draw_tx += target / draw_scale;
}
}
}
if(new_line_status != LineStatus::NONE)
{
// have to open a new tag
if(new_line_status == LineStatus::SPAN)
{
html_fout << "<span class=\"";
}
else if (new_line_status == LineStatus::DIV)
{
// TODO: recheck descent/ascent
double x,y; // in user space
state->transform(state->getCurX(), state->getCurY(), &x, &y);
html_fout << format("<div style=\"bottom:%1%px;top:%2%px;left:%3%px;\" class=\"l ")
% (y + state->getFont()->getDescent() * draw_font_size)
% (pageHeight - y - state->getFont()->getAscent() * draw_font_size)
% x;
// "t0" is the id_matrix
if(cur_tm_id != 0)
html_fout << format("t%|1$x| ") % cur_tm_id;
}
else
{
assert(false && "Bad value of new_line_status");
}
// TODO: only show changed styles for <span>
html_fout << format("f%|1$x| s%|2$x| c%|3$x|") % cur_fn_id % cur_fs_id % cur_color_id;
if(cur_ls_id != 0)
html_fout << format(" l%|1$x|") % cur_ls_id;
if(cur_ws_id != 0)
html_fout << format(" w%|1$x|") % cur_ws_id;
html_fout << "\">";
line_status = new_line_status;
}
reset_state_track();
if(close_line)
close_cur_line();
}
void HTMLRenderer::reset_state_track()
{
all_changed = false;
@ -242,24 +332,32 @@ void HTMLRenderer::reset_state_track()
color_changed = false;
}
void HTMLRenderer::close_cur_line()
void HTMLRenderer::prepare_line()
{
switch(line_status)
if((line_status == LineStatus::NONE) || (new_line_status == LineStatus::NONE))
return;
if(new_line_status == LineStatus::DIV)
{
case LineStatus::SPAN:
html_fout << "</span>";
// fall through
case LineStatus::DIV:
html_fout << "</div>" << endl;
line_status = LineStatus::CLOSED;
break;
case LineStatus::CLOSED:
default:
break;
close_line();
}
else
{
assert(new_line_status == LineStatus::SPAN);
if(line_status == LineStatus::SPAN)
html_fout << "</span>";
else
assert(line_status == LineStatus::DIV);
// don't change line_status
}
draw_ty = cur_ty + cur_rise;
draw_tx = cur_tx;
}
void HTMLRenderer::close_line()
{
if(line_status == LineStatus::SPAN)
html_fout << "</span>";
else
assert(line_status == LineStatus::DIV);
html_fout << "</div>";
line_status = LineStatus::NONE;
}

74
src/HTMLRenderer/text.cc
View File

@ -155,80 +155,6 @@ void HTMLRenderer::drawString(GfxState * state, GooString * s)
// see if the line has to be closed due to state change
check_state_change(state);
// if the line is still open, try to merge with it
if(line_status != LineStatus::CLOSED)
{
double target = (cur_tx - draw_tx) * draw_scale;
if(abs(target) < param->h_eps)
{
// ignore it
}
else
{
// don't close a pending span here, keep the styling
if(target > param->h_eps)
{
double w;
auto wid = install_whitespace(target, w);
html_fout << format("<span class=\"_ _%|1$x|\"> </span>") % wid;
draw_tx += w / draw_scale;
}
else
{
// shift left
// TODO, create a class for this
html_fout << format("<span style=\"margin-left:%1%px\"></span>") % target;
draw_tx += target / draw_scale;
}
}
}
else
{
// have to open a new line
// classes
html_fout << "<div class=\"l "
<< format("f%|1$x| s%|2$x| c%|3$x|") % cur_fn_id % cur_fs_id % cur_color_id;
// "t0" is the id_matrix
if(cur_tm_id != 0)
html_fout << format(" t%|1$x|") % cur_tm_id;
if(cur_ls_id != 0)
html_fout << format(" l%|1$x|") % cur_ls_id;
if(cur_ws_id != 0)
html_fout << format(" w%|1$x|") % cur_ws_id;
{
double x,y; // in user space
state->transform(state->getCurX(), state->getCurY(), &x, &y);
// TODO: recheck descent/ascent
html_fout << "\" style=\""
<< "bottom:" << (y + state->getFont()->getDescent() * draw_font_size) << "px;"
<< "top:" << (pageHeight - y - state->getFont()->getAscent() * draw_font_size) << "px;"
<< "left:" << x << "px;"
;
}
//debug
{
#if 0
html_fout << "\"";
double x,y;
state->transform(state->getCurX(), state->getCurY(), &x, &y);
html_fout << format("data-lx=\"%5%\" data-ly=\"%6%\" data-draw-x=\"%7%\" data-draw-y=\"%8%\" data-drawscale=\"%4%\" data-x=\"%1%\" data-y=\"%2%\" data-hs=\"%3%")
%x%y%(state->getHorizScaling())%draw_scale%state->getLineX()%state->getLineY()%draw_tx%draw_ty;
#endif
}
html_fout << "\">";
line_status = LineStatus::DIV;
}
// Now ready to output
// get the unicodes