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

Improve rendering accuracy

Browse Source
This commit is contained in:
Lu Wang 2012-08-05 19:39:37 +08:00
parent ab345ac753
commit c7ddf60bac
4 changed files with 169 additions and 93 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
bin/pfa2otf.pe
View File

@ -1,2 +1,2 @@
Open($1); Open($1);
Generate($1:r+".otf"); Generate($1:r+".woff");

220
src/HTMLRenderer.cc
View File

@ -37,6 +37,7 @@
* p - Page * p - Page
* t - Transform * t - Transform
* l - Line * l - Line
* w - White space
* *
* *
* Reusable CSS classes * Reusable CSS classes
@ -61,14 +62,14 @@ const char * HTML_HEAD = "<!DOCTYPE html>\n\
overflow:auto;\ overflow:auto;\
background-color:grey;\ background-color:grey;\
}\ }\
#pdf-main .p {\ #pdf-main > .p {\
position:relative;\ position:relative;\
margin:13px auto;\ margin:13px auto;\
background-color:white;\ background-color:white;\
overflow:hidden;\ overflow:hidden;\
display:none;\ display:none;\
}\ }\
.p .t {\ .p > .t {\
position:absolute;\ position:absolute;\
top:0;\ top:0;\
left:0;\ left:0;\
@ -80,7 +81,7 @@ const char * HTML_HEAD = "<!DOCTYPE html>\n\
-webkit-transform-origin:0% 100%;\ -webkit-transform-origin:0% 100%;\
-o-transform-origin:0% 100%;\ -o-transform-origin:0% 100%;\
}\ }\
.t .l {\ .t > .l {\
position:absolute; \ position:absolute; \
white-space:pre;\ white-space:pre;\
transform-origin:0% 100%;\ transform-origin:0% 100%;\
@ -89,7 +90,7 @@ const char * HTML_HEAD = "<!DOCTYPE html>\n\
-webkit-transform-origin:0% 100%;\ -webkit-transform-origin:0% 100%;\
-o-transform-origin:0% 100%;\ -o-transform-origin:0% 100%;\
}\ }\
.l > span{\ .l > .w{\
display:inline-block;\ display:inline-block;\
}\ }\
::selection{\ ::selection{\
@ -131,7 +132,8 @@ TextString::TextString(GfxState *state)
,x(state->getCurX()), y(state->getCurY()) ,x(state->getCurX()), y(state->getCurY())
,width(0),height(0) ,width(0),height(0)
,state(state) ,state(state)
{ } {
}
TextString::~TextString() TextString::~TextString()
{ {
@ -161,8 +163,6 @@ void TextString::addChar(GfxState *state, double x, double y,
HTMLRenderer::HTMLRenderer(const Param * param) HTMLRenderer::HTMLRenderer(const Param * param)
:cur_string(nullptr), cur_line(nullptr) :cur_string(nullptr), cur_line(nullptr)
,cur_line_x_offset(0)
,cur_fs_id(0), cur_fn_id(0)
,html_fout(param->output_filename.c_str(), ofstream::binary), allcss_fout("all.css") ,html_fout(param->output_filename.c_str(), ofstream::binary), allcss_fout("all.css")
,param(param) ,param(param)
{ {
@ -172,10 +172,6 @@ HTMLRenderer::HTMLRenderer(const Param * param)
html_fout << HTML_HEAD; html_fout << HTML_HEAD;
if(param->readable) html_fout << endl; if(param->readable) html_fout << endl;
for(int i = 0; i < 6; ++i)
ctm[i] = text_mat[i] = 0.0;
ctm[0] = text_mat[0] = ctm[3] = text_mat[3] = 1.0;
} }
HTMLRenderer::~HTMLRenderer() HTMLRenderer::~HTMLRenderer()
@ -197,23 +193,35 @@ void HTMLRenderer::process(PDFDoc *doc)
void HTMLRenderer::startPage(int pageNum, GfxState *state) void HTMLRenderer::startPage(int pageNum, GfxState *state)
{ {
this->pageNum = pageNum; this->pageNum = pageNum;
this->pageWidth=static_cast<int>(state->getPageWidth()); this->pageWidth = state->getPageWidth();
this->pageHeight=static_cast<int>(state->getPageHeight()); this->pageHeight = state->getPageHeight();
assert(cur_line == nullptr); assert(cur_line == nullptr);
assert(cur_string == nullptr);
html_fout << boost::format("<div id=\"page-%3%\" class=\"p\" style=\"width:%1%px;height:%2%px;") % pageWidth % pageHeight % pageNum; html_fout << boost::format("<div id=\"page-%3%\" class=\"p\" style=\"width:%1%px;height:%2%px;") % pageWidth % pageHeight % pageNum;
#if 0
// TODO:background // TODO:background
html_fout << boost::format("background-image:url(p%3%.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%.png);background-position:0 0;background-size:%1%px %2%px;background-repeat:no-repeat;") % pageWidth % pageHeight % pageNum;
#endif
html_fout << "\">"; html_fout << "\">";
if(param->readable) html_fout << endl; if(param->readable) html_fout << endl;
cur_x = cur_y = 0;
cur_fn_id = cur_fs_id = 0;
cur_line_x_offset = 0;
for(int i = 0; i < 6; ++i)
cur_ctm[i] = cur_text_mat[i] = 0.0;
cur_ctm[0] = cur_text_mat[0] = cur_ctm[3] = cur_text_mat[3] = 1.0;
pos_changed = false;
ctm_changed = false;
text_mat_changed = false;
font_changed = false;
// default CTM // default CTM
html_fout << "<div class=\"t\">"; html_fout << boost::format("<div class=\"t t%|1$x|\">") % install_transform_matrix(cur_ctm);
if(param->readable) html_fout << endl; if(param->readable) html_fout << endl;
} }
@ -227,13 +235,6 @@ void HTMLRenderer::endPage() {
if(param->readable) html_fout << endl; if(param->readable) html_fout << endl;
} }
void HTMLRenderer::convert_transform_matrix(double * tm)
{
tm[1] = -tm[1];
tm[2] = -tm[2];
tm[5] = -tm[5];
}
bool HTMLRenderer::at_same_line(const TextString * ts1, const TextString * ts2) const bool HTMLRenderer::at_same_line(const TextString * ts1, const TextString * ts2) const
{ {
if(!(std::abs(ts1->getY() - ts2->getY()) < param->v_eps)) if(!(std::abs(ts1->getY() - ts2->getY()) < param->v_eps))
@ -301,54 +302,47 @@ void HTMLRenderer::outputTextString(TextString * str)
} }
} }
void HTMLRenderer::updateCTM(GfxState * state, double m11, double m12, double m21, double m22, double m31, double m32) void HTMLRenderer::updateAll(GfxState *state)
{ {
double new_ctm[6]; font_changed = true;
memcpy(new_ctm, state->getCTM(), sizeof(new_ctm)); text_mat_changed = true;
convert_transform_matrix(new_ctm); ctm_changed = true;
pos_changed = true;
if(!_tm_equal(ctm, new_ctm))
{
close_cur_line();
memcpy(ctm, new_ctm, sizeof(ctm));
// close old CTM div and create a new one
html_fout << "</div>";
if(param->readable) html_fout << endl;
html_fout << boost::format("<div class=\"t t%|1$x|\">") % install_transform_matrix(ctm);
if(param->readable) html_fout << endl;
}
} }
void HTMLRenderer::updateFont(GfxState *state) { void HTMLRenderer::updateFont(GfxState *state)
long long new_fn_id = install_font(state->getFont()); {
long long new_fs_id = install_font_size(state->getFontSize()); font_changed = true;
if(!((new_fn_id == cur_fn_id) && (new_fs_id == cur_fs_id)))
{
close_cur_line();
cur_fn_id = new_fn_id;
cur_fs_id = new_fs_id;
}
} }
void HTMLRenderer::updateTextMat(GfxState * state) void HTMLRenderer::updateTextMat(GfxState * state)
{ {
double new_text_mat[6]; text_mat_changed = true;
memcpy(new_text_mat, state->getTextMat(), sizeof(new_text_mat)); }
convert_transform_matrix(new_text_mat);
if(!_tm_equal(text_mat, new_text_mat)) void HTMLRenderer::updateCTM(GfxState * state, double m11, double m12, double m21, double m22, double m31, double m32)
{ {
close_cur_line(); ctm_changed = true;
memcpy(text_mat, new_text_mat, sizeof(text_mat)); }
//debug void HTMLRenderer::updateTextPos(GfxState * state)
//TODO: why {
text_mat[4] = text_mat[5] = 0.0; pos_changed = true;
} }
void HTMLRenderer::saveTextPos(GfxState * state)
{
cout << "save" << endl;
}
void HTMLRenderer::restoreTextPos(GfxState * state)
{
cout << "restore" << endl;
} }
void HTMLRenderer::beginString(GfxState *state, GooString *s) { void HTMLRenderer::beginString(GfxState *state, GooString *s) {
check_state_change(state);
// TODO: remove this // TODO: remove this
GfxState * new_state = state->copy(gTrue); GfxState * new_state = state->copy(gTrue);
@ -377,7 +371,7 @@ void HTMLRenderer::endString(GfxState *state) {
double w; double w;
auto wid = install_whitespace(target, w); auto wid = install_whitespace(target, w);
cur_line_x_offset = w-target; cur_line_x_offset = w-target;
html_fout << boost::format("<span class=\"w%|1$x|\"> </span>") % wid; html_fout << boost::format("<span class=\"w w%|1$x|\"> </span>") % wid;
} }
else else
{ {
@ -396,21 +390,23 @@ void HTMLRenderer::endString(GfxState *state) {
close_cur_line(); close_cur_line();
GfxState * cur_state = cur_string -> getState();
// TODO: optimize text matrix search/install // TODO: optimize text matrix search/install
html_fout << boost::format("<div class=\"l f%|1$x| s%|2$x| t%|3$x|\" style=\"") % cur_fn_id % cur_fs_id % install_transform_matrix(text_mat) html_fout << boost::format("<div class=\"l f%|1$x| s%|2$x| t%|3$x|\" style=\"") % cur_fn_id % cur_fs_id % install_transform_matrix(cur_text_mat)
<< "top:" << (pageHeight - cur_string->getY()) << "px;" << "bottom:" << cur_string->getY() << "px;"
<< "left:" << cur_string->getX() << "px;" << "left:" << cur_string->getX() << "px;"
// << "height:" << cur_string->getHeight() << "px;" << "line-height:" << (cur_state->getFont()->getAscent() * cur_state->getFontSize()) << "px;"
; ;
// letter & word spacing // letter & word spacing
GfxState * cur_state = cur_string -> getState();
if(_is_positive(cur_state->getCharSpace())) if(_is_positive(cur_state->getCharSpace()))
html_fout << "letter-spacing:" << cur_state->getCharSpace() << "px;"; html_fout << "letter-spacing:" << cur_state->getCharSpace() << "px;";
if(_is_positive(cur_state->getWordSpace())) if(_is_positive(cur_state->getWordSpace()))
html_fout << "word-spacing:" << cur_state->getWordSpace() << "px;"; html_fout << "word-spacing:" << cur_state->getWordSpace() << "px;";
//debug //debug
//real pos
{ {
html_fout << "\""; html_fout << "\"";
double x,y; double x,y;
@ -425,10 +421,6 @@ void HTMLRenderer::endString(GfxState *state) {
cur_line = cur_string; cur_line = cur_string;
cur_string = nullptr; cur_string = nullptr;
cur_line_x_offset = 0; cur_line_x_offset = 0;
// HERE
//debug
// close_cur_line();
} }
void HTMLRenderer::drawChar(GfxState *state, double x, double y, void HTMLRenderer::drawChar(GfxState *state, double x, double y,
@ -720,7 +712,7 @@ void HTMLRenderer::install_embedded_type1_font (Ref * id, long long fn_id)
tmpf.write(CTM, strlen(CTM)); tmpf.write(CTM, strlen(CTM));
} }
export_remote_font(fn_id, "otf"); export_remote_font(fn_id, "woff");
err: err:
str_obj.streamClose(); str_obj.streamClose();
@ -898,21 +890,91 @@ void HTMLRenderer::export_whitespace (long long ws_id, double ws_width)
void HTMLRenderer::export_transform_matrix (long long tm_id, double * tm) void HTMLRenderer::export_transform_matrix (long long tm_id, double * tm)
{ {
// TODO: recognize common matices
allcss_fout << boost::format(".t%|1$x|{") % tm_id; allcss_fout << boost::format(".t%|1$x|{") % tm_id;
for(const std::string & prefix : {"", "-ms-", "-moz-", "-webkit-", "-o-"})
// TODO: recognize common matices
static const double id_matrix[6] = {1.0, 0.0, 0.0, 1.0, 0.0, 0.0};
if(_tm_equal(tm, id_matrix))
{ {
allcss_fout << prefix << "transform:matrix("; // no need to output anything
for(int i = 0; i < 4; ++i) }
allcss_fout << tm[i] << ','; else
if(prefix == "-moz-") {
allcss_fout << boost::format("%1%px,%2%px);") % tm[4] % tm[5]; for(const std::string & prefix : {"", "-ms-", "-moz-", "-webkit-", "-o-"})
else {
allcss_fout << boost::format("%1%,%2%);") % tm[4] % tm[5]; // PDF use a different coordinate system from Web
allcss_fout << prefix << "transform:matrix("
<< tm[0] << ','
<< -tm[1] << ','
<< -tm[2] << ','
<< tm[3] << ',';
if(prefix == "-moz-")
allcss_fout << boost::format("%1%px,%2%px);") % tm[4] % -tm[5];
else
allcss_fout << boost::format("%1%,%2%);") % tm[4] % -tm[5];
}
} }
allcss_fout << "}"; allcss_fout << "}";
if(param->readable) allcss_fout << endl; if(param->readable) allcss_fout << endl;
} }
void HTMLRenderer::check_state_change(GfxState * state)
{
if(pos_changed)
{
if(!(_equal(state->getCurX(), cur_x) && _equal(state->getCurY(), cur_y)))
{
close_cur_line();
cur_x = state->getCurX();
cur_y = state->getCurY();
}
pos_changed = false;
}
if(font_changed)
{
long long new_fn_id = install_font(state->getFont());
long long new_fs_id = install_font_size(state->getFontSize());
cur_font_size = state->getFontSize();
if(!((new_fn_id == cur_fn_id) && (new_fs_id == cur_fs_id)))
{
close_cur_line();
cur_fn_id = new_fn_id;
cur_fs_id = new_fs_id;
}
font_changed = false;
}
if(text_mat_changed)
{
if(!_tm_equal(cur_text_mat, state->getTextMat(), 4))
{
close_cur_line();
memcpy(cur_text_mat, state->getTextMat(), sizeof(cur_text_mat));
// we've already shift the text to the correct posstion
// so later in css we need to ignore the these offsets
cur_text_mat[4] = cur_text_mat[5] = 0.0;
}
text_mat_changed = false;
}
if(ctm_changed)
{
if(!_tm_equal(cur_ctm, state->getCTM()))
{
close_cur_line();
memcpy(cur_ctm, state->getCTM(), sizeof(cur_ctm));
// close old CTM div and create a new one
html_fout << "</div>";
if(param->readable) html_fout << endl;
html_fout << boost::format("<div class=\"t t%|1$x|\">") % install_transform_matrix(cur_ctm);
if(param->readable) html_fout << endl;
}
ctm_changed = false;
}
}

39
src/HTMLRenderer.h
View File

@ -35,9 +35,9 @@ using namespace std;
static const double EPS = 1e-6; static const double EPS = 1e-6;
inline bool _equal(double x, double y) { return std::abs(x-y) < EPS; } inline bool _equal(double x, double y) { return std::abs(x-y) < EPS; }
inline bool _is_positive(double x) { return x > EPS; } inline bool _is_positive(double x) { return x > EPS; }
inline bool _tm_equal(const double * tm1, const double * tm2) inline bool _tm_equal(const double * tm1, const double * tm2, int size = 6)
{ {
for(int i = 0; i < 6; ++i) for(int i = 0; i < size; ++i)
if(!_equal(tm1[i], tm2[i])) if(!_equal(tm1[i], tm2[i]))
return false; return false;
return true; return true;
@ -55,7 +55,6 @@ class TextString
Unicode u); Unicode u);
double getX() const {return x;} double getX() const {return x;}
double getY() const {return y;} double getY() const {return y;}
double getWidth() const {return width;} double getWidth() const {return width;}
double getHeight() const {return height;} double getHeight() const {return height;}
@ -100,7 +99,6 @@ class HTMLRenderer : public OutputDev
virtual GBool needNonText() { return gFalse; } virtual GBool needNonText() { return gFalse; }
//----- initialization and control //----- initialization and control
virtual GBool checkPageSlice(Page *page, double hDPI, double vDPI, virtual GBool checkPageSlice(Page *page, double hDPI, double vDPI,
int rotate, GBool useMediaBox, GBool crop, int rotate, GBool useMediaBox, GBool crop,
int sliceX, int sliceY, int sliceW, int sliceH, int sliceX, int sliceY, int sliceW, int sliceH,
@ -121,9 +119,13 @@ class HTMLRenderer : public OutputDev
virtual void endPage(); virtual void endPage();
//----- update state //----- update state
virtual void updateCTM(GfxState * state, double m11, double m12, double m21, double m22, double m31, double m32); virtual void updateAll(GfxState * state);
virtual void updateFont(GfxState * state); virtual void updateFont(GfxState * state);
virtual void updateTextMat(GfxState * state); virtual void updateTextMat(GfxState * state);
virtual void updateCTM(GfxState * state, double m11, double m12, double m21, double m22, double m31, double m32);
virtual void updateTextPos(GfxState * state);
virtual void saveTextPos(GfxState * state);
virtual void restoreTextPos(GfxState * state);
//----- text drawing //----- text drawing
virtual void beginString(GfxState *state, GooString *s); virtual void beginString(GfxState *state, GooString *s);
@ -138,9 +140,6 @@ class HTMLRenderer : public OutputDev
private: private:
bool at_same_line(const TextString * ts1, const TextString * ts2) const; bool at_same_line(const TextString * ts1, const TextString * ts2) const;
// CSS use a different coordinate system from PDF
void convert_transform_matrix(double * tm);
void close_cur_line(); void close_cur_line();
void outputTextString(TextString * str); void outputTextString(TextString * str);
@ -176,11 +175,18 @@ class HTMLRenderer : public OutputDev
// page info // page info
int pageNum ; int pageNum ;
int pageWidth ; double pageWidth ;
int pageHeight ; double pageHeight ;
// state maintained when processing pdf // state maintained when processing pdf
void check_state_change(GfxState * state);
// current position
double cur_x, cur_y;
bool pos_changed;
// the string being processed // the string being processed
TextString * cur_string; TextString * cur_string;
// the last word of current line // the last word of current line
@ -188,8 +194,17 @@ class HTMLRenderer : public OutputDev
TextString * cur_line; TextString * cur_line;
// (actual x) - (supposed x) // (actual x) - (supposed x)
double cur_line_x_offset; double cur_line_x_offset;
double ctm[6], text_mat[6];
long long cur_fs_id, cur_fn_id; double cur_ctm[6];
bool ctm_changed;
double cur_text_mat[6];
bool text_mat_changed;
long long cur_fn_id;
double cur_font_size;
long long cur_fs_id;
bool font_changed;
ofstream html_fout, allcss_fout; ofstream html_fout, allcss_fout;

1
src/pdftohtmlEX.cc
View File

@ -124,7 +124,6 @@ po::variables_map parse_options (int argc, char **argv)
("hdpi", po::value<double>(&param.h_dpi)->default_value(72.0), "horizontal DPI") ("hdpi", po::value<double>(&param.h_dpi)->default_value(72.0), "horizontal DPI")
("vdpi", po::value<double>(&param.v_dpi)->default_value(72.0), "vertical DPI") ("vdpi", po::value<double>(&param.v_dpi)->default_value(72.0), "vertical DPI")
("heps", po::value<double>(&param.h_eps)->default_value(1.0), "max tolerated horizontal offset (in pixels)") ("heps", po::value<double>(&param.h_eps)->default_value(1.0), "max tolerated horizontal offset (in pixels)")
("heps", po::value<double>(&param.h_eps)->default_value(1.0), "max tolerated horizontal offset (in pixels)")
("veps", po::value<double>(&param.v_eps)->default_value(1.0), "max tolerated vertical offset (in pixels)") ("veps", po::value<double>(&param.v_eps)->default_value(1.0), "max tolerated vertical offset (in pixels)")
("readable", po::value<int>(&param.readable)->default_value(0), "make the ouptut human readable") ("readable", po::value<int>(&param.readable)->default_value(0), "make the ouptut human readable")
; ;