DenysMedvid/pdf2htmlEX
1
0
Fork 0
mirror of https://github.com/pdf2htmlEX/pdf2htmlEX.git synced 2024-06-30 23:49:00 +00:00
Code

separate TextLineBuffer

Browse Source
This commit is contained in:
Lu Wang 2012-12-11 20:48:01 +08:00
parent add1d1e28e
commit 5842d2279d
6 changed files with 31 additions and 89 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
CMakeLists.txt
View File

@ -151,7 +151,8 @@ add_executable(pdf2htmlEX
src/HTMLRenderer/general.cc
src/HTMLRenderer/image.cc
src/HTMLRenderer/install.cc
src/HTMLRenderer/LineBuffer.cc
src/HTMLRenderer/TextLineBuffer.h
src/HTMLRenderer/TextLineBuffer.cc
src/HTMLRenderer/link.cc
src/HTMLRenderer/state.cc
src/HTMLRenderer/text.cc

70
src/HTMLRenderer/HTMLRenderer.h
View File

@ -379,73 +379,9 @@ class HTMLRenderer : public OutputDev
double draw_tx, draw_ty;
// some metrics have to be determined after all elements in the lines have been seen
class LineBuffer {
public:
LineBuffer (HTMLRenderer * renderer) : renderer(renderer) { }
class State {
public:
void begin(std::ostream & out, const State * prev_state);
void end(std::ostream & out) const;
void hash(void);
int diff(const State & s) const;
enum {
FONT_ID,
FONT_SIZE_ID,
COLOR_ID,
LETTER_SPACE_ID,
WORD_SPACE_ID,
RISE_ID,
ID_COUNT
};
long long ids[ID_COUNT];
double ascent;
double descent;
double draw_font_size;
size_t start_idx; // index of the first Text using this state
// for optimzation
long long hash_value;
bool need_close;
static const char * format_str; // class names for each id
};
class Offset {
public:
size_t start_idx; // should put this idx before text[start_idx];
double width;
};
void reset(GfxState * state);
void append_unicodes(const Unicode * u, int l);
void append_offset(double width);
void append_state(void);
void flush(void);
private:
// retrieve state from renderer
void set_state(State & state);
HTMLRenderer * renderer;
double x, y;
long long tm_id;
std::vector<State> states;
std::vector<Offset> offsets;
std::vector<Unicode> text;
// for flush
std::vector<State*> stack;
} line_buf;
friend class LineBuffer;
class TextLineBuffer;
friend class TextLineBuffer;
TextLineBuffer * text_line_buf;
// for font reencoding
int32_t * cur_mapping;

25
src/HTMLRenderer/LineBuffer.cc → src/HTMLRenderer/TextLineBuffer.cc
View File

@ -1,5 +1,5 @@
/*
* LineBuffer.cc
* TextLineBuffer.cc
*
* Generate and optimized HTML for one line
*
@ -10,6 +10,7 @@
#include <vector>
#include "HTMLRenderer.h"
#include "TextLineBuffer.h"
#include "util/namespace.h"
#include "util/unicode.h"
#include "util/math.h"
@ -23,18 +24,18 @@ using std::ostream;
using std::cerr;
using std::endl;
void HTMLRenderer::LineBuffer::reset(GfxState * state)
void HTMLRenderer::TextLineBuffer::reset(GfxState * state)
{
state->transform(state->getCurX(), state->getCurY(), &x, &y);
tm_id = renderer->cur_ttm_id;
}
void HTMLRenderer::LineBuffer::append_unicodes(const Unicode * u, int l)
void HTMLRenderer::TextLineBuffer::append_unicodes(const Unicode * u, int l)
{
text.insert(text.end(), u, u+l);
}
void HTMLRenderer::LineBuffer::append_offset(double width)
void HTMLRenderer::TextLineBuffer::append_offset(double width)
{
if((!offsets.empty()) && (offsets.back().start_idx == text.size()))
offsets.back().width += width;
@ -42,7 +43,7 @@ void HTMLRenderer::LineBuffer::append_offset(double width)
offsets.push_back(Offset({text.size(), width}));
}
void HTMLRenderer::LineBuffer::append_state(void)
void HTMLRenderer::TextLineBuffer::append_state(void)
{
if(states.empty() || (states.back().start_idx != text.size()))
{
@ -53,7 +54,7 @@ void HTMLRenderer::LineBuffer::append_state(void)
set_state(states.back());
}
void HTMLRenderer::LineBuffer::flush(void)
void HTMLRenderer::TextLineBuffer::flush(void)
{
/*
* Each Line is an independent absolute positioined block
@ -181,7 +182,7 @@ void HTMLRenderer::LineBuffer::flush(void)
}
void HTMLRenderer::LineBuffer::set_state (State & state)
void HTMLRenderer::TextLineBuffer::set_state (State & state)
{
state.ids[State::FONT_ID] = renderer->cur_font_info->id;
state.ids[State::FONT_SIZE_ID] = renderer->cur_fs_id;
@ -196,7 +197,7 @@ void HTMLRenderer::LineBuffer::set_state (State & state)
state.draw_font_size = renderer->draw_font_size;
}
void HTMLRenderer::LineBuffer::State::begin (ostream & out, const State * prev_state)
void HTMLRenderer::TextLineBuffer::State::begin (ostream & out, const State * prev_state)
{
bool first = true;
for(int i = 0; i < ID_COUNT; ++i)
@ -229,13 +230,13 @@ void HTMLRenderer::LineBuffer::State::begin (ostream & out, const State * prev_s
}
}
void HTMLRenderer::LineBuffer::State::end(ostream & out) const
void HTMLRenderer::TextLineBuffer::State::end(ostream & out) const
{
if(need_close)
out << "</span>";
}
void HTMLRenderer::LineBuffer::State::hash(void)
void HTMLRenderer::TextLineBuffer::State::hash(void)
{
hash_value = 0;
for(int i = 0; i < ID_COUNT; ++i)
@ -244,7 +245,7 @@ void HTMLRenderer::LineBuffer::State::hash(void)
}
}
int HTMLRenderer::LineBuffer::State::diff(const State & s) const
int HTMLRenderer::TextLineBuffer::State::diff(const State & s) const
{
/*
* A quick check based on hash_value
@ -260,5 +261,5 @@ int HTMLRenderer::LineBuffer::State::diff(const State & s) const
return d;
}
const char * HTMLRenderer::LineBuffer::State::format_str = "fsclwr";
const char * HTMLRenderer::TextLineBuffer::State::format_str = "fsclwr";
} //namespace pdf2htmlEX

4
src/HTMLRenderer/general.cc
View File

@ -14,6 +14,7 @@
#include <vector>
#include "HTMLRenderer.h"
#include "TextLineBuffer.h"
#include "pdf2htmlEX-config.h"
#include "BackgroundRenderer/BackgroundRenderer.h"
#include "util/namespace.h"
@ -41,7 +42,7 @@ static void dummy(void *, enum ErrorCategory, int pos, char *)
HTMLRenderer::HTMLRenderer(const Param * param)
:OutputDev()
,line_opened(false)
,line_buf(this)
,text_line_buf(new TextLineBuffer(this))
,preprocessor(param)
,tmp_files(*param)
,param(param)
@ -60,6 +61,7 @@ HTMLRenderer::HTMLRenderer(const Param * param)
HTMLRenderer::~HTMLRenderer()
{
delete text_line_buf;
ffw_finalize();
delete [] cur_mapping;
delete [] cur_mapping2;

9
src/HTMLRenderer/state.cc
View File

@ -16,6 +16,7 @@
#include <algorithm>
#include "HTMLRenderer.h"
#include "TextLineBuffer.h"
#include "util/namespace.h"
#include "util/math.h"
@ -342,7 +343,7 @@ void HTMLRenderer::prepare_text_line(GfxState * state)
{
close_text_line();
line_buf.reset(state);
text_line_buf->reset(state);
//resync position
draw_ty = cur_ty;
@ -359,14 +360,14 @@ void HTMLRenderer::prepare_text_line(GfxState * state)
}
else
{
line_buf.append_offset(target);
text_line_buf->append_offset(target);
draw_tx += target / draw_text_scale;
}
}
if(new_line_state != NLS_NONE)
{
line_buf.append_state();
text_line_buf->append_state();
}
line_opened = true;
@ -377,7 +378,7 @@ void HTMLRenderer::close_text_line()
if(line_opened)
{
line_opened = false;
line_buf.flush();
text_line_buf->flush();
}
}

9
src/HTMLRenderer/text.cc
View File

@ -16,6 +16,7 @@
#include <fofi/FoFiTrueType.h>
#include "HTMLRenderer.h"
#include "TextLineBuffer.h"
#include "util/ffw.h"
#include "util/namespace.h"
#include "util/unicode.h"
@ -555,25 +556,25 @@ void HTMLRenderer::drawString(GfxState * state, GooString * s)
if(is_space && (param->space_as_offset))
{
// ignore horiz_scaling, as it's merged in CTM
line_buf.append_offset((dx1 * cur_font_size + cur_letter_space + cur_word_space) * draw_text_scale);
text_line_buf->append_offset((dx1 * cur_font_size + cur_letter_space + cur_word_space) * draw_text_scale);
}
else
{
if((param->decompose_ligature) && (uLen > 1) && all_of(u, u+uLen, isLegalUnicode))
{
line_buf.append_unicodes(u, uLen);
text_line_buf->append_unicodes(u, uLen);
}
else
{
if(cur_font_info->use_tounicode)
{
Unicode uu = check_unicode(u, uLen, code, font);
line_buf.append_unicodes(&uu, 1);
text_line_buf->append_unicodes(&uu, 1);
}
else
{
Unicode uu = unicode_from_font(code, font);
line_buf.append_unicodes(&uu, 1);
text_line_buf->append_unicodes(&uu, 1);
}
}
}