mewin/nana
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity
nana/source/paint/detail/image_processor.hpp

697 lines
22 KiB
C++
Raw Blame History

/*
* Image Processor Algorithm Implementation
* Nana C++ Library(http://www.nanapro.org)
* Copyright(C) 2003-2018 Jinhao(cnjinhao@hotmail.com)
*
* Distributed under the Boost Software License, Version 1.0.
* (See accompanying file LICENSE_1_0.txt or copy at
* http://www.boost.org/LICENSE_1_0.txt)
*
* @file: nana/paint/detail/image_processor.hpp
* @brief: This header file implements the algorithms of image processor
*
* DON'T INCLUDE THIS HEADER FILE DIRECTLY TO YOUR SOURCE FILE.
*/
#ifndef NANA_PAINT_DETAIL_IMAGE_PROCESSOR_HPP
#define NANA_PAINT_DETAIL_IMAGE_PROCESSOR_HPP
#include <nana/paint/pixel_buffer.hpp>
#include <nana/paint/image_process_interface.hpp>
#include <nana/paint/detail/native_paint_interface.hpp>
#include <algorithm>
namespace nana
{
namespace paint
{
namespace detail
{
namespace algorithms
{
///@brief Seek a pixel address by using offset bytes
///@return the specified pixel address
inline pixel_color_t * pixel_at(pixel_color_t * p, std::size_t bytes)
{
return reinterpret_cast<pixel_color_t*>(reinterpret_cast<char*>(p) + bytes);
}
inline const pixel_color_t * pixel_at(const pixel_color_t * p, std::size_t bytes)
{
return reinterpret_cast<const pixel_color_t*>(reinterpret_cast<const char*>(p) + bytes);
}
class proximal_interoplation
: public image_process::stretch_interface
{
void process(const paint::pixel_buffer& s_pixbuf, const nana::rectangle& r_src, paint::pixel_buffer & pixbuf, const nana::rectangle& r_dst) const
{
const auto bytes_per_line = s_pixbuf.bytes_per_line();
double rate_x = double(r_src.width) / r_dst.width;
double rate_y = double(r_src.height) / r_dst.height;
pixel_argb_t * s_raw_pixbuf = s_pixbuf.raw_ptr(0);
if(s_pixbuf.alpha_channel())
{
for(std::size_t row = 0; row < r_dst.height; ++row)
{
const pixel_argb_t * s_line = pixel_at(s_raw_pixbuf, (static_cast<int>(row * rate_y) + r_src.y) * bytes_per_line);
pixel_argb_t * i = pixbuf.raw_ptr(r_dst.y + row);
for(std::size_t x = 0; x < r_dst.width; ++x, ++i)
{
const pixel_argb_t * s = s_line + static_cast<int>(x * rate_x) + r_src.x;
if(0 == s->element.alpha_channel)
continue;
if(s->element.alpha_channel != 255)
{
i->element.red = unsigned(i->element.red * (255 - s->element.alpha_channel) + s->element.red * s->element.alpha_channel) / 255;
i->element.green = unsigned(i->element.green * (255 - s->element.alpha_channel) + s->element.green * s->element.alpha_channel) / 255;
i->element.blue = unsigned(i->element.blue * (255 - s->element.alpha_channel) + s->element.blue * s->element.alpha_channel) / 255;
}
else
{
unsigned alpha_chn = i->element.alpha_channel;
*i = *s;
i->element.alpha_channel = alpha_chn;
}
}
}
}
else
{
for(std::size_t row = 0; row < r_dst.height; ++row)
{
const pixel_argb_t * s_line = pixel_at(s_raw_pixbuf, (static_cast<int>(row * rate_y) + r_src.y) * bytes_per_line);
pixel_argb_t * i = pixbuf.raw_ptr(r_dst.y + row);
for(std::size_t x = 0; x < r_dst.width; ++x, ++i)
*i = s_line[static_cast<int>(x * rate_x) + r_src.x];
}
}
}
};
class bilinear_interoplation
: public image_process::stretch_interface
{
struct x_u_table_tag
{
int x;
int iu;
int iu_minus_coef;
};
void process(const paint::pixel_buffer & s_pixbuf, const nana::rectangle& r_src, paint::pixel_buffer & pixbuf, const nana::rectangle& r_dst) const
{
const auto s_bytes_per_line = s_pixbuf.bytes_per_line();
const int shift_size = 8;
const std::size_t coef = 1 << shift_size;
const int double_shift_size = shift_size << 1;
double rate_x = double(r_src.width) / r_dst.width;
double rate_y = double(r_src.height) / r_dst.height;
const int right_bound = static_cast<int>(r_src.width) - 1 + r_src.x;
const nana::pixel_argb_t * s_raw_pixel_buffer = s_pixbuf.raw_ptr(0);
const int bottom = r_src.y + static_cast<int>(r_src.height - 1);
x_u_table_tag * x_u_table = new x_u_table_tag[r_dst.width];
for(std::size_t x = 0; x < r_dst.width; ++x)
{
double u = (int(x) + 0.5) * rate_x - 0.5;
x_u_table_tag el;
el.x = r_src.x;
if(u < 0)
{
u = 0;
}
else
{
int ipart = static_cast<int>(u);
el.x += ipart;
u -= ipart;
}
el.iu = static_cast<int>(u * coef);
el.iu_minus_coef = coef - el.iu;
x_u_table[x] = el;
}
const bool is_alpha_channel = s_pixbuf.alpha_channel();
for(std::size_t row = 0; row < r_dst.height; ++row)
{
double v = (int(row) + 0.5) * rate_y - 0.5;
int sy = r_src.y;
if(v < 0)
{
v = 0;
}
else
{
int ipart = static_cast<int>(v);
sy += ipart;
v -= ipart;
}
std::size_t iv = static_cast<size_t>(v * coef);
const std::size_t iv_minus_coef = coef - iv;
const nana::pixel_argb_t * s_line = pixel_at(s_raw_pixel_buffer, sy * s_bytes_per_line);
const nana::pixel_argb_t * next_s_line = pixel_at(s_line, (sy < bottom ? s_bytes_per_line : 0));
nana::pixel_argb_t col0;
nana::pixel_argb_t col1;
nana::pixel_argb_t col2;
nana::pixel_argb_t col3;
pixel_argb_t * i = pixbuf.raw_ptr(row + r_dst.y) + r_dst.x;
if(is_alpha_channel)
{
for(std::size_t x = 0; x < r_dst.width; ++x, ++i)
{
x_u_table_tag el = x_u_table[x];
col0 = s_line[el.x];
col1 = next_s_line[el.x];
if(el.x < right_bound)
{
col2 = s_line[el.x + 1];
col3 = next_s_line[el.x + 1];
}
else
{
col2 = col0;
col3 = col1;
}
std::size_t coef0 = el.iu_minus_coef * iv_minus_coef;
std::size_t coef1 = el.iu_minus_coef * iv;
std::size_t coef2 = el.iu * iv_minus_coef;
std::size_t coef3 = el.iu * iv;
unsigned alpha_chn = static_cast<unsigned>((coef0 * col0.element.alpha_channel + coef1 * col1.element.alpha_channel + (coef2 * col2.element.alpha_channel + coef3 * col3.element.alpha_channel)) >> double_shift_size);
unsigned s_red = static_cast<unsigned>((coef0 * col0.element.red + coef1 * col1.element.red + (coef2 * col2.element.red + coef3 * col3.element.red)) >> double_shift_size);
unsigned s_green = static_cast<unsigned>((coef0 * col0.element.green + coef1 * col1.element.green + (coef2 * col2.element.green + coef3 * col3.element.green)) >> double_shift_size);
unsigned s_blue = static_cast<unsigned>((coef0 * col0.element.blue + coef1 * col1.element.blue + (coef2 * col2.element.blue + coef3 * col3.element.blue)) >> double_shift_size);
if(alpha_chn)
{
if(alpha_chn != 255)
{
i->element.red = unsigned(i->element.red * (255 - alpha_chn) + s_red * alpha_chn) / 255;
i->element.green = unsigned(i->element.green * (255 - alpha_chn) + s_green * alpha_chn) / 255;
i->element.blue = unsigned(i->element.blue * (255 - alpha_chn) + s_blue * alpha_chn) / 255;
}
else
{
i->element.red = s_red;
i->element.green = s_green;
i->element.blue = s_blue;
}
}
}
}
else
{
for(std::size_t x = 0; x < r_dst.width; ++x, ++i)
{
x_u_table_tag el = x_u_table[x];
col0 = s_line[el.x];
col1 = next_s_line[el.x];
if(el.x < right_bound)
{
col2 = s_line[el.x + 1];
col3 = next_s_line[el.x + 1];
}
else
{
col2 = col0;
col3 = col1;
}
std::size_t coef0 = el.iu_minus_coef * iv_minus_coef;
std::size_t coef1 = el.iu_minus_coef * iv;
std::size_t coef2 = el.iu * iv_minus_coef;
std::size_t coef3 = el.iu * iv;
i->element.red = static_cast<unsigned char>((coef0 * col0.element.red + coef1 * col1.element.red + (coef2 * col2.element.red + coef3 * col3.element.red)) >> double_shift_size);
i->element.green = static_cast<unsigned char>((coef0 * col0.element.green + coef1 * col1.element.green + (coef2 * col2.element.green + coef3 * col3.element.green)) >> double_shift_size);
i->element.blue = static_cast<unsigned char>((coef0 * col0.element.blue + coef1 * col1.element.blue + (coef2 * col2.element.blue + coef3 * col3.element.blue)) >> double_shift_size);
}
}
}
delete [] x_u_table;
}
};
//alpha_blend
class alpha_blend
: public image_process::alpha_blend_interface
{
//process
virtual void process(const paint::pixel_buffer& s_pixbuf, const nana::rectangle& s_r, paint::pixel_buffer& d_pixbuf, const nana::point& d_pos) const
{
auto d_rgb = d_pixbuf.at(d_pos);
auto s_rgb = s_pixbuf.raw_ptr(s_r.y) + s_r.x;
if(d_rgb && s_rgb)
{
const unsigned rest = s_r.width & 0x3;
const unsigned length_align4 = s_r.width - rest;
std::size_t d_step_bytes = d_pixbuf.bytes_per_line() - (s_r.width - rest) * sizeof(pixel_argb_t);
std::size_t s_step_bytes = s_pixbuf.bytes_per_line() - (s_r.width - rest) * sizeof(pixel_argb_t);
for(unsigned line = 0; line < s_r.height; ++line)
{
const auto end = d_rgb + length_align4;
for(; d_rgb < end; d_rgb += 4, s_rgb += 4)
{
//0
if(s_rgb->element.alpha_channel)
{
if(s_rgb->element.alpha_channel != 255)
{
d_rgb->element.red = unsigned(d_rgb->element.red * (255 - s_rgb[0].element.alpha_channel) + s_rgb[0].element.red * s_rgb[0].element.alpha_channel) / 255;
d_rgb->element.green = unsigned(d_rgb->element.green * (255 - s_rgb[0].element.alpha_channel) + s_rgb[0].element.green * s_rgb[0].element.alpha_channel) / 255;
d_rgb->element.blue = unsigned(d_rgb->element.blue * (255 - s_rgb[0].element.alpha_channel) + s_rgb[0].element.blue * s_rgb[0].element.alpha_channel) / 255;
}
else
*d_rgb = *s_rgb;
}
//1
if(s_rgb[1].element.alpha_channel)
{
if(s_rgb[1].element.alpha_channel != 255)
{
d_rgb[1].element.red = unsigned(d_rgb[1].element.red * (255 - s_rgb[1].element.alpha_channel) + s_rgb[1].element.red * s_rgb[1].element.alpha_channel) / 255;
d_rgb[1].element.green = unsigned(d_rgb[1].element.green * (255 - s_rgb[1].element.alpha_channel) + s_rgb[1].element.green * s_rgb[1].element.alpha_channel) / 255;
d_rgb[1].element.blue = unsigned(d_rgb[1].element.blue * (255 - s_rgb[1].element.alpha_channel) + s_rgb[1].element.blue * s_rgb[1].element.alpha_channel) / 255;
}
else
d_rgb[1] = s_rgb[1];
}
//2
if(s_rgb[2].element.alpha_channel)
{
if(s_rgb[2].element.alpha_channel != 255)
{
d_rgb[2].element.red = unsigned(d_rgb[2].element.red * (255 - s_rgb[2].element.alpha_channel) + s_rgb[2].element.red * s_rgb[2].element.alpha_channel) / 255;
d_rgb[2].element.green = unsigned(d_rgb[2].element.green * (255 - s_rgb[2].element.alpha_channel) + s_rgb[2].element.green * s_rgb[2].element.alpha_channel) / 255;
d_rgb[2].element.blue = unsigned(d_rgb[2].element.blue * (255 - s_rgb[2].element.alpha_channel) + s_rgb[2].element.blue * s_rgb[2].element.alpha_channel) / 255;
}
else
d_rgb[2] = s_rgb[2];
}
//3
if(s_rgb[3].element.alpha_channel)
{
if(s_rgb[3].element.alpha_channel != 255)
{
d_rgb[3].element.red = unsigned(d_rgb[3].element.red * (255 - s_rgb[3].element.alpha_channel) + s_rgb[3].element.red * s_rgb[3].element.alpha_channel) / 255;
d_rgb[3].element.green = unsigned(d_rgb[3].element.green * (255 - s_rgb[3].element.alpha_channel) + s_rgb[3].element.green * s_rgb[3].element.alpha_channel) / 255;
d_rgb[3].element.blue = unsigned(d_rgb[3].element.blue * (255 - s_rgb[3].element.alpha_channel) + s_rgb[3].element.blue * s_rgb[3].element.alpha_channel) / 255;
}
else
d_rgb[3] = s_rgb[3];
}
}
const pixel_argb_t * s_end = s_rgb + rest;
auto rest_d_rgb = d_rgb;
for(auto i = s_rgb; i != s_end; ++i)
{
if(i->element.alpha_channel)
{
if(i->element.alpha_channel != 255)
{
rest_d_rgb->element.red = unsigned(rest_d_rgb->element.red * (255 - i->element.alpha_channel) + i->element.red * i->element.alpha_channel) / 255;
rest_d_rgb->element.green = unsigned(rest_d_rgb->element.green * (255 - i->element.alpha_channel) + i->element.green * i->element.alpha_channel) / 255;
rest_d_rgb->element.blue = unsigned(rest_d_rgb->element.blue * (255 - i->element.alpha_channel) + i->element.blue * i->element.alpha_channel) / 255;
}
else
*rest_d_rgb = *i;
}
++rest_d_rgb;
}
d_rgb = pixel_at(d_rgb, d_step_bytes);
s_rgb = pixel_at(s_rgb, s_step_bytes);
}
}
}
};
//blend
class blend
: public image_process::blend_interface
{
//process
virtual void process(const paint::pixel_buffer& s_pixbuf, const nana::rectangle& s_r, paint::pixel_buffer& d_pixbuf, const nana::point& d_pos, double fade_rate) const
{
auto d_rgb = d_pixbuf.raw_ptr(d_pos.y) + d_pos.x;
auto s_rgb = s_pixbuf.raw_ptr(s_r.y) + s_r.x;
if(d_rgb && s_rgb)
{
auto ptr = detail::alloc_fade_table(fade_rate);//new unsigned char[0x100 * 2];
unsigned char* d_table = ptr.get();
unsigned char* s_table = d_table + 0x100;
const unsigned rest = s_r.width & 0x3;
const unsigned length_align4 = s_r.width - rest;
std::size_t d_step_bytes = d_pixbuf.bytes_per_line() - (s_r.width - rest) * sizeof(pixel_argb_t);
std::size_t s_step_bytes = s_pixbuf.bytes_per_line() - (s_r.width - rest) * sizeof(pixel_argb_t);
for(unsigned line = 0; line < s_r.height; ++line)
{
const auto end = d_rgb + length_align4;
for(; d_rgb < end; d_rgb += 4, s_rgb += 4)
{
//0
d_rgb[0].element.red = unsigned(d_table[d_rgb[0].element.red] + s_table[s_rgb[0].element.red]);
d_rgb[0].element.green = unsigned(d_table[d_rgb[0].element.green] + s_table[s_rgb[0].element.green]);
d_rgb[0].element.blue = unsigned(d_table[d_rgb[0].element.blue] + s_table[s_rgb[0].element.blue]);
//1
d_rgb[1].element.red = unsigned(d_table[d_rgb[1].element.red] + s_table[s_rgb[1].element.red]);
d_rgb[1].element.green = unsigned(d_table[d_rgb[1].element.green] + s_table[s_rgb[1].element.green]);
d_rgb[1].element.blue = unsigned(d_table[d_rgb[1].element.blue] + s_table[s_rgb[1].element.blue]);
//2
d_rgb[2].element.red = unsigned(d_table[d_rgb[2].element.red] + s_table[s_rgb[2].element.red]);
d_rgb[2].element.green = unsigned(d_table[d_rgb[2].element.green] + s_table[s_rgb[2].element.green]);
d_rgb[2].element.blue = unsigned(d_table[d_rgb[2].element.blue] + s_table[s_rgb[2].element.blue]);
//3
d_rgb[3].element.red = unsigned(d_table[d_rgb[3].element.red] + s_table[s_rgb[3].element.red]);
d_rgb[3].element.green = unsigned(d_table[d_rgb[3].element.green] + s_table[s_rgb[3].element.green]);
d_rgb[3].element.blue = unsigned(d_table[d_rgb[3].element.blue] + s_table[s_rgb[3].element.blue]);
}
for(unsigned i = 0; i < rest; ++i)
{
d_rgb[i].element.red = unsigned(d_table[d_rgb[i].element.red] + s_table[s_rgb[i].element.red]);
d_rgb[i].element.green = unsigned(d_table[d_rgb[i].element.green] + s_table[s_rgb[i].element.green]);
d_rgb[i].element.blue = unsigned(d_table[d_rgb[i].element.blue] + s_table[s_rgb[i].element.blue]);
}
d_rgb = pixel_at(d_rgb, d_step_bytes);
s_rgb = pixel_at(s_rgb, s_step_bytes);
}
}
}
};
//class line
class bresenham_line
: public image_process::line_interface
{
virtual void process(paint::pixel_buffer & pixbuf, const nana::point& pos_beg, const nana::point& pos_end, const ::nana::color& clr, double fade_rate) const
{
//Return if it is completely transparent
if (fade_rate <= 0)
return;
auto rgb_color = clr.px_color().value;
const std::size_t bytes_pl = pixbuf.bytes_per_line();
unsigned char * fade_table = nullptr;
std::unique_ptr<unsigned char[]> autoptr;
nana::pixel_argb_t rgb_imd = {};
if(fade_rate < 1)
{
autoptr = detail::alloc_fade_table(1.0 - fade_rate);
fade_table = autoptr.get();
rgb_imd.value = rgb_color;
rgb_imd = detail::fade_color_intermedia(rgb_imd, fade_table);
}
auto i = pixel_at(pixbuf.raw_ptr(0), pos_beg.y * bytes_pl) + pos_beg.x;
auto delta = pos_end - pos_beg;
int step_bytes;
if(delta.y < 0)
{
delta.y = -delta.y;
step_bytes = -static_cast<int>(bytes_pl);
}
else
step_bytes = static_cast<int>(bytes_pl);
if(delta.x == delta.y)
{
step_bytes += sizeof(pixel_argb_t);
++delta.x;
if(fade_table)
{
for(int x = 0; x < delta.x; ++x)
{
*i = detail::fade_color_by_intermedia(*i, rgb_imd, fade_table);
i = pixel_at(i, step_bytes);
}
}
else
{
for(int x = 0; x < delta.x; ++x)
{
i->value = rgb_color;
i = pixel_at(i, step_bytes);
}
}
}
else
{
int dx_2 = delta.x << 1;
int dy_2 = delta.y << 1;
if(delta.x > delta.y)
{
int error = dy_2 - delta.x;
++delta.x; //Include the end poing
if(fade_table)
{
for(int x = 0; x < delta.x; ++x)
{
*i = detail::fade_color_by_intermedia(*i, rgb_imd, fade_table);
if(error >= 0)
{
error -= dx_2;
i = pixel_at(i, step_bytes);
}
error += dy_2;
++i;
}
}
else
{
for(int x = 0; x < delta.x; ++x)
{
i->value = rgb_color;
if(error >= 0)
{
error -= dx_2;
i = pixel_at(i, step_bytes);
}
error += dy_2;
++i;
}
}
}
else
{
int error = dx_2 - delta.y;
++delta.y; //Include the end point
if(fade_table)
{
for (int y = 0; y < delta.y; ++y)
{
*i = detail::fade_color_by_intermedia(*i, rgb_imd, fade_table);
if(error >= 0)
{
error -= dy_2;
++i;
}
error += dx_2;
i = pixel_at(i, step_bytes);
}
}
else
{
for (int y = 0; y < delta.y; ++y)
{
i->value = rgb_color;
if(error >= 0)
{
error -= dy_2;
++i;
}
error += dx_2;
i = pixel_at(i, step_bytes);
}
}
}
}
}
};
class superfast_blur
: public image_process::blur_interface
{
void process(pixel_buffer& pixbuf, const nana::rectangle& area, std::size_t u_radius) const
{
int radius = static_cast<int>(u_radius);
int w = area.width;
int h = area.height;
int wm = w - 1;
int hm = h - 1;
int wh = w * h;
int div = (radius << 1) + 1;
int large_edge = (w > h ? w : h);
const int div_256 = div * 256;
std::unique_ptr<int[]> table_rgb(new int[(wh << 1) + wh + (large_edge << 1) + div_256]);
int * tbl_r = table_rgb.get();
int * tbl_g = tbl_r + wh;
int * tbl_b = tbl_g + wh;
int * vmin = tbl_b + wh;
int * vmax = vmin + large_edge;
int * dv = vmax + large_edge;
const int end_div = div - 1;
for(int i = 0, *dv_block = dv; i < 256; ++i)
{
for(int u = 0; u < end_div; u += 2)
{
dv_block[u] = i;
dv_block[u + 1] = i;
}
dv_block[div - 1] = i;
dv_block += div;
}
auto linepix = pixbuf.raw_ptr(area.y) + area.x;
int yi = 0;
for(int y = 0; y < h; ++y)
{
int sum_r = 0, sum_g = 0, sum_b = 0;
if(radius <= wm)
{
for(int i = - radius; i <= radius; ++i)
{
auto px = linepix[(i > 0 ? i : 0)];
sum_r += px.element.red;
sum_g += px.element.green;
sum_b += px.element.blue;
}
}
else
{
for(int i = - radius; i <= radius; ++i)
{
auto px = linepix[std::min(wm, (i > 0 ? i : 0))];
sum_r += px.element.red;
sum_g += px.element.green;
sum_b += px.element.blue;
}
}
for(int x = 0; x < w; ++x)
{
tbl_r[yi] = dv[sum_r];
tbl_g[yi] = dv[sum_g];
tbl_b[yi] = dv[sum_b];
if(0 == y)
{
vmin[x] = std::min(x + radius + 1, wm);
vmax[x] = std::max(x - radius, 0);
}
auto p1 = linepix[vmin[x]];
auto p2 = linepix[vmax[x]];
sum_r += p1.element.red - p2.element.red;
sum_g += p1.element.green - p2.element.green;
sum_b += p1.element.blue - p2.element.blue;
++yi;
}
linepix = pixbuf.raw_ptr(area.y + y) + area.x;
}
const int yp_init = -radius * w;
const std::size_t bytes_pl = pixbuf.bytes_per_line();
for(int x = 0; x < w; ++x)
{
int sum_r = 0, sum_g = 0, sum_b = 0;
int yp = yp_init;
for(int i = -radius; i <= radius; ++i)
{
if(yp < 1)
{
sum_r += tbl_r[x];
sum_g += tbl_g[x];
sum_b += tbl_b[x];
}
else
{
int yi = yp + x;
sum_r += tbl_r[yi];
sum_g += tbl_g[yi];
sum_b += tbl_b[yi];
}
yp += w;
}
linepix = pixbuf.raw_ptr(area.y) + x + area.x;
for(int y = 0; y < h; ++y)
{
linepix->value = 0xFF000000 | (dv[sum_r] << 16) | (dv[sum_g] << 8) | dv[sum_b];
if(x == 0)
{
vmin[y] = std::min(y + radius + 1, hm) * w;
vmax[y] = std::max(y - radius, 0) * w;
}
int pt1 = x + vmin[y];
int pt2 = x + vmax[y];
sum_r += tbl_r[pt1] - tbl_r[pt2];
sum_g += tbl_g[pt1] - tbl_g[pt2];
sum_b += tbl_b[pt1] - tbl_b[pt2];
linepix = pixel_at(linepix, bytes_pl);
}
}
}
};//end class superfast_blur
}
}
}
}
#endif
Reference in New Issue View Git Blame Copy Permalink