char *
fz_copy_selection(fz_context *ctx, fz_stext_page *page, fz_rect rect)
{
fz_buffer *buffer;
fz_rect hitbox;
int c, i, block_num, seen = 0;
unsigned char *s;
float x0 = rect.x0;
float x1 = rect.x1;
float y0 = rect.y0;
float y1 = rect.y1;
buffer = fz_new_buffer(ctx, 1024);
for (block_num = 0; block_num < page->len; block_num++)
{
fz_stext_block *block;
fz_stext_line *line;
fz_stext_span *span;
if (page->blocks[block_num].type != FZ_PAGE_BLOCK_TEXT)
continue;
block = page->blocks[block_num].u.text;
for (line = block->lines; line < block->lines + block->len; line++)
{
for (span = line->first_span; span; span = span->next)
{
if (seen)
{
fz_write_buffer_byte(ctx, buffer, '\n');
}
seen = 0;
for (i = 0; i < span->len; i++)
{
fz_stext_char_bbox(ctx, &hitbox, span, i);
c = span->text[i].c;
if (c < 32)
c = 0xFFFD;
if (hitbox.x1 >= x0 && hitbox.x0 <= x1 && hitbox.y1 >= y0 && hitbox.y0 <= y1)
{
fz_write_buffer_rune(ctx, buffer, c);
seen = 1;
}
}
seen = (seen && span == line->last_span);
}
}
}
fz_write_buffer_byte(ctx, buffer, 0);
fz_buffer_extract(ctx, buffer, &s);
fz_drop_buffer(ctx, buffer);
return (char*)s;
}
/**
* @brief Get bounding box of a unicode char.
*
* @param index index of the unicode char(start from 0)
*/
QRectF TextBox::charBoundingBox(int index) const
{
fz_rect rect;
fz_stext_char_bbox(d->ctx, &rect, d->text_span, index);
return QRectF(QPointF(rect.x0, rect.y0), QPointF(rect.x1, rect.y1));
}
char *
fz_copy_selection(fz_context *ctx, fz_stext_page *page, fz_rect rect)
{
fz_buffer *buffer;
fz_rect hitbox;
int c, seen = 0;
unsigned char *s;
fz_stext_block *block;
fz_stext_line *line;
fz_stext_char *ch;
float x0 = rect.x0;
float x1 = rect.x1;
float y0 = rect.y0;
float y1 = rect.y1;
buffer = fz_new_buffer(ctx, 1024);
for (block = page->first_block; block; block = block->next)
{
if (block->type != FZ_STEXT_BLOCK_TEXT)
continue;
for (line = block->u.t.first_line; line; line = line->next)
{
if (seen)
{
fz_append_byte(ctx, buffer, '\n');
}
seen = 0;
for (ch = line->first_char; ch; ch = ch->next)
{
fz_stext_char_bbox(ctx, &hitbox, line, ch);
c = ch->c;
if (c < 32)
c = FZ_REPLACEMENT_CHARACTER;
if (hitbox.x1 >= x0 && hitbox.x0 <= x1 && hitbox.y1 >= y0 && hitbox.y0 <= y1)
{
fz_append_rune(ctx, buffer, c);
seen = 1;
}
}
seen = (seen && line == block->u.t.last_line);
}
}
fz_terminate_buffer(ctx, buffer);
fz_buffer_extract(ctx, buffer, &s); /* take over the data */
fz_drop_buffer(ctx, buffer);
return (char*)s;
}
int
fz_highlight_selection(fz_context *ctx, fz_stext_page *page, fz_rect rect, fz_rect *hit_bbox, int hit_max)
{
fz_rect linebox, charbox;
fz_stext_block *block;
fz_stext_line *line;
fz_stext_span *span;
int i, block_num, hit_count;
float x0 = rect.x0;
float x1 = rect.x1;
float y0 = rect.y0;
float y1 = rect.y1;
hit_count = 0;
for (block_num = 0; block_num < page->len; block_num++)
{
if (page->blocks[block_num].type != FZ_PAGE_BLOCK_TEXT)
continue;
block = page->blocks[block_num].u.text;
for (line = block->lines; line < block->lines + block->len; line++)
{
linebox = fz_empty_rect;
for (span = line->first_span; span; span = span->next)
{
for (i = 0; i < span->len; i++)
{
fz_stext_char_bbox(ctx, &charbox, span, i);
if (charbox.x1 >= x0 && charbox.x0 <= x1 && charbox.y1 >= y0 && charbox.y0 <= y1)
{
if (charbox.y0 != linebox.y0 || fz_abs(charbox.x0 - linebox.x1) > 5)
{
if (!fz_is_empty_rect(&linebox) && hit_count < hit_max)
hit_bbox[hit_count++] = linebox;
linebox = charbox;
}
else
{
fz_union_rect(&linebox, &charbox);
}
}
}
}
if (!fz_is_empty_rect(&linebox) && hit_count < hit_max)
hit_bbox[hit_count++] = linebox;
}
}
return hit_count;
}
int
fz_highlight_selection(fz_context *ctx, fz_stext_page *page, fz_rect rect, fz_rect *hit_bbox, int hit_max)
{
fz_rect linebox, charbox;
fz_stext_block *block;
fz_stext_line *line;
fz_stext_char *ch;
int hit_count;
float x0 = rect.x0;
float x1 = rect.x1;
float y0 = rect.y0;
float y1 = rect.y1;
hit_count = 0;
for (block = page->first_block; block; block = block->next)
{
if (block->type != FZ_STEXT_BLOCK_TEXT)
continue;
for (line = block->u.t.first_line; line; line = line->next)
{
linebox = fz_empty_rect;
for (ch = line->first_char; ch; ch = ch->next)
{
fz_stext_char_bbox(ctx, &charbox, line, ch);
if (charbox.x1 >= x0 && charbox.x0 <= x1 && charbox.y1 >= y0 && charbox.y0 <= y1)
{
if (charbox.y0 != linebox.y0 || fz_abs(charbox.x0 - linebox.x1) > 5)
{
if (!fz_is_empty_rect(&linebox) && hit_count < hit_max)
hit_bbox[hit_count++] = linebox;
linebox = charbox;
}
else
{
fz_union_rect(&linebox, &charbox);
}
}
}
if (!fz_is_empty_rect(&linebox) && hit_count < hit_max)
hit_bbox[hit_count++] = linebox;
}
}
return hit_count;
}
fz_char_and_box *fz_stext_char_at(fz_context *ctx, fz_char_and_box *cab, fz_stext_page *page, int idx)
{
int block_num;
int ofs = 0;
for (block_num = 0; block_num < page->len; block_num++)
{
fz_stext_block *block;
fz_stext_line *line;
fz_stext_span *span;
if (page->blocks[block_num].type != FZ_PAGE_BLOCK_TEXT)
continue;
block = page->blocks[block_num].u.text;
for (line = block->lines; line < block->lines + block->len; line++)
{
for (span = line->first_span; span; span = span->next)
{
if (idx < ofs + span->len)
{
cab->c = span->text[idx - ofs].c;
fz_stext_char_bbox(ctx, &cab->bbox, span, idx - ofs);
return cab;
}
ofs += span->len;
}
/* pseudo-newline */
if (idx == ofs)
{
cab->bbox = fz_empty_rect;
cab->c = ' ';
return cab;
}
ofs++;
}
}
cab->bbox = fz_empty_rect;
cab->c = 0;
return cab;
}
fz_char_and_box *fz_stext_char_at(fz_context *ctx, fz_char_and_box *cab, fz_stext_page *page, int idx)
{
fz_stext_block *block;
fz_stext_line *line;
fz_stext_char *ch;
int ofs = 0;
for (block = page->first_block; block; block = block->next)
{
if (block->type != FZ_STEXT_BLOCK_TEXT)
continue;
for (line = block->u.t.first_line; line; line = line->next)
{
for (ch = line->first_char; ch; ch = ch->next)
{
if (ofs == idx)
{
cab->c = ch->c;
fz_stext_char_bbox(ctx, &cab->bbox, line, ch);
return cab;
}
++ofs;
}
/* pseudo-newline */
if (idx == ofs)
{
cab->bbox = fz_empty_rect;
cab->c = ' ';
return cab;
}
++ofs;
}
}
cab->bbox = fz_empty_rect;
cab->c = 0;
return cab;
}
fz_buffer *
fz_new_buffer_from_stext_page(fz_context *ctx, fz_stext_page *text, const fz_rect *sel, int crlf)
{
fz_buffer *buf;
fz_rect hitbox;
float x0, y0, x1, y1;
int block_num;
int need_newline;
int i;
need_newline = 0;
if (fz_is_infinite_rect(sel))
{
x0 = y0 = -FLT_MAX;
x1 = y1 = FLT_MAX;
}
else
{
x0 = sel->x0;
y0 = sel->y0;
x1 = sel->x1;
y1 = sel->y1;
}
buf = fz_new_buffer(ctx, 256);
fz_try(ctx)
{
for (block_num = 0; block_num < text->len; block_num++)
{
fz_stext_line *line;
fz_stext_block *block;
fz_stext_span *span;
if (text->blocks[block_num].type != FZ_PAGE_BLOCK_TEXT)
continue;
block = text->blocks[block_num].u.text;
for (line = block->lines; line < block->lines + block->len; line++)
{
int saw_text = 0;
for (span = line->first_span; span; span = span->next)
{
for (i = 0; i < span->len; i++)
{
int c;
fz_stext_char_bbox(ctx, &hitbox, span, i);
c = span->text[i].c;
if (c < 32)
c = '?';
if (hitbox.x1 >= x0 && hitbox.x0 <= x1 && hitbox.y1 >= y0 && hitbox.y0 <= y1)
{
saw_text = 1;
if (need_newline)
{
if (crlf)
fz_write_buffer_rune(ctx, buf, '\r');
fz_write_buffer_rune(ctx, buf, '\n');
need_newline = 0;
}
fz_write_buffer_rune(ctx, buf, c);
}
}
}
if (saw_text)
need_newline = 1;
}
}
}
fz_catch(ctx)
{
fz_drop_buffer(ctx, buf);
fz_rethrow(ctx);
}
return buf;
}
void
fz_print_stext_page_as_xml(fz_context *ctx, fz_output *out, fz_stext_page *page)
{
fz_stext_block *block;
fz_stext_line *line;
fz_stext_char *ch;
fz_write_printf(ctx, out, "<page width=\"%g\" height=\"%g\">\n",
page->mediabox.x1 - page->mediabox.x0,
page->mediabox.y1 - page->mediabox.y0);
for (block = page->first_block; block; block = block->next)
{
switch (block->type)
{
case FZ_STEXT_BLOCK_TEXT:
fz_write_printf(ctx, out, "<block bbox=\"%g %g %g %g\">\n",
block->bbox.x0, block->bbox.y0, block->bbox.x1, block->bbox.y1);
for (line = block->u.t.first_line; line; line = line->next)
{
fz_font *font = NULL;
float size = 0;
const char *name = NULL;
fz_rect rect;
fz_write_printf(ctx, out, "<line bbox=\"%g %g %g %g\" wmode=\"%d\" dir=\"%g %g\">\n",
line->bbox.x0, line->bbox.y0, line->bbox.x1, line->bbox.y1,
line->wmode,
line->dir.x, line->dir.y);
for (ch = line->first_char; ch; ch = ch->next)
{
if (ch->font != font || ch->size != size)
{
if (font)
fz_write_string(ctx, out, "</font>\n");
font = ch->font;
size = ch->size;
name = font_full_name(ctx, font);
fz_write_printf(ctx, out, "<font name=\"%s\" size=\"%g\">\n", name, size);
}
fz_stext_char_bbox(ctx, &rect, line, ch);
fz_write_printf(ctx, out, "<char bbox=\"%g %g %g %g\" x=\"%g\" y=\"%g\" c=\"",
rect.x0, rect.y0, rect.x1, rect.y1, ch->origin.x, ch->origin.y);
switch (ch->c)
{
case '<': fz_write_string(ctx, out, "<"); break;
case '>': fz_write_string(ctx, out, ">"); break;
case '&': fz_write_string(ctx, out, "&"); break;
case '"': fz_write_string(ctx, out, """); break;
case '\'': fz_write_string(ctx, out, "'"); break;
default:
if (ch->c >= 32 && ch->c <= 127)
fz_write_printf(ctx, out, "%c", ch->c);
else
fz_write_printf(ctx, out, "&#x%x;", ch->c);
break;
}
fz_write_string(ctx, out, "\"/>\n");
}
if (font)
fz_write_string(ctx, out, "</font>\n");
fz_write_string(ctx, out, "</line>\n");
}
fz_write_string(ctx, out, "</block>\n");
break;
case FZ_STEXT_BLOCK_IMAGE:
fz_write_printf(ctx, out, "<image bbox=\"%g %g %g %g\" />\n",
block->bbox.x0, block->bbox.y0, block->bbox.x1, block->bbox.y1);
break;
}
}
fz_write_string(ctx, out, "</page>\n");
}
void
fz_print_stext_page_xml(fz_context *ctx, fz_output *out, fz_stext_page *page)
{
int block_n;
fz_printf(ctx, out, "<page width=\"%g\" height=\"%g\">\n",
page->mediabox.x1 - page->mediabox.x0,
page->mediabox.y1 - page->mediabox.y0);
for (block_n = 0; block_n < page->len; block_n++)
{
switch (page->blocks[block_n].type)
{
case FZ_PAGE_BLOCK_TEXT:
{
fz_stext_block *block = page->blocks[block_n].u.text;
fz_stext_line *line;
const char *s;
fz_printf(ctx, out, "<block bbox=\"%g %g %g %g\">\n",
block->bbox.x0, block->bbox.y0, block->bbox.x1, block->bbox.y1);
for (line = block->lines; line < block->lines + block->len; line++)
{
fz_stext_span *span;
fz_printf(ctx, out, "<line bbox=\"%g %g %g %g\">\n",
line->bbox.x0, line->bbox.y0, line->bbox.x1, line->bbox.y1);
for (span = line->first_span; span; span = span->next)
{
fz_stext_style *style = NULL;
const char *name = NULL;
int char_num;
for (char_num = 0; char_num < span->len; char_num++)
{
fz_stext_char *ch = &span->text[char_num];
if (ch->style != style)
{
if (style)
{
fz_printf(ctx, out, "</span>\n");
}
style = ch->style;
name = fz_font_name(ctx, style->font);
s = strchr(name, '+');
s = s ? s + 1 : name;
fz_printf(ctx, out, "<span bbox=\"%g %g %g %g\" font=\"%s\" size=\"%g\">\n",
span->bbox.x0, span->bbox.y0, span->bbox.x1, span->bbox.y1,
s, style->size);
}
{
fz_rect rect;
fz_stext_char_bbox(ctx, &rect, span, char_num);
fz_printf(ctx, out, "<char bbox=\"%g %g %g %g\" x=\"%g\" y=\"%g\" c=\"",
rect.x0, rect.y0, rect.x1, rect.y1, ch->p.x, ch->p.y);
}
switch (ch->c)
{
case '<': fz_printf(ctx, out, "<"); break;
case '>': fz_printf(ctx, out, ">"); break;
case '&': fz_printf(ctx, out, "&"); break;
case '"': fz_printf(ctx, out, """); break;
case '\'': fz_printf(ctx, out, "'"); break;
default:
if (ch->c >= 32 && ch->c <= 127)
fz_printf(ctx, out, "%c", ch->c);
else
fz_printf(ctx, out, "&#x%x;", ch->c);
break;
}
fz_printf(ctx, out, "\"/>\n");
}
if (style)
fz_printf(ctx, out, "</span>\n");
}
fz_printf(ctx, out, "</line>\n");
}
fz_printf(ctx, out, "</block>\n");
break;
}
case FZ_PAGE_BLOCK_IMAGE:
{
break;
}
}
}
fz_printf(ctx, out, "</page>\n");
}
