PageInfo *analyse_page (PopplerDocument *doc, guint page_num)
{
PopplerPage *page;
PageInfo *info;
GdkPixbuf *image;
double width_points, height_points;
int width, height;
gboolean *white_rows, *white_cols;
page = poppler_document_get_page (doc, page_num);
if (!page) {
g_error ("Couldn't open page %d of document", page_num);
}
/* There are 72 points in an inch. So width and height should be
* multiplied by settings.dpi / 72.0 */
poppler_page_get_size (page, &width_points, &height_points);
width = (int) ((width_points * settings.dpi / 72.0) + 0.5);
height = (int) ((height_points * settings.dpi / 72.0) + 0.5);
image = gdk_pixbuf_new(GDK_COLORSPACE_RGB, TRUE, 8, width, height);
if (!image) {
g_error ("Couldn't create an image (size %d x %d) for page %d",
width, height, page_num);
}
poppler_page_render_to_pixbuf (page, 0, 0, width, height,
settings.dpi / 72.0, 0, image);
g_object_unref (page);
find_white (image, &white_rows, &white_cols);
g_object_unref (image);
guint firstrow, lastrow, hunkscount;
HunkData* hunks = find_hunks (white_rows, height,
&firstrow, &lastrow, &hunkscount);
info = g_new (PageInfo, 1);
info->bbox.x = first_zero (white_cols, width);
info->bbox.width = last_zero (white_cols, width) - info->bbox.x;
if (info->bbox.width <= 0) {
g_error ("Empty page (%d)? Couldn't find a nonwhite column.",
page_num);
}
info->bbox.y = firstrow;
info->bbox.height = lastrow - firstrow;
info->num_hunks = hunkscount;
info->hunks = hunks;
g_free (white_rows);
g_free (white_cols);
return info;
}
int bitfield_get_first_zero(struct bitfield *bf)
{
size_t i;
for (i = 0; i < (bf->max_bits + 7) / 8; i++) {
if (bf->bits[i] != 0xff)
break;
}
if (i == (bf->max_bits + 7) / 8)
return -1;
i = i * 8 + first_zero(bf->bits[i]);
if (i >= bf->max_bits)
return -1;
return i;
}
static HunkData* find_hunks (const gboolean *lines, guint length,
guint* bbstart, guint* bbend, guint* count)
{
g_assert(lines && bbstart && bbend && count);
*count = num_enclosed_white_regions (lines, length) + 1;
guint i, start, hunk_n = 0;
gboolean last = FALSE;
HunkData *hunks = g_new(HunkData, *count);
// Set i to be the first non-white line.
i = first_zero (lines, length);
*bbstart = i;
start = i;
// Walk down the lines, setting HunkData as we go.
for (; i < length; i++) {
if (*(lines + i)) {
if (!last) {
(hunks+hunk_n)->start = start;
(hunks+hunk_n)->end = i;
hunk_n++;
*bbend = i-1;
}
} else {
if (last) {
start = i;
}
}
last = *(lines + i);
}
// If hunk_n == 0 but *count > 0 then *count == 1 and there were no
// white lines at all. In which case, set (hunks+0)->start and end.
if ((hunk_n == 0) && (*count > 0)) {
g_assert (*count == 1);
(hunks+0)->start = 0;
(hunks+0)->end = length;
}
return hunks;
}
