static pdf_obj*
pdf_get_page_obj (pdf_file *pf, int page_no,
pdf_obj **ret_bbox, pdf_obj **ret_resources)
{
pdf_obj *page_tree;
pdf_obj *bbox = NULL, *resources = NULL, *rotate = NULL;
int page_idx;
/*
* Get Page Tree.
*/
page_tree = NULL;
{
pdf_obj *trailer, *catalog;
pdf_obj *markinfo, *tmp;
trailer = pdf_file_get_trailer(pf);
if (pdf_lookup_dict(trailer, "Encrypt")) {
WARN("This PDF document is encrypted.");
pdf_release_obj(trailer);
return NULL;
}
catalog = pdf_deref_obj(pdf_lookup_dict(trailer, "Root"));
if (!PDF_OBJ_DICTTYPE(catalog)) {
WARN("Can't read document catalog.");
pdf_release_obj(trailer);
if (catalog)
pdf_release_obj(catalog);
return NULL;
}
pdf_release_obj(trailer);
markinfo = pdf_deref_obj(pdf_lookup_dict(catalog, "MarkInfo"));
if (markinfo) {
tmp = pdf_lookup_dict(markinfo, "Marked");
if (PDF_OBJ_BOOLEANTYPE(tmp) && pdf_boolean_value(tmp))
WARN("PDF file is tagged... Ignoring tags.");
pdf_release_obj(markinfo);
}
page_tree = pdf_deref_obj(pdf_lookup_dict(catalog, "Pages"));
pdf_release_obj(catalog);
}
if (!page_tree) {
WARN("Page tree not found.");
return NULL;
}
/*
* Negative page numbers are counted from the back.
*/
{
int count = pdf_number_value(pdf_lookup_dict(page_tree, "Count"));
page_idx = page_no + (page_no >= 0 ? -1 : count);
if (page_idx < 0 || page_idx >= count) {
WARN("Page %ld does not exist.", page_no);
pdf_release_obj(page_tree);
return NULL;
}
page_no = page_idx+1;
}
/*
* Seek correct page. Get Media/Crop Box.
* Media box and resources can be inherited.
*/
{
pdf_obj *kids_ref, *kids;
pdf_obj *crop_box = NULL;
pdf_obj *tmp;
tmp = pdf_lookup_dict(page_tree, "Resources");
resources = tmp ? pdf_deref_obj(tmp) : pdf_new_dict();
while (1) {
int kids_length, i;
if ((tmp = pdf_deref_obj(pdf_lookup_dict(page_tree, "MediaBox")))) {
if (bbox)
pdf_release_obj(bbox);
bbox = tmp;
}
if ((tmp = pdf_deref_obj(pdf_lookup_dict(page_tree, "BleedBox")))) {
if (!rect_equal(tmp, bbox)) {
if (bbox)
pdf_release_obj(bbox);
bbox = tmp;
} else {
pdf_release_obj(tmp);
}
}
if ((tmp = pdf_deref_obj(pdf_lookup_dict(page_tree, "TrimBox")))) {
if (!rect_equal(tmp, bbox)) {
if (bbox)
pdf_release_obj(bbox);
bbox = tmp;
} else {
pdf_release_obj(tmp);
}
}
if ((tmp = pdf_deref_obj(pdf_lookup_dict(page_tree, "ArtBox")))) {
if (!rect_equal(tmp, bbox)) {
if (bbox)
pdf_release_obj(bbox);
bbox = tmp;
} else {
pdf_release_obj(tmp);
}
}
if ((tmp = pdf_deref_obj(pdf_lookup_dict(page_tree, "CropBox")))) {
if (crop_box)
pdf_release_obj(crop_box);
crop_box = tmp;
}
if ((tmp = pdf_deref_obj(pdf_lookup_dict(page_tree, "Rotate")))) {
if (rotate)
pdf_release_obj(rotate);
rotate = tmp;
}
if ((tmp = pdf_deref_obj(pdf_lookup_dict(page_tree, "Resources")))) {
#if 0
pdf_merge_dict(tmp, resources);
#endif
if (resources)
pdf_release_obj(resources);
resources = tmp;
}
kids_ref = pdf_lookup_dict(page_tree, "Kids");
if (!kids_ref)
break;
kids = pdf_deref_obj(kids_ref);
kids_length = pdf_array_length(kids);
for (i = 0; i < kids_length; i++) {
int count;
pdf_release_obj(page_tree);
page_tree = pdf_deref_obj(pdf_get_array(kids, i));
tmp = pdf_deref_obj(pdf_lookup_dict(page_tree, "Count"));
if (tmp) {
/* Pages object */
count = pdf_number_value(tmp);
pdf_release_obj(tmp);
} else {
/* Page object */
count = 1;
}
if (page_idx < count)
break;
page_idx -= count;
}
pdf_release_obj(kids);
if (i == kids_length) {
WARN("Page %ld not found! Broken PDF file?", page_no);
if (bbox)
pdf_release_obj(bbox);
if (crop_box)
pdf_release_obj(crop_box);
if (rotate)
pdf_release_obj(rotate);
pdf_release_obj(resources);
pdf_release_obj(page_tree);
return NULL;
}
}
if (crop_box) {
pdf_release_obj(bbox);
bbox = crop_box;
}
}
if (!bbox) {
WARN("No BoundingBox information available.");
pdf_release_obj(page_tree);
pdf_release_obj(resources);
if (rotate)
pdf_release_obj(rotate);
return NULL;
}
if (rotate) {
if (pdf_number_value(rotate) != 0.0)
WARN("<< /Rotate %d >> found. (Not supported yet)",
(int)pdf_number_value(rotate));
pdf_release_obj(rotate);
rotate = NULL;
}
if (ret_bbox != NULL)
*ret_bbox = bbox;
if (ret_resources != NULL)
*ret_resources = resources;
return page_tree;
}
pdf_obj *pdf_include_page(FILE *image_file, struct xform_info *p, char *res_name)
{
pdf_obj *trailer = NULL, *catalog = NULL, *page_tree = NULL;
pdf_obj *kids_ref, *kids;
pdf_obj *media_box = NULL, *crop_box = NULL, *resources = NULL,
*contents = NULL, *contents_ref = NULL;
pdf_obj *tmp1;
#ifdef MEM_DEBUG
MEM_START
#endif
if (!(trailer = pdf_open (image_file))) {
fprintf (stderr, "\nCorrupt PDF file?\n");
}
/* Now just lookup catalog location */
/* Deref catalog */
if (trailer && (catalog = pdf_deref_obj(pdf_lookup_dict (trailer,"Root"))) ==
NULL) {
fprintf (stderr, "\nCatalog isn't where I expect it.\n");
}
if (trailer)
pdf_release_obj (trailer);
/* Lookup page tree in catalog */
if (catalog) {
page_tree = pdf_deref_obj (pdf_lookup_dict (catalog, "Pages"));
/* Should be finished with catalog */
pdf_release_obj (catalog);
}
/* Media box and resources can be inherited so start looking for
them here */
if (page_tree) {
if ((tmp1 = pdf_lookup_dict (page_tree, "CropBox")))
crop_box = pdf_deref_obj (tmp1);
if ((tmp1 = pdf_lookup_dict (page_tree, "MediaBox")))
media_box = pdf_deref_obj (tmp1);
resources = pdf_deref_obj (pdf_lookup_dict (page_tree, "Resources"));
if (resources == NULL) {
resources = pdf_new_dict();
}
while ((kids_ref = pdf_lookup_dict (page_tree, "Kids")) != NULL) {
kids = pdf_deref_obj (kids_ref);
pdf_release_obj (page_tree);
page_tree = pdf_deref_obj (pdf_get_array(kids, 0));
pdf_release_obj (kids);
/* Replace MediaBox if it's here */
tmp1 = pdf_deref_obj(pdf_lookup_dict (page_tree, "MediaBox"));
if (tmp1 && media_box)
pdf_release_obj (media_box);
if (tmp1)
media_box = tmp1;
/* Do same for CropBox */
tmp1 = pdf_deref_obj(pdf_lookup_dict (page_tree, "CropBox"));
if (tmp1 && crop_box)
pdf_release_obj (crop_box);
if (tmp1)
crop_box = tmp1;
/* Add resources if they're here */
tmp1 = pdf_deref_obj (pdf_lookup_dict (page_tree, "Resources"));
if (tmp1) {
pdf_merge_dict (tmp1, resources);
pdf_release_obj (resources);
resources = tmp1;
}
}
/* At this point, page_tree contains the first page. media_box,
crop_box, and resources should also be set. */
/* If there's a crop_box, replace media_box with crop_box.
The rest of this routine assumes crop_box has been released */
if (crop_box) {
pdf_release_obj (media_box);
media_box = crop_box;
crop_box = NULL;
}
/* This gets bit confusing. In the following code,
media_box is the box the image is cropped to.
The bounding box is the box the image is scaled to */
/* If user did not supply bounding box, use media_box
(which may really be cropbox) as bounding box */
/* Set the crop box parameters in the xform_info structure */
p->c_llx = pdf_number_value (pdf_get_array (media_box, 0));
p->c_lly = pdf_number_value (pdf_get_array (media_box, 1));
p->c_urx = pdf_number_value (pdf_get_array (media_box, 2));
p->c_ury = pdf_number_value (pdf_get_array (media_box, 3));
/* Adjust scaling and clipping information as necessary */
pdf_scale_image (p);
/* Set the media box to whatever pdf_scale_image() decided
for the crop box (which may be unchanged) */
pdf_release_obj (media_box);
media_box = pdf_new_array ();
pdf_add_array (media_box, pdf_new_number (p->c_llx));
pdf_add_array (media_box, pdf_new_number (p->c_lly));
pdf_add_array (media_box, pdf_new_number (p->c_urx));
pdf_add_array (media_box, pdf_new_number (p->c_ury));
if ((contents =
pdf_deref_obj(pdf_lookup_dict(page_tree,"Contents")))==NULL) {
fprintf (stderr, "\nNo Contents found\n");
return NULL;
}
pdf_release_obj (page_tree);
}
/* Arrays of contents must be handled very differently (not implemented) */
if (contents && contents -> type != PDF_ARRAY) {
doc_make_form_xobj (contents, media_box,
p->user_bbox? p->u_llx: 0.0,
p->user_bbox? p->u_lly: 0.0,
1.0, 1.0,
resources, res_name);
} else {
fprintf (stderr, "\nIgnoring stream with with multiple segments\n");
contents = NULL;
}
if (contents) {
contents_ref = pdf_ref_obj (contents);
pdf_release_obj (contents);
}
pdf_close ();
#ifdef MEM_DEBUG
MEM_END
#endif
return (contents_ref);
}
