/* ximage here is the result. DONT USE IT FOR PASSING OPTIONS! */
int
pdf_include_page (pdf_ximage *ximage,
FILE *image_file,
const char *ident,
load_options options)
{
pdf_file *pf;
xform_info info;
pdf_obj *contents = NULL, *catalog;
pdf_obj *page = NULL, *resources = NULL, *markinfo = NULL;
pf = pdf_open(ident, image_file);
if (!pf)
return -1;
if (pdf_file_get_version(pf) > pdf_get_version()) {
WARN("Trying to include PDF file which has newer version number " \
"than output PDF: 1.%d.", pdf_get_version());
}
pdf_ximage_init_form_info(&info);
if (options.page_no == 0)
options.page_no = 1;
page = pdf_doc_get_page(pf,
options.page_no, options.bbox_type,
&info.bbox, &resources);
if(!page)
goto error_silent;
catalog = pdf_file_get_catalog(pf);
markinfo = pdf_deref_obj(pdf_lookup_dict(catalog, "MarkInfo"));
if (markinfo) {
pdf_obj *tmp = pdf_deref_obj(pdf_lookup_dict(markinfo, "Marked"));
pdf_release_obj(markinfo);
if (!PDF_OBJ_BOOLEANTYPE(tmp)) {
if (tmp)
pdf_release_obj(tmp);
goto error;
} else if (pdf_boolean_value(tmp)) {
WARN("PDF file is tagged... Ignoring tags.");
}
pdf_release_obj(tmp);
}
contents = pdf_deref_obj(pdf_lookup_dict(page, "Contents"));
pdf_release_obj(page);
page = NULL;
/*
* Handle page content stream.
*/
{
pdf_obj *content_new;
if (!contents) {
/*
* Empty page
*/
content_new = pdf_new_stream(0);
/* TODO: better don't include anything if the page is empty */
} else if (PDF_OBJ_STREAMTYPE(contents)) {
/*
* We must import the stream because its dictionary
* may contain indirect references.
*/
content_new = pdf_import_object(contents);
} else if (PDF_OBJ_ARRAYTYPE(contents)) {
/*
* Concatenate all content streams.
*/
int idx, len = pdf_array_length(contents);
content_new = pdf_new_stream(STREAM_COMPRESS);
for (idx = 0; idx < len; idx++) {
pdf_obj *content_seg = pdf_deref_obj(pdf_get_array(contents, idx));
if (!PDF_OBJ_STREAMTYPE(content_seg) ||
pdf_concat_stream(content_new, content_seg) < 0) {
pdf_release_obj(content_seg);
pdf_release_obj(content_new);
goto error;
}
pdf_release_obj(content_seg);
}
} else {
goto error;
}
if (contents)
pdf_release_obj(contents);
contents = content_new;
}
/*
* Add entries to contents stream dictionary.
*/
{
pdf_obj *contents_dict, *bbox, *matrix;
contents_dict = pdf_stream_dict(contents);
pdf_add_dict(contents_dict,
pdf_new_name("Type"), pdf_new_name("XObject"));
pdf_add_dict(contents_dict,
pdf_new_name("Subtype"), pdf_new_name("Form"));
pdf_add_dict(contents_dict,
pdf_new_name("FormType"), pdf_new_number(1.0));
bbox = pdf_new_array();
pdf_add_array(bbox, pdf_new_number(info.bbox.llx));
pdf_add_array(bbox, pdf_new_number(info.bbox.lly));
pdf_add_array(bbox, pdf_new_number(info.bbox.urx));
pdf_add_array(bbox, pdf_new_number(info.bbox.ury));
pdf_add_dict(contents_dict, pdf_new_name("BBox"), bbox);
matrix = pdf_new_array();
pdf_add_array(matrix, pdf_new_number(1.0));
pdf_add_array(matrix, pdf_new_number(0.0));
pdf_add_array(matrix, pdf_new_number(0.0));
pdf_add_array(matrix, pdf_new_number(1.0));
pdf_add_array(matrix, pdf_new_number(0.0));
pdf_add_array(matrix, pdf_new_number(0.0));
pdf_add_dict(contents_dict, pdf_new_name("Matrix"), matrix);
pdf_add_dict(contents_dict, pdf_new_name("Resources"),
pdf_import_object(resources));
pdf_release_obj(resources);
}
pdf_close(pf);
pdf_ximage_set_form(ximage, &info, contents);
return 0;
error:
WARN("Cannot parse document. Broken PDF file?");
error_silent:
if (resources)
pdf_release_obj(resources);
if (markinfo)
pdf_release_obj(markinfo);
if (page)
pdf_release_obj(page);
if (contents)
pdf_release_obj(contents);
pdf_close(pf);
return -1;
}
int
pdf_copy_clip (FILE *image_file, int pageNo, double x_user, double y_user)
{
pdf_obj *page_tree, *contents;
int depth = 0, top = -1;
const char *clip_path, *end_path;
char *save_path, *temp;
pdf_tmatrix M;
double stack[6];
pdf_file *pf;
pf = pdf_open(NULL, image_file);
if (!pf)
return -1;
pdf_dev_currentmatrix(&M);
pdf_invertmatrix(&M);
M.e += x_user; M.f += y_user;
page_tree = pdf_get_page_obj (pf, pageNo, NULL, NULL);
if (!page_tree) {
pdf_close(pf);
return -1;
}
contents = pdf_get_page_content(page_tree);
pdf_release_obj(page_tree);
if (!contents) {
pdf_close(pf);
return -1;
}
pdf_doc_add_page_content(" ", 1);
save_path = malloc(pdf_stream_length(contents) + 1);
strncpy(save_path, (const char *) pdf_stream_dataptr(contents), pdf_stream_length(contents));
clip_path = save_path;
end_path = clip_path + pdf_stream_length(contents);
depth = 0;
for (; clip_path < end_path; clip_path++) {
int color_dimen = 0; /* silence uninitialized warning */
char *token;
skip_white(&clip_path, end_path);
if (clip_path == end_path)
break;
if (depth > 1) {
if (*clip_path == 'q')
depth++;
if (*clip_path == 'Q')
depth--;
parse_ident(&clip_path, end_path);
continue;
} else if (*clip_path == '-'
|| *clip_path == '+'
|| *clip_path == '.'
|| isdigit((unsigned char)*clip_path)) {
stack[++top] = strtod(clip_path, &temp);
clip_path = temp;
} else if (*clip_path == '[') {
/* Ignore, but put a dummy value on the stack (in case of d operator) */
parse_pdf_array(&clip_path, end_path, pf);
stack[++top] = 0;
} else if (*clip_path == '/') {
if (strncmp("/DeviceGray", clip_path, 11) == 0
|| strncmp("/Indexed", clip_path, 8) == 0
|| strncmp("/CalGray", clip_path, 8) == 0) {
color_dimen = 1;
continue;
}
else if (strncmp("/DeviceRGB", clip_path, 10) == 0
|| strncmp("/CalRGB", clip_path, 7) == 0
|| strncmp("/Lab", clip_path, 4) == 0) {
color_dimen = 3;
continue;
}
else if (strncmp("/DeviceCMYK", clip_path, 11) == 0) {
color_dimen = 4;
continue;
}
else {
clip_path++;
parse_ident(&clip_path, end_path);
skip_white(&clip_path, end_path);
token = parse_ident(&clip_path, end_path);
if (strcmp(token, "gs") == 0) {
continue;
}
return -1;
}
} else {
int j;
pdf_tmatrix T;
pdf_coord p0, p1, p2, p3;
token = parse_ident(&clip_path, end_path);
for (j = 0; j < sizeof(pdf_operators) / sizeof(pdf_operators[0]); j++)
if (strcmp(token, pdf_operators[j].token) == 0)
break;
if (j == sizeof(pdf_operators) / sizeof(pdf_operators[0])) {
return -1;
}
switch (pdf_operators[j].opcode) {
case 0:
case -1:
case -2:
case -3:
case -4:
/* Just pop the stack and do nothing. */
top += pdf_operators[j].opcode;
if (top < -1)
return -1;
break;
case OP_SETCOLOR:
top -= color_dimen;
if (top < -1)
return -1;
break;
case OP_CLOSEandCLIP:
pdf_dev_closepath();
case OP_CLIP:
#if 0
pdf_dev_clip();
#else
pdf_dev_flushpath('W', PDF_FILL_RULE_NONZERO);
#endif
break;
case OP_CONCATMATRIX:
if (top < 5)
return -1;
T.f = stack[top--];
T.e = stack[top--];
T.d = stack[top--];
T.c = stack[top--];
T.b = stack[top--];
T.a = stack[top--];
pdf_concatmatrix(&M, &T);
break;
case OP_SETCOLORSPACE:
/* Do nothing. */
break;
case OP_RECTANGLE:
if (top < 3)
return -1;
p1.y = stack[top--];
p1.x = stack[top--];
p0.y = stack[top--];
p0.x = stack[top--];
if (M.b == 0 && M.c == 0) {
pdf_tmatrix M0;
M0.a = M.a; M0.b = M.b; M0.c = M.c; M0.d = M.d;
M0.e = 0; M0.f = 0;
pdf_dev_transform(&p0, &M);
pdf_dev_transform(&p1, &M0);
pdf_dev_rectadd(p0.x, p0.y, p1.x, p1.y);
} else {
p2.x = p0.x + p1.x; p2.y = p0.y + p1.y;
p3.x = p0.x; p3.y = p0.y + p1.y;
p1.x += p0.x; p1.y = p0.y;
pdf_dev_transform(&p0, &M);
pdf_dev_transform(&p1, &M);
pdf_dev_transform(&p2, &M);
pdf_dev_transform(&p3, &M);
pdf_dev_moveto(p0.x, p0.y);
pdf_dev_lineto(p1.x, p1.y);
pdf_dev_lineto(p2.x, p2.y);
pdf_dev_lineto(p3.x, p3.y);
pdf_dev_closepath();
}
break;
case OP_CURVETO:
if (top < 5)
return -1;
p0.y = stack[top--];
p0.x = stack[top--];
pdf_dev_transform(&p0, &M);
p1.y = stack[top--];
p1.x = stack[top--];
pdf_dev_transform(&p1, &M);
p2.y = stack[top--];
p2.x = stack[top--];
pdf_dev_transform(&p2, &M);
pdf_dev_curveto(p2.x, p2.y, p1.x, p1.y, p0.x, p0.y);
break;
case OP_CLOSEPATH:
pdf_dev_closepath();
break;
case OP_LINETO:
if (top < 1)
return -1;
p0.y = stack[top--];
p0.x = stack[top--];
pdf_dev_transform(&p0, &M);
pdf_dev_lineto(p0.x, p0.y);
break;
case OP_MOVETO:
if (top < 1)
return -1;
p0.y = stack[top--];
p0.x = stack[top--];
pdf_dev_transform(&p0, &M);
pdf_dev_moveto(p0.x, p0.y);
break;
case OP_NOOP:
pdf_doc_add_page_content(" n", 2);
break;
case OP_GSAVE:
depth++;
break;
case OP_GRESTORE:
depth--;
break;
case OP_CURVETO1:
if (top < 3)
return -1;
p0.y = stack[top--];
p0.x = stack[top--];
pdf_dev_transform(&p0, &M);
p1.y = stack[top--];
p1.x = stack[top--];
pdf_dev_transform(&p1, &M);
pdf_dev_vcurveto(p1.x, p1.y, p0.x, p0.y);
break;
case OP_CURVETO2:
if (top < 3)
return -1;
p0.y = stack[top--];
p0.x = stack[top--];
pdf_dev_transform(&p0, &M);
p1.y = stack[top--];
p1.x = stack[top--];
pdf_dev_transform(&p1, &M);
pdf_dev_ycurveto(p1.x, p1.y, p0.x, p0.y);
break;
default:
return -1;
}
}
}
free(save_path);
pdf_release_obj(contents);
pdf_close(pf);
return 0;
}
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);
}
