static int strip_outlines(fz_context *ctx, pdf_document *doc, pdf_obj *outlines, int page_count, int *page_object_nums, pdf_obj *names_list)
{
int nc;
pdf_obj *first;
pdf_obj *last;
first = pdf_dict_get(ctx, outlines, PDF_NAME_First);
if (first == NULL)
nc = 0;
else
nc = strip_outline(ctx, doc, first, page_count, page_object_nums, names_list, &first, &last);
if (nc == 0)
{
pdf_dict_del(ctx, outlines, PDF_NAME_First);
pdf_dict_del(ctx, outlines, PDF_NAME_Last);
pdf_dict_del(ctx, outlines, PDF_NAME_Count);
}
else
{
int old_count = pdf_to_int(ctx, pdf_dict_get(ctx, outlines, PDF_NAME_Count));
pdf_dict_put(ctx, outlines, PDF_NAME_First, first);
pdf_dict_put(ctx, outlines, PDF_NAME_Last, last);
pdf_dict_put(ctx, outlines, PDF_NAME_Count, pdf_new_int(ctx, doc, old_count > 0 ? nc : -nc));
}
return nc;
}
void pdf_delete_portfolio_schema(fz_context *ctx, pdf_document *doc, int entry)
{
pdf_portfolio **pp;
pdf_portfolio *p;
pdf_obj *s;
if (!doc)
fz_throw(ctx, FZ_ERROR_GENERIC, "Bad pdf_delete_portfolio_schema call");
if (doc->portfolio == NULL)
load_portfolio(ctx, doc);
pp = &doc->portfolio;
while (*pp && entry > 0)
pp = &(*pp)->next, entry--;
p = *pp;
if (p == NULL || entry)
fz_throw(ctx, FZ_ERROR_GENERIC, "entry out of range in pdf_delete_portfolio_schema");
*pp = p->next;
/* Delete the key from the schema */
s = pdf_dict_getl(ctx, pdf_trailer(ctx, doc), PDF_NAME_Root, PDF_NAME_Collection, PDF_NAME_Schema, NULL);
pdf_dict_del(ctx, s, p->key);
/* Delete this entry from all the collection entries */
s = pdf_dict_getl(ctx, pdf_trailer(ctx, doc), PDF_NAME_Root, PDF_NAME_Names, PDF_NAME_EmbeddedFiles, NULL);
pdf_name_tree_map(ctx, s, delete_from_node, p->key);
pdf_drop_obj(ctx, p->entry.name);
pdf_drop_obj(ctx, p->key);
pdf_drop_obj(ctx, p->val);
fz_free(ctx, p);
}
void
pdf_clear_annot_quad_points(fz_context *ctx, pdf_annot *annot)
{
check_allowed_subtypes(ctx, annot, PDF_NAME(QuadPoints), quad_point_subtypes);
pdf_dict_del(ctx, annot->obj, PDF_NAME(QuadPoints));
pdf_dirty_annot(ctx, annot);
}
void
pdf_clear_annot_ink_list(fz_context *ctx, pdf_annot *annot)
{
check_allowed_subtypes(ctx, annot, PDF_NAME(InkList), ink_list_subtypes);
pdf_dict_del(ctx, annot->obj, PDF_NAME(InkList));
pdf_dirty_annot(ctx, annot);
}
static int delete_from_node(fz_context *ctx, pdf_obj *container, pdf_obj *key, pdf_obj *val, void *arg)
{
pdf_obj *delete_key = (pdf_obj *)arg;
pdf_dict_del(ctx, pdf_dict_get(ctx, val, PDF_NAME_CI), delete_key);
return 0;
}
void pdf_set_annot_opacity(fz_context *ctx, pdf_annot *annot, float opacity)
{
if (opacity != 1)
pdf_dict_put_real(ctx, annot->obj, PDF_NAME(CA), opacity);
else
pdf_dict_del(ctx, annot->obj, PDF_NAME(CA));
pdf_dirty_annot(ctx, annot);
}
void
pdf_update_xobject(fz_context *ctx, pdf_document *doc, pdf_obj *form, fz_rect bbox, fz_matrix matrix, pdf_obj *res, fz_buffer *contents)
{
pdf_dict_put_rect(ctx, form, PDF_NAME(BBox), bbox);
pdf_dict_put_matrix(ctx, form, PDF_NAME(Matrix), matrix);
if (res)
pdf_dict_put(ctx, form, PDF_NAME(Resources), res);
else
pdf_dict_del(ctx, form, PDF_NAME(Resources));
pdf_update_stream(ctx, doc, form, contents, 0);
}
void
pdf_set_annot_border(fz_context *ctx, pdf_annot *annot, float w)
{
pdf_document *doc = annot->page->doc;
pdf_obj *border = pdf_dict_get(ctx, annot->obj, PDF_NAME_Border);
if (pdf_is_array(ctx, border))
pdf_array_put_drop(ctx, border, 2, pdf_new_real(ctx, doc, w));
else
{
border = pdf_new_array(ctx, doc, 3);
pdf_array_push_drop(ctx, border, pdf_new_real(ctx, doc, 0));
pdf_array_push_drop(ctx, border, pdf_new_real(ctx, doc, 0));
pdf_array_push_drop(ctx, border, pdf_new_real(ctx, doc, w));
pdf_dict_put_drop(ctx, annot->obj, PDF_NAME_Border, border);
}
/* Remove border style and effect dictionaries so they won't interfere. */
pdf_dict_del(ctx, annot->obj, PDF_NAME_BS);
pdf_dict_del(ctx, annot->obj, PDF_NAME_BE);
annot->changed = 1;
}
/*
When resetting or submitting a form, the fields to act upon are defined
by an array of either field references or field names, plus a flag determining
whether to act upon the fields in the array, or all fields other than those in
the array. specified_fields interprets this information and produces the array
of fields to be acted upon.
*/
static pdf_obj *specified_fields(fz_context *ctx, pdf_document *doc, pdf_obj *fields, int exclude)
{
pdf_obj *form = pdf_dict_getl(ctx, pdf_trailer(ctx, doc), PDF_NAME_Root, PDF_NAME_AcroForm, PDF_NAME_Fields, NULL);
int i, n;
pdf_obj *result = pdf_new_array(ctx, doc, 0);
pdf_obj *nil = NULL;
fz_var(nil);
fz_try(ctx)
{
/* The 'fields' array not being present signals that all fields
* should be acted upon, so handle it using the exclude case - excluding none */
if (exclude || !fields)
{
/* mark the fields we don't want to act upon */
nil = pdf_new_null(ctx, doc);
n = pdf_array_len(ctx, fields);
for (i = 0; i < n; i++)
{
pdf_obj *field = pdf_array_get(ctx, fields, i);
if (pdf_is_string(ctx, field))
field = pdf_lookup_field(ctx, form, pdf_to_str_buf(ctx, field));
if (field)
pdf_dict_put(ctx, field, PDF_NAME_Exclude, nil);
}
/* Act upon all unmarked fields */
n = pdf_array_len(ctx, form);
for (i = 0; i < n; i++)
add_field_hierarchy_to_array(ctx, result, pdf_array_get(ctx, form, i));
/* Unmark the marked fields */
n = pdf_array_len(ctx, fields);
for (i = 0; i < n; i++)
{
pdf_obj *field = pdf_array_get(ctx, fields, i);
if (pdf_is_string(ctx, field))
field = pdf_lookup_field(ctx, form, pdf_to_str_buf(ctx, field));
if (field)
pdf_dict_del(ctx, field, PDF_NAME_Exclude);
}
}
else
{
n = pdf_array_len(ctx, fields);
for (i = 0; i < n; i++)
{
pdf_obj *field = pdf_array_get(ctx, fields, i);
if (pdf_is_string(ctx, field))
field = pdf_lookup_field(ctx, form, pdf_to_str_buf(ctx, field));
if (field)
add_field_hierarchy_to_array(ctx, result, field);
}
}
}
fz_always(ctx)
{
pdf_drop_obj(ctx, nil);
}
fz_catch(ctx)
{
pdf_drop_obj(ctx, result);
fz_rethrow(ctx);
}
return result;
}
static void reset_field(fz_context *ctx, pdf_document *doc, pdf_obj *field)
{
/* Set V to DV whereever DV is present, and delete V where DV is not.
* FIXME: we assume for now that V has not been set unequal
* to DV higher in the hierarchy than "field".
*
* At the bottom of the hierarchy we may find widget annotations
* that aren't also fields, but DV and V will not be present in their
* dictionaries, and attempts to remove V will be harmless. */
pdf_obj *dv = pdf_dict_get(ctx, field, PDF_NAME_DV);
pdf_obj *kids = pdf_dict_get(ctx, field, PDF_NAME_Kids);
if (dv)
pdf_dict_put(ctx, field, PDF_NAME_V, dv);
else
pdf_dict_del(ctx, field, PDF_NAME_V);
if (kids == NULL)
{
/* The leaves of the tree are widget annotations
* In some cases we need to update the appearance state;
* in others we need to mark the field as dirty so that
* the appearance stream will be regenerated. */
switch (pdf_field_type(ctx, doc, field))
{
case PDF_WIDGET_TYPE_RADIOBUTTON:
case PDF_WIDGET_TYPE_CHECKBOX:
{
pdf_obj *leafv = pdf_get_inheritable(ctx, doc, field, PDF_NAME_V);
if (leafv)
pdf_keep_obj(ctx, leafv);
else
leafv = PDF_NAME_Off;
fz_try(ctx)
{
pdf_dict_put(ctx, field, PDF_NAME_AS, leafv);
}
fz_always(ctx)
{
pdf_drop_obj(ctx, leafv);
}
fz_catch(ctx)
{
fz_rethrow(ctx);
}
}
break;
case PDF_WIDGET_TYPE_PUSHBUTTON:
break;
default:
pdf_field_mark_dirty(ctx, doc, field);
break;
}
}
if (pdf_field_dirties_document(ctx, doc, field))
doc->dirty = 1;
}
void pdf_clear_annot_vertices(fz_context *ctx, pdf_annot *annot)
{
check_allowed_subtypes(ctx, annot, PDF_NAME(Vertices), vertices_subtypes);
pdf_dict_del(ctx, annot->obj, PDF_NAME(Vertices));
pdf_dirty_annot(ctx, annot);
}
static void wmupdf_convert_single_page_to_form(pdf_document *xref,fz_context *ctx,int pageno)
{
pdf_obj *array,*srcpageobj,*srcpagecontents;
int i,len,streamlen,pageref,pagegen,compressed;
double bbox_array[4];
double matrix[6];
/* New source page, so get the source page objects */
srcpageobj = xref->page_objs[pageno-1];
pageref=pdf_to_num(xref->page_refs[pageno-1]);
pagegen=pdf_to_gen(xref->page_refs[pageno-1]);
wmupdf_page_bbox(srcpageobj,bbox_array);
for (i=0;i<6;i++)
matrix[i]=0.;
matrix[0]=matrix[3]=1.;
srcpagecontents=pdf_dict_gets(srcpageobj,"Contents");
/* Concatenate all indirect streams from source page directly into it. */
// printf("Adding streams to source page %d (pageref=%d, pagegen=%d)...\n",pageno,pageref,pagegen);
streamlen=0;
if (pdf_is_array(srcpagecontents))
{
int k;
for (k=0;k<pdf_array_len(srcpagecontents);k++)
{
pdf_obj *obj;
obj=pdf_array_get(srcpagecontents,k);
if (pdf_is_indirect(obj))
pdf_resolve_indirect(obj);
streamlen=add_to_srcpage_stream(xref,ctx,pageref,pagegen,obj);
}
}
else
{
if (pdf_is_indirect(srcpagecontents))
pdf_resolve_indirect(srcpagecontents);
streamlen=add_to_srcpage_stream(xref,ctx,pageref,pagegen,srcpagecontents);
}
compressed=stream_deflate(xref,ctx,pageref,pagegen,&streamlen);
srcpageobj = xref->page_objs[pageno-1];
pageref=pdf_to_num(xref->page_refs[pageno-1]);
len=pdf_dict_len(srcpageobj);
for (i=0;i<len;i++)
{
pdf_obj *key; /* *value */
key=pdf_dict_get_key(srcpageobj,i);
/*
if (pdf_is_name(key))
printf("key[%d] = name = %s\n",i,pdf_to_name(key));
else
printf("key[%d] = ??\n",i);
*/
/* value=pdf_dict_get_val(srcpageobj,i); */
/* Keep same resources */
if (!pdf_is_name(key))
continue;
if (pdf_is_name(key) && !stricmp("Resources",pdf_to_name(key)))
continue;
/* Drop dictionary entry otherwise */
// printf("Deleting key %s.\n",pdf_to_name(key));
pdf_dict_del(srcpageobj,key);
i=-1;
len=pdf_dict_len(srcpageobj);
}
pdf_dict_puts(srcpageobj,"Type",pdf_new_name(ctx,"XObject"));
pdf_dict_puts(srcpageobj,"Subtype",pdf_new_name(ctx,"Form"));
pdf_dict_puts(srcpageobj,"FormType",pdf_new_int(ctx,1));
if (compressed)
pdf_dict_puts(srcpageobj,"Filter",pdf_new_name(ctx,"FlateDecode"));
pdf_dict_puts(srcpageobj,"Length",pdf_new_int(ctx,streamlen));
array=pdf_new_array(ctx,4);
for (i=0;i<4;i++)
pdf_array_push(array,pdf_new_real(ctx,bbox_array[i]));
pdf_dict_puts(srcpageobj,"BBox",array);
array=pdf_new_array(ctx,6);
for (i=0;i<6;i++)
pdf_array_push(array,pdf_new_real(ctx,matrix[i]));
pdf_dict_puts(srcpageobj,"Matrix",array);
/* (It's no longer a "page"--it's a Form-type XObject) */
/* I don't think this call should be made since it will call fz_drop_object on srcpageobj */
/* pdf_update_object(xref,pageref,srcpageobj); */
}
void pdf_clean_page_contents(fz_context *ctx, pdf_document *doc, pdf_page *page, fz_cookie *cookie, pdf_page_contents_process_fn *proc_fn, void *proc_arg)
{
pdf_processor *proc_buffer = NULL;
pdf_processor *proc_filter = NULL;
pdf_obj *new_obj = NULL;
pdf_obj *new_ref = NULL;
pdf_obj *res = NULL;
pdf_obj *ref = NULL;
pdf_obj *obj;
pdf_obj *contents;
fz_buffer *buffer;
fz_var(new_obj);
fz_var(new_ref);
fz_var(res);
fz_var(ref);
fz_var(proc_buffer);
fz_var(proc_filter);
buffer = fz_new_buffer(ctx, 1024);
fz_try(ctx)
{
res = pdf_new_dict(ctx, doc, 1);
proc_buffer = pdf_new_buffer_processor(ctx, buffer);
proc_filter = pdf_new_filter_processor(ctx, proc_buffer, doc, page->resources, res);
pdf_process_contents(ctx, proc_filter, doc, page->resources, page->contents, cookie);
contents = page->contents;
if (pdf_is_array(ctx, contents))
{
/* create a new object to replace the array */
new_obj = pdf_new_dict(ctx, doc, 1);
new_ref = pdf_new_ref(ctx, doc, new_obj);
page->contents = contents = new_ref;
}
else
{
pdf_dict_del(ctx, contents, PDF_NAME_Filter);
pdf_dict_del(ctx, contents, PDF_NAME_DecodeParms);
}
/* Now deal with resources. The spec allows for Type3 fonts and form
* XObjects to omit a resource dictionary and look in the parent.
* Avoid that by flattening here as part of the cleaning. This could
* conceivably cause changes in rendering, but we don't care. */
/* ExtGState */
obj = pdf_dict_get(ctx, res, PDF_NAME_ExtGState);
if (obj)
{
int i, l;
l = pdf_dict_len(ctx, obj);
for (i = 0; i < l; i++)
{
pdf_obj *o = pdf_dict_get(ctx, pdf_dict_get_val(ctx, obj, i), PDF_NAME_SMask);
if (!o)
continue;
o = pdf_dict_get(ctx, o, PDF_NAME_G);
if (!o)
continue;
/* Transparency group XObject */
pdf_clean_stream_object(ctx, doc, o, page->resources, cookie, 1);
}
}
/* ColorSpace - no cleaning possible */
/* Pattern */
obj = pdf_dict_get(ctx, res, PDF_NAME_Pattern);
if (obj)
{
int i, l;
l = pdf_dict_len(ctx, obj);
for (i = 0; i < l; i++)
{
pdf_obj *pat = pdf_dict_get_val(ctx, obj, i);
if (!pat)
continue;
if (pdf_to_int(ctx, pdf_dict_get(ctx, pat, PDF_NAME_PatternType)) == 1)
pdf_clean_stream_object(ctx, doc, pat, page->resources, cookie, 0);
}
}
/* Shading - no cleaning possible */
/* XObject */
obj = pdf_dict_get(ctx, res, PDF_NAME_XObject);
if (obj)
{
int i, l;
l = pdf_dict_len(ctx, obj);
for (i = 0; i < l; i++)
{
pdf_obj *xobj = pdf_dict_get_val(ctx, obj, i);
if (!pdf_name_eq(ctx, PDF_NAME_Form, pdf_dict_get(ctx, xobj, PDF_NAME_Subtype)))
continue;
pdf_clean_stream_object(ctx, doc, xobj, page->resources, cookie, 1);
}
}
/* Font */
obj = pdf_dict_get(ctx, res, PDF_NAME_Font);
if (obj)
{
int i, l;
l = pdf_dict_len(ctx, obj);
for (i = 0; i < l; i++)
{
pdf_obj *o = pdf_dict_get_val(ctx, obj, i);
if (pdf_name_eq(ctx, PDF_NAME_Type3, pdf_dict_get(ctx, o, PDF_NAME_Subtype)))
{
pdf_clean_type3(ctx, doc, o, page->resources, cookie);
}
}
}
/* ProcSet - no cleaning possible. Inherit this from the old dict. */
obj = pdf_dict_get(ctx, page->resources, PDF_NAME_ProcSet);
if (obj)
pdf_dict_put(ctx, res, PDF_NAME_ProcSet, obj);
/* Properties - no cleaning possible. */
if (proc_fn)
(*proc_fn)(ctx, buffer, res, proc_arg);
pdf_update_stream(ctx, doc, contents, buffer, 0);
pdf_drop_obj(ctx, page->resources);
ref = pdf_new_ref(ctx, doc, res);
page->resources = pdf_keep_obj(ctx, ref);
pdf_dict_put(ctx, page->me, PDF_NAME_Resources, ref);
}
fz_always(ctx)
{
pdf_drop_processor(ctx, proc_filter);
pdf_drop_processor(ctx, proc_buffer);
fz_drop_buffer(ctx, buffer);
pdf_drop_obj(ctx, new_obj);
pdf_drop_obj(ctx, new_ref);
pdf_drop_obj(ctx, res);
pdf_drop_obj(ctx, ref);
}
fz_catch(ctx)
{
fz_rethrow_message(ctx, "Failed while cleaning page");
}
}
static int strip_outline(fz_context *ctx, pdf_document *doc, pdf_obj *outlines, int page_count, int *page_object_nums, pdf_obj *names_list, pdf_obj **pfirst, pdf_obj **plast)
{
pdf_obj *prev = NULL;
pdf_obj *first = NULL;
pdf_obj *current;
int count = 0;
for (current = outlines; current != NULL; )
{
int nc;
/* Strip any children to start with. This takes care of
* First/Last/Count for us. */
nc = strip_outlines(ctx, doc, current, page_count, page_object_nums, names_list);
if (!dest_is_valid(ctx, current, page_count, page_object_nums, names_list))
{
if (nc == 0)
{
/* Outline with invalid dest and no children. Drop it by
* pulling the next one in here. */
pdf_obj *next = pdf_dict_get(ctx, current, PDF_NAME_Next);
if (next == NULL)
{
/* There is no next one to pull in */
if (prev != NULL)
pdf_dict_del(ctx, prev, PDF_NAME_Next);
}
else if (prev != NULL)
{
pdf_dict_put(ctx, prev, PDF_NAME_Next, next);
pdf_dict_put(ctx, next, PDF_NAME_Prev, prev);
}
else
{
pdf_dict_del(ctx, next, PDF_NAME_Prev);
}
current = next;
}
else
{
/* Outline with invalid dest, but children. Just drop the dest. */
pdf_dict_del(ctx, current, PDF_NAME_Dest);
pdf_dict_del(ctx, current, PDF_NAME_A);
current = pdf_dict_get(ctx, current, PDF_NAME_Next);
}
}
else
{
/* Keep this one */
if (first == NULL)
first = current;
prev = current;
current = pdf_dict_get(ctx, current, PDF_NAME_Next);
count++;
}
}
*pfirst = first;
*plast = prev;
return count;
}
static void retainpages(fz_context *ctx, globals *glo, int argc, char **argv)
{
pdf_obj *oldroot, *root, *pages, *kids, *countobj, *parent, *olddests;
pdf_document *doc = glo->doc;
int argidx = 0;
pdf_obj *names_list = NULL;
pdf_obj *outlines;
int pagecount;
int i;
int *page_object_nums;
/* Keep only pages/type and (reduced) dest entries to avoid
* references to unretained pages */
oldroot = pdf_dict_get(ctx, pdf_trailer(ctx, doc), PDF_NAME_Root);
pages = pdf_dict_get(ctx, oldroot, PDF_NAME_Pages);
olddests = pdf_load_name_tree(ctx, doc, PDF_NAME_Dests);
outlines = pdf_dict_get(ctx, oldroot, PDF_NAME_Outlines);
root = pdf_new_dict(ctx, doc, 3);
pdf_dict_put(ctx, root, PDF_NAME_Type, pdf_dict_get(ctx, oldroot, PDF_NAME_Type));
pdf_dict_put(ctx, root, PDF_NAME_Pages, pdf_dict_get(ctx, oldroot, PDF_NAME_Pages));
pdf_dict_put(ctx, root, PDF_NAME_Outlines, outlines);
pdf_update_object(ctx, doc, pdf_to_num(ctx, oldroot), root);
/* Create a new kids array with only the pages we want to keep */
parent = pdf_new_indirect(ctx, doc, pdf_to_num(ctx, pages), pdf_to_gen(ctx, pages));
kids = pdf_new_array(ctx, doc, 1);
/* Retain pages specified */
while (argc - argidx)
{
int page, spage, epage;
char *spec, *dash;
char *pagelist = argv[argidx];
pagecount = pdf_count_pages(ctx, doc);
spec = fz_strsep(&pagelist, ",");
while (spec)
{
dash = strchr(spec, '-');
if (dash == spec)
spage = epage = pagecount;
else
spage = epage = atoi(spec);
if (dash)
{
if (strlen(dash) > 1)
epage = atoi(dash + 1);
else
epage = pagecount;
}
spage = fz_clampi(spage, 1, pagecount);
epage = fz_clampi(epage, 1, pagecount);
if (spage < epage)
for (page = spage; page <= epage; ++page)
retainpage(ctx, doc, parent, kids, page);
else
for (page = spage; page >= epage; --page)
retainpage(ctx, doc, parent, kids, page);
spec = fz_strsep(&pagelist, ",");
}
argidx++;
}
pdf_drop_obj(ctx, parent);
/* Update page count and kids array */
countobj = pdf_new_int(ctx, doc, pdf_array_len(ctx, kids));
pdf_dict_put(ctx, pages, PDF_NAME_Count, countobj);
pdf_drop_obj(ctx, countobj);
pdf_dict_put(ctx, pages, PDF_NAME_Kids, kids);
pdf_drop_obj(ctx, kids);
/* Force the next call to pdf_count_pages to recount */
glo->doc->page_count = 0;
pagecount = pdf_count_pages(ctx, doc);
page_object_nums = fz_calloc(ctx, pagecount, sizeof(*page_object_nums));
for (i = 0; i < pagecount; i++)
{
pdf_obj *pageref = pdf_lookup_page_obj(ctx, doc, i);
page_object_nums[i] = pdf_to_num(ctx, pageref);
}
/* If we had an old Dests tree (now reformed as an olddests
* dictionary), keep any entries in there that point to
* valid pages. This may mean we keep more than we need, but
* it's safe at least. */
if (olddests)
{
pdf_obj *names = pdf_new_dict(ctx, doc, 1);
pdf_obj *dests = pdf_new_dict(ctx, doc, 1);
int len = pdf_dict_len(ctx, olddests);
names_list = pdf_new_array(ctx, doc, 32);
for (i = 0; i < len; i++)
{
pdf_obj *key = pdf_dict_get_key(ctx, olddests, i);
pdf_obj *val = pdf_dict_get_val(ctx, olddests, i);
pdf_obj *dest = pdf_dict_get(ctx, val, PDF_NAME_D);
dest = pdf_array_get(ctx, dest ? dest : val, 0);
if (dest_is_valid_page(ctx, dest, page_object_nums, pagecount))
{
pdf_obj *key_str = pdf_new_string(ctx, doc, pdf_to_name(ctx, key), strlen(pdf_to_name(ctx, key)));
pdf_array_push(ctx, names_list, key_str);
pdf_array_push(ctx, names_list, val);
pdf_drop_obj(ctx, key_str);
}
}
pdf_dict_put(ctx, dests, PDF_NAME_Names, names_list);
pdf_dict_put(ctx, names, PDF_NAME_Dests, dests);
pdf_dict_put(ctx, root, PDF_NAME_Names, names);
pdf_drop_obj(ctx, names);
pdf_drop_obj(ctx, dests);
pdf_drop_obj(ctx, olddests);
}
/* Edit each pages /Annot list to remove any links that point to
* nowhere. */
for (i = 0; i < pagecount; i++)
{
pdf_obj *pageref = pdf_lookup_page_obj(ctx, doc, i);
pdf_obj *pageobj = pdf_resolve_indirect(ctx, pageref);
pdf_obj *annots = pdf_dict_get(ctx, pageobj, PDF_NAME_Annots);
int len = pdf_array_len(ctx, annots);
int j;
for (j = 0; j < len; j++)
{
pdf_obj *o = pdf_array_get(ctx, annots, j);
if (!pdf_name_eq(ctx, pdf_dict_get(ctx, o, PDF_NAME_Subtype), PDF_NAME_Link))
continue;
if (!dest_is_valid(ctx, o, pagecount, page_object_nums, names_list))
{
/* Remove this annotation */
pdf_array_delete(ctx, annots, j);
j--;
}
}
}
if (strip_outlines(ctx, doc, outlines, pagecount, page_object_nums, names_list) == 0)
{
pdf_dict_del(ctx, root, PDF_NAME_Outlines);
}
fz_free(ctx, page_object_nums);
pdf_drop_obj(ctx, names_list);
pdf_drop_obj(ctx, root);
}
/*
Performs the same task as
pdf_clean_page_contents, but with an optional text filter
function.
text_filter: Function to assess whether a given character
should be kept (return 0) or removed (return 1).
after_text: Function called after each text object is closed
to allow other output to be sent.
arg: Opaque value to be passed to callback functions.
*/
void pdf_filter_page_contents(fz_context *ctx, pdf_document *doc, pdf_page *page, fz_cookie *cookie,
pdf_page_contents_process_fn *proc_fn, pdf_text_filter_fn *text_filter, pdf_after_text_object_fn *after_text, void *proc_arg,
int sanitize, int ascii)
{
pdf_processor *proc_buffer = NULL;
pdf_processor *proc_filter = NULL;
pdf_obj *new_obj = NULL;
pdf_obj *new_ref = NULL;
pdf_obj *res = NULL;
pdf_obj *obj;
pdf_obj *contents;
pdf_obj *resources;
fz_buffer *buffer;
fz_var(new_obj);
fz_var(new_ref);
fz_var(res);
fz_var(proc_buffer);
fz_var(proc_filter);
buffer = fz_new_buffer(ctx, 1024);
fz_try(ctx)
{
contents = pdf_page_contents(ctx, page);
resources = pdf_page_resources(ctx, page);
proc_buffer = pdf_new_buffer_processor(ctx, buffer, ascii);
if (sanitize)
{
res = pdf_new_dict(ctx, doc, 1);
proc_filter = pdf_new_filter_processor_with_text_filter(ctx, doc, proc_buffer, resources, res, text_filter, after_text, proc_arg);
pdf_process_contents(ctx, proc_filter, doc, resources, contents, cookie);
pdf_close_processor(ctx, proc_filter);
}
else
{
res = pdf_keep_obj(ctx, resources);
pdf_process_contents(ctx, proc_buffer, doc, resources, contents, cookie);
}
pdf_close_processor(ctx, proc_buffer);
/* Deal with page content stream. */
if (pdf_is_array(ctx, contents))
{
/* create a new object to replace the array */
new_obj = pdf_new_dict(ctx, doc, 1);
new_ref = pdf_add_object(ctx, doc, new_obj);
contents = new_ref;
pdf_dict_put(ctx, page->obj, PDF_NAME(Contents), contents);
}
else
{
pdf_dict_del(ctx, contents, PDF_NAME(Filter));
pdf_dict_del(ctx, contents, PDF_NAME(DecodeParms));
}
pdf_update_stream(ctx, doc, contents, buffer, 0);
/* Now deal with resources. The spec allows for Type3 fonts and form
* XObjects to omit a resource dictionary and look in the parent.
* Avoid that by flattening here as part of the cleaning. This could
* conceivably cause changes in rendering, but we don't care. */
/* ExtGState */
obj = pdf_dict_get(ctx, res, PDF_NAME(ExtGState));
if (obj)
{
int i, l;
l = pdf_dict_len(ctx, obj);
for (i = 0; i < l; i++)
{
pdf_obj *o = pdf_dict_get(ctx, pdf_dict_get_val(ctx, obj, i), PDF_NAME(SMask));
if (!o)
continue;
o = pdf_dict_get(ctx, o, PDF_NAME(G));
if (!o)
continue;
/* Transparency group XObject */
pdf_clean_stream_object(ctx, doc, o, resources, cookie, 1, text_filter, after_text, proc_arg, sanitize, ascii);
}
}
/* Pattern */
obj = pdf_dict_get(ctx, res, PDF_NAME(Pattern));
if (obj)
{
int i, l;
l = pdf_dict_len(ctx, obj);
for (i = 0; i < l; i++)
{
pdf_obj *pat_res;
pdf_obj *pat = pdf_dict_get_val(ctx, obj, i);
if (!pat)
continue;
pat_res = pdf_dict_get(ctx, pat, PDF_NAME(Resources));
if (pat_res == NULL)
pat_res = resources;
if (pdf_dict_get_int(ctx, pat, PDF_NAME(PatternType)) == 1)
pdf_clean_stream_object(ctx, doc, pat, pat_res, cookie, 0, text_filter, after_text, proc_arg, sanitize, ascii);
}
}
/* XObject */
obj = pdf_dict_get(ctx, res, PDF_NAME(XObject));
if (obj)
{
int i, l;
l = pdf_dict_len(ctx, obj);
for (i = 0; i < l; i++)
{
pdf_obj *xobj_res;
pdf_obj *xobj = pdf_dict_get_val(ctx, obj, i);
if (!xobj)
continue;
xobj_res = pdf_dict_get(ctx, xobj, PDF_NAME(Resources));
if (xobj_res == NULL)
xobj_res = resources;
if (pdf_name_eq(ctx, PDF_NAME(Form), pdf_dict_get(ctx, xobj, PDF_NAME(Subtype))))
pdf_clean_stream_object(ctx, doc, xobj, xobj_res, cookie, 1, text_filter, after_text, proc_arg, sanitize, ascii);
}
}
/* Font */
obj = pdf_dict_get(ctx, res, PDF_NAME(Font));
if (obj)
{
int i, l;
l = pdf_dict_len(ctx, obj);
for (i = 0; i < l; i++)
{
pdf_obj *o = pdf_dict_get_val(ctx, obj, i);
if (!o)
continue;
if (pdf_name_eq(ctx, PDF_NAME(Type3), pdf_dict_get(ctx, o, PDF_NAME(Subtype))))
pdf_clean_type3(ctx, doc, o, resources, cookie, sanitize, ascii);
}
}
/* ProcSet - no cleaning possible. Inherit this from the old dict. */
obj = pdf_dict_get(ctx, resources, PDF_NAME(ProcSet));
if (obj)
pdf_dict_put(ctx, res, PDF_NAME(ProcSet), obj);
/* ColorSpace - no cleaning possible. */
/* Properties - no cleaning possible. */
if (proc_fn)
(*proc_fn)(ctx, buffer, res, proc_arg);
/* Update resource dictionary */
if (sanitize)
{
pdf_dict_put(ctx, page->obj, PDF_NAME(Resources), res);
}
}
fz_always(ctx)
{
pdf_drop_processor(ctx, proc_filter);
pdf_drop_processor(ctx, proc_buffer);
fz_drop_buffer(ctx, buffer);
pdf_drop_obj(ctx, new_obj);
pdf_drop_obj(ctx, new_ref);
pdf_drop_obj(ctx, res);
}
fz_catch(ctx)
{
fz_rethrow(ctx);
}
}
pdf_obj *
pdf_add_image(fz_context *ctx, pdf_document *doc, fz_image *image)
{
fz_pixmap *pixmap = NULL;
pdf_obj *imobj = NULL;
pdf_obj *dp;
fz_buffer *buffer = NULL;
pdf_obj *imref = NULL;
fz_compressed_buffer *cbuffer;
unsigned char digest[16];
int i, n;
/* If we can maintain compression, do so */
cbuffer = fz_compressed_image_buffer(ctx, image);
fz_var(pixmap);
fz_var(buffer);
fz_var(imobj);
fz_var(imref);
/* Check if the same image already exists in this doc. */
imref = pdf_find_image_resource(ctx, doc, image, digest);
if (imref)
return imref;
imobj = pdf_add_new_dict(ctx, doc, 3);
fz_try(ctx)
{
dp = pdf_dict_put_dict(ctx, imobj, PDF_NAME(DecodeParms), 3);
pdf_dict_put(ctx, imobj, PDF_NAME(Type), PDF_NAME(XObject));
pdf_dict_put(ctx, imobj, PDF_NAME(Subtype), PDF_NAME(Image));
if (cbuffer)
{
fz_compression_params *cp = &cbuffer->params;
switch (cp ? cp->type : FZ_IMAGE_UNKNOWN)
{
default:
goto raw_or_unknown_compression;
case FZ_IMAGE_JPEG:
if (cp->u.jpeg.color_transform != -1)
pdf_dict_put_int(ctx, dp, PDF_NAME(ColorTransform), cp->u.jpeg.color_transform);
pdf_dict_put(ctx, imobj, PDF_NAME(Filter), PDF_NAME(DCTDecode));
break;
case FZ_IMAGE_JPX:
if (cp->u.jpx.smask_in_data)
pdf_dict_put_int(ctx, dp, PDF_NAME(SMaskInData), cp->u.jpx.smask_in_data);
pdf_dict_put(ctx, imobj, PDF_NAME(Filter), PDF_NAME(JPXDecode));
break;
case FZ_IMAGE_FAX:
if (cp->u.fax.columns)
pdf_dict_put_int(ctx, dp, PDF_NAME(Columns), cp->u.fax.columns);
if (cp->u.fax.rows)
pdf_dict_put_int(ctx, dp, PDF_NAME(Rows), cp->u.fax.rows);
if (cp->u.fax.k)
pdf_dict_put_int(ctx, dp, PDF_NAME(K), cp->u.fax.k);
if (cp->u.fax.end_of_line)
pdf_dict_put_bool(ctx, dp, PDF_NAME(EndOfLine), cp->u.fax.end_of_line);
if (cp->u.fax.encoded_byte_align)
pdf_dict_put_bool(ctx, dp, PDF_NAME(EncodedByteAlign), cp->u.fax.encoded_byte_align);
if (cp->u.fax.end_of_block)
pdf_dict_put_bool(ctx, dp, PDF_NAME(EndOfBlock), cp->u.fax.end_of_block);
if (cp->u.fax.black_is_1)
pdf_dict_put_bool(ctx, dp, PDF_NAME(BlackIs1), cp->u.fax.black_is_1);
if (cp->u.fax.damaged_rows_before_error)
pdf_dict_put_int(ctx, dp, PDF_NAME(DamagedRowsBeforeError), cp->u.fax.damaged_rows_before_error);
pdf_dict_put(ctx, imobj, PDF_NAME(Filter), PDF_NAME(CCITTFaxDecode));
break;
case FZ_IMAGE_FLATE:
if (cp->u.flate.columns)
pdf_dict_put_int(ctx, dp, PDF_NAME(Columns), cp->u.flate.columns);
if (cp->u.flate.colors)
pdf_dict_put_int(ctx, dp, PDF_NAME(Colors), cp->u.flate.colors);
if (cp->u.flate.predictor)
pdf_dict_put_int(ctx, dp, PDF_NAME(Predictor), cp->u.flate.predictor);
if (cp->u.flate.bpc)
pdf_dict_put_int(ctx, dp, PDF_NAME(BitsPerComponent), cp->u.flate.bpc);
pdf_dict_put(ctx, imobj, PDF_NAME(Filter), PDF_NAME(FlateDecode));
pdf_dict_put_int(ctx, imobj, PDF_NAME(BitsPerComponent), image->bpc);
break;
case FZ_IMAGE_LZW:
if (cp->u.lzw.columns)
pdf_dict_put_int(ctx, dp, PDF_NAME(Columns), cp->u.lzw.columns);
if (cp->u.lzw.colors)
pdf_dict_put_int(ctx, dp, PDF_NAME(Colors), cp->u.lzw.colors);
if (cp->u.lzw.predictor)
pdf_dict_put_int(ctx, dp, PDF_NAME(Predictor), cp->u.lzw.predictor);
if (cp->u.lzw.early_change)
pdf_dict_put_int(ctx, dp, PDF_NAME(EarlyChange), cp->u.lzw.early_change);
if (cp->u.lzw.bpc)
pdf_dict_put_int(ctx, dp, PDF_NAME(BitsPerComponent), cp->u.lzw.bpc);
pdf_dict_put(ctx, imobj, PDF_NAME(Filter), PDF_NAME(LZWDecode));
break;
case FZ_IMAGE_RLD:
pdf_dict_put(ctx, imobj, PDF_NAME(Filter), PDF_NAME(RunLengthDecode));
break;
}
if (!pdf_dict_len(ctx, dp))
pdf_dict_del(ctx, imobj, PDF_NAME(DecodeParms));
buffer = fz_keep_buffer(ctx, cbuffer->buffer);
if (image->use_decode)
{
pdf_obj *ary = pdf_dict_put_array(ctx, imobj, PDF_NAME(Decode), image->n * 2);
for (i = 0; i < image->n * 2; ++i)
pdf_array_push_real(ctx, ary, image->decode[i]);
}
}
else
{
unsigned int size;
int h;
unsigned char *d, *s;
raw_or_unknown_compression:
/* Currently, set to maintain resolution; should we consider
* subsampling here according to desired output res? */
pixmap = fz_get_pixmap_from_image(ctx, image, NULL, NULL, NULL, NULL);
n = pixmap->n - pixmap->alpha - pixmap->s; /* number of colorants */
if (n == 0)
n = 1; /* treat pixmaps with only alpha or spots as grayscale */
size = image->w * n;
h = image->h;
s = pixmap->samples;
d = fz_malloc(ctx, size * h);
buffer = fz_new_buffer_from_data(ctx, d, size * h);
if (n == pixmap->n)
{
/* If we use all channels, we can copy the data as is. */
while (h--)
{
memcpy(d, s, size);
d += size;
s += pixmap->stride;
}
}
else
{
/* Need to remove the alpha and spot planes. */
/* TODO: extract alpha plane to a soft mask. */
/* TODO: convert spots to colors. */
int line_skip = pixmap->stride - pixmap->w * pixmap->n;
int skip = pixmap->n - n;
while (h--)
{
int w = pixmap->w;
while (w--)
{
int k;
for (k = 0; k < n; ++k)
*d++ = *s++;
s += skip;
}
s += line_skip;
}
}
}
pdf_dict_put_int(ctx, imobj, PDF_NAME(Width), pixmap ? pixmap->w : image->w);
pdf_dict_put_int(ctx, imobj, PDF_NAME(Height), pixmap ? pixmap->h : image->h);
if (image->imagemask)
{
pdf_dict_put_bool(ctx, imobj, PDF_NAME(ImageMask), 1);
}
else
{
fz_colorspace *cs;
pdf_dict_put_int(ctx, imobj, PDF_NAME(BitsPerComponent), image->bpc);
cs = pixmap ? pixmap->colorspace : image->colorspace;
switch (fz_colorspace_type(ctx, cs))
{
case FZ_COLORSPACE_INDEXED:
{
fz_colorspace *basecs;
unsigned char *lookup = NULL;
int high = 0;
int basen;
pdf_obj *arr;
lookup = fz_indexed_colorspace_palette(ctx, cs, &high);
basecs = fz_colorspace_base(ctx, cs);
basen = fz_colorspace_n(ctx, basecs);
arr = pdf_dict_put_array(ctx, imobj, PDF_NAME(ColorSpace), 4);
pdf_array_push(ctx, arr, PDF_NAME(Indexed));
switch (fz_colorspace_type(ctx, basecs))
{
case FZ_COLORSPACE_GRAY:
pdf_array_push(ctx, arr, PDF_NAME(DeviceGray));
break;
case FZ_COLORSPACE_RGB:
pdf_array_push(ctx, arr, PDF_NAME(DeviceRGB));
break;
case FZ_COLORSPACE_CMYK:
pdf_array_push(ctx, arr, PDF_NAME(DeviceCMYK));
break;
default:
// TODO: convert to RGB!
fz_throw(ctx, FZ_ERROR_GENERIC, "only indexed Gray, RGB, and CMYK colorspaces supported");
break;
}
pdf_array_push_int(ctx, arr, high);
pdf_array_push_string(ctx, arr, (char *) lookup, basen * (high + 1));
}
break;
case FZ_COLORSPACE_NONE:
case FZ_COLORSPACE_GRAY:
pdf_dict_put(ctx, imobj, PDF_NAME(ColorSpace), PDF_NAME(DeviceGray));
break;
case FZ_COLORSPACE_RGB:
pdf_dict_put(ctx, imobj, PDF_NAME(ColorSpace), PDF_NAME(DeviceRGB));
break;
case FZ_COLORSPACE_CMYK:
pdf_dict_put(ctx, imobj, PDF_NAME(ColorSpace), PDF_NAME(DeviceCMYK));
break;
default:
// TODO: convert to RGB!
fz_throw(ctx, FZ_ERROR_GENERIC, "only Gray, RGB, and CMYK colorspaces supported");
break;
}
}
if (image->mask)
{
pdf_dict_put_drop(ctx, imobj, PDF_NAME(SMask), pdf_add_image(ctx, doc, image->mask));
}
pdf_update_stream(ctx, doc, imobj, buffer, 1);
/* Add ref to our image resource hash table. */
imref = pdf_insert_image_resource(ctx, doc, digest, imobj);
}
fz_always(ctx)
{
fz_drop_pixmap(ctx, pixmap);
fz_drop_buffer(ctx, buffer);
pdf_drop_obj(ctx, imobj);
}
fz_catch(ctx)
fz_rethrow(ctx);
return imref;
}
