static int
pdf_resources_use_blending(pdf_document *doc, pdf_obj *rdb)
{
fz_context *ctx = doc->ctx;
pdf_obj *obj;
int i, n, useBM = 0;
if (!rdb)
return 0;
/* Have we been here before and remembered an answer? */
if (pdf_obj_memo(rdb, &useBM))
return useBM;
/* stop on cyclic resource dependencies */
if (pdf_mark_obj(rdb))
return 0;
fz_try(ctx)
{
obj = pdf_dict_gets(rdb, "ExtGState");
n = pdf_dict_len(obj);
for (i = 0; i < n; i++)
if (pdf_extgstate_uses_blending(doc, pdf_dict_get_val(obj, i)))
goto found;
obj = pdf_dict_gets(rdb, "Pattern");
n = pdf_dict_len(obj);
for (i = 0; i < n; i++)
if (pdf_pattern_uses_blending(doc, pdf_dict_get_val(obj, i)))
goto found;
obj = pdf_dict_gets(rdb, "XObject");
n = pdf_dict_len(obj);
for (i = 0; i < n; i++)
if (pdf_xobject_uses_blending(doc, pdf_dict_get_val(obj, i)))
goto found;
if (0)
{
found:
useBM = 1;
}
}
fz_always(ctx)
{
pdf_unmark_obj(rdb);
}
fz_catch(ctx)
{
fz_rethrow(ctx);
}
pdf_set_obj_memo(rdb, useBM);
return useBM;
}
pdf_obj *
pdf_copy_dict(fz_context *ctx, pdf_obj *obj)
{
pdf_document *doc;
pdf_obj *dict;
int i, n;
RESOLVE(obj);
if (!OBJ_IS_DICT(obj))
fz_throw(ctx, FZ_ERROR_GENERIC, "not a dict (%s)", pdf_objkindstr(obj));
doc = DICT(obj)->doc;
n = pdf_dict_len(ctx, obj);
dict = pdf_new_dict(ctx, doc, n);
fz_try(ctx)
for (i = 0; i < n; i++)
pdf_dict_put(ctx, dict, pdf_dict_get_key(ctx, obj, i), pdf_dict_get_val(ctx, obj, i));
fz_catch(ctx)
{
pdf_drop_obj(ctx, dict);
fz_rethrow(ctx);
}
return dict;
}
static void
load_portfolio(fz_context *ctx, pdf_document *doc)
{
pdf_obj *obj;
int i, n;
pdf_portfolio **pp;
obj = pdf_dict_getl(ctx, pdf_trailer(ctx, doc), PDF_NAME_Root, PDF_NAME_Collection, PDF_NAME_Schema, NULL);
n = pdf_dict_len(ctx, obj);
for (i = 0; i < n; i++)
{
pdf_obj *k = pdf_dict_get_key(ctx, obj, i);
pdf_obj *v = pdf_dict_get_val(ctx, obj, i);
int sort = pdf_to_int(ctx, pdf_dict_get(ctx, v, PDF_NAME_O));
pdf_obj *eo = pdf_dict_get(ctx, v, PDF_NAME_E);
int editable = eo ? pdf_to_bool(ctx, eo) : 0;
pdf_obj *vo = pdf_dict_get(ctx, v, PDF_NAME_V);
int visible = vo ? pdf_to_bool(ctx, vo) : 1;
char *subtype = pdf_to_name(ctx, pdf_dict_get(ctx, v, PDF_NAME_Subtype));
pdf_obj *name = pdf_dict_get(ctx, v, PDF_NAME_N);
pdf_portfolio *p = fz_malloc_struct(ctx, pdf_portfolio);
p->key = pdf_keep_obj(ctx, k);
p->val = pdf_keep_obj(ctx, v);
p->sort = sort;
p->entry.visible = visible;
p->entry.editable = editable;
p->entry.name = pdf_keep_obj(ctx, name);
if (!strcmp(subtype, "S"))
p->entry.type = PDF_SCHEMA_TEXT;
else if (!strcmp(subtype, "D"))
p->entry.type = PDF_SCHEMA_DATE;
else if (!strcmp(subtype, "N"))
p->entry.type = PDF_SCHEMA_NUMBER;
else if (!strcmp(subtype, "F"))
p->entry.type = PDF_SCHEMA_FILENAME;
else if (!strcmp(subtype, "Desc"))
p->entry.type = PDF_SCHEMA_DESC;
else if (!strcmp(subtype, "ModDate"))
p->entry.type = PDF_SCHEMA_MODDATE;
else if (!strcmp(subtype, "CreationDate"))
p->entry.type = PDF_SCHEMA_CREATIONDATE;
else if (!strcmp(subtype, "Size"))
p->entry.type = PDF_SCHEMA_SIZE;
else
p->entry.type = PDF_SCHEMA_UNKNOWN;
/* Now insert p */
pp = &doc->portfolio;
while (*pp && (*pp)->sort <= p->sort)
pp = &(*pp)->next;
p->next = *pp;
*pp = p;
}
}
static void
gatherforms(int page, pdf_obj *pageref, pdf_obj *pageobj, pdf_obj *dict)
{
int i, n;
n = pdf_dict_len(dict);
for (i = 0; i < n; i++)
{
pdf_obj *xobjdict;
pdf_obj *type;
pdf_obj *subtype;
pdf_obj *group;
pdf_obj *groupsubtype;
pdf_obj *reference;
int k;
xobjdict = pdf_dict_get_val(dict, i);
if (!pdf_is_dict(xobjdict))
{
fz_warn(ctx, "not a xobject dict (%d %d R)", pdf_to_num(xobjdict), pdf_to_gen(xobjdict));
continue;
}
type = pdf_dict_gets(xobjdict, "Subtype");
if (strcmp(pdf_to_name(type), "Form"))
continue;
subtype = pdf_dict_gets(xobjdict, "Subtype2");
if (!strcmp(pdf_to_name(subtype), "PS"))
continue;
group = pdf_dict_gets(xobjdict, "Group");
groupsubtype = pdf_dict_gets(group, "S");
reference = pdf_dict_gets(xobjdict, "Ref");
for (k = 0; k < forms; k++)
if (!pdf_objcmp(form[k].u.form.obj, xobjdict))
break;
if (k < forms)
continue;
form = fz_resize_array(ctx, form, forms+1, sizeof(struct info));
forms++;
form[forms - 1].page = page;
form[forms - 1].pageref = pageref;
form[forms - 1].pageobj = pageobj;
form[forms - 1].u.form.obj = xobjdict;
form[forms - 1].u.form.groupsubtype = groupsubtype;
form[forms - 1].u.form.reference = reference;
}
}
/*
** Merge srcdict into dstdict.
*/
static void wmupdf_dict_merge(fz_context *ctx,char *dictname,pdf_obj *dstdict,pdf_obj *srcdict)
{
int i,len;
/*
printf(" Merging %s dictionaries (%d <-- %d)\n",dictname,pdf_to_num(dstdict),pdf_to_num(srcdict));
*/
len=pdf_dict_len(srcdict);
for (i=0;i<len;i++)
{
pdf_obj *key,*value;
key=pdf_dict_get_key(srcdict,i);
value=pdf_dict_get_val(srcdict,i);
wmupdf_dict_merge_keyval(ctx,dstdict,key,value);
}
}
static void
gatherfonts(int page, pdf_obj *pageref, pdf_obj *pageobj, pdf_obj *dict)
{
int i, n;
n = pdf_dict_len(dict);
for (i = 0; i < n; i++)
{
pdf_obj *fontdict = NULL;
pdf_obj *subtype = NULL;
pdf_obj *basefont = NULL;
pdf_obj *name = NULL;
int k;
fontdict = pdf_dict_get_val(dict, i);
if (!pdf_is_dict(fontdict))
{
fz_warn(ctx, "not a font dict (%d %d R)", pdf_to_num(fontdict), pdf_to_gen(fontdict));
continue;
}
subtype = pdf_dict_gets(fontdict, "Subtype");
basefont = pdf_dict_gets(fontdict, "BaseFont");
if (!basefont || pdf_is_null(basefont))
name = pdf_dict_gets(fontdict, "Name");
for (k = 0; k < fonts; k++)
if (!pdf_objcmp(font[k].u.font.obj, fontdict))
break;
if (k < fonts)
continue;
font = fz_resize_array(ctx, font, fonts+1, sizeof(struct info));
fonts++;
font[fonts - 1].page = page;
font[fonts - 1].pageref = pageref;
font[fonts - 1].pageobj = pageobj;
font[fonts - 1].u.font.obj = fontdict;
font[fonts - 1].u.font.subtype = subtype;
font[fonts - 1].u.font.name = basefont ? basefont : name;
}
}
pdf_obj *
pdf_copy_dict(fz_context *ctx, pdf_obj *obj)
{
pdf_obj *dict;
int i, n;
RESOLVE(obj);
if (!obj)
return NULL; /* Can't warn :( */
if (obj->kind != PDF_DICT)
fz_warn(ctx, "assert: not a dict (%s)", pdf_objkindstr(obj));
n = pdf_dict_len(obj);
dict = pdf_new_dict(ctx, n);
for (i = 0; i < n; i++)
fz_dict_put(dict, pdf_dict_get_key(obj, i), pdf_dict_get_val(obj, i));
return dict;
}
static void
gathershadings(int page, pdf_obj *pageref, pdf_obj *pageobj, pdf_obj *dict)
{
int i, n;
n = pdf_dict_len(dict);
for (i = 0; i < n; i++)
{
pdf_obj *shade;
pdf_obj *type;
int k;
shade = pdf_dict_get_val(dict, i);
if (!pdf_is_dict(shade))
{
fz_warn(ctx, "not a shading dict (%d %d R)", pdf_to_num(shade), pdf_to_gen(shade));
continue;
}
type = pdf_dict_gets(shade, "ShadingType");
if (!pdf_is_int(type) || pdf_to_int(type) < 1 || pdf_to_int(type) > 7)
{
fz_warn(ctx, "not a shading type (%d %d R)", pdf_to_num(shade), pdf_to_gen(shade));
type = NULL;
}
for (k = 0; k < shadings; k++)
if (!pdf_objcmp(shading[k].u.shading.obj, shade))
break;
if (k < shadings)
continue;
shading = fz_resize_array(ctx, shading, shadings+1, sizeof(struct info));
shadings++;
shading[shadings - 1].page = page;
shading[shadings - 1].pageref = pageref;
shading[shadings - 1].pageobj = pageobj;
shading[shadings - 1].u.shading.obj = shade;
shading[shadings - 1].u.shading.type = type;
}
}
static void
gatherpsobjs(int page, pdf_obj *pageref, pdf_obj *pageobj, pdf_obj *dict)
{
int i, n;
n = pdf_dict_len(dict);
for (i = 0; i < n; i++)
{
pdf_obj *xobjdict;
pdf_obj *type;
pdf_obj *subtype;
int k;
xobjdict = pdf_dict_get_val(dict, i);
if (!pdf_is_dict(xobjdict))
{
fz_warn(ctx, "not a xobject dict (%d %d R)", pdf_to_num(xobjdict), pdf_to_gen(xobjdict));
continue;
}
type = pdf_dict_gets(xobjdict, "Subtype");
subtype = pdf_dict_gets(xobjdict, "Subtype2");
if (strcmp(pdf_to_name(type), "PS") &&
(strcmp(pdf_to_name(type), "Form") || strcmp(pdf_to_name(subtype), "PS")))
continue;
for (k = 0; k < psobjs; k++)
if (!pdf_objcmp(psobj[k].u.form.obj, xobjdict))
break;
if (k < psobjs)
continue;
psobj = fz_resize_array(ctx, psobj, psobjs+1, sizeof(struct info));
psobjs++;
psobj[psobjs - 1].page = page;
psobj[psobjs - 1].pageref = pageref;
psobj[psobjs - 1].pageobj = pageobj;
psobj[psobjs - 1].u.form.obj = xobjdict;
}
}
/*
Performs the same task as
pdf_clean_annot_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_annot_contents(fz_context *ctx, pdf_document *doc, pdf_annot *annot, fz_cookie *cookie,
pdf_page_contents_process_fn *proc, pdf_text_filter_fn *text_filter, pdf_after_text_object_fn *after_text, void *arg, int sanitize, int ascii)
{
pdf_obj *ap;
int i, n;
ap = pdf_dict_get(ctx, annot->obj, PDF_NAME(AP));
if (ap == NULL)
return;
n = pdf_dict_len(ctx, ap);
for (i = 0; i < n; i++)
{
pdf_obj *v = pdf_dict_get_val(ctx, ap, i);
if (v == NULL)
continue;
pdf_clean_stream_object(ctx, doc, v, NULL, cookie, 1, text_filter, after_text, arg, sanitize, ascii);
}
}
pdf_obj *
pdf_copy_dict(fz_context *ctx, pdf_obj *obj)
{
pdf_obj *dict;
int i, n;
RESOLVE(obj);
if (obj >= PDF_OBJ__LIMIT)
{
pdf_document *doc = DICT(obj)->doc;
if (obj->kind != PDF_DICT)
fz_warn(ctx, "assert: not a dict (%s)", pdf_objkindstr(obj));
n = pdf_dict_len(ctx, obj);
dict = pdf_new_dict(ctx, doc, n);
for (i = 0; i < n; i++)
pdf_dict_put(ctx, dict, pdf_dict_get_key(ctx, obj, i), pdf_dict_get_val(ctx, obj, i));
return dict;
}
return NULL; /* Can't warn :( */
}
/*
** From MuPDF pdfclean.c
*/
static void wmupdf_preserve_old_dests(pdf_obj *olddests,fz_context *ctx,pdf_document *xref,
pdf_obj *pages)
{
int i;
pdf_obj *names = pdf_new_dict(ctx,1);
pdf_obj *dests = pdf_new_dict(ctx,1);
pdf_obj *names_list = pdf_new_array(ctx,32);
int len = pdf_dict_len(olddests);
pdf_obj *root;
for (i=0;i<len;i++)
{
pdf_obj *key = pdf_dict_get_key(olddests,i);
pdf_obj *val = pdf_dict_get_val(olddests,i);
pdf_obj *key_str = pdf_new_string(ctx,pdf_to_name(key),strlen(pdf_to_name(key)));
pdf_obj *dest = pdf_dict_gets(val,"D");
dest = pdf_array_get(dest ? dest : val, 0);
if (pdf_array_contains(pdf_dict_gets(pages,"Kids"),dest))
{
pdf_array_push(names_list, key_str);
pdf_array_push(names_list, val);
}
pdf_drop_obj(key_str);
}
root = pdf_dict_gets(xref->trailer,"Root");
pdf_dict_puts(dests,"Names",names_list);
pdf_dict_puts(names,"Dests",dests);
pdf_dict_puts(root,"Names",names);
pdf_drop_obj(names);
pdf_drop_obj(dests);
pdf_drop_obj(names_list);
pdf_drop_obj(olddests);
}
static void
gatherpatterns(int page, pdf_obj *pageref, pdf_obj *pageobj, pdf_obj *dict)
{
int i, n;
n = pdf_dict_len(dict);
for (i = 0; i < n; i++)
{
pdf_obj *patterndict;
pdf_obj *type;
pdf_obj *paint = NULL;
pdf_obj *tiling = NULL;
pdf_obj *shading = NULL;
int k;
patterndict = pdf_dict_get_val(dict, i);
if (!pdf_is_dict(patterndict))
{
fz_warn(ctx, "not a pattern dict (%d %d R)", pdf_to_num(patterndict), pdf_to_gen(patterndict));
continue;
}
type = pdf_dict_gets(patterndict, "PatternType");
if (!pdf_is_int(type) || pdf_to_int(type) < 1 || pdf_to_int(type) > 2)
{
fz_warn(ctx, "not a pattern type (%d %d R)", pdf_to_num(patterndict), pdf_to_gen(patterndict));
type = NULL;
}
if (pdf_to_int(type) == 1)
{
paint = pdf_dict_gets(patterndict, "PaintType");
if (!pdf_is_int(paint) || pdf_to_int(paint) < 1 || pdf_to_int(paint) > 2)
{
fz_warn(ctx, "not a pattern paint type (%d %d R)", pdf_to_num(patterndict), pdf_to_gen(patterndict));
paint = NULL;
}
tiling = pdf_dict_gets(patterndict, "TilingType");
if (!pdf_is_int(tiling) || pdf_to_int(tiling) < 1 || pdf_to_int(tiling) > 3)
{
fz_warn(ctx, "not a pattern tiling type (%d %d R)", pdf_to_num(patterndict), pdf_to_gen(patterndict));
tiling = NULL;
}
}
else
{
shading = pdf_dict_gets(patterndict, "Shading");
}
for (k = 0; k < patterns; k++)
if (!pdf_objcmp(pattern[k].u.pattern.obj, patterndict))
break;
if (k < patterns)
continue;
pattern = fz_resize_array(ctx, pattern, patterns+1, sizeof(struct info));
patterns++;
pattern[patterns - 1].page = page;
pattern[patterns - 1].pageref = pageref;
pattern[patterns - 1].pageobj = pageobj;
pattern[patterns - 1].u.pattern.obj = patterndict;
pattern[patterns - 1].u.pattern.type = type;
pattern[patterns - 1].u.pattern.paint = paint;
pattern[patterns - 1].u.pattern.tiling = tiling;
pattern[patterns - 1].u.pattern.shading = shading;
}
}
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); */
}
static void
gatherimages(int page, pdf_obj *pageref, pdf_obj *pageobj, pdf_obj *dict)
{
int i, n;
n = pdf_dict_len(dict);
for (i = 0; i < n; i++)
{
pdf_obj *imagedict;
pdf_obj *type;
pdf_obj *width;
pdf_obj *height;
pdf_obj *bpc = NULL;
pdf_obj *filter = NULL;
pdf_obj *cs = NULL;
pdf_obj *altcs;
int k;
imagedict = pdf_dict_get_val(dict, i);
if (!pdf_is_dict(imagedict))
{
fz_warn(ctx, "not an image dict (%d %d R)", pdf_to_num(imagedict), pdf_to_gen(imagedict));
continue;
}
type = pdf_dict_gets(imagedict, "Subtype");
if (strcmp(pdf_to_name(type), "Image"))
continue;
filter = pdf_dict_gets(imagedict, "Filter");
altcs = NULL;
cs = pdf_dict_gets(imagedict, "ColorSpace");
if (pdf_is_array(cs))
{
pdf_obj *cses = cs;
cs = pdf_array_get(cses, 0);
if (pdf_is_name(cs) && (!strcmp(pdf_to_name(cs), "DeviceN") || !strcmp(pdf_to_name(cs), "Separation")))
{
altcs = pdf_array_get(cses, 2);
if (pdf_is_array(altcs))
altcs = pdf_array_get(altcs, 0);
}
}
width = pdf_dict_gets(imagedict, "Width");
height = pdf_dict_gets(imagedict, "Height");
bpc = pdf_dict_gets(imagedict, "BitsPerComponent");
for (k = 0; k < images; k++)
if (!pdf_objcmp(image[k].u.image.obj, imagedict))
break;
if (k < images)
continue;
image = fz_resize_array(ctx, image, images+1, sizeof(struct info));
images++;
image[images - 1].page = page;
image[images - 1].pageref = pageref;
image[images - 1].pageobj = pageobj;
image[images - 1].u.image.obj = imagedict;
image[images - 1].u.image.width = width;
image[images - 1].u.image.height = height;
image[images - 1].u.image.bpc = bpc;
image[images - 1].u.image.filter = filter;
image[images - 1].u.image.cs = cs;
image[images - 1].u.image.altcs = altcs;
}
}
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);
}
static void retainpages(int argc, char **argv)
{
pdf_obj *oldroot, *root, *pages, *kids, *countobj, *parent, *olddests;
/* Keep only pages/type and (reduced) dest entries to avoid
* references to unretained pages */
oldroot = pdf_dict_gets(xref->trailer, "Root");
pages = pdf_dict_gets(oldroot, "Pages");
olddests = pdf_load_name_tree(xref, "Dests");
root = pdf_new_dict(ctx, 2);
pdf_dict_puts(root, "Type", pdf_dict_gets(oldroot, "Type"));
pdf_dict_puts(root, "Pages", pdf_dict_gets(oldroot, "Pages"));
pdf_update_object(xref, pdf_to_num(oldroot), root);
pdf_drop_obj(root);
/* Create a new kids array with only the pages we want to keep */
parent = pdf_new_indirect(ctx, pdf_to_num(pages), pdf_to_gen(pages), xref);
kids = pdf_new_array(ctx, 1);
/* Retain pages specified */
while (argc - fz_optind)
{
int page, spage, epage, pagecount;
char *spec, *dash;
char *pagelist = argv[fz_optind];
pagecount = pdf_count_pages(xref);
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;
}
if (spage > epage)
page = spage, spage = epage, epage = page;
spage = fz_clampi(spage, 1, pagecount);
epage = fz_clampi(epage, 1, pagecount);
for (page = spage; page <= epage; page++)
{
pdf_obj *pageobj = xref->page_objs[page-1];
pdf_obj *pageref = xref->page_refs[page-1];
pdf_dict_puts(pageobj, "Parent", parent);
/* Store page object in new kids array */
pdf_array_push(kids, pageref);
}
spec = fz_strsep(&pagelist, ",");
}
fz_optind++;
}
pdf_drop_obj(parent);
/* Update page count and kids array */
countobj = pdf_new_int(ctx, pdf_array_len(kids));
pdf_dict_puts(pages, "Count", countobj);
pdf_drop_obj(countobj);
pdf_dict_puts(pages, "Kids", kids);
pdf_drop_obj(kids);
/* Also preserve the (partial) Dests name tree */
if (olddests)
{
int i;
pdf_obj *names = pdf_new_dict(ctx, 1);
pdf_obj *dests = pdf_new_dict(ctx, 1);
pdf_obj *names_list = pdf_new_array(ctx, 32);
int len = pdf_dict_len(olddests);
for (i = 0; i < len; i++)
{
pdf_obj *key = pdf_dict_get_key(olddests, i);
pdf_obj *val = pdf_dict_get_val(olddests, i);
pdf_obj *key_str = pdf_new_string(ctx, pdf_to_name(key), strlen(pdf_to_name(key)));
pdf_obj *dest = pdf_dict_gets(val, "D");
dest = pdf_array_get(dest ? dest : val, 0);
if (pdf_array_contains(pdf_dict_gets(pages, "Kids"), dest))
{
pdf_array_push(names_list, key_str);
pdf_array_push(names_list, val);
}
pdf_drop_obj(key_str);
}
root = pdf_dict_gets(xref->trailer, "Root");
pdf_dict_puts(dests, "Names", names_list);
pdf_dict_puts(names, "Dests", dests);
pdf_dict_puts(root, "Names", names);
pdf_drop_obj(names);
pdf_drop_obj(dests);
pdf_drop_obj(names_list);
pdf_drop_obj(olddests);
}
}
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 void
gatherresourceinfo(int page, pdf_obj *rsrc)
{
pdf_obj *pageobj;
pdf_obj *pageref;
pdf_obj *font;
pdf_obj *xobj;
pdf_obj *shade;
pdf_obj *pattern;
pdf_obj *subrsrc;
int i;
pageobj = xref->page_objs[page-1];
pageref = xref->page_refs[page-1];
if (!pageobj)
fz_throw(ctx, "cannot retrieve info from page %d", page);
font = pdf_dict_gets(rsrc, "Font");
if (font)
{
int n;
gatherfonts(page, pageref, pageobj, font);
n = pdf_dict_len(font);
for (i = 0; i < n; i++)
{
pdf_obj *obj = pdf_dict_get_val(font, i);
subrsrc = pdf_dict_gets(obj, "Resources");
if (subrsrc && pdf_objcmp(rsrc, subrsrc))
gatherresourceinfo(page, subrsrc);
}
}
xobj = pdf_dict_gets(rsrc, "XObject");
if (xobj)
{
int n;
gatherimages(page, pageref, pageobj, xobj);
gatherforms(page, pageref, pageobj, xobj);
gatherpsobjs(page, pageref, pageobj, xobj);
n = pdf_dict_len(xobj);
for (i = 0; i < n; i++)
{
pdf_obj *obj = pdf_dict_get_val(xobj, i);
subrsrc = pdf_dict_gets(obj, "Resources");
if (subrsrc && pdf_objcmp(rsrc, subrsrc))
gatherresourceinfo(page, subrsrc);
}
}
shade = pdf_dict_gets(rsrc, "Shading");
if (shade)
gathershadings(page, pageref, pageobj, shade);
pattern = pdf_dict_gets(rsrc, "Pattern");
if (pattern)
{
int n;
gatherpatterns(page, pageref, pageobj, pattern);
n = pdf_dict_len(pattern);
for (i = 0; i < n; i++)
{
pdf_obj *obj = pdf_dict_get_val(pattern, i);
subrsrc = pdf_dict_gets(obj, "Resources");
if (subrsrc && pdf_objcmp(rsrc, subrsrc))
gatherresourceinfo(page, subrsrc);
}
}
}
// just copy one of the resource sub-entries (e.g. /Font)
static int copy_and_rename_resource(fz_context *dest_ctx, pdf_obj *dest,
fz_context *src_ctx, pdf_obj *src, char *prefix, struct put_info *info)
{
char new_name[64]; /* this buffer is big enough up to hold all digits for two 16-bit numbers */
int i;
for(i = 0; i < pdf_dict_len(src_ctx, src); i++) {
pdf_obj *src_key = pdf_dict_get_key(src_ctx, src, i);
pdf_obj *src_val = pdf_dict_get_val(src_ctx, src, i);
if(!pdf_is_name(src_ctx, src_key)) {
return(2);
}
/* if this is an inline resource, just copy this object into the new
resource dict */
if(!pdf_is_indirect(src_ctx, src_val)) {
if(snprintf(new_name, sizeof(new_name) / sizeof(new_name[0]),
"%sinline_%d", prefix, info->next_inline_id++) >= sizeof(new_name) / sizeof(new_name[0]))
return(1); // not enough space
pdf_obj *new_res = copy_unassigned_object_continue(dest_ctx,
info->dest_doc, src_ctx, info->src_doc, src_val, &info->new_ids);
//pdf_obj *new_res = pdf_new_dict(dest_ctx, info->dest_doc, 10);
printf("dump it...\n");
pdf_fprint_obj(dest_ctx, stdout, new_res, 0);
/* now reference this new object in the resource object of this sheet */
pdf_obj *dest_key = pdf_new_name(dest_ctx, info->dest_doc, new_name);
pdf_dict_put(dest_ctx, dest, dest_key, new_res);
pdf_drop_obj(dest_ctx, dest_key);
pdf_drop_obj(dest_ctx, new_res);
} else {
/* The new name of resource objects is always the num/gen of the
referenced object in the src-file. Thus we can check by that name
if the object was already referenced by another page of this sheet. */
if(snprintf(new_name, sizeof(new_name) / sizeof(new_name[0]),
"%s%d_%d", prefix, pdf_to_num(dest_ctx, src_val), pdf_to_gen(dest_ctx, src_val)) >=
sizeof(new_name) / sizeof(new_name[0]))
return(1); // not enough space
if(pdf_dict_gets(dest_ctx, dest, new_name) == NULL) {
/* if this resource is not inlined and not already in the resource-dict
of the current sheet... */
/* ...copy the referenced resource to the new document!
If this object has copied already (for another sheet in dest_doc),
copy_object_continue() will do nothing */
pdf_obj *new_res = copy_object_continue(dest_ctx, info->dest_doc,
src_ctx, info->src_doc, src_val, &info->new_ids);
/* now reference this new object in the resource object of this sheet */
pdf_obj *dest_key = pdf_new_name(dest_ctx, info->dest_doc, new_name);
pdf_dict_put(dest_ctx, dest, dest_key, new_res);
pdf_drop_obj(dest_ctx, dest_key);
pdf_drop_obj(dest_ctx, new_res);
}
}
/* even if it was used on another sheet or on this sheet, add it to the
rename-dict for this sheet! Because it could have different names
on different source-pages */
pdf_obj *rename_key = pdf_new_name(dest_ctx, info->dest_doc, pdf_to_name(dest_ctx, src_key));
pdf_obj *rename_val = pdf_new_name(dest_ctx, info->dest_doc, new_name);
pdf_dict_put(dest_ctx, info->rename_dict, rename_key, rename_val);
pdf_drop_obj(dest_ctx, rename_key);
pdf_drop_obj(dest_ctx, rename_val);
}
return(0);
}
static void
gatherresourceinfo(int page, pdf_obj *rsrc, int show)
{
pdf_obj *pageobj;
pdf_obj *pageref;
pdf_obj *font;
pdf_obj *xobj;
pdf_obj *shade;
pdf_obj *pattern;
pdf_obj *subrsrc;
int i;
pageref = pdf_lookup_page_obj(doc, page-1);
pageobj = pdf_resolve_indirect(pageref);
if (!pageobj)
fz_throw(ctx, FZ_ERROR_GENERIC, "cannot retrieve info from page %d", page);
font = pdf_dict_gets(rsrc, "Font");
if (show & FONTS && font)
{
int n;
gatherfonts(page, pageref, pageobj, font);
n = pdf_dict_len(font);
for (i = 0; i < n; i++)
{
pdf_obj *obj = pdf_dict_get_val(font, i);
subrsrc = pdf_dict_gets(obj, "Resources");
if (subrsrc && pdf_objcmp(rsrc, subrsrc))
gatherresourceinfo(page, subrsrc, show);
}
}
xobj = pdf_dict_gets(rsrc, "XObject");
if (show & XOBJS && xobj)
{
int n;
gatherimages(page, pageref, pageobj, xobj);
gatherforms(page, pageref, pageobj, xobj);
gatherpsobjs(page, pageref, pageobj, xobj);
n = pdf_dict_len(xobj);
for (i = 0; i < n; i++)
{
pdf_obj *obj = pdf_dict_get_val(xobj, i);
subrsrc = pdf_dict_gets(obj, "Resources");
if (subrsrc && pdf_objcmp(rsrc, subrsrc))
gatherresourceinfo(page, subrsrc, show);
}
}
shade = pdf_dict_gets(rsrc, "Shading");
if (show & SHADINGS && shade)
gathershadings(page, pageref, pageobj, shade);
pattern = pdf_dict_gets(rsrc, "Pattern");
if (show & PATTERNS && pattern)
{
int n;
gatherpatterns(page, pageref, pageobj, pattern);
n = pdf_dict_len(pattern);
for (i = 0; i < n; i++)
{
pdf_obj *obj = pdf_dict_get_val(pattern, i);
subrsrc = pdf_dict_gets(obj, "Resources");
if (subrsrc && pdf_objcmp(rsrc, subrsrc))
gatherresourceinfo(page, subrsrc, show);
}
}
}
/*
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);
}
}
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;
int pagecount;
int i;
/* 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);
root = pdf_new_dict(ctx, doc, 2);
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_update_object(ctx, doc, pdf_to_num(ctx, oldroot), root);
pdf_drop_obj(ctx, 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);
/* Also preserve the (partial) Dests name tree */
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 (pdf_array_contains(ctx, pdf_dict_get(ctx, pages, PDF_NAME_Kids), dest))
{
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);
}
}
root = pdf_dict_get(ctx, pdf_trailer(ctx, doc), PDF_NAME_Root);
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, names_list);
pdf_drop_obj(ctx, olddests);
}
/* Force the next call to pdf_count_pages to recount */
glo->doc->page_count = 0;
/* Edit each pages /Annot list to remove any links that point to
* nowhere. */
pagecount = pdf_count_pages(ctx, doc);
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);
pdf_obj *p;
if (!pdf_name_eq(ctx, pdf_dict_get(ctx, o, PDF_NAME_Subtype), PDF_NAME_Link))
continue;
p = pdf_dict_get(ctx, o, PDF_NAME_A);
if (!pdf_name_eq(ctx, pdf_dict_get(ctx, p, PDF_NAME_S), PDF_NAME_GoTo))
continue;
if (string_in_names_list(ctx, pdf_dict_get(ctx, p, PDF_NAME_D), names_list))
continue;
/* FIXME: Should probably look at Next too */
/* Remove this annotation */
pdf_array_delete(ctx, annots, j);
j--;
}
}
}
static void
pdf_clean_type3(fz_context *ctx, pdf_document *doc, pdf_obj *obj, pdf_obj *orig_res, fz_cookie *cookie, int sanitize, int ascii)
{
pdf_processor *proc_buffer = NULL;
pdf_processor *proc_filter = NULL;
pdf_obj *res = NULL;
pdf_obj *ref;
pdf_obj *charprocs;
int i, l;
fz_var(res);
fz_var(proc_buffer);
fz_var(proc_filter);
fz_try(ctx)
{
res = pdf_dict_get(ctx, obj, PDF_NAME(Resources));
if (res)
orig_res = res;
res = NULL;
res = pdf_new_dict(ctx, doc, 1);
charprocs = pdf_dict_get(ctx, obj, PDF_NAME(CharProcs));
l = pdf_dict_len(ctx, charprocs);
for (i = 0; i < l; i++)
{
pdf_obj *val = pdf_dict_get_val(ctx, charprocs, i);
fz_buffer *buffer = fz_new_buffer(ctx, 1024);
fz_try(ctx)
{
proc_buffer = pdf_new_buffer_processor(ctx, buffer, ascii);
if (sanitize)
{
proc_filter = pdf_new_filter_processor(ctx, doc, proc_buffer, orig_res, res);
pdf_process_contents(ctx, proc_filter, doc, orig_res, val, cookie);
pdf_close_processor(ctx, proc_filter);
}
else
{
pdf_process_contents(ctx, proc_filter, doc, orig_res, val, cookie);
}
pdf_close_processor(ctx, proc_buffer);
pdf_update_stream(ctx, doc, val, buffer, 0);
}
fz_always(ctx)
{
pdf_drop_processor(ctx, proc_filter);
pdf_drop_processor(ctx, proc_buffer);
fz_drop_buffer(ctx, buffer);
}
fz_catch(ctx)
{
fz_rethrow(ctx);
}
}
/* ProcSet - no cleaning possible. Inherit this from the old dict. */
pdf_dict_put(ctx, res, PDF_NAME(ProcSet), pdf_dict_get(ctx, orig_res, PDF_NAME(ProcSet)));
ref = pdf_add_object(ctx, doc, res);
pdf_dict_put_drop(ctx, obj, PDF_NAME(Resources), ref);
}
fz_always(ctx)
{
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;
}
