/*
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);
}
}
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");
}
}
