/*
* Create a filter given a name and param dictionary.
*/
static fz_stream *
build_filter(fz_context *ctx, fz_stream *chain, pdf_document *doc, pdf_obj *f, pdf_obj *p, int num, int gen, fz_compression_params *params)
{
fz_compression_params local_params;
if (params == NULL)
params = &local_params;
build_compression_params(ctx, f, p, params);
/* If we were using params we were passed in, and we successfully
* recognised the image type, we can use the existing filter and
* shortstop here. */
if (params != &local_params && params->type != FZ_IMAGE_RAW)
return chain;
if (params->type != FZ_IMAGE_RAW)
return fz_open_image_decomp_stream(ctx, chain, params, NULL);
if (pdf_name_eq(ctx, f, PDF_NAME_ASCIIHexDecode) || pdf_name_eq(ctx, f, PDF_NAME_AHx))
return fz_open_ahxd(ctx, chain);
else if (pdf_name_eq(ctx, f, PDF_NAME_ASCII85Decode) || pdf_name_eq(ctx, f, PDF_NAME_A85))
return fz_open_a85d(ctx, chain);
else if (pdf_name_eq(ctx, f, PDF_NAME_JBIG2Decode))
{
fz_jbig2_globals *globals = NULL;
pdf_obj *obj = pdf_dict_get(ctx, p, PDF_NAME_JBIG2Globals);
if (pdf_is_indirect(ctx, obj))
globals = pdf_load_jbig2_globals(ctx, doc, obj);
/* fz_open_jbig2d takes possession of globals */
return fz_open_jbig2d(ctx, chain, globals);
}
else if (pdf_name_eq(ctx, f, PDF_NAME_JPXDecode))
return chain; /* JPX decoding is special cased in the image loading code */
else if (pdf_name_eq(ctx, f, PDF_NAME_Crypt))
{
pdf_obj *name;
if (!doc->crypt)
{
fz_warn(ctx, "crypt filter in unencrypted document");
return chain;
}
name = pdf_dict_get(ctx, p, PDF_NAME_Name);
if (pdf_is_name(ctx, name))
return pdf_open_crypt_with_filter(ctx, chain, doc->crypt, name, num, gen);
return chain;
}
fz_warn(ctx, "unknown filter name (%s)", pdf_to_name(ctx, f));
return chain;
}
static void
pdf_write_widget_appearance(fz_context *ctx, pdf_annot *annot, fz_buffer *buf,
fz_rect *rect, fz_rect *bbox, fz_matrix *matrix, pdf_obj **res)
{
pdf_obj *ft = pdf_dict_get_inheritable(ctx, annot->obj, PDF_NAME(FT));
if (pdf_name_eq(ctx, ft, PDF_NAME(Tx)))
{
int ff = pdf_field_flags(ctx, annot->obj);
char *format = NULL;
const char *text = NULL;
if (!annot->ignore_trigger_events)
{
format = pdf_field_event_format(ctx, annot->page->doc, annot->obj);
if (format)
text = format;
else
text = pdf_field_value(ctx, annot->obj);
}
else
{
text = pdf_field_value(ctx, annot->obj);
}
fz_try(ctx)
pdf_write_tx_widget_appearance(ctx, annot, buf, rect, bbox, matrix, res, text, ff);
fz_always(ctx)
fz_free(ctx, format);
fz_catch(ctx)
fz_rethrow(ctx);
}
else if (pdf_name_eq(ctx, ft, PDF_NAME(Ch)))
{
pdf_write_ch_widget_appearance(ctx, annot, buf, rect, bbox, matrix, res);
}
else if (pdf_name_eq(ctx, ft, PDF_NAME(Sig)))
{
pdf_write_sig_widget_appearance(ctx, annot, buf, rect, bbox, matrix, res);
}
else
{
fz_throw(ctx, FZ_ERROR_GENERIC, "cannot create appearance stream for %s widgets", pdf_to_name(ctx, ft));
}
}
static int is_allowed_subtype(fz_context *ctx, pdf_annot *annot, pdf_obj *property, pdf_obj **allowed)
{
pdf_obj *subtype = pdf_dict_get(ctx, annot->obj, PDF_NAME(Subtype));
while (*allowed) {
if (pdf_name_eq(ctx, subtype, *allowed))
return 1;
allowed++;
}
return 0;
}
static void check_allowed_subtypes(fz_context *ctx, pdf_annot *annot, pdf_obj *property, pdf_obj **allowed)
{
pdf_obj *subtype = pdf_dict_get(ctx, annot->obj, PDF_NAME_Subtype);
while (allowed) {
if (pdf_name_eq(ctx, subtype, *allowed))
return;
allowed++;
}
fz_throw(ctx, FZ_ERROR_GENERIC, "%s annotations have no %s property", pdf_to_name(ctx, subtype), pdf_to_name(ctx, property));
}
const char *
pdf_annot_icon_name(fz_context *ctx, pdf_annot *annot)
{
pdf_obj *name;
check_allowed_subtypes(ctx, annot, PDF_NAME(Name), icon_name_subtypes);
name = pdf_dict_get(ctx, annot->obj, PDF_NAME(Name));
if (!name)
{
pdf_obj *subtype = pdf_dict_get(ctx, annot->obj, PDF_NAME(Subtype));
if (pdf_name_eq(ctx, subtype, PDF_NAME(Text)))
return "Note";
if (pdf_name_eq(ctx, subtype, PDF_NAME(Stamp)))
return "Draft";
if (pdf_name_eq(ctx, subtype, PDF_NAME(FileAttachment)))
return "PushPin";
if (pdf_name_eq(ctx, subtype, PDF_NAME(Sound)))
return "Speaker";
}
return pdf_to_name(ctx, name);
}
/* Image specific methods */
static void
res_image_init(fz_context *ctx, pdf_document *doc, pdf_res_table *table)
{
int len, k;
pdf_obj *obj;
pdf_obj *type;
pdf_obj *res = NULL;
fz_image *image = NULL;
unsigned char digest[16];
fz_var(obj);
fz_var(image);
fz_var(res);
fz_try(ctx)
{
table->hash = fz_new_hash_table(ctx, 4096, 16, -1);
len = pdf_count_objects(ctx, doc);
for (k = 1; k < len; k++)
{
obj = pdf_load_object(ctx, doc, k, 0);
type = pdf_dict_get(ctx, obj, PDF_NAME_Subtype);
if (pdf_name_eq(ctx, type, PDF_NAME_Image))
{
image = pdf_load_image(ctx, doc, obj);
res_image_get_md5(ctx, image, digest);
fz_drop_image(ctx, image);
image = NULL;
/* Don't allow overwrites. */
if (fz_hash_find(ctx, table->hash, digest) == NULL)
fz_hash_insert(ctx, table->hash, digest, obj);
}
else
{
pdf_drop_obj(ctx, obj);
}
obj = NULL;
}
}
fz_always(ctx)
{
fz_drop_image(ctx, image);
pdf_drop_obj(ctx, obj);
}
fz_catch(ctx)
{
res_table_free(ctx, table);
fz_rethrow(ctx);
}
}
char *
pdf_parse_file_spec(fz_context *ctx, pdf_document *doc, pdf_obj *file_spec, pdf_obj *dest)
{
pdf_obj *filename = NULL;
const char *path;
char *uri;
char frag[256];
if (pdf_is_string(ctx, file_spec))
filename = file_spec;
if (pdf_is_dict(ctx, file_spec)) {
#ifdef _WIN32
filename = pdf_dict_get(ctx, file_spec, PDF_NAME(DOS));
#else
filename = pdf_dict_get(ctx, file_spec, PDF_NAME(Unix));
#endif
if (!filename)
filename = pdf_dict_geta(ctx, file_spec, PDF_NAME(UF), PDF_NAME(F));
}
if (!pdf_is_string(ctx, filename))
{
fz_warn(ctx, "cannot parse file specification");
return NULL;
}
if (pdf_is_array(ctx, dest))
fz_snprintf(frag, sizeof frag, "#page=%d", pdf_array_get_int(ctx, dest, 0) + 1);
else if (pdf_is_name(ctx, dest))
fz_snprintf(frag, sizeof frag, "#%s", pdf_to_name(ctx, dest));
else if (pdf_is_string(ctx, dest))
fz_snprintf(frag, sizeof frag, "#%s", pdf_to_str_buf(ctx, dest));
else
frag[0] = 0;
path = pdf_to_text_string(ctx, filename);
uri = NULL;
#ifdef _WIN32
if (!pdf_name_eq(ctx, pdf_dict_get(ctx, file_spec, PDF_NAME(FS)), PDF_NAME(URL)))
{
/* Fix up the drive letter (change "/C/Documents/Foo" to "C:/Documents/Foo") */
if (path[0] == '/' && (('A' <= path[1] && path[1] <= 'Z') || ('a' <= path[1] && path[1] <= 'z')) && path[2] == '/')
uri = fz_asprintf(ctx, "file://%c:%s%s", path[1], path+2, frag);
}
#endif
if (!uri)
uri = fz_asprintf(ctx, "file://%s%s", path, frag);
return uri;
}
static void execute_action(fz_context *ctx, pdf_document *doc, pdf_obj *obj, pdf_obj *a)
{
if (a)
{
pdf_obj *type = pdf_dict_get(ctx, a, PDF_NAME_S);
if (pdf_name_eq(ctx, type, PDF_NAME_JavaScript))
{
pdf_obj *js = pdf_dict_get(ctx, a, PDF_NAME_JS);
if (js)
{
char *code = pdf_to_utf8(ctx, doc, js);
fz_try(ctx)
{
pdf_js_execute(doc->js, code);
}
fz_always(ctx)
{
fz_free(ctx, code);
}
fz_catch(ctx)
{
fz_rethrow(ctx);
}
}
}
else if (pdf_name_eq(ctx, type, PDF_NAME_ResetForm))
{
reset_form(ctx, doc, pdf_dict_get(ctx, a, PDF_NAME_Fields), pdf_to_int(ctx, pdf_dict_get(ctx, a, PDF_NAME_Flags)) & 1);
}
else if (pdf_name_eq(ctx, type, PDF_NAME_Named))
{
pdf_obj *name = pdf_dict_get(ctx, a, PDF_NAME_N);
if (pdf_name_eq(ctx, name, PDF_NAME_Print))
pdf_event_issue_print(ctx, doc);
}
}
static fz_link *
pdf_load_link(fz_context *ctx, pdf_document *doc, pdf_obj *dict, int pagenum, const fz_matrix *page_ctm)
{
pdf_obj *action;
pdf_obj *obj;
fz_rect bbox;
char *uri;
fz_link *link = NULL;
obj = pdf_dict_get(ctx, dict, PDF_NAME_Subtype);
if (!pdf_name_eq(ctx, obj, PDF_NAME_Link))
return NULL;
obj = pdf_dict_get(ctx, dict, PDF_NAME_Rect);
if (!obj)
return NULL;
pdf_to_rect(ctx, obj, &bbox);
fz_transform_rect(&bbox, page_ctm);
obj = pdf_dict_get(ctx, dict, PDF_NAME_Dest);
if (obj)
uri = pdf_parse_link_dest(ctx, doc, obj);
else
{
action = pdf_dict_get(ctx, dict, PDF_NAME_A);
/* fall back to additional action button's down/up action */
if (!action)
action = pdf_dict_geta(ctx, pdf_dict_get(ctx, dict, PDF_NAME_AA), PDF_NAME_U, PDF_NAME_D);
uri = pdf_parse_link_action(ctx, doc, action, pagenum);
}
if (!uri)
return NULL;
fz_try(ctx)
link = fz_new_link(ctx, &bbox, doc, uri);
fz_always(ctx)
fz_free(ctx, uri);
fz_catch(ctx)
fz_rethrow(ctx);
return link;
}
static int dest_is_valid(fz_context *ctx, pdf_obj *o, int page_count, int *page_object_nums, pdf_obj *names_list)
{
pdf_obj *p;
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) &&
!string_in_names_list(ctx, pdf_dict_get(ctx, p, PDF_NAME_D), names_list))
return 0;
p = pdf_dict_get(ctx, o, PDF_NAME_Dest);
if (p == NULL)
{}
else if (pdf_is_string(ctx, p))
{
return string_in_names_list(ctx, p, names_list);
}
else if (!dest_is_valid_page(ctx, pdf_array_get(ctx, p, 0), page_object_nums, page_count))
return 0;
return 1;
}
/*
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 int isfontdesc(pdf_obj *obj)
{
pdf_obj *type = pdf_dict_get(ctx, obj, PDF_NAME_Type);
return pdf_name_eq(ctx, type, PDF_NAME_FontDescriptor);
}
static void savefont(pdf_obj *dict, int num)
{
char namebuf[1024];
fz_buffer *buf;
pdf_obj *stream = NULL;
pdf_obj *obj;
char *ext = "";
fz_output *out;
char *fontname = "font";
size_t len;
unsigned char *data;
obj = pdf_dict_get(ctx, dict, PDF_NAME_FontName);
if (obj)
fontname = pdf_to_name(ctx, obj);
obj = pdf_dict_get(ctx, dict, PDF_NAME_FontFile);
if (obj)
{
stream = obj;
ext = "pfa";
}
obj = pdf_dict_get(ctx, dict, PDF_NAME_FontFile2);
if (obj)
{
stream = obj;
ext = "ttf";
}
obj = pdf_dict_get(ctx, dict, PDF_NAME_FontFile3);
if (obj)
{
stream = obj;
obj = pdf_dict_get(ctx, obj, PDF_NAME_Subtype);
if (obj && !pdf_is_name(ctx, obj))
fz_throw(ctx, FZ_ERROR_GENERIC, "invalid font descriptor subtype");
if (pdf_name_eq(ctx, obj, PDF_NAME_Type1C))
ext = "cff";
else if (pdf_name_eq(ctx, obj, PDF_NAME_CIDFontType0C))
ext = "cid";
else if (pdf_name_eq(ctx, obj, PDF_NAME_OpenType))
ext = "otf";
else
fz_throw(ctx, FZ_ERROR_GENERIC, "unhandled font type '%s'", pdf_to_name(ctx, obj));
}
if (!stream)
{
fz_warn(ctx, "unhandled font type");
return;
}
buf = pdf_load_stream(ctx, doc, pdf_to_num(ctx, stream));
len = fz_buffer_storage(ctx, buf, &data);
fz_try(ctx)
{
snprintf(namebuf, sizeof(namebuf), "%s-%04d.%s", fontname, num, ext);
printf("extracting font %s\n", namebuf);
out = fz_new_output_with_path(ctx, namebuf, 0);
fz_try(ctx)
fz_write(ctx, out, data, len);
fz_always(ctx)
fz_drop_output(ctx, out);
fz_catch(ctx)
fz_rethrow(ctx);
}
fz_always(ctx)
fz_drop_buffer(ctx, buf);
fz_catch(ctx)
fz_rethrow(ctx);
}
static void
pdf_process_extgstate(fz_context *ctx, pdf_processor *proc, pdf_csi *csi, pdf_obj *dict)
{
pdf_obj *obj;
obj = pdf_dict_get(ctx, dict, PDF_NAME_LW);
if (pdf_is_number(ctx, obj) && proc->op_w)
proc->op_w(ctx, proc, pdf_to_real(ctx, obj));
obj = pdf_dict_get(ctx, dict, PDF_NAME_LC);
if (pdf_is_int(ctx, obj) && proc->op_J)
proc->op_J(ctx, proc, fz_clampi(pdf_to_int(ctx, obj), 0, 2));
obj = pdf_dict_get(ctx, dict, PDF_NAME_LJ);
if (pdf_is_int(ctx, obj) && proc->op_j)
proc->op_j(ctx, proc, fz_clampi(pdf_to_int(ctx, obj), 0, 2));
obj = pdf_dict_get(ctx, dict, PDF_NAME_ML);
if (pdf_is_number(ctx, obj) && proc->op_M)
proc->op_M(ctx, proc, pdf_to_real(ctx, obj));
obj = pdf_dict_get(ctx, dict, PDF_NAME_D);
if (pdf_is_array(ctx, obj) && proc->op_d)
{
pdf_obj *dash_array = pdf_array_get(ctx, obj, 0);
pdf_obj *dash_phase = pdf_array_get(ctx, obj, 1);
proc->op_d(ctx, proc, dash_array, pdf_to_real(ctx, dash_phase));
}
obj = pdf_dict_get(ctx, dict, PDF_NAME_RI);
if (pdf_is_name(ctx, obj) && proc->op_ri)
proc->op_ri(ctx, proc, pdf_to_name(ctx, obj));
obj = pdf_dict_get(ctx, dict, PDF_NAME_FL);
if (pdf_is_number(ctx, obj) && proc->op_i)
proc->op_i(ctx, proc, pdf_to_real(ctx, obj));
obj = pdf_dict_get(ctx, dict, PDF_NAME_Font);
if (pdf_is_array(ctx, obj) && proc->op_Tf)
{
pdf_obj *font_ref = pdf_array_get(ctx, obj, 0);
pdf_obj *font_size = pdf_array_get(ctx, obj, 1);
pdf_font_desc *font = load_font_or_hail_mary(ctx, csi->doc, csi->rdb, font_ref, 0, csi->cookie);
fz_try(ctx)
proc->op_Tf(ctx, proc, "ExtGState", font, pdf_to_real(ctx, font_size));
fz_always(ctx)
pdf_drop_font(ctx, font);
fz_catch(ctx)
fz_rethrow(ctx);
}
/* transfer functions */
obj = pdf_dict_get(ctx, dict, PDF_NAME_TR2);
if (pdf_is_name(ctx, obj))
if (!pdf_name_eq(ctx, obj, PDF_NAME_Identity) && !pdf_name_eq(ctx, obj, PDF_NAME_Default))
fz_warn(ctx, "ignoring transfer function");
if (!obj) /* TR is ignored in the presence of TR2 */
{
pdf_obj *tr = pdf_dict_get(ctx, dict, PDF_NAME_TR);
if (pdf_is_name(ctx, tr))
if (!pdf_name_eq(ctx, tr, PDF_NAME_Identity))
fz_warn(ctx, "ignoring transfer function");
}
/* transparency state */
obj = pdf_dict_get(ctx, dict, PDF_NAME_CA);
if (pdf_is_number(ctx, obj) && proc->op_gs_CA)
proc->op_gs_CA(ctx, proc, pdf_to_real(ctx, obj));
obj = pdf_dict_get(ctx, dict, PDF_NAME_ca);
if (pdf_is_number(ctx, obj) && proc->op_gs_ca)
proc->op_gs_ca(ctx, proc, pdf_to_real(ctx, obj));
obj = pdf_dict_get(ctx, dict, PDF_NAME_BM);
if (pdf_is_array(ctx, obj))
obj = pdf_array_get(ctx, obj, 0);
if (pdf_is_name(ctx, obj) && proc->op_gs_BM)
proc->op_gs_BM(ctx, proc, pdf_to_name(ctx, obj));
obj = pdf_dict_get(ctx, dict, PDF_NAME_SMask);
if (proc->op_gs_SMask)
{
if (pdf_is_dict(ctx, obj))
{
pdf_xobject *xobj;
pdf_obj *group, *s, *bc, *tr;
float softmask_bc[FZ_MAX_COLORS];
fz_colorspace *colorspace;
int colorspace_n = 1;
int k, luminosity;
fz_var(xobj);
group = pdf_dict_get(ctx, obj, PDF_NAME_G);
if (!group)
fz_throw(ctx, FZ_ERROR_GENERIC, "cannot load softmask xobject (%d 0 R)", pdf_to_num(ctx, obj));
xobj = pdf_load_xobject(ctx, csi->doc, group);
fz_try(ctx)
{
colorspace = pdf_xobject_colorspace(ctx, xobj);
if (colorspace)
{
colorspace_n = fz_colorspace_n(ctx, colorspace);
fz_drop_colorspace(ctx, colorspace);
}
/* Default background color is black. */
for (k = 0; k < colorspace_n; k++)
softmask_bc[k] = 0;
/* Which in CMYK means not all zeros! This should really be
* a test for subtractive color spaces, but this will have
* to do for now. */
if (colorspace == fz_device_cmyk(ctx))
softmask_bc[3] = 1.0;
bc = pdf_dict_get(ctx, obj, PDF_NAME_BC);
if (pdf_is_array(ctx, bc))
{
for (k = 0; k < colorspace_n; k++)
softmask_bc[k] = pdf_to_real(ctx, pdf_array_get(ctx, bc, k));
}
s = pdf_dict_get(ctx, obj, PDF_NAME_S);
if (pdf_name_eq(ctx, s, PDF_NAME_Luminosity))
luminosity = 1;
else
luminosity = 0;
tr = pdf_dict_get(ctx, obj, PDF_NAME_TR);
if (tr && !pdf_name_eq(ctx, tr, PDF_NAME_Identity))
fz_warn(ctx, "ignoring transfer function");
proc->op_gs_SMask(ctx, proc, xobj, csi->rdb, softmask_bc, luminosity);
}
fz_always(ctx)
{
pdf_drop_xobject(ctx, xobj);
}
fz_catch(ctx)
{
fz_rethrow(ctx);
}
}
else if (pdf_is_name(ctx, obj) && pdf_name_eq(ctx, obj, PDF_NAME_None))
{
proc->op_gs_SMask(ctx, proc, NULL, NULL, NULL, 0);
}
}
static int isimage(pdf_obj *obj)
{
pdf_obj *type = pdf_dict_get(ctx, obj, PDF_NAME_Subtype);
return pdf_name_eq(ctx, type, PDF_NAME_Image);
}
static void savefont(pdf_obj *dict, int num)
{
char namebuf[1024];
fz_buffer *buf;
pdf_obj *stream = NULL;
pdf_obj *obj;
char *ext = "";
FILE *f;
char *fontname = "font";
int n, len;
unsigned char *data;
obj = pdf_dict_get(ctx, dict, PDF_NAME_FontName);
if (obj)
fontname = pdf_to_name(ctx, obj);
obj = pdf_dict_get(ctx, dict, PDF_NAME_FontFile);
if (obj)
{
stream = obj;
ext = "pfa";
}
obj = pdf_dict_get(ctx, dict, PDF_NAME_FontFile2);
if (obj)
{
stream = obj;
ext = "ttf";
}
obj = pdf_dict_get(ctx, dict, PDF_NAME_FontFile3);
if (obj)
{
stream = obj;
obj = pdf_dict_get(ctx, obj, PDF_NAME_Subtype);
if (obj && !pdf_is_name(ctx, obj))
fz_throw(ctx, FZ_ERROR_GENERIC, "invalid font descriptor subtype");
if (pdf_name_eq(ctx, obj, PDF_NAME_Type1C))
ext = "cff";
else if (pdf_name_eq(ctx, obj, PDF_NAME_CIDFontType0C))
ext = "cid";
else if (pdf_name_eq(ctx, obj, PDF_NAME_OpenType))
ext = "otf";
else
fz_throw(ctx, FZ_ERROR_GENERIC, "unhandled font type '%s'", pdf_to_name(ctx, obj));
}
if (!stream)
{
fz_warn(ctx, "unhandled font type");
return;
}
buf = pdf_load_stream(ctx, doc, pdf_to_num(ctx, stream), pdf_to_gen(ctx, stream));
snprintf(namebuf, sizeof(namebuf), "%s-%04d.%s", fontname, num, ext);
printf("extracting font %s\n", namebuf);
f = fz_fopen(namebuf, "wb");
if (!f)
fz_throw(ctx, FZ_ERROR_GENERIC, "cannot create font file");
len = fz_buffer_storage(ctx, buf, &data);
n = fwrite(data, 1, len, f);
if (n < len)
fz_throw(ctx, FZ_ERROR_GENERIC, "cannot write font file");
if (fclose(f) < 0)
fz_throw(ctx, FZ_ERROR_GENERIC, "cannot close font file");
fz_drop_buffer(ctx, buf);
}
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--;
}
}
}
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");
}
}
int
pdf_repair_obj(fz_context *ctx, pdf_document *doc, pdf_lexbuf *buf, fz_off_t *stmofsp, int *stmlenp, pdf_obj **encrypt, pdf_obj **id, pdf_obj **page, fz_off_t *tmpofs, pdf_obj **root)
{
fz_stream *file = doc->file;
pdf_token tok;
int stm_len;
*stmofsp = 0;
if (stmlenp)
*stmlenp = -1;
stm_len = 0;
/* On entry to this function, we know that we've just seen
* '<int> <int> obj'. We expect the next thing we see to be a
* pdf object. Regardless of the type of thing we meet next
* we only need to fully parse it if it is a dictionary. */
tok = pdf_lex(ctx, file, buf);
if (tok == PDF_TOK_OPEN_DICT)
{
pdf_obj *dict, *obj;
fz_try(ctx)
{
dict = pdf_parse_dict(ctx, doc, file, buf);
}
fz_catch(ctx)
{
fz_rethrow_if(ctx, FZ_ERROR_TRYLATER);
/* Don't let a broken object at EOF overwrite a good one */
if (file->eof)
fz_rethrow(ctx);
/* Silently swallow the error */
dict = pdf_new_dict(ctx, NULL, 2);
}
/* We must be careful not to try to resolve any indirections
* here. We have just read dict, so we know it to be a non
* indirected dictionary. Before we look at any values that
* we get back from looking up in it, we need to check they
* aren't indirected. */
if (encrypt || id || root)
{
obj = pdf_dict_get(ctx, dict, PDF_NAME_Type);
if (!pdf_is_indirect(ctx, obj) && pdf_name_eq(ctx, obj, PDF_NAME_XRef))
{
if (encrypt)
{
obj = pdf_dict_get(ctx, dict, PDF_NAME_Encrypt);
if (obj)
{
pdf_drop_obj(ctx, *encrypt);
*encrypt = pdf_keep_obj(ctx, obj);
}
}
if (id)
{
obj = pdf_dict_get(ctx, dict, PDF_NAME_ID);
if (obj)
{
pdf_drop_obj(ctx, *id);
*id = pdf_keep_obj(ctx, obj);
}
}
if (root)
*root = pdf_keep_obj(ctx, pdf_dict_get(ctx, dict, PDF_NAME_Root));
}
}
obj = pdf_dict_get(ctx, dict, PDF_NAME_Length);
if (!pdf_is_indirect(ctx, obj) && pdf_is_int(ctx, obj))
stm_len = pdf_to_int(ctx, obj);
if (doc->file_reading_linearly && page)
{
obj = pdf_dict_get(ctx, dict, PDF_NAME_Type);
if (!pdf_is_indirect(ctx, obj) && pdf_name_eq(ctx, obj, PDF_NAME_Page))
{
pdf_drop_obj(ctx, *page);
*page = pdf_keep_obj(ctx, dict);
}
}
pdf_drop_obj(ctx, dict);
}
void pdf_add_portfolio_schema(fz_context *ctx, pdf_document *doc, int entry, const pdf_portfolio_schema *info)
{
pdf_portfolio **pp;
pdf_portfolio *p;
pdf_obj *s;
pdf_obj *sc;
int num;
char str_name[32];
pdf_obj *num_name = NULL;
if (!doc)
fz_throw(ctx, FZ_ERROR_GENERIC, "Bad pdf_add_portfolio_schema call");
if (doc->portfolio == NULL)
load_portfolio(ctx, doc);
fz_var(num_name);
pp = &doc->portfolio;
while (*pp && entry > 0)
pp = &(*pp)->next, entry--;
fz_try(ctx)
{
/* Find a name for the new schema entry */
num = 0;
do
{
pdf_drop_obj(ctx, num_name);
num_name = NULL;
num++;
sprintf(str_name, "%d", num);
num_name = pdf_new_name(ctx, doc, str_name);
p = doc->portfolio;
for (p = doc->portfolio; p; p = p->next)
if (pdf_name_eq(ctx, num_name, p->key))
break;
}
while (p);
sc = pdf_new_dict(ctx, doc, 4);
pdf_dict_put_drop(ctx, sc, PDF_NAME_E, pdf_new_bool(ctx, doc, !!info->editable));
pdf_dict_put_drop(ctx, sc, PDF_NAME_V, pdf_new_bool(ctx, doc, !!info->visible));
pdf_dict_put_drop(ctx, sc, PDF_NAME_N, info->name);
pdf_dict_put(ctx, sc, PDF_NAME_Subtype, PDF_NAME_S);
/* Add to our linked list (in the correct sorted place) */
p = fz_malloc_struct(ctx, pdf_portfolio);
p->entry = *info;
p->sort = 0; /* Will be rewritten in a mo */
p->key = pdf_keep_obj(ctx, num_name);
p->val = pdf_keep_obj(ctx, sc);
p->next = *pp;
*pp = p;
/* Add the key to the schema */
s = pdf_dict_getl(ctx, pdf_trailer(ctx, doc), PDF_NAME_Root, PDF_NAME_Collection, PDF_NAME_Schema, NULL);
pdf_dict_put(ctx, s, num_name, sc);
/* Renumber the schema entries */
for (num = 0, p = doc->portfolio; p; num++, p = p->next)
{
pdf_dict_put_drop(ctx, p->val, PDF_NAME_O, pdf_new_int(ctx, doc, num));
p->sort = num;
}
}
fz_always(ctx)
pdf_drop_obj(ctx, num_name);
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;
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);
}
char *
pdf_parse_link_dest(fz_context *ctx, pdf_document *doc, pdf_obj *dest)
{
pdf_obj *obj, *pageobj;
fz_rect mediabox;
fz_matrix pagectm;
const char *ld;
int page, x, y, h;
dest = resolve_dest(ctx, doc, dest);
if (dest == NULL)
{
fz_warn(ctx, "undefined link destination");
return NULL;
}
if (pdf_is_name(ctx, dest))
{
ld = pdf_to_name(ctx, dest);
return fz_strdup(ctx, ld);
}
else if (pdf_is_string(ctx, dest))
{
ld = pdf_to_str_buf(ctx, dest);
return fz_strdup(ctx, ld);
}
pageobj = pdf_array_get(ctx, dest, 0);
if (pdf_is_int(ctx, pageobj))
{
page = pdf_to_int(ctx, pageobj);
pageobj = pdf_lookup_page_obj(ctx, doc, page);
}
else
{
fz_try(ctx)
page = pdf_lookup_page_number(ctx, doc, pageobj);
fz_catch(ctx)
page = -1;
}
if (page < 0)
return NULL;
obj = pdf_array_get(ctx, dest, 1);
if (obj)
{
/* Link coords use a coordinate space that does not seem to respect Rotate or UserUnit. */
/* All we need to do is figure out the page height to flip the coordinate space. */
pdf_page_obj_transform(ctx, pageobj, &mediabox, &pagectm);
mediabox = fz_transform_rect(mediabox, pagectm);
h = mediabox.y1 - mediabox.y0;
if (pdf_name_eq(ctx, obj, PDF_NAME(XYZ)))
{
x = pdf_array_get_int(ctx, dest, 2);
y = h - pdf_array_get_int(ctx, dest, 3);
}
else if (pdf_name_eq(ctx, obj, PDF_NAME(FitR)))
{
x = pdf_array_get_int(ctx, dest, 2);
y = h - pdf_array_get_int(ctx, dest, 5);
}
else if (pdf_name_eq(ctx, obj, PDF_NAME(FitH)) || pdf_name_eq(ctx, obj, PDF_NAME(FitBH)))
{
x = 0;
y = h - pdf_array_get_int(ctx, dest, 2);
}
else if (pdf_name_eq(ctx, obj, PDF_NAME(FitV)) || pdf_name_eq(ctx, obj, PDF_NAME(FitBV)))
{
x = pdf_array_get_int(ctx, dest, 2);
y = 0;
}
else
{
x = 0;
y = 0;
}
return fz_asprintf(ctx, "#%d,%d,%d", page + 1, x, y);
}
return fz_asprintf(ctx, "#%d", page + 1);
}
char *
pdf_parse_link_dest(fz_context *ctx, pdf_document *doc, pdf_obj *dest)
{
pdf_obj *obj;
char buf[256];
const char *ld;
int page;
int x, y;
dest = resolve_dest(ctx, doc, dest);
if (dest == NULL)
{
fz_warn(ctx, "undefined link destination");
return NULL;
}
if (pdf_is_name(ctx, dest))
{
ld = pdf_to_name(ctx, dest);
return fz_strdup(ctx, ld);
}
else if (pdf_is_string(ctx, dest))
{
ld = pdf_to_str_buf(ctx, dest);
return fz_strdup(ctx, ld);
}
obj = pdf_array_get(ctx, dest, 0);
if (pdf_is_int(ctx, obj))
page = pdf_to_int(ctx, obj);
else
{
fz_try(ctx)
page = pdf_lookup_page_number(ctx, doc, obj);
fz_catch(ctx)
page = -1;
}
x = y = 0;
obj = pdf_array_get(ctx, dest, 1);
if (pdf_name_eq(ctx, obj, PDF_NAME_XYZ))
{
x = pdf_to_int(ctx, pdf_array_get(ctx, dest, 2));
y = pdf_to_int(ctx, pdf_array_get(ctx, dest, 3));
}
else if (pdf_name_eq(ctx, obj, PDF_NAME_FitR))
{
x = pdf_to_int(ctx, pdf_array_get(ctx, dest, 2));
y = pdf_to_int(ctx, pdf_array_get(ctx, dest, 5));
}
else if (pdf_name_eq(ctx, obj, PDF_NAME_FitH) || pdf_name_eq(ctx, obj, PDF_NAME_FitBH))
y = pdf_to_int(ctx, pdf_array_get(ctx, dest, 2));
else if (pdf_name_eq(ctx, obj, PDF_NAME_FitV) || pdf_name_eq(ctx, obj, PDF_NAME_FitBV))
x = pdf_to_int(ctx, pdf_array_get(ctx, dest, 2));
if (page >= 0)
{
if (x != 0 || y != 0)
fz_snprintf(buf, sizeof buf, "#%d,%d,%d", page + 1, x, y);
else
fz_snprintf(buf, sizeof buf, "#%d", page + 1);
return fz_strdup(ctx, buf);
}
return NULL;
}
char *
pdf_parse_link_action(fz_context *ctx, pdf_document *doc, pdf_obj *action, int pagenum)
{
pdf_obj *obj, *dest, *file_spec;
if (!action)
return NULL;
obj = pdf_dict_get(ctx, action, PDF_NAME_S);
if (pdf_name_eq(ctx, PDF_NAME_GoTo, obj))
{
dest = pdf_dict_get(ctx, action, PDF_NAME_D);
return pdf_parse_link_dest(ctx, doc, dest);
}
else if (pdf_name_eq(ctx, PDF_NAME_URI, obj))
{
/* URI entries are ASCII strings */
const char *uri = pdf_to_str_buf(ctx, pdf_dict_get(ctx, action, PDF_NAME_URI));
if (!fz_is_external_link(ctx, uri))
{
pdf_obj *uri_base_obj = pdf_dict_getp(ctx, pdf_trailer(ctx, doc), "Root/URI/Base");
const char *uri_base = uri_base_obj ? pdf_to_str_buf(ctx, uri_base_obj) : "file://";
char *new_uri = fz_malloc(ctx, strlen(uri_base) + strlen(uri) + 1);
strcpy(new_uri, uri_base);
strcat(new_uri, uri);
return new_uri;
}
return fz_strdup(ctx, uri);
}
else if (pdf_name_eq(ctx, PDF_NAME_Launch, obj))
{
file_spec = pdf_dict_get(ctx, action, PDF_NAME_F);
return pdf_parse_file_spec(ctx, doc, file_spec, NULL);
}
else if (pdf_name_eq(ctx, PDF_NAME_GoToR, obj))
{
dest = pdf_dict_get(ctx, action, PDF_NAME_D);
file_spec = pdf_dict_get(ctx, action, PDF_NAME_F);
return pdf_parse_file_spec(ctx, doc, file_spec, dest);
}
else if (pdf_name_eq(ctx, PDF_NAME_Named, obj))
{
dest = pdf_dict_get(ctx, action, PDF_NAME_N);
if (pdf_name_eq(ctx, PDF_NAME_FirstPage, dest))
pagenum = 0;
else if (pdf_name_eq(ctx, PDF_NAME_LastPage, dest))
pagenum = pdf_count_pages(ctx, doc) - 1;
else if (pdf_name_eq(ctx, PDF_NAME_PrevPage, dest) && pagenum >= 0)
{
if (pagenum > 0)
pagenum--;
}
else if (pdf_name_eq(ctx, PDF_NAME_NextPage, dest) && pagenum >= 0)
{
if (pagenum < pdf_count_pages(ctx, doc) - 1)
pagenum++;
}
else
return NULL;
return fz_asprintf(ctx, "#%d", pagenum + 1);
}
return NULL;
}
