static void sweepref(fz_obj *obj)
{
int num = fz_to_num(obj);
int gen = fz_to_gen(obj);
if (num < 0 || num >= xref->len)
return;
if (uselist[num])
return;
uselist[num] = 1;
/* Bake in /Length in stream objects */
fz_try(ctx)
{
if (pdf_is_stream(xref, num, gen))
{
fz_obj *len = fz_dict_gets(obj, "Length");
if (fz_is_indirect(len))
{
uselist[fz_to_num(len)] = 0;
len = fz_resolve_indirect(len);
fz_dict_puts(obj, "Length", len);
}
}
}
fz_catch(ctx)
{
/* Leave broken */
}
sweepobj(fz_resolve_indirect(obj));
}
fz_error
pdf_load_shading(fz_shade **shadep, pdf_xref *xref, fz_obj *dict)
{
fz_error error;
fz_matrix mat;
fz_obj *obj;
fz_context *ctx = xref->ctx;
if ((*shadep = pdf_find_item(ctx, xref->store, fz_drop_shade, dict)))
{
fz_keep_shade(*shadep);
return fz_okay;
}
/* Type 2 pattern dictionary */
if (fz_dict_gets(ctx, dict, "PatternType"))
{
obj = fz_dict_gets(ctx, dict, "Matrix");
if (obj)
mat = pdf_to_matrix(ctx, obj);
else
mat = fz_identity;
obj = fz_dict_gets(ctx, dict, "ExtGState");
if (obj)
{
if (fz_dict_gets(ctx, obj, "CA") || fz_dict_gets(ctx, obj, "ca"))
{
fz_warn(ctx, "shading with alpha not supported");
}
}
obj = fz_dict_gets(ctx, dict, "Shading");
if (!obj)
return fz_error_make(ctx, "syntaxerror: missing shading dictionary");
error = pdf_load_shading_dict(shadep, xref, obj, mat);
if (error)
return fz_error_note(ctx, error, "cannot load shading dictionary (%d %d R)", fz_to_num(obj), fz_to_gen(obj));
}
/* Naked shading dictionary */
else
{
error = pdf_load_shading_dict(shadep, xref, dict, fz_identity);
if (error)
return fz_error_note(ctx, error, "cannot load shading dictionary (%d %d R)", fz_to_num(dict), fz_to_gen(dict));
}
pdf_store_item(ctx, xref->store, fz_keep_shade, fz_drop_shade, dict, *shadep);
return fz_okay;
}
fz_error
pdf_load_pattern(pdf_pattern **patp, pdf_xref *xref, fz_obj *dict)
{
fz_error error;
pdf_pattern *pat;
fz_obj *obj;
if ((*patp = pdf_find_item(xref->store, pdf_drop_pattern, dict)))
{
pdf_keep_pattern(*patp);
return fz_okay;
}
pat = fz_malloc(sizeof(pdf_pattern));
pat->refs = 1;
pat->resources = NULL;
pat->contents = NULL;
/* Store pattern now, to avoid possible recursion if objects refer back to this one */
pdf_store_item(xref->store, pdf_keep_pattern, pdf_drop_pattern, dict, pat);
pat->ismask = fz_to_int(fz_dict_gets(dict, "PaintType")) == 2;
pat->xstep = fz_to_real(fz_dict_gets(dict, "XStep"));
pat->ystep = fz_to_real(fz_dict_gets(dict, "YStep"));
obj = fz_dict_gets(dict, "BBox");
pat->bbox = pdf_to_rect(obj);
obj = fz_dict_gets(dict, "Matrix");
if (obj)
pat->matrix = pdf_to_matrix(obj);
else
pat->matrix = fz_identity;
pat->resources = fz_dict_gets(dict, "Resources");
if (pat->resources)
fz_keep_obj(pat->resources);
error = pdf_load_stream(&pat->contents, xref, fz_to_num(dict), fz_to_gen(dict));
if (error)
{
pdf_remove_item(xref->store, pdf_drop_pattern, dict);
pdf_drop_pattern(pat);
return fz_rethrow(error, "cannot load pattern stream (%d %d R)", fz_to_num(dict), fz_to_gen(dict));
}
*patp = pat;
return fz_okay;
}
static fz_obj *
pdf_get_ap_stream(pdf_xref *xref, fz_obj *obj)
{
fz_obj *ap = fz_dict_gets(xref->ctx, obj, "AP");
if (!fz_is_dict(xref->ctx, ap))
return NULL;
ap = fz_dict_gets(xref->ctx, ap, "N");
if (!pdf_is_stream(xref, fz_to_num(ap), fz_to_gen(ap)))
ap = fz_dict_get(xref->ctx, ap, fz_dict_gets(xref->ctx, obj, "AS"));
if (!pdf_is_stream(xref, fz_to_num(ap), fz_to_gen(ap)))
return NULL;
return ap;
}
static fz_error
pdf_read_ocg(pdf_xref *xref)
{
fz_obj *obj, *ocg;
int len, i;
pdf_ocg_descriptor *desc;
obj = fz_dict_gets(fz_dict_gets(xref->trailer, "Root"), "OCProperties");
if (obj == NULL)
return fz_okay;
ocg = fz_dict_gets(obj, "OCGs");
if (ocg == NULL || !fz_is_array(ocg))
/* Not ever supposed to happen, but live with it. */
return fz_okay;
len = fz_array_len(ocg);
desc = fz_malloc(sizeof(*desc));
desc->len = len;
desc->ocgs = fz_calloc(len, sizeof(*desc->ocgs));
desc->intent = NULL;
for (i=0; i < len; i++)
{
fz_obj *o = fz_array_get(ocg, i);
desc->ocgs[i].num = fz_to_num(o);
desc->ocgs[i].gen = fz_to_gen(o);
desc->ocgs[i].state = 0;
}
xref->ocg = desc;
return pdf_ocg_set_config(xref, 0);
}
static void renumberobj(fz_obj *obj)
{
int i;
fz_context *ctx = xref->ctx;
if (fz_is_dict(obj))
{
int n = fz_dict_len(obj);
for (i = 0; i < n; i++)
{
fz_obj *key = fz_dict_get_key(obj, i);
fz_obj *val = fz_dict_get_val(obj, i);
if (fz_is_indirect(val))
{
val = fz_new_indirect(ctx, renumbermap[fz_to_num(val)], 0, xref);
fz_dict_put(obj, key, val);
fz_drop_obj(val);
}
else
{
renumberobj(val);
}
}
}
else if (fz_is_array(obj))
{
int n = fz_array_len(obj);
for (i = 0; i < n; i++)
{
fz_obj *val = fz_array_get(obj, i);
if (fz_is_indirect(val))
{
val = fz_new_indirect(ctx, renumbermap[fz_to_num(val)], 0, xref);
fz_array_put(obj, i, val);
fz_drop_obj(val);
}
else
{
renumberobj(val);
}
}
}
}
static fz_error
load_separation(fz_colorspace **csp, pdf_xref *xref, fz_obj *array)
{
fz_error error;
fz_colorspace *cs;
struct separation *sep;
fz_obj *nameobj = fz_array_get(array, 1);
fz_obj *baseobj = fz_array_get(array, 2);
fz_obj *tintobj = fz_array_get(array, 3);
fz_colorspace *base;
pdf_function *tint;
int n;
if (fz_is_array(nameobj))
n = fz_array_len(nameobj);
else
n = 1;
if (n > FZ_MAX_COLORS)
return fz_throw("too many components in colorspace");
error = pdf_load_colorspace(&base, xref, baseobj);
if (error)
return fz_rethrow(error, "cannot load base colorspace (%d %d R)", fz_to_num(baseobj), fz_to_gen(baseobj));
error = pdf_load_function(&tint, xref, tintobj);
if (error)
{
fz_drop_colorspace(base);
return fz_rethrow(error, "cannot load tint function (%d %d R)", fz_to_num(tintobj), fz_to_gen(tintobj));
}
sep = fz_malloc(sizeof(struct separation));
sep->base = base;
sep->tint = tint;
cs = fz_new_colorspace(n == 1 ? "Separation" : "DeviceN", n);
cs->to_rgb = separation_to_rgb;
cs->free_data = free_separation;
cs->data = sep;
*csp = cs;
return fz_okay;
}
static void renumberobjs(void)
{
pdf_xref_entry *oldxref;
int newlen;
int num;
/* Apply renumber map to indirect references in all objects in xref */
renumberobj(xref->trailer);
for (num = 0; num < xref->len; num++)
{
fz_obj *obj = xref->table[num].obj;
if (fz_is_indirect(obj))
{
obj = fz_new_indirect(ctx, renumbermap[fz_to_num(obj)], 0, xref);
pdf_update_object(xref, num, 0, obj);
fz_drop_obj(obj);
}
else
{
renumberobj(obj);
}
}
/* Create new table for the reordered, compacted xref */
oldxref = xref->table;
xref->table = fz_malloc_array(xref->ctx, xref->len, sizeof(pdf_xref_entry));
xref->table[0] = oldxref[0];
/* Move used objects into the new compacted xref */
newlen = 0;
for (num = 1; num < xref->len; num++)
{
if (uselist[num])
{
if (newlen < renumbermap[num])
newlen = renumbermap[num];
xref->table[renumbermap[num]] = oldxref[num];
}
else
{
if (oldxref[num].obj)
fz_drop_obj(oldxref[num].obj);
}
}
fz_free(xref->ctx, oldxref);
/* Update the used objects count in compacted xref */
xref->len = newlen + 1;
/* Update list of used objects to fit with compacted xref */
for (num = 1; num < xref->len; num++)
uselist[num] = 1;
}
static fz_error
pdf_load_jpx_image(fz_pixmap **imgp, pdf_xref *xref, fz_obj *dict)
{
fz_error error;
fz_buffer *buf;
fz_colorspace *colorspace;
fz_pixmap *img;
fz_obj *obj;
colorspace = NULL;
error = pdf_load_stream(&buf, xref, fz_to_num(dict), fz_to_gen(dict));
if (error)
return fz_rethrow(error, "cannot load jpx image data");
obj = fz_dict_gets(dict, "ColorSpace");
if (obj)
{
error = pdf_load_colorspace(&colorspace, xref, obj);
if (error)
fz_catch(error, "cannot load image colorspace");
}
error = fz_load_jpx_image(&img, buf->data, buf->len, colorspace);
if (error)
{
if (colorspace)
fz_drop_colorspace(colorspace);
fz_drop_buffer(buf);
return fz_rethrow(error, "cannot load jpx image");
}
if (colorspace)
fz_drop_colorspace(colorspace);
fz_drop_buffer(buf);
obj = fz_dict_getsa(dict, "SMask", "Mask");
if (fz_is_dict(obj))
{
error = pdf_load_image_imp(&img->mask, xref, NULL, obj, NULL, 1);
if (error)
{
fz_drop_pixmap(img);
return fz_rethrow(error, "cannot load image mask/softmask");
}
}
*imgp = img;
return fz_okay;
}
fz_error
pdf_load_colorspace(fz_colorspace **csp, pdf_xref *xref, fz_obj *obj)
{
fz_error error;
if ((*csp = pdf_find_item(xref->store, fz_drop_colorspace, obj)))
{
fz_keep_colorspace(*csp);
return fz_okay;
}
error = pdf_load_colorspace_imp(csp, xref, obj);
if (error)
return fz_rethrow(error, "cannot load colorspace (%d %d R)", fz_to_num(obj), fz_to_gen(obj));
pdf_store_item(xref->store, fz_keep_colorspace, fz_drop_colorspace, obj, *csp);
return fz_okay;
}
fz_error
pdf_load_image(fz_pixmap **pixp, pdf_xref *xref, fz_obj *dict)
{
fz_error error;
if ((*pixp = pdf_find_item(xref->store, fz_drop_pixmap, dict)))
{
fz_keep_pixmap(*pixp);
return fz_okay;
}
error = pdf_load_image_imp(pixp, xref, NULL, dict, NULL, 0);
if (error)
return fz_rethrow(error, "cannot load image (%d 0 R)", fz_to_num(dict));
// EBookDroid: Commented out for saving A LOTS of memory. Caching is not needed for embedded solutions
// pdf_store_item(xref->store, fz_keep_pixmap, fz_drop_pixmap, dict, *pixp);
return fz_okay;
}
fz_obj *
pdf_resolve_indirect(fz_obj *ref)
{
if (fz_is_indirect(ref))
{
pdf_xref *xref = fz_get_indirect_xref(ref);
int num = fz_to_num(ref);
int gen = fz_to_gen(ref);
if (xref)
{
fz_error error = pdf_cache_object(xref, num, gen);
if (error)
{
fz_catch(error, "cannot load object (%d %d R) into cache", num, gen);
return ref;
}
if (xref->table[num].obj)
return xref->table[num].obj;
}
}
return ref;
}
static fz_error
pdf_load_colorspace_imp(fz_colorspace **csp, pdf_xref *xref, fz_obj *obj)
{
if (fz_is_name(obj))
{
if (!strcmp(fz_to_name(obj), "Pattern"))
*csp = fz_device_gray;
else if (!strcmp(fz_to_name(obj), "G"))
*csp = fz_device_gray;
else if (!strcmp(fz_to_name(obj), "RGB"))
*csp = fz_device_rgb;
else if (!strcmp(fz_to_name(obj), "CMYK"))
*csp = fz_device_cmyk;
else if (!strcmp(fz_to_name(obj), "DeviceGray"))
*csp = fz_device_gray;
else if (!strcmp(fz_to_name(obj), "DeviceRGB"))
*csp = fz_device_rgb;
else if (!strcmp(fz_to_name(obj), "DeviceCMYK"))
*csp = fz_device_cmyk;
else
return fz_throw("unknown colorspace: %s", fz_to_name(obj));
return fz_okay;
}
else if (fz_is_array(obj))
{
fz_obj *name = fz_array_get(obj, 0);
if (fz_is_name(name))
{
/* load base colorspace instead */
if (!strcmp(fz_to_name(name), "Pattern"))
{
fz_error error;
obj = fz_array_get(obj, 1);
if (!obj)
{
*csp = fz_device_gray;
return fz_okay;
}
error = pdf_load_colorspace(csp, xref, obj);
if (error)
return fz_rethrow(error, "cannot load pattern (%d %d R)", fz_to_num(obj), fz_to_gen(obj));
}
else if (!strcmp(fz_to_name(name), "G"))
*csp = fz_device_gray;
else if (!strcmp(fz_to_name(name), "RGB"))
*csp = fz_device_rgb;
else if (!strcmp(fz_to_name(name), "CMYK"))
*csp = fz_device_cmyk;
else if (!strcmp(fz_to_name(name), "DeviceGray"))
*csp = fz_device_gray;
else if (!strcmp(fz_to_name(name), "DeviceRGB"))
*csp = fz_device_rgb;
else if (!strcmp(fz_to_name(name), "DeviceCMYK"))
*csp = fz_device_cmyk;
else if (!strcmp(fz_to_name(name), "CalGray"))
*csp = fz_device_gray;
else if (!strcmp(fz_to_name(name), "CalRGB"))
*csp = fz_device_rgb;
else if (!strcmp(fz_to_name(name), "CalCMYK"))
*csp = fz_device_cmyk;
else if (!strcmp(fz_to_name(name), "Lab"))
*csp = fz_device_lab;
else if (!strcmp(fz_to_name(name), "ICCBased"))
return load_icc_based(csp, xref, fz_array_get(obj, 1));
else if (!strcmp(fz_to_name(name), "Indexed"))
return load_indexed(csp, xref, obj);
else if (!strcmp(fz_to_name(name), "I"))
return load_indexed(csp, xref, obj);
else if (!strcmp(fz_to_name(name), "Separation"))
return load_separation(csp, xref, obj);
else if (!strcmp(fz_to_name(name), "DeviceN"))
return load_separation(csp, xref, obj);
else
return fz_throw("syntaxerror: unknown colorspace %s", fz_to_name(name));
return fz_okay;
}
}
return fz_throw("syntaxerror: could not parse color space (%d %d R)", fz_to_num(obj), fz_to_gen(obj));
}
static fz_error
load_indexed(fz_colorspace **csp, pdf_xref *xref, fz_obj *array)
{
fz_error error;
fz_colorspace *cs;
struct indexed *idx;
fz_obj *baseobj = fz_array_get(array, 1);
fz_obj *highobj = fz_array_get(array, 2);
fz_obj *lookup = fz_array_get(array, 3);
fz_colorspace *base;
int i, n;
error = pdf_load_colorspace(&base, xref, baseobj);
if (error)
return fz_rethrow(error, "cannot load base colorspace (%d %d R)", fz_to_num(baseobj), fz_to_gen(baseobj));
idx = fz_malloc(sizeof(struct indexed));
idx->base = base;
idx->high = fz_to_int(highobj);
idx->high = CLAMP(idx->high, 0, 255);
n = base->n * (idx->high + 1);
idx->lookup = fz_malloc(n);
memset(idx->lookup, 0, n);
cs = fz_new_colorspace("Indexed", 1);
cs->to_rgb = indexed_to_rgb;
cs->free_data = free_indexed;
cs->data = idx;
if (fz_is_string(lookup) && fz_to_str_len(lookup) == n)
{
unsigned char *buf = (unsigned char *) fz_to_str_buf(lookup);
for (i = 0; i < n; i++)
idx->lookup[i] = buf[i];
}
else if (fz_is_indirect(lookup))
{
fz_stream *file;
error = pdf_open_stream(&file, xref, fz_to_num(lookup), fz_to_gen(lookup));
if (error)
{
fz_drop_colorspace(cs);
return fz_rethrow(error, "cannot open colorspace lookup table (%d 0 R)", fz_to_num(lookup));
}
i = fz_read(file, idx->lookup, n);
if (i < 0)
{
fz_drop_colorspace(cs);
return fz_throw("cannot read colorspace lookup table (%d 0 R)", fz_to_num(lookup));
}
fz_close(file);
}
else
{
fz_drop_colorspace(cs);
return fz_throw("cannot parse colorspace lookup table");
}
*csp = cs;
return fz_okay;
}
static fz_colorspace *
load_indexed(pdf_document *xref, fz_obj *array)
{
struct indexed *idx = NULL;
fz_context *ctx = xref->ctx;
fz_obj *baseobj = fz_array_get(array, 1);
fz_obj *highobj = fz_array_get(array, 2);
fz_obj *lookup = fz_array_get(array, 3);
fz_colorspace *base = NULL;
fz_colorspace *cs = NULL;
int i, n;
fz_var(idx);
fz_var(base);
fz_var(cs);
fz_try(ctx)
{
base = pdf_load_colorspace(xref, baseobj);
/* "cannot load base colorspace (%d %d R)", fz_to_num(baseobj), fz_to_gen(baseobj) */
idx = fz_malloc_struct(ctx, struct indexed);
idx->lookup = NULL;
idx->base = base;
idx->high = fz_to_int(highobj);
idx->high = CLAMP(idx->high, 0, 255);
n = base->n * (idx->high + 1);
idx->lookup = fz_malloc_array(ctx, 1, n);
cs = fz_new_colorspace(ctx, "Indexed", 1);
cs->to_rgb = indexed_to_rgb;
cs->free_data = free_indexed;
cs->data = idx;
cs->size += sizeof(*idx) + n + (base ? base->size : 0);
if (fz_is_string(lookup) && fz_to_str_len(lookup) == n)
{
unsigned char *buf = (unsigned char *) fz_to_str_buf(lookup);
for (i = 0; i < n; i++)
idx->lookup[i] = buf[i];
}
else if (fz_is_indirect(lookup))
{
fz_stream *file = NULL;
fz_try(ctx)
{
file = pdf_open_stream(xref, fz_to_num(lookup), fz_to_gen(lookup));
}
fz_catch(ctx)
{
fz_throw(ctx, "cannot open colorspace lookup table (%d 0 R)", fz_to_num(lookup));
}
i = fz_read(file, idx->lookup, n);
if (i < 0)
{
fz_close(file);
fz_throw(ctx, "cannot read colorspace lookup table (%d 0 R)", fz_to_num(lookup));
}
fz_close(file);
}
else
{
fz_throw(ctx, "cannot parse colorspace lookup table");
}
}
fz_error
pdf_load_to_unicode(pdf_font_desc *font, pdf_xref *xref,
char **strings, char *collection, fz_obj *cmapstm)
{
fz_error error = fz_okay;
pdf_cmap *cmap;
int cid;
int ucsbuf[8];
int ucslen;
int i;
if (pdf_is_stream(xref, fz_to_num(cmapstm), fz_to_gen(cmapstm)))
{
error = pdf_load_embedded_cmap(&cmap, xref, cmapstm);
if (error)
return fz_rethrow(error, "cannot load embedded cmap (%d %d R)", fz_to_num(cmapstm), fz_to_gen(cmapstm));
font->to_unicode = pdf_new_cmap();
for (i = 0; i < (strings ? 256 : 65536); i++)
{
cid = pdf_lookup_cmap(font->encoding, i);
if (cid >= 0)
{
ucslen = pdf_lookup_cmap_full(cmap, i, ucsbuf);
if (ucslen == 1)
pdf_map_range_to_range(font->to_unicode, cid, cid, ucsbuf[0]);
if (ucslen > 1)
pdf_map_one_to_many(font->to_unicode, cid, ucsbuf, ucslen);
}
}
pdf_sort_cmap(font->to_unicode);
pdf_drop_cmap(cmap);
}
else if (collection)
{
error = fz_okay;
if (!strcmp(collection, "Adobe-CNS1"))
error = pdf_load_system_cmap(&font->to_unicode, "Adobe-CNS1-UCS2");
else if (!strcmp(collection, "Adobe-GB1"))
error = pdf_load_system_cmap(&font->to_unicode, "Adobe-GB1-UCS2");
else if (!strcmp(collection, "Adobe-Japan1"))
error = pdf_load_system_cmap(&font->to_unicode, "Adobe-Japan1-UCS2");
else if (!strcmp(collection, "Adobe-Korea1"))
error = pdf_load_system_cmap(&font->to_unicode, "Adobe-Korea1-UCS2");
if (error)
return fz_rethrow(error, "cannot load ToUnicode system cmap %s-UCS2", collection);
}
if (strings)
{
/* TODO one-to-many mappings */
font->cid_to_ucs_len = 256;
font->cid_to_ucs = fz_calloc(256, sizeof(unsigned short));
for (i = 0; i < 256; i++)
{
if (strings[i])
font->cid_to_ucs[i] = pdf_lookup_agl(strings[i]);
else
font->cid_to_ucs[i] = '?';
}
}
if (!font->to_unicode && !font->cid_to_ucs)
{
/* TODO: synthesize a ToUnicode if it's a freetype font with
* cmap and/or post tables or if it has glyph names. */
}
return fz_okay;
}
static void retainpages(int argc, char **argv)
{
fz_obj *oldroot, *root, *pages, *kids, *countobj, *parent, *olddests;
/* Keep only pages/type and (reduced) dest entries to avoid
* references to unretained pages */
oldroot = fz_dict_gets(xref->trailer, "Root");
pages = fz_dict_gets(oldroot, "Pages");
olddests = pdf_load_name_tree(xref, "Dests");
root = fz_new_dict(ctx, 2);
fz_dict_puts(root, "Type", fz_dict_gets(oldroot, "Type"));
fz_dict_puts(root, "Pages", fz_dict_gets(oldroot, "Pages"));
pdf_update_object(xref, fz_to_num(oldroot), fz_to_gen(oldroot), root);
fz_drop_obj(root);
/* Create a new kids array with only the pages we want to keep */
parent = fz_new_indirect(ctx, fz_to_num(pages), fz_to_gen(pages), xref);
kids = fz_new_array(ctx, 1);
/* Retain pages specified */
while (argc - fz_optind)
{
int page, spage, epage;
char *spec, *dash;
char *pagelist = argv[fz_optind];
spec = fz_strsep(&pagelist, ",");
while (spec)
{
dash = strchr(spec, '-');
if (dash == spec)
spage = epage = pdf_count_pages(xref);
else
spage = epage = atoi(spec);
if (dash)
{
if (strlen(dash) > 1)
epage = atoi(dash + 1);
else
epage = pdf_count_pages(xref);
}
if (spage > epage)
page = spage, spage = epage, epage = page;
if (spage < 1)
spage = 1;
if (epage > pdf_count_pages(xref))
epage = pdf_count_pages(xref);
for (page = spage; page <= epage; page++)
{
fz_obj *pageobj = xref->page_objs[page-1];
fz_obj *pageref = xref->page_refs[page-1];
fz_dict_puts(pageobj, "Parent", parent);
/* Store page object in new kids array */
fz_array_push(kids, pageref);
}
spec = fz_strsep(&pagelist, ",");
}
fz_optind++;
}
fz_drop_obj(parent);
/* Update page count and kids array */
countobj = fz_new_int(ctx, fz_array_len(kids));
fz_dict_puts(pages, "Count", countobj);
fz_drop_obj(countobj);
fz_dict_puts(pages, "Kids", kids);
fz_drop_obj(kids);
/* Also preserve the (partial) Dests name tree */
if (olddests)
{
int i;
fz_obj *names = fz_new_dict(ctx, 1);
fz_obj *dests = fz_new_dict(ctx, 1);
fz_obj *names_list = fz_new_array(ctx, 32);
for (i = 0; i < fz_dict_len(olddests); i++)
{
fz_obj *key = fz_dict_get_key(olddests, i);
fz_obj *val = fz_dict_get_val(olddests, i);
fz_obj *key_str = fz_new_string(ctx, fz_to_name(key), strlen(fz_to_name(key)));
fz_obj *dest = fz_dict_gets(val, "D");
dest = fz_array_get(dest ? dest : val, 0);
if (fz_array_contains(fz_dict_gets(pages, "Kids"), dest))
{
fz_array_push(names_list, key_str);
fz_array_push(names_list, val);
}
fz_drop_obj(key_str);
}
root = fz_dict_gets(xref->trailer, "Root");
fz_dict_puts(dests, "Names", names_list);
fz_dict_puts(names, "Dests", dests);
fz_dict_puts(root, "Names", names);
fz_drop_obj(names);
fz_drop_obj(dests);
fz_drop_obj(names_list);
fz_drop_obj(olddests);
}
}
static void savefont(fz_obj *dict, int num)
{
fz_error error;
char name[1024];
char *subtype;
fz_buffer *buf;
fz_obj *stream = NULL;
fz_obj *obj;
char *ext = "";
FILE *f;
char *fontname = "font";
int n;
obj = fz_dict_gets(ctx, dict, "FontName");
if (obj)
fontname = fz_to_name(ctx, obj);
obj = fz_dict_gets(ctx, dict, "FontFile");
if (obj)
{
stream = obj;
ext = "pfa";
}
obj = fz_dict_gets(ctx, dict, "FontFile2");
if (obj)
{
stream = obj;
ext = "ttf";
}
obj = fz_dict_gets(ctx, dict, "FontFile3");
if (obj)
{
stream = obj;
obj = fz_dict_gets(ctx, obj, "Subtype");
if (obj && !fz_is_name(ctx, obj))
die(fz_error_make(ctx, "Invalid font descriptor subtype"));
subtype = fz_to_name(ctx, obj);
if (!strcmp(subtype, "Type1C"))
ext = "cff";
else if (!strcmp(subtype, "CIDFontType0C"))
ext = "cid";
else
die(fz_error_make(ctx, "Unhandled font type '%s'", subtype));
}
if (!stream)
{
fz_warn(ctx, "Unhandled font type");
return;
}
buf = fz_new_buffer(ctx, 0);
error = pdf_load_stream(&buf, xref, fz_to_num(stream), fz_to_gen(stream));
if (error)
die(error);
sprintf(name, "%s-%04d.%s", fontname, num, ext);
printf("extracting font %s\n", name);
f = fopen(name, "wb");
if (f == NULL)
die(fz_error_make(ctx, "Error creating font file"));
n = fwrite(buf->data, 1, buf->len, f);
if (n < buf->len)
die(fz_error_make(ctx, "Error writing font file"));
if (fclose(f) < 0)
die(fz_error_make(ctx, "Error closing font file"));
fz_drop_buffer(ctx, buf);
}
static fz_error
pdf_load_image_imp(fz_pixmap **imgp, pdf_xref *xref, fz_obj *rdb, fz_obj *dict, fz_stream *cstm, int forcemask)
{
fz_stream *stm;
fz_pixmap *tile;
fz_obj *obj, *res;
fz_error error;
int w, h, bpc, n;
int imagemask;
int interpolate;
int indexed;
fz_colorspace *colorspace;
fz_pixmap *mask; /* explicit mask/softmask image */
int usecolorkey;
int colorkey[FZ_MAX_COLORS * 2];
float decode[FZ_MAX_COLORS * 2];
int stride;
unsigned char *samples;
int i, len;
/* special case for JPEG2000 images */
if (pdf_is_jpx_image(dict))
{
tile = NULL;
error = pdf_load_jpx_image(&tile, xref, dict);
if (error)
return fz_rethrow(error, "cannot load jpx image");
if (forcemask)
{
if (tile->n != 2)
{
fz_drop_pixmap(tile);
return fz_throw("softmask must be grayscale");
}
mask = fz_alpha_from_gray(tile, 1);
fz_drop_pixmap(tile);
*imgp = mask;
return fz_okay;
}
*imgp = tile;
return fz_okay;
}
w = fz_to_int(fz_dict_getsa(dict, "Width", "W"));
h = fz_to_int(fz_dict_getsa(dict, "Height", "H"));
bpc = fz_to_int(fz_dict_getsa(dict, "BitsPerComponent", "BPC"));
imagemask = fz_to_bool(fz_dict_getsa(dict, "ImageMask", "IM"));
interpolate = fz_to_bool(fz_dict_getsa(dict, "Interpolate", "I"));
indexed = 0;
usecolorkey = 0;
colorspace = NULL;
mask = NULL;
if (imagemask)
bpc = 1;
if (w == 0)
return fz_throw("image width is zero");
if (h == 0)
return fz_throw("image height is zero");
if (bpc == 0)
return fz_throw("image depth is zero");
if (w > (1 << 16))
return fz_throw("image is too wide");
if (h > (1 << 16))
return fz_throw("image is too high");
obj = fz_dict_getsa(dict, "ColorSpace", "CS");
if (obj && !imagemask && !forcemask)
{
/* colorspace resource lookup is only done for inline images */
if (fz_is_name(obj))
{
res = fz_dict_get(fz_dict_gets(rdb, "ColorSpace"), obj);
if (res)
obj = res;
}
error = pdf_load_colorspace(&colorspace, xref, obj);
if (error)
return fz_rethrow(error, "cannot load image colorspace");
if (!strcmp(colorspace->name, "Indexed"))
indexed = 1;
n = colorspace->n;
}
else
{
n = 1;
}
obj = fz_dict_getsa(dict, "Decode", "D");
if (obj)
{
for (i = 0; i < n * 2; i++)
decode[i] = fz_to_real(fz_array_get(obj, i));
}
else
{
float maxval = indexed ? (1 << bpc) - 1 : 1;
for (i = 0; i < n * 2; i++)
decode[i] = i & 1 ? maxval : 0;
}
obj = fz_dict_getsa(dict, "SMask", "Mask");
if (fz_is_dict(obj))
{
/* Not allowed for inline images */
if (!cstm)
{
error = pdf_load_image_imp(&mask, xref, rdb, obj, NULL, 1);
if (error)
{
if (colorspace)
fz_drop_colorspace(colorspace);
return fz_rethrow(error, "cannot load image mask/softmask");
}
}
}
else if (fz_is_array(obj))
{
usecolorkey = 1;
for (i = 0; i < n * 2; i++)
colorkey[i] = fz_to_int(fz_array_get(obj, i));
}
/* Allocate now, to fail early if we run out of memory */
tile = fz_new_pixmap_with_limit(colorspace, w, h);
if (!tile)
{
if (colorspace)
fz_drop_colorspace(colorspace);
if (mask)
fz_drop_pixmap(mask);
return fz_throw("out of memory");
}
if (colorspace)
fz_drop_colorspace(colorspace);
tile->mask = mask;
tile->interpolate = interpolate;
stride = (w * n * bpc + 7) / 8;
if (cstm)
{
stm = pdf_open_inline_stream(cstm, xref, dict, stride * h);
}
else
{
error = pdf_open_stream(&stm, xref, fz_to_num(dict), fz_to_gen(dict));
if (error)
{
fz_drop_pixmap(tile);
return fz_rethrow(error, "cannot open image data stream (%d 0 R)", fz_to_num(dict));
}
}
samples = fz_calloc(h, stride);
len = fz_read(stm, samples, h * stride);
if (len < 0)
{
fz_close(stm);
fz_free(samples);
fz_drop_pixmap(tile);
return fz_rethrow(len, "cannot read image data");
}
/* Make sure we read the EOF marker (for inline images only) */
if (cstm)
{
unsigned char tbuf[512];
int tlen = fz_read(stm, tbuf, sizeof tbuf);
if (tlen < 0)
fz_catch(tlen, "ignoring error at end of image");
if (tlen > 0)
fz_warn("ignoring garbage at end of image");
}
fz_close(stm);
/* Pad truncated images */
if (len < stride * h)
{
fz_warn("padding truncated image (%d 0 R)", fz_to_num(dict));
memset(samples + len, 0, stride * h - len);
}
/* Invert 1-bit image masks */
if (imagemask)
{
/* 0=opaque and 1=transparent so we need to invert */
unsigned char *p = samples;
len = h * stride;
for (i = 0; i < len; i++)
p[i] = ~p[i];
}
fz_unpack_tile(tile, samples, n, bpc, stride, indexed);
fz_free(samples);
if (usecolorkey)
pdf_mask_color_key(tile, n, colorkey);
if (indexed)
{
fz_pixmap *conv;
fz_decode_indexed_tile(tile, decode, (1 << bpc) - 1);
conv = pdf_expand_indexed_pixmap(tile);
fz_drop_pixmap(tile);
tile = conv;
}
else
{
fz_decode_tile(tile, decode);
}
*imgp = tile;
return fz_okay;
}
static fz_error
pdf_load_shading_dict(fz_shade **shadep, pdf_xref *xref, fz_obj *dict, fz_matrix transform)
{
fz_error error;
fz_shade *shade;
pdf_function *func[FZ_MAX_COLORS] = { NULL };
fz_stream *stream = NULL;
fz_obj *obj;
int funcs;
int type;
int i;
shade = fz_malloc(sizeof(fz_shade));
shade->refs = 1;
shade->type = FZ_MESH;
shade->use_background = 0;
shade->use_function = 0;
shade->matrix = transform;
shade->bbox = fz_infinite_rect;
shade->extend[0] = 0;
shade->extend[1] = 0;
shade->mesh_len = 0;
shade->mesh_cap = 0;
shade->mesh = NULL;
shade->colorspace = NULL;
funcs = 0;
obj = fz_dict_gets(dict, "ShadingType");
type = fz_to_int(obj);
obj = fz_dict_gets(dict, "ColorSpace");
if (!obj)
{
fz_drop_shade(shade);
return fz_throw("shading colorspace is missing");
}
error = pdf_load_colorspace(&shade->colorspace, xref, obj);
if (error)
{
fz_drop_shade(shade);
return fz_rethrow(error, "cannot load colorspace (%d %d R)", fz_to_num(obj), fz_to_gen(obj));
}
obj = fz_dict_gets(dict, "Background");
if (obj)
{
shade->use_background = 1;
for (i = 0; i < shade->colorspace->n; i++)
shade->background[i] = fz_to_real(fz_array_get(obj, i));
}
obj = fz_dict_gets(dict, "BBox");
if (fz_is_array(obj))
{
shade->bbox = pdf_to_rect(obj);
}
obj = fz_dict_gets(dict, "Function");
if (fz_is_dict(obj))
{
funcs = 1;
error = pdf_load_function(&func[0], xref, obj);
if (error)
{
error = fz_rethrow(error, "cannot load shading function (%d %d R)", fz_to_num(obj), fz_to_gen(obj));
goto cleanup;
}
}
else if (fz_is_array(obj))
{
funcs = fz_array_len(obj);
if (funcs != 1 && funcs != shade->colorspace->n)
{
error = fz_throw("incorrect number of shading functions");
goto cleanup;
}
for (i = 0; i < funcs; i++)
{
error = pdf_load_function(&func[i], xref, fz_array_get(obj, i));
if (error)
{
error = fz_rethrow(error, "cannot load shading function (%d %d R)", fz_to_num(obj), fz_to_gen(obj));
goto cleanup;
}
}
}
if (type >= 4 && type <= 7)
{
error = pdf_open_stream(&stream, xref, fz_to_num(dict), fz_to_gen(dict));
if (error)
{
error = fz_rethrow(error, "cannot open shading stream (%d %d R)", fz_to_num(dict), fz_to_gen(dict));
goto cleanup;
}
}
switch (type)
{
case 1: pdf_load_function_based_shading(shade, xref, dict, func[0]); break;
case 2: pdf_load_axial_shading(shade, xref, dict, funcs, func); break;
case 3: pdf_load_radial_shading(shade, xref, dict, funcs, func); break;
case 4: pdf_load_type4_shade(shade, xref, dict, funcs, func, stream); break;
case 5: pdf_load_type5_shade(shade, xref, dict, funcs, func, stream); break;
case 6: pdf_load_type6_shade(shade, xref, dict, funcs, func, stream); break;
case 7: pdf_load_type7_shade(shade, xref, dict, funcs, func, stream); break;
default:
error = fz_throw("unknown shading type: %d", type);
goto cleanup;
}
if (stream)
fz_close(stream);
for (i = 0; i < funcs; i++)
if (func[i])
pdf_drop_function(func[i]);
*shadep = shade;
return fz_okay;
cleanup:
if (stream)
fz_close(stream);
for (i = 0; i < funcs; i++)
if (func[i])
pdf_drop_function(func[i]);
fz_drop_shade(shade);
return fz_rethrow(error, "cannot load shading type %d (%d %d R)", type, fz_to_num(dict), fz_to_gen(dict));
}
/*
* Create a filter given a name and param dictionary.
*/
static fz_stream *
build_filter(fz_stream *chain, pdf_xref * xref, fz_obj * f, fz_obj * p, int num, int gen)
{
fz_error error;
char *s;
s = fz_to_name(f);
if (!strcmp(s, "ASCIIHexDecode") || !strcmp(s, "AHx"))
return fz_open_ahxd(chain);
else if (!strcmp(s, "ASCII85Decode") || !strcmp(s, "A85"))
return fz_open_a85d(chain);
else if (!strcmp(s, "CCITTFaxDecode") || !strcmp(s, "CCF"))
return fz_open_faxd(chain, p);
else if (!strcmp(s, "DCTDecode") || !strcmp(s, "DCT"))
return fz_open_dctd(chain, p);
else if (!strcmp(s, "RunLengthDecode") || !strcmp(s, "RL"))
return fz_open_rld(chain);
else if (!strcmp(s, "FlateDecode") || !strcmp(s, "Fl"))
{
fz_obj *obj = fz_dict_gets(p, "Predictor");
if (fz_to_int(obj) > 1)
return fz_open_predict(fz_open_flated(chain), p);
return fz_open_flated(chain);
}
else if (!strcmp(s, "LZWDecode") || !strcmp(s, "LZW"))
{
fz_obj *obj = fz_dict_gets(p, "Predictor");
if (fz_to_int(obj) > 1)
return fz_open_predict(fz_open_lzwd(chain, p), p);
return fz_open_lzwd(chain, p);
}
else if (!strcmp(s, "JBIG2Decode"))
{
fz_obj *obj = fz_dict_gets(p, "JBIG2Globals");
if (obj)
{
fz_buffer *globals;
error = pdf_load_stream(&globals, xref, fz_to_num(obj), fz_to_gen(obj));
if (error)
fz_catch(error, "cannot load jbig2 global segments");
chain = fz_open_jbig2d(chain, globals);
fz_drop_buffer(globals);
return chain;
}
return fz_open_jbig2d(chain, NULL);
}
else if (!strcmp(s, "JPXDecode"))
return chain; /* JPX decoding is special cased in the image loading code */
else if (!strcmp(s, "Crypt"))
{
fz_obj *name;
if (!xref->crypt)
{
fz_warn("crypt filter in unencrypted document");
return chain;
}
name = fz_dict_gets(p, "Name");
if (fz_is_name(name))
return pdf_open_crypt_with_filter(chain, xref->crypt, fz_to_name(name), num, gen);
return chain;
}
fz_warn("unknown filter name (%s)", s);
return chain;
}
void
pdf_load_annots(pdf_annot **annotp, pdf_xref *xref, fz_obj *annots)
{
pdf_annot *annot, *head, *tail;
fz_obj *obj, *ap, *as, *n, *rect;
pdf_xobject *form;
fz_error error;
int i;
fz_context *ctx = xref->ctx;
head = tail = NULL;
annot = NULL;
for (i = 0; i < fz_array_len(ctx, annots); i++)
{
obj = fz_array_get(ctx, annots, i);
/* cf. http://bugs.ghostscript.com/show_bug.cgi?id=692078 */
if ((annot = pdf_update_tx_widget_annot(xref, obj)))
{
if (!head)
head = tail = annot;
else
{
tail->next = annot;
tail = annot;
}
continue;
}
rect = fz_dict_gets(ctx, obj, "Rect");
ap = fz_dict_gets(ctx, obj, "AP");
as = fz_dict_gets(ctx, obj, "AS");
if (fz_is_dict(ctx, ap))
{
n = fz_dict_gets(ctx, ap, "N"); /* normal state */
/* lookup current state in sub-dictionary */
if (!pdf_is_stream(xref, fz_to_num(n), fz_to_gen(n)))
n = fz_dict_get(ctx, n, as);
if (pdf_is_stream(xref, fz_to_num(n), fz_to_gen(n)))
{
error = pdf_load_xobject(&form, xref, n);
if (error)
{
fz_error_handle(ctx, error, "ignoring broken annotation");
continue;
}
annot = fz_malloc(ctx, sizeof(pdf_annot));
annot->obj = fz_keep_obj(obj);
annot->rect = pdf_to_rect(ctx, rect);
annot->ap = form;
annot->next = NULL;
pdf_transform_annot(annot);
if (annot)
{
if (!head)
head = tail = annot;
else
{
tail->next = annot;
tail = annot;
}
}
}
}
/* SumatraPDF: synthesize appearance streams for a few more annotations */
else if ((annot = pdf_create_annot_with_appearance(xref, obj)))
{
if (!head)
head = tail = annot;
else
{
tail->next = annot;
tail = annot;
}
}
}
*annotp = head;
}
/*
* Load CMap stream in PDF file
*/
fz_error
pdf_load_embedded_cmap(pdf_cmap **cmapp, pdf_xref *xref, fz_obj *stmobj)
{
fz_error error = fz_okay;
fz_stream *file = NULL;
pdf_cmap *cmap = NULL;
pdf_cmap *usecmap;
fz_obj *wmode;
fz_obj *obj;
if ((*cmapp = pdf_find_item(xref->store, pdf_drop_cmap, stmobj)))
{
pdf_keep_cmap(*cmapp);
return fz_okay;
}
error = pdf_open_stream(&file, xref, fz_to_num(stmobj), fz_to_gen(stmobj));
if (error)
{
error = fz_rethrow(error, "cannot open cmap stream (%d %d R)", fz_to_num(stmobj), fz_to_gen(stmobj));
goto cleanup;
}
error = pdf_parse_cmap(&cmap, file);
if (error)
{
error = fz_rethrow(error, "cannot parse cmap stream (%d %d R)", fz_to_num(stmobj), fz_to_gen(stmobj));
goto cleanup;
}
fz_close(file);
wmode = fz_dict_gets(stmobj, "WMode");
if (fz_is_int(wmode))
pdf_set_wmode(cmap, fz_to_int(wmode));
obj = fz_dict_gets(stmobj, "UseCMap");
if (fz_is_name(obj))
{
error = pdf_load_system_cmap(&usecmap, fz_to_name(obj));
if (error)
{
error = fz_rethrow(error, "cannot load system usecmap '%s'", fz_to_name(obj));
goto cleanup;
}
pdf_set_usecmap(cmap, usecmap);
pdf_drop_cmap(usecmap);
}
else if (fz_is_indirect(obj))
{
error = pdf_load_embedded_cmap(&usecmap, xref, obj);
if (error)
{
error = fz_rethrow(error, "cannot load embedded usecmap (%d %d R)", fz_to_num(obj), fz_to_gen(obj));
goto cleanup;
}
pdf_set_usecmap(cmap, usecmap);
pdf_drop_cmap(usecmap);
}
pdf_store_item(xref->store, pdf_keep_cmap, pdf_drop_cmap, stmobj, cmap);
*cmapp = cmap;
return fz_okay;
cleanup:
if (file)
fz_close(file);
if (cmap)
pdf_drop_cmap(cmap);
return error; /* already rethrown */
}
fz_error
pdf_ocg_set_config(pdf_xref *xref, int config)
{
int i, j, len, len2;
pdf_ocg_descriptor *desc = xref->ocg;
fz_obj *obj, *cobj;
char *name;
obj = fz_dict_gets(fz_dict_gets(xref->trailer, "Root"), "OCProperties");
if (obj == NULL)
{
if (config == 0)
return fz_okay;
else
return fz_throw("Unknown OCG config (None known!)");
}
if (config == 0)
{
cobj = fz_dict_gets(obj, "D");
if (cobj == NULL)
return fz_throw("No default OCG config");
}
else
{
cobj = fz_array_get(fz_dict_gets(obj, "Configs"), config);
if (cobj == NULL)
return fz_throw("Illegal OCG config");
}
if (desc->intent != NULL)
fz_drop_obj(desc->intent);
desc->intent = fz_dict_gets(cobj, "Intent");
if (desc->intent != NULL)
fz_keep_obj(desc->intent);
len = desc->len;
name = fz_to_name(fz_dict_gets(cobj, "BaseState"));
if (strcmp(name, "Unchanged") == 0)
{
/* Do nothing */
}
else if (strcmp(name, "OFF") == 0)
{
for (i = 0; i < len; i++)
{
desc->ocgs[i].state = 0;
}
}
else /* Default to ON */
{
for (i = 0; i < len; i++)
{
desc->ocgs[i].state = 1;
}
}
obj = fz_dict_gets(cobj, "ON");
len2 = fz_array_len(obj);
for (i = 0; i < len2; i++)
{
fz_obj *o = fz_array_get(obj, i);
int n = fz_to_num(o);
int g = fz_to_gen(o);
for (j=0; j < len; j++)
{
if (desc->ocgs[j].num == n && desc->ocgs[j].gen == g)
{
desc->ocgs[j].state = 1;
break;
}
}
}
obj = fz_dict_gets(cobj, "OFF");
len2 = fz_array_len(obj);
for (i = 0; i < len2; i++)
{
fz_obj *o = fz_array_get(obj, i);
int n = fz_to_num(o);
int g = fz_to_gen(o);
for (j=0; j < len; j++)
{
if (desc->ocgs[j].num == n && desc->ocgs[j].gen == g)
{
desc->ocgs[j].state = 0;
break;
}
}
}
/* FIXME: Should make 'num configs' available in the descriptor. */
/* FIXME: Should copy out 'Intent' here into the descriptor, and remove
* csi->intent in favour of that. */
/* FIXME: Should copy 'AS' into the descriptor, and visibility
* decisions should respect it. */
/* FIXME: Make 'Order' available via the descriptor (when we have an
* app that needs it) */
/* FIXME: Make 'ListMode' available via the descriptor (when we have
* an app that needs it) */
/* FIXME: Make 'RBGroups' available via the descriptor (when we have
* an app that needs it) */
/* FIXME: Make 'Locked' available via the descriptor (when we have
* an app that needs it) */
return fz_okay;
}