static void
pdf_load_mesh_params(pdf_xref *xref, fz_obj *dict, struct mesh_params *p)
{
fz_obj *obj;
int i, n;
p->x0 = p->y0 = 0;
p->x1 = p->y1 = 1;
for (i = 0; i < FZ_MAX_COLORS; i++)
{
p->c0[i] = 0;
p->c1[i] = 1;
}
p->vprow = fz_to_int(fz_dict_gets(dict, "VerticesPerRow"));
p->bpflag = fz_to_int(fz_dict_gets(dict, "BitsPerFlag"));
p->bpcoord = fz_to_int(fz_dict_gets(dict, "BitsPerCoordinate"));
p->bpcomp = fz_to_int(fz_dict_gets(dict, "BitsPerComponent"));
obj = fz_dict_gets(dict, "Decode");
if (fz_array_len(obj) >= 6)
{
n = (fz_array_len(obj) - 4) / 2;
p->x0 = fz_to_real(fz_array_get(obj, 0));
p->x1 = fz_to_real(fz_array_get(obj, 1));
p->y0 = fz_to_real(fz_array_get(obj, 2));
p->y1 = fz_to_real(fz_array_get(obj, 3));
for (i = 0; i < n; i++)
{
p->c0[i] = fz_to_real(fz_array_get(obj, 4 + i * 2));
p->c1[i] = fz_to_real(fz_array_get(obj, 5 + i * 2));
}
}
if (p->vprow < 2)
p->vprow = 2;
if (p->bpflag != 2 && p->bpflag != 4 && p->bpflag != 8)
p->bpflag = 8;
if (p->bpcoord != 1 && p->bpcoord != 2 && p->bpcoord != 4 &&
p->bpcoord != 8 && p->bpcoord != 12 && p->bpcoord != 16 &&
p->bpcoord != 24 && p->bpcoord != 32)
p->bpcoord = 8;
if (p->bpcomp != 1 && p->bpcomp != 2 && p->bpcomp != 4 &&
p->bpcomp != 8 && p->bpcomp != 12 && p->bpcomp != 16)
p->bpcomp = 8;
}
/*
* Load raw (compressed but decrypted) contents of a stream into buf.
*/
fz_error
pdf_load_raw_stream(fz_buffer **bufp, pdf_xref *xref, int num, int gen)
{
fz_error error;
fz_stream *stm;
fz_obj *dict;
int len;
error = pdf_load_object(&dict, xref, num, gen);
if (error)
return fz_rethrow(error, "cannot load stream dictionary (%d %d R)", num, gen);
len = fz_to_int(fz_dict_gets(dict, "Length"));
fz_drop_obj(dict);
error = pdf_open_raw_stream(&stm, xref, num, gen);
if (error)
return fz_rethrow(error, "cannot open raw stream (%d %d R)", num, gen);
error = fz_read_all(bufp, stm, len);
if (error)
{
fz_close(stm);
return fz_rethrow(error, "cannot read raw stream (%d %d R)", num, gen);
}
fz_close(stm);
return fz_okay;
}
/* SumatraPDF: synthesize appearance streams for a few more annotations */
static pdf_annot *
pdf_create_annot(fz_context *ctx, fz_rect rect, fz_obj *base_obj, fz_buffer *content, fz_obj *resources, int transparency)
{
pdf_annot *annot;
pdf_xobject *form;
int rotate = fz_to_int(ctx, fz_dict_gets(ctx, fz_dict_gets(ctx, base_obj, "MK"), "R"));
form = fz_malloc(ctx, sizeof(pdf_xobject));
memset(form, 0, sizeof(pdf_xobject));
form->refs = 1;
form->matrix = fz_rotate(rotate);
form->bbox.x1 = (rotate % 180 == 0) ? rect.x1 - rect.x0 : rect.y1 - rect.y0;
form->bbox.y1 = (rotate % 180 == 0) ? rect.y1 - rect.y0 : rect.x1 - rect.x0;
form->transparency = transparency;
form->isolated = !transparency;
form->contents = content;
form->resources = resources;
annot = fz_malloc(ctx, sizeof(pdf_annot));
annot->obj = base_obj;
annot->rect = rect;
annot->ap = form;
annot->next = NULL;
pdf_transform_annot(annot);
return annot;
}
/*
* Load uncompressed contents of a stream into buf.
*/
fz_error
pdf_load_stream(fz_buffer **bufp, pdf_xref *xref, int num, int gen)
{
fz_error error;
fz_stream *stm;
fz_obj *dict, *obj;
int i, len;
error = pdf_open_stream(&stm, xref, num, gen);
if (error)
return fz_rethrow(error, "cannot open stream (%d %d R)", num, gen);
error = pdf_load_object(&dict, xref, num, gen);
if (error)
return fz_rethrow(error, "cannot load stream dictionary (%d %d R)", num, gen);
len = fz_to_int(fz_dict_gets(dict, "Length"));
obj = fz_dict_gets(dict, "Filter");
len = pdf_guess_filter_length(len, fz_to_name(obj));
for (i = 0; i < fz_array_len(obj); i++)
len = pdf_guess_filter_length(len, fz_to_name(fz_array_get(obj, i)));
fz_drop_obj(dict);
error = fz_read_all(bufp, stm, len);
if (error)
{
fz_close(stm);
return fz_rethrow(error, "cannot read raw stream (%d %d R)", num, gen);
}
fz_close(stm);
return fz_okay;
}
static pdf_annot *
pdf_create_freetext_annot(pdf_xref *xref, fz_obj *obj)
{
fz_context *ctx = xref->ctx;
fz_buffer *content = fz_new_buffer(ctx, 256);
fz_buffer *base_ap = fz_new_buffer(ctx, 256);
fz_obj *ap = fz_dict_gets(ctx, obj, "DA");
fz_obj *value = fz_dict_gets(ctx, obj, "Contents");
fz_rect rect = pdf_to_rect(ctx, fz_dict_gets(ctx, obj, "Rect"));
int align = fz_to_int(ctx, fz_dict_gets(ctx, obj, "Q"));
fz_obj *res = pdf_dict_from_string(xref, ANNOT_FREETEXT_AP_RESOURCES);
unsigned short *ucs2, *rest;
float x;
char *font_name = NULL;
float font_size = pdf_extract_font_size(xref, fz_to_str_buf(ctx, ap), &font_name);
if (!font_size)
font_size = 10;
/* TODO: what resource dictionary does this font name refer to? */
if (font_name)
{
fz_obj *font = fz_dict_gets(ctx, res, "Font");
fz_dict_puts(ctx, font, font_name, fz_dict_gets(ctx, font, "Default"));
fz_free(ctx, font_name);
}
fz_buffer_printf(ctx, content, "q 1 1 %.4f %.4f re W n BT %s ",
rect.x1 - rect.x0 - 2.0f, rect.y1 - rect.y0 - 2.0f, fz_to_str_buf(ctx, ap));
fz_buffer_printf(ctx, base_ap, "q BT %s ", fz_to_str_buf(ctx, ap));
fz_buffer_printf(ctx, content, "/Default %.4f Tf ", font_size);
fz_buffer_printf(ctx, base_ap, "/Default %.4f Tf ", font_size);
fz_buffer_printf(ctx, content, "1 0 0 1 2 %.4f Tm ", rect.y1 - rect.y0 - 2);
/* Adobe Reader seems to consider "[1 0 0] r" and "1 0 0 rg" to mean the same(?) */
if (strchr(base_ap->data, '['))
{
float r, g, b;
if (sscanf(strchr(base_ap->data, '['), "[%f %f %f] r", &r, &g, &b) == 3)
fz_buffer_printf(ctx, content, "%.4f %.4f %.4f rg ", r, g, b);
}
ucs2 = pdf_to_ucs2(ctx, value);
for (rest = ucs2; *rest; rest++)
if (*rest > 0xFF)
*rest = '?';
x = 0;
rest = ucs2;
while (*rest)
rest = pdf_append_line(xref, res, content, base_ap, rest, font_size, align, rect.x1 - rect.x0 - 4.0f, 1, &x);
fz_free(ctx, ucs2);
fz_buffer_printf(ctx, content, "ET Q");
fz_drop_buffer(ctx, base_ap);
return pdf_create_annot(ctx, rect, fz_keep_obj(obj), content, res, 0);
}
static fz_error
pdf_read_xref_sections(pdf_xref *xref, int ofs, char *buf, int cap)
{
fz_error error;
fz_obj *trailer;
fz_obj *prev;
fz_obj *xrefstm;
error = pdf_read_xref(&trailer, xref, ofs, buf, cap);
if (error)
return fz_rethrow(error, "cannot read xref section");
/* FIXME: do we overwrite free entries properly? */
xrefstm = fz_dict_gets(trailer, "XRefStm");
if (xrefstm)
{
error = pdf_read_xref_sections(xref, fz_to_int(xrefstm), buf, cap);
if (error)
{
fz_drop_obj(trailer);
return fz_rethrow(error, "cannot read /XRefStm xref section");
}
}
prev = fz_dict_gets(trailer, "Prev");
if (prev)
{
error = pdf_read_xref_sections(xref, fz_to_int(prev), buf, cap);
if (error)
{
fz_drop_obj(trailer);
return fz_rethrow(error, "cannot read /Prev xref section");
}
}
fz_drop_obj(trailer);
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_error
load_icc_based(fz_colorspace **csp, pdf_xref *xref, fz_obj *dict)
{
int n;
n = fz_to_int(fz_dict_gets(dict, "N"));
switch (n)
{
case 1: *csp = fz_device_gray; return fz_okay;
case 3: *csp = fz_device_rgb; return fz_okay;
case 4: *csp = fz_device_cmyk; return fz_okay;
}
return fz_throw("syntaxerror: ICCBased must have 1, 3 or 4 components");
}
static fz_error
pdf_load_xref(pdf_xref *xref, char *buf, int bufsize)
{
fz_error error;
fz_obj *size;
int i;
error = pdf_load_version(xref);
if (error)
return fz_rethrow(error, "cannot read version marker");
error = pdf_read_start_xref(xref);
if (error)
return fz_rethrow(error, "cannot read startxref");
error = pdf_read_trailer(xref, buf, bufsize);
if (error)
return fz_rethrow(error, "cannot read trailer");
size = fz_dict_gets(xref->trailer, "Size");
if (!size)
return fz_throw("trailer missing Size entry");
pdf_resize_xref(xref, fz_to_int(size));
error = pdf_read_xref_sections(xref, xref->startxref, buf, bufsize);
if (error)
return fz_rethrow(error, "cannot read xref");
/* broken pdfs where first object is not free */
if (xref->table[0].type != 'f')
return fz_throw("first object in xref is not free");
/* broken pdfs where object offsets are out of range */
for (i = 0; i < xref->len; i++)
{
if (xref->table[i].type == 'n')
if (xref->table[i].ofs <= 0 || xref->table[i].ofs >= xref->file_size)
return fz_throw("object offset out of range: %d (%d 0 R)", xref->table[i].ofs, i);
if (xref->table[i].type == 'o')
if (xref->table[i].ofs <= 0 || xref->table[i].ofs >= xref->len || xref->table[xref->table[i].ofs].type != 'n')
return fz_throw("invalid reference to an objstm that does not exist: %d (%d 0 R)", xref->table[i].ofs, i);
}
return fz_okay;
}
static fz_colorspace *
load_icc_based(pdf_document *xref, fz_obj *dict)
{
int n;
n = fz_to_int(fz_dict_gets(dict, "N"));
switch (n)
{
case 1: return fz_device_gray;
case 3: return fz_device_rgb;
case 4: return fz_device_cmyk;
}
fz_throw(xref->ctx, "syntaxerror: ICCBased must have 1, 3 or 4 components");
return NULL; /* Stupid MSVC */
}
static int _pdf_doc_find_page_no(
struct _pdf_doc *self, fz_obj *dest)
{
if (fz_is_dict(dest)) {
/* The destination is linked from a Go-To action's D array. */
fz_obj * D = fz_dict_gets(dest, "D");
if (D && fz_is_array(D))
dest = D;
}
if (fz_is_array(dest))
dest = fz_array_get(dest, 0);
if (fz_is_int(dest))
return fz_to_int(dest);
return pdf_find_page_number(self->xref, dest);
}
/*
* Build a filter for reading raw stream data.
* This is a null filter to constrain reading to the
* stream length, followed by a decryption filter.
*/
static fz_stream *
pdf_open_raw_filter(fz_stream *chain, pdf_xref *xref, fz_obj *stmobj, int num, int gen)
{
int hascrypt;
int len;
/* don't close chain when we close this filter */
fz_keep_stream(chain);
len = fz_to_int(fz_dict_gets(stmobj, "Length"));
chain = fz_open_null(chain, len);
hascrypt = pdf_stream_has_crypt(stmobj);
if (xref->crypt && !hascrypt)
chain = pdf_open_crypt(chain, xref->crypt, num, gen);
return chain;
}
fz_stream *
fz_open_dctd(fz_stream *chain, fz_obj *params)
{
fz_dctd *state;
fz_obj *obj;
state = fz_malloc(chain->ctx, sizeof(fz_dctd));
memset(state, 0, sizeof(fz_dctd));
state->chain = chain;
state->color_transform = -1; /* unset */
state->init = 0;
obj = fz_dict_gets(chain->ctx, params, "ColorTransform");
if (obj)
state->color_transform = fz_to_int(chain->ctx, obj);
return fz_new_stream(chain->ctx, state, read_dctd, close_dctd);
}
/*
* 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 */
}
/*
* 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;
}
static pdf_annot *
pdf_update_tx_widget_annot(pdf_xref *xref, fz_obj *obj)
{
fz_obj *ap, *res, *value;
fz_rect rect;
fz_buffer *content, *base_ap;
int flags, align, rotate, is_multiline;
float font_size, x, y;
char *font_name;
unsigned short *ucs2, *rest;
fz_context *ctx = xref->ctx;
if (strcmp(fz_to_name(ctx, fz_dict_gets(ctx, obj, "Subtype")), "Widget") != 0)
return NULL;
if (!fz_to_bool(ctx, pdf_dict_get_inheritable(xref, NULL, "NeedAppearances")) && pdf_get_ap_stream(xref, obj))
return NULL;
value = pdf_dict_get_inheritable(xref, obj, "FT");
if (strcmp(fz_to_name(ctx, value), "Tx") != 0)
return NULL;
ap = pdf_dict_get_inheritable(xref, obj, "DA");
value = pdf_dict_get_inheritable(xref, obj, "V");
if (!ap || !value)
return NULL;
res = pdf_dict_get_inheritable(xref, obj, "DR");
rect = pdf_to_rect(ctx, fz_dict_gets(ctx, obj, "Rect"));
rotate = fz_to_int(ctx, fz_dict_gets(ctx, fz_dict_gets(ctx, obj, "MK"), "R"));
rect = fz_transform_rect(fz_rotate(rotate), rect);
flags = fz_to_int(ctx, fz_dict_gets(ctx, obj, "Ff"));
is_multiline = (flags & (1 << 12)) != 0;
if ((flags & (1 << 25) /* richtext */))
fz_warn(ctx, "missing support for richtext fields");
align = fz_to_int(ctx, fz_dict_gets(ctx, obj, "Q"));
font_size = pdf_extract_font_size(xref, fz_to_str_buf(ctx, ap), &font_name);
if (!font_size || !font_name)
font_size = is_multiline ? 10 /* FIXME */ : floor(rect.y1 - rect.y0 - 2);
content = fz_new_buffer(ctx, 256);
base_ap = fz_new_buffer(ctx, 256);
pdf_prepend_ap_background(content, xref, obj);
fz_buffer_printf(ctx, content, "/Tx BMC q 1 1 %.4f %.4f re W n BT %s ",
rect.x1 - rect.x0 - 2.0f, rect.y1 - rect.y0 - 2.0f, fz_to_str_buf(ctx, ap));
fz_buffer_printf(ctx, base_ap, "/Tx BMC q BT %s ", fz_to_str_buf(ctx, ap));
if (font_name)
{
fz_buffer_printf(ctx, content, "/%s %.4f Tf ", font_name, font_size);
fz_buffer_printf(ctx, base_ap, "/%s %.4f Tf ", font_name, font_size);
fz_free(ctx, font_name);
}
y = 0.5f * (rect.y1 - rect.y0) + 0.6f * font_size;
if (is_multiline)
y = rect.y1 - rect.y0 - 2;
fz_buffer_printf(ctx, content, "1 0 0 1 2 %.4f Tm ", y);
ucs2 = pdf_to_ucs2(ctx, value);
for (rest = ucs2; *rest; rest++)
if (*rest > 0xFF)
*rest = '?';
if ((flags & (1 << 13) /* password */))
for (rest = ucs2; *rest; rest++)
*rest = '*';
x = 0;
rest = ucs2;
if ((flags & (1 << 24) /* comb */))
{
pdf_append_combed_line(xref, res, content, base_ap, ucs2, font_size, rect.x1 - rect.x0, fz_to_int(ctx, pdf_dict_get_inheritable(xref, obj, "MaxLen")));
rest = L"";
}
while (*rest)
rest = pdf_append_line(xref, res, content, base_ap, rest, font_size, align, rect.x1 - rect.x0 - 4.0f, is_multiline, &x);
fz_free(ctx, ucs2);
fz_buffer_printf(ctx, content, "ET Q EMC");
fz_drop_buffer(ctx, base_ap);
rect = fz_transform_rect(fz_rotate(-rotate), rect);
return pdf_create_annot(ctx, rect, fz_keep_obj(obj), content, res ? fz_keep_obj(res) : NULL, 0);
}
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_error
pdf_read_new_xref(fz_obj **trailerp, pdf_xref *xref, char *buf, int cap)
{
fz_error error;
fz_stream *stm;
fz_obj *trailer;
fz_obj *index;
fz_obj *obj;
int num, gen, stm_ofs;
int size, w0, w1, w2;
int t;
error = pdf_parse_ind_obj(&trailer, xref, xref->file, buf, cap, &num, &gen, &stm_ofs);
if (error)
return fz_rethrow(error, "cannot parse compressed xref stream object");
obj = fz_dict_gets(trailer, "Size");
if (!obj)
{
fz_drop_obj(trailer);
return fz_throw("xref stream missing Size entry (%d %d R)", num, gen);
}
size = fz_to_int(obj);
if (size > xref->len)
{
pdf_resize_xref(xref, size);
}
if (num < 0 || num >= xref->len)
{
fz_drop_obj(trailer);
return fz_throw("object id (%d %d R) out of range (0..%d)", num, gen, xref->len - 1);
}
obj = fz_dict_gets(trailer, "W");
if (!obj) {
fz_drop_obj(trailer);
return fz_throw("xref stream missing W entry (%d %d R)", num, gen);
}
w0 = fz_to_int(fz_array_get(obj, 0));
w1 = fz_to_int(fz_array_get(obj, 1));
w2 = fz_to_int(fz_array_get(obj, 2));
index = fz_dict_gets(trailer, "Index");
error = pdf_open_stream_at(&stm, xref, num, gen, trailer, stm_ofs);
if (error)
{
fz_drop_obj(trailer);
return fz_rethrow(error, "cannot open compressed xref stream (%d %d R)", num, gen);
}
if (!index)
{
error = pdf_read_new_xref_section(xref, stm, 0, size, w0, w1, w2);
if (error)
{
fz_close(stm);
fz_drop_obj(trailer);
return fz_rethrow(error, "cannot read xref stream (%d %d R)", num, gen);
}
}
else
{
for (t = 0; t < fz_array_len(index); t += 2)
{
int i0 = fz_to_int(fz_array_get(index, t + 0));
int i1 = fz_to_int(fz_array_get(index, t + 1));
error = pdf_read_new_xref_section(xref, stm, i0, i1, w0, w1, w2);
if (error)
{
fz_close(stm);
fz_drop_obj(trailer);
return fz_rethrow(error, "cannot read xref stream section (%d %d R)", num, gen);
}
}
}
fz_close(stm);
*trailerp = trailer;
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");
}
}
static fz_outline *
pdf_load_outline_imp(pdf_xref *xref, fz_obj *dict)
{
pdf_link *link;
fz_outline *node;
fz_obj *obj;
/* SumatraPDF: prevent potential stack overflow */
fz_outline *prev, *root = NULL;
fz_obj *origDict = dict;
fz_context *ctx = xref->ctx;
if (fz_is_null(ctx, dict))
return NULL;
/* SumatraPDF: prevent cyclic outlines */
do
{
if (fz_dict_gets(ctx, dict, ".seen"))
break;
obj = fz_new_null(ctx);
fz_dict_puts(ctx, dict, ".seen", obj);
fz_drop_obj(ctx, obj);
node = fz_malloc(ctx, sizeof(fz_outline));
node->title = NULL;
node->page = -1;
node->down = NULL;
node->next = NULL;
obj = fz_dict_gets(ctx, dict, "Title");
if (obj)
node->title = pdf_to_utf8(ctx, obj);
/* SumatraPDF: support expansion states */
node->is_open = fz_to_int(ctx, fz_dict_gets(ctx, dict, "Count")) >= 0;
/* SumatraPDF: extended outline actions */
node->data = node->free_data = NULL;
if (fz_dict_gets(ctx, dict, "Dest") || fz_dict_gets(ctx, dict, "A"))
{
link = pdf_load_link(xref, dict);
if (link) /* SumatraPDF: don't crash if it's no link after all */
{
if (link->kind == PDF_LINK_GOTO)
node->page = pdf_find_page_number(xref, fz_array_get(ctx, link->dest, 0));
/* SumatraPDF: extended outline actions */
node->data = link;
node->free_data = pdf_free_link;
}
}
obj = fz_dict_gets(ctx, dict, "First");
if (obj)
node->down = pdf_load_outline_imp(xref, obj);
/* SumatraPDF: prevent potential stack overflow */
if (!root)
prev = root = node;
else
prev = prev->next = node;
dict = fz_dict_gets(ctx, dict, "Next");
} while (dict && !fz_is_null(ctx, dict));
node = root;
/* SumatraPDF: prevent cyclic outlines */
for (dict = origDict; dict && fz_dict_gets(ctx, dict, ".seen"); dict = fz_dict_gets(ctx, dict, "Next"))
fz_dict_dels(ctx, dict, ".seen");
return node;
}
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_obj_stm(pdf_xref *xref, int num, int gen, char *buf, int cap)
{
fz_error error;
fz_stream *stm;
fz_obj *objstm;
int *numbuf;
int *ofsbuf;
fz_obj *obj;
int first;
int count;
int i, n;
int tok;
error = pdf_load_object(&objstm, xref, num, gen);
if (error)
return fz_rethrow(error, "cannot load object stream object (%d %d R)", num, gen);
count = fz_to_int(fz_dict_gets(objstm, "N"));
first = fz_to_int(fz_dict_gets(objstm, "First"));
numbuf = fz_calloc(count, sizeof(int));
ofsbuf = fz_calloc(count, sizeof(int));
error = pdf_open_stream(&stm, xref, num, gen);
if (error)
{
error = fz_rethrow(error, "cannot open object stream (%d %d R)", num, gen);
goto cleanupbuf;
}
for (i = 0; i < count; i++)
{
error = pdf_lex(&tok, stm, buf, cap, &n);
if (error || tok != PDF_TOK_INT)
{
error = fz_rethrow(error, "corrupt object stream (%d %d R)", num, gen);
goto cleanupstm;
}
numbuf[i] = atoi(buf);
error = pdf_lex(&tok, stm, buf, cap, &n);
if (error || tok != PDF_TOK_INT)
{
error = fz_rethrow(error, "corrupt object stream (%d %d R)", num, gen);
goto cleanupstm;
}
ofsbuf[i] = atoi(buf);
}
fz_seek(stm, first, 0);
for (i = 0; i < count; i++)
{
fz_seek(stm, first + ofsbuf[i], 0);
error = pdf_parse_stm_obj(&obj, xref, stm, buf, cap);
if (error)
{
error = fz_rethrow(error, "cannot parse object %d in stream (%d %d R)", i, num, gen);
goto cleanupstm;
}
if (numbuf[i] < 1 || numbuf[i] >= xref->len)
{
fz_drop_obj(obj);
error = fz_throw("object id (%d 0 R) out of range (0..%d)", numbuf[i], xref->len - 1);
goto cleanupstm;
}
if (xref->table[numbuf[i]].type == 'o' && xref->table[numbuf[i]].ofs == num)
{
if (xref->table[numbuf[i]].obj)
fz_drop_obj(xref->table[numbuf[i]].obj);
xref->table[numbuf[i]].obj = obj;
}
else
{
fz_drop_obj(obj);
}
}
fz_close(stm);
fz_free(ofsbuf);
fz_free(numbuf);
fz_drop_obj(objstm);
return fz_okay;
cleanupstm:
fz_close(stm);
cleanupbuf:
fz_free(ofsbuf);
fz_free(numbuf);
fz_drop_obj(objstm);
return error; /* already rethrown */
}
static int _pdf_doc_load(struct _pdf_doc *self, mume_stream_t *stm)
{
fz_error error;
fz_stream *fzstm;
fz_rect *mbox;
fz_obj *page_obj, *box_obj;
int i, c;
_pdf_doc_clear(self);
fzstm = fz_new_stream(stm, _pdf_stream_read, _pdf_stream_close);
mume_stream_reference(stm);
fzstm->seek = _pdf_stream_seek;
error = pdf_open_xref_with_stream(&self->xref, fzstm, NULL);
fz_close(fzstm);
if (error) {
mume_error(("Read xref failed\n", error));
return 0;
}
assert(!pdf_needs_password(self->xref));
/* Load meta information. */
error = pdf_load_page_tree(self->xref);
if (error) {
mume_error(("Cannot load page tree\n"));
return 0;
}
c = pdf_count_pages(self->xref);
self->glyph_cache = fz_new_glyph_cache();
self->pages = calloc_abort(c, sizeof(pdf_page*));
self->disps = calloc_abort(c, sizeof(fz_display_list*));
self->media_boxes = malloc_abort(c * sizeof(fz_rect));
self->page_rotates = malloc_abort(c * sizeof(int));
/* Extract each pages' media box and rotation. */
for (i = 0; i < c; ++i) {
mbox = self->media_boxes + i;
page_obj = self->xref->page_objs[i];
if (!page_obj) {
*mbox = fz_empty_rect;
continue;
}
box_obj = fz_dict_gets(page_obj, "MediaBox");
*mbox = pdf_to_rect(box_obj);
if (fz_is_empty_rect(*mbox)) {
fz_warn("Cannot find page bounds, guessing page bounds.");
mbox->x1 = 612;
mbox->y1 = 792;
}
box_obj = fz_dict_gets(page_obj, "CropBox");
if (fz_is_array(box_obj))
*mbox = fz_intersect_rect(*mbox, pdf_to_rect(box_obj));
self->page_rotates[i] = fz_to_int(
fz_dict_gets(page_obj, "Rotate"));
if (self->page_rotates[i] % 90)
self->page_rotates[i] = 0;
}
return 1;
}
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));
}