/* Used while reading the individual xref sections from a file */
pdf_xref_entry *pdf_get_populating_xref_entry(pdf_document *doc, int num)
{
/* Return an entry within the xref currently being populated */
pdf_xref *xref;
int i;
if (doc->num_xref_sections == 0)
{
doc->xref_sections = fz_calloc(doc->ctx, 1, sizeof(pdf_xref));
doc->num_xref_sections = 1;
}
/* Ensure all xref sections map this entry */
for (i = doc->num_xref_sections - 1; i >= 0; i--)
{
xref = &doc->xref_sections[i];
if (num >= xref->len)
pdf_resize_xref(doc->ctx, xref, num+1);
else
break; /* Remaining sections already of sufficient size */
}
/* Loop leaves xref pointing at the populating section */
return &doc->xref_sections[doc->num_xref_sections-1].table[num];
}
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_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_error
pdf_read_old_xref(fz_obj **trailerp, pdf_xref *xref, char *buf, int cap)
{
fz_error error;
int ofs, len;
char *s;
int n;
int tok;
int i;
int c;
fz_read_line(xref->file, buf, cap);
if (strncmp(buf, "xref", 4) != 0)
return fz_throw("cannot find xref marker");
while (1)
{
c = fz_peek_byte(xref->file);
if (!(c >= '0' && c <= '9'))
break;
fz_read_line(xref->file, buf, cap);
s = buf;
ofs = atoi(fz_strsep(&s, " "));
len = atoi(fz_strsep(&s, " "));
/* broken pdfs where the section is not on a separate line */
if (s && *s != '\0')
{
fz_warn("broken xref section. proceeding anyway.");
fz_seek(xref->file, -(2 + (int)strlen(s)), 1);
}
/* broken pdfs where size in trailer undershoots entries in xref sections */
if (ofs + len > xref->len)
{
fz_warn("broken xref section, proceeding anyway.");
pdf_resize_xref(xref, ofs + len);
}
for (i = ofs; i < ofs + len; i++)
{
n = fz_read(xref->file, (unsigned char *) buf, 20);
if (n < 0)
return fz_rethrow(n, "cannot read xref table");
if (!xref->table[i].type)
{
s = buf;
/* broken pdfs where line start with white space */
while (*s != '\0' && iswhite(*s))
s++;
xref->table[i].ofs = atoi(s);
xref->table[i].gen = atoi(s + 11);
xref->table[i].type = s[17];
if (s[17] != 'f' && s[17] != 'n' && s[17] != 'o')
return fz_throw("unexpected xref type: %#x (%d %d R)", s[17], i, xref->table[i].gen);
}
}
}
error = pdf_lex(&tok, xref->file, buf, cap, &n);
if (error)
return fz_rethrow(error, "cannot parse trailer");
if (tok != PDF_TOK_TRAILER)
return fz_throw("expected trailer marker");
error = pdf_lex(&tok, xref->file, buf, cap, &n);
if (error)
return fz_rethrow(error, "cannot parse trailer");
if (tok != PDF_TOK_OPEN_DICT)
return fz_throw("expected trailer dictionary");
error = pdf_parse_dict(trailerp, xref, xref->file, buf, cap);
if (error)
return fz_rethrow(error, "cannot parse trailer");
return fz_okay;
}