pdf_obj *
pdf_parse_stm_obj(pdf_document *doc, fz_stream *file, pdf_lexbuf *buf)
{
pdf_token tok;
fz_context *ctx = file->ctx;
tok = pdf_lex(file, buf);
switch (tok)
{
case PDF_TOK_OPEN_ARRAY:
return pdf_parse_array(doc, file, buf);
case PDF_TOK_OPEN_DICT:
return pdf_parse_dict(doc, file, buf);
case PDF_TOK_NAME: return pdf_new_name(doc, buf->scratch); break;
case PDF_TOK_REAL: return pdf_new_real(doc, buf->f); break;
case PDF_TOK_STRING: return pdf_new_string(doc, buf->scratch, buf->len); break;
case PDF_TOK_TRUE: return pdf_new_bool(doc, 1); break;
case PDF_TOK_FALSE: return pdf_new_bool(doc, 0); break;
case PDF_TOK_NULL: return pdf_new_null(doc); break;
case PDF_TOK_INT: return pdf_new_int(doc, buf->i); break;
default: fz_throw(ctx, FZ_ERROR_GENERIC, "unknown token in object stream");
}
return NULL; /* Stupid MSVC */
}
static fz_image *
parse_inline_image(fz_context *ctx, pdf_csi *csi, fz_stream *stm)
{
pdf_document *doc = csi->doc;
pdf_obj *rdb = csi->rdb;
pdf_obj *obj = NULL;
fz_image *img = NULL;
int ch, found;
fz_var(obj);
fz_var(img);
fz_try(ctx)
{
obj = pdf_parse_dict(ctx, doc, stm, &doc->lexbuf.base);
/* read whitespace after ID keyword */
ch = fz_read_byte(ctx, stm);
if (ch == '\r')
if (fz_peek_byte(ctx, stm) == '\n')
fz_read_byte(ctx, stm);
img = pdf_load_inline_image(ctx, doc, rdb, obj, stm);
/* find EI */
found = 0;
ch = fz_read_byte(ctx, stm);
do
{
while (ch != 'E' && ch != EOF)
ch = fz_read_byte(ctx, stm);
if (ch == 'E')
{
ch = fz_read_byte(ctx, stm);
if (ch == 'I')
{
ch = fz_peek_byte(ctx, stm);
if (ch == ' ' || ch <= 32 || ch == EOF || ch == '<' || ch == '/')
{
found = 1;
break;
}
}
}
} while (ch != EOF);
if (!found)
fz_throw(ctx, FZ_ERROR_GENERIC, "syntax error after inline image");
}
fz_always(ctx)
{
pdf_drop_obj(ctx, obj);
}
fz_catch(ctx)
{
fz_drop_image(ctx, img);
fz_rethrow(ctx);
}
return img;
}
fz_error
pdf_parse_stm_obj(fz_obj **op, pdf_xref *xref, fz_stream *file, char *buf, int cap)
{
fz_error error;
int tok;
int len;
error = pdf_lex(&tok, file, buf, cap, &len);
if (error)
return fz_rethrow(error, "cannot parse token in object stream");
switch (tok)
{
case PDF_TOK_OPEN_ARRAY:
error = pdf_parse_array(op, xref, file, buf, cap);
if (error)
return fz_rethrow(error, "cannot parse object stream");
break;
case PDF_TOK_OPEN_DICT:
error = pdf_parse_dict(op, xref, file, buf, cap);
if (error)
return fz_rethrow(error, "cannot parse object stream");
break;
case PDF_TOK_NAME: *op = fz_new_name(buf); break;
case PDF_TOK_REAL: *op = fz_new_real(fz_atof(buf)); break;
case PDF_TOK_STRING: *op = fz_new_string(buf, len); break;
case PDF_TOK_TRUE: *op = fz_new_bool(1); break;
case PDF_TOK_FALSE: *op = fz_new_bool(0); break;
case PDF_TOK_NULL: *op = fz_new_null(); break;
case PDF_TOK_INT: *op = fz_new_int(atoi(buf)); break;
default: return fz_throw("unknown token in object stream");
}
return fz_okay;
}
static fz_error
pdf_read_old_trailer(pdf_xref *xref, char *buf, int cap)
{
fz_error error;
int len;
char *s;
int n;
int t;
int tok;
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;
fz_strsep(&s, " "); /* ignore ofs */
if (!s)
return fz_throw("invalid range marker in xref");
len = atoi(fz_strsep(&s, " "));
/* broken pdfs where the section is not on a separate line */
if (s && *s != '\0')
fz_seek(xref->file, -(2 + (int)strlen(s)), 1);
t = fz_tell(xref->file);
if (t < 0)
return fz_throw("cannot tell in file");
fz_seek(xref->file, t + 20 * len, 0);
}
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(&xref->trailer, xref, xref->file, buf, cap);
if (error)
return fz_rethrow(error, "cannot parse trailer");
return fz_okay;
}
pdf_obj *
pdf_parse_stm_obj(pdf_document *xref, fz_stream *file, pdf_lexbuf *buf)
{
int tok;
fz_context *ctx = file->ctx;
tok = pdf_lex(file, buf);
/* RJW: "cannot parse token in object stream") */
switch (tok)
{
case PDF_TOK_OPEN_ARRAY:
return pdf_parse_array(xref, file, buf);
/* RJW: "cannot parse object stream" */
case PDF_TOK_OPEN_DICT:
return pdf_parse_dict(xref, file, buf);
/* RJW: "cannot parse object stream" */
case PDF_TOK_NAME:
return fz_new_name(ctx, buf->scratch);
break;
case PDF_TOK_REAL:
return pdf_new_real(ctx, buf->f);
break;
case PDF_TOK_STRING:
return pdf_new_string(ctx, buf->scratch, buf->len);
break;
case PDF_TOK_TRUE:
return pdf_new_bool(ctx, 1);
break;
case PDF_TOK_FALSE:
return pdf_new_bool(ctx, 0);
break;
case PDF_TOK_NULL:
return pdf_new_null(ctx);
break;
case PDF_TOK_INT:
return pdf_new_int(ctx, buf->i);
break;
default:
fz_throw(ctx, "unknown token in object stream");
}
return NULL; /* Stupid MSVC */
}
pdf_obj *
pdf_parse_dict(pdf_document *xref, fz_stream *file, pdf_lexbuf *buf)
{
pdf_obj *dict;
pdf_obj *key = NULL;
pdf_obj *val = NULL;
int tok;
int a, b;
fz_context *ctx = file->ctx;
dict = pdf_new_dict(ctx, 8);
fz_var(key);
fz_var(val);
fz_try(ctx)
{
while (1)
{
tok = pdf_lex(file, buf);
skip:
if (tok == PDF_TOK_CLOSE_DICT)
break;
/* for BI .. ID .. EI in content streams */
if (tok == PDF_TOK_KEYWORD && !strcmp(buf->scratch, "ID"))
break;
if (tok != PDF_TOK_NAME)
fz_throw(ctx, "invalid key in dict");
key = fz_new_name(ctx, buf->scratch);
tok = pdf_lex(file, buf);
switch (tok)
{
case PDF_TOK_OPEN_ARRAY:
/* cf. http://code.google.com/p/sumatrapdf/issues/detail?id=1643 */
fz_try(ctx)
{
val = pdf_parse_array(xref, file, buf);
}
fz_catch(ctx)
{
fz_warn(ctx, "ignoring broken array for '%s'", pdf_to_name(key));
pdf_drop_obj(key);
val = key = NULL;
do
tok = pdf_lex(file, buf);
while (tok != PDF_TOK_CLOSE_DICT && tok != PDF_TOK_CLOSE_ARRAY &&
tok != PDF_TOK_EOF && tok != PDF_TOK_OPEN_ARRAY && tok != PDF_TOK_OPEN_DICT);
if (tok == PDF_TOK_CLOSE_DICT)
goto skip;
if (tok == PDF_TOK_CLOSE_ARRAY)
continue;
fz_throw(ctx, "cannot make sense of broken array after all");
}
break;
case PDF_TOK_OPEN_DICT:
val = pdf_parse_dict(xref, file, buf);
break;
case PDF_TOK_NAME:
val = fz_new_name(ctx, buf->scratch);
break;
case PDF_TOK_REAL:
val = pdf_new_real(ctx, buf->f);
break;
case PDF_TOK_STRING:
val = pdf_new_string(ctx, buf->scratch, buf->len);
break;
case PDF_TOK_TRUE:
val = pdf_new_bool(ctx, 1);
break;
case PDF_TOK_FALSE:
val = pdf_new_bool(ctx, 0);
break;
case PDF_TOK_NULL:
val = pdf_new_null(ctx);
break;
case PDF_TOK_INT:
/* 64-bit to allow for numbers > INT_MAX and overflow */
a = buf->i;
tok = pdf_lex(file, buf);
if (tok == PDF_TOK_CLOSE_DICT || tok == PDF_TOK_NAME ||
(tok == PDF_TOK_KEYWORD && !strcmp(buf->scratch, "ID")))
{
val = pdf_new_int(ctx, a);
fz_dict_put(dict, key, val);
pdf_drop_obj(val);
val = NULL;
pdf_drop_obj(key);
key = NULL;
goto skip;
}
if (tok == PDF_TOK_INT)
{
b = buf->i;
tok = pdf_lex(file, buf);
if (tok == PDF_TOK_R)
{
val = pdf_new_indirect(ctx, a, b, xref);
break;
}
}
fz_throw(ctx, "invalid indirect reference in dict");
default:
fz_throw(ctx, "unknown token in dict");
}
fz_dict_put(dict, key, val);
pdf_drop_obj(val);
val = NULL;
pdf_drop_obj(key);
key = NULL;
}
}
fz_catch(ctx)
{
pdf_drop_obj(dict);
pdf_drop_obj(key);
pdf_drop_obj(val);
fz_throw(ctx, "cannot parse dict");
}
return dict;
}
pdf_obj *
pdf_parse_array(pdf_document *xref, fz_stream *file, pdf_lexbuf *buf)
{
pdf_obj *ary = NULL;
pdf_obj *obj = NULL;
int a = 0, b = 0, n = 0;
int tok;
fz_context *ctx = file->ctx;
pdf_obj *op;
fz_var(obj);
ary = pdf_new_array(ctx, 4);
fz_try(ctx)
{
while (1)
{
tok = pdf_lex(file, buf);
if (tok != PDF_TOK_INT && tok != PDF_TOK_R)
{
if (n > 0)
{
obj = pdf_new_int(ctx, a);
pdf_array_push(ary, obj);
pdf_drop_obj(obj);
obj = NULL;
}
if (n > 1)
{
obj = pdf_new_int(ctx, b);
pdf_array_push(ary, obj);
pdf_drop_obj(obj);
obj = NULL;
}
n = 0;
}
if (tok == PDF_TOK_INT && n == 2)
{
obj = pdf_new_int(ctx, a);
pdf_array_push(ary, obj);
pdf_drop_obj(obj);
obj = NULL;
a = b;
n --;
}
switch (tok)
{
case PDF_TOK_CLOSE_ARRAY:
op = ary;
goto end;
case PDF_TOK_INT:
if (n == 0)
a = buf->i;
if (n == 1)
b = buf->i;
n ++;
break;
case PDF_TOK_R:
if (n != 2)
fz_throw(ctx, "cannot parse indirect reference in array");
obj = pdf_new_indirect(ctx, a, b, xref);
pdf_array_push(ary, obj);
pdf_drop_obj(obj);
obj = NULL;
n = 0;
break;
case PDF_TOK_OPEN_ARRAY:
obj = pdf_parse_array(xref, file, buf);
pdf_array_push(ary, obj);
pdf_drop_obj(obj);
obj = NULL;
break;
case PDF_TOK_OPEN_DICT:
obj = pdf_parse_dict(xref, file, buf);
pdf_array_push(ary, obj);
pdf_drop_obj(obj);
obj = NULL;
break;
case PDF_TOK_NAME:
obj = fz_new_name(ctx, buf->scratch);
pdf_array_push(ary, obj);
pdf_drop_obj(obj);
obj = NULL;
break;
case PDF_TOK_REAL:
obj = pdf_new_real(ctx, buf->f);
pdf_array_push(ary, obj);
pdf_drop_obj(obj);
obj = NULL;
break;
case PDF_TOK_STRING:
obj = pdf_new_string(ctx, buf->scratch, buf->len);
pdf_array_push(ary, obj);
pdf_drop_obj(obj);
obj = NULL;
break;
case PDF_TOK_TRUE:
obj = pdf_new_bool(ctx, 1);
pdf_array_push(ary, obj);
pdf_drop_obj(obj);
obj = NULL;
break;
case PDF_TOK_FALSE:
obj = pdf_new_bool(ctx, 0);
pdf_array_push(ary, obj);
pdf_drop_obj(obj);
obj = NULL;
break;
case PDF_TOK_NULL:
obj = pdf_new_null(ctx);
pdf_array_push(ary, obj);
pdf_drop_obj(obj);
obj = NULL;
break;
default:
fz_throw(ctx, "cannot parse token in array");
}
}
end:
{}
}
fz_catch(ctx)
{
pdf_drop_obj(obj);
pdf_drop_obj(ary);
fz_throw(ctx, "cannot parse array");
}
return op;
}
struct pdf_array *pdf_parse_array(struct pdf_struct *pdf, struct pdf_obj *obj, size_t objsz, char *begin, char **endchar)
{
struct pdf_array *res=NULL;
struct pdf_array_node *node=NULL;
const char *objstart;
char *end;
int in_string=0, ninner=0;
/* Sanity checking */
if (!(pdf) || !(obj) || !(begin))
return NULL;
objstart = obj->start + pdf->map;
if (begin < objstart || (size_t)(begin - objstart) >= objsz)
return NULL;
if (begin[0] != '[')
return NULL;
/* Find the end of the array */
end = begin;
while ((size_t)(end - objstart) < objsz) {
if (in_string) {
if (*end == '\\') {
end += 2;
continue;
}
if (*end == ')')
in_string = 0;
end++;
continue;
}
switch (*end) {
case '(':
in_string=1;
break;
case '[':
ninner++;
break;
case ']':
ninner--;
break;
}
if (*end == ']' && ninner == 0)
break;
end++;
}
/* More sanity checking */
if ((size_t)(end - objstart) == objsz)
return NULL;
if (*end != ']')
return NULL;
res = cli_calloc(1, sizeof(struct pdf_array));
if (!(res))
return NULL;
begin++;
while (begin < end) {
char *val=NULL, *p1;
struct pdf_array *arr=NULL;
struct pdf_dict *dict=NULL;
while (begin < end && isspace(begin[0]))
begin++;
if (begin == end)
break;
switch (begin[0]) {
case '<':
if ((size_t)(begin - objstart) < objsz - 2 && begin[1] == '<') {
dict = pdf_parse_dict(pdf, obj, objsz, begin, &begin);
begin+=2;
break;
}
/* Not a dictionary. Intentially fall through. */
case '(':
val = pdf_parse_string(pdf, obj, begin, objsz, NULL, &begin, NULL);
begin += 2;
break;
case '[':
/* XXX We should have a recursion counter here */
arr = pdf_parse_array(pdf, obj, objsz, begin, &begin);
begin+=1;
break;
default:
p1 = end;
if (!is_object_reference(begin, &p1, NULL)) {
p1 = begin+1;
while (p1 < end && !isspace(p1[0]))
p1++;
}
val = cli_calloc((p1 - begin) + 2, 1);
if (!(val))
break;
strncpy(val, begin, p1 - begin);
val[p1 - begin] = '\0';
begin = p1;
break;
}
/* Parse error, just return what we could */
if (!(val) && !(arr) && !(dict))
break;
if (!(node)) {
res->nodes = res->tail = node = calloc(1, sizeof(struct pdf_array_node));
if (!(node)) {
if (dict)
pdf_free_dict(dict);
if (val)
free(val);
if (arr)
pdf_free_array(arr);
break;
}
} else {
node = calloc(1, sizeof(struct pdf_array_node));
if (!(node)) {
if (dict)
pdf_free_dict(dict);
if (val)
free(val);
if (arr)
pdf_free_array(arr);
break;
}
node->prev = res->tail;
if (res->tail)
res->tail->next = node;
res->tail = node;
}
if (val != NULL) {
node->type = PDF_ARR_STRING;
node->data = val;
node->datasz = strlen(val);
} else if (dict != NULL) {
node->type = PDF_ARR_DICT;
node->data = dict;
node->datasz = sizeof(struct pdf_dict);
} else {
node->type = PDF_ARR_ARRAY;
node->data = arr;
node->datasz = sizeof(struct pdf_array);
}
}
if (endchar)
*endchar = end;
return res;
}
struct pdf_dict *pdf_parse_dict(struct pdf_struct *pdf, struct pdf_obj *obj, size_t objsz, char *begin, char **endchar)
{
struct pdf_dict *res=NULL;
struct pdf_dict_node *node=NULL;
const char *objstart;
char *end;
unsigned int in_string=0, ninner=0;
/* Sanity checking */
if (!(pdf) || !(obj) || !(begin))
return NULL;
objstart = (const char *)(obj->start + pdf->map);
if (begin < objstart || (size_t)(begin - objstart) >= objsz - 2)
return NULL;
if (begin[0] != '<' || begin[1] != '<')
return NULL;
/* Find the end of the dictionary */
end = begin;
while ((size_t)(end - objstart) < objsz) {
int increment=1;
if (in_string) {
if (*end == '\\') {
end += 2;
continue;
}
if (*end == ')')
in_string = 0;
end++;
continue;
}
switch (*end) {
case '(':
in_string=1;
break;
case '<':
if ((size_t)(end - objstart) <= objsz - 2 && end[1] == '<')
ninner++;
increment=2;
break;
case '>':
if ((size_t)(end - objstart) <= objsz - 2 && end[1] == '>')
ninner--;
increment=2;
break;
}
if ((size_t)(end - objstart) <= objsz - 2)
if (end[0] == '>' && end[1] == '>' && ninner == 0)
break;
end += increment;
}
/* More sanity checking */
if ((size_t)(end - objstart) >= objsz - 2)
return NULL;
if (end[0] != '>' || end[1] != '>')
return NULL;
res = cli_calloc(1, sizeof(struct pdf_dict));
if (!(res))
return NULL;
/* Loop through each element of the dictionary */
begin += 2;
while (begin < end) {
char *val=NULL, *key=NULL, *p1, *p2;
struct pdf_dict *dict=NULL;
struct pdf_array *arr=NULL;
unsigned int nhex=0, i;
/* Skip any whitespaces */
while (begin < end && isspace(begin[0]))
begin++;
if (begin == end)
break;
/* Get the key */
p1 = begin+1;
while (p1 < end && !isspace(p1[0])) {
int breakout=0;
switch (*p1) {
case '<':
case '[':
case '(':
case '/':
case '\r':
case '\n':
case ' ':
case '\t':
breakout=1;
break;
case '#':
/* Key name obfuscated with hex characters */
nhex++;
if (p1 > end-3) {
return res;
}
break;
}
if (breakout)
break;
p1++;
}
if (p1 == end)
break;
key = cli_calloc((p1 - begin) + 2, 1);
if (!(key))
break;
if (nhex == 0) {
/* Key isn't obfuscated with hex. Just copy the string */
strncpy(key, begin, p1 - begin);
key[p1 - begin] = '\0';
} else {
for (i=0, p2 = begin; p2 < p1; p2++, i++) {
if (*p2 == '#') {
cli_hex2str_to(p2+1, key+i, 2);
p2 += 2;
} else {
key[i] = *p2;
}
}
}
/* Now for the value */
begin = p1;
/* Skip any whitespaces */
while (begin < end && isspace(begin[0]))
begin++;
if (begin == end) {
free(key);
break;
}
switch (begin[0]) {
case '(':
val = pdf_parse_string(pdf, obj, begin, objsz, NULL, &p1, NULL);
begin = p1+2;
break;
case '[':
arr = pdf_parse_array(pdf, obj, objsz, begin, &p1);
begin = p1+1;
break;
case '<':
if ((size_t)(begin - objstart) < objsz - 2) {
if (begin[1] == '<') {
dict = pdf_parse_dict(pdf, obj, objsz, begin, &p1);
begin = p1+2;
break;
}
}
val = pdf_parse_string(pdf, obj, begin, objsz, NULL, &p1, NULL);
begin = p1+2;
break;
default:
p1 = (begin[0] == '/') ? begin+1 : begin;
while (p1 < end) {
int shouldbreak = 0;
switch (p1[0]) {
case '>':
case '/':
shouldbreak=1;
break;
}
if (shouldbreak)
break;
p1++;
}
is_object_reference(begin, &p1, NULL);
val = cli_calloc((p1 - begin) + 2, 1);
if (!(val))
break;
strncpy(val, begin, p1 - begin);
val[p1 - begin] = '\0';
if (p1[0] != '/')
begin = p1+1;
else
begin = p1;
break;
}
if (!(val) && !(dict) && !(arr)) {
free(key);
break;
}
if (!(res->nodes)) {
res->nodes = res->tail = node = cli_calloc(1, sizeof(struct pdf_dict_node));
if (!(node)) {
free(key);
if (dict)
pdf_free_dict(dict);
if (val)
free(val);
if (arr)
pdf_free_array(arr);
break;
}
} else {
node = calloc(1, sizeof(struct pdf_dict_node));
if (!(node)) {
free(key);
if (dict)
pdf_free_dict(dict);
if (val)
free(val);
if (arr)
pdf_free_array(arr);
break;
}
node->prev = res->tail;
if (res->tail)
res->tail->next = node;
res->tail = node;
}
node->key = key;
if ((val)) {
node->value = val;
node->valuesz = strlen(val);
node->type = PDF_DICT_STRING;
} else if ((arr)) {
node->value = arr;
node->valuesz = sizeof(struct pdf_array);
node->type = PDF_DICT_ARRAY;
} else if ((dict)) {
node->value = dict;
node->valuesz = sizeof(struct pdf_dict);
node->type = PDF_DICT_DICT;
}
}
if (endchar)
*endchar = end;
return res;
}
fz_error
pdf_parse_dict(fz_obj **op, pdf_xref *xref, fz_stream *file, char *buf, int cap)
{
fz_error error = fz_okay;
fz_obj *dict = NULL;
fz_obj *key = NULL;
fz_obj *val = NULL;
int tok;
int len;
int a, b;
dict = fz_new_dict(8);
while (1)
{
error = pdf_lex(&tok, file, buf, cap, &len);
if (error)
{
fz_drop_obj(dict);
return fz_rethrow(error, "cannot parse dict");
}
skip:
if (tok == PDF_TOK_CLOSE_DICT)
{
*op = dict;
return fz_okay;
}
/* for BI .. ID .. EI in content streams */
if (tok == PDF_TOK_KEYWORD && !strcmp(buf, "ID"))
{
*op = dict;
return fz_okay;
}
if (tok != PDF_TOK_NAME)
{
fz_drop_obj(dict);
return fz_throw("invalid key in dict");
}
key = fz_new_name(buf);
error = pdf_lex(&tok, file, buf, cap, &len);
if (error)
{
fz_drop_obj(key);
fz_drop_obj(dict);
return fz_rethrow(error, "cannot parse dict");
}
switch (tok)
{
case PDF_TOK_OPEN_ARRAY:
error = pdf_parse_array(&val, xref, file, buf, cap);
if (error)
{
fz_drop_obj(key);
fz_drop_obj(dict);
return fz_rethrow(error, "cannot parse dict");
}
break;
case PDF_TOK_OPEN_DICT:
error = pdf_parse_dict(&val, xref, file, buf, cap);
if (error)
{
fz_drop_obj(key);
fz_drop_obj(dict);
return fz_rethrow(error, "cannot parse dict");
}
break;
case PDF_TOK_NAME: val = fz_new_name(buf); break;
case PDF_TOK_REAL: val = fz_new_real(fz_atof(buf)); break;
case PDF_TOK_STRING: val = fz_new_string(buf, len); break;
case PDF_TOK_TRUE: val = fz_new_bool(1); break;
case PDF_TOK_FALSE: val = fz_new_bool(0); break;
case PDF_TOK_NULL: val = fz_new_null(); break;
case PDF_TOK_INT:
/* 64-bit to allow for numbers > INT_MAX and overflow */
a = (int) strtoll(buf, 0, 10);
error = pdf_lex(&tok, file, buf, cap, &len);
if (error)
{
fz_drop_obj(key);
fz_drop_obj(dict);
return fz_rethrow(error, "cannot parse dict");
}
if (tok == PDF_TOK_CLOSE_DICT || tok == PDF_TOK_NAME ||
(tok == PDF_TOK_KEYWORD && !strcmp(buf, "ID")))
{
val = fz_new_int(a);
fz_dict_put(dict, key, val);
fz_drop_obj(val);
fz_drop_obj(key);
goto skip;
}
if (tok == PDF_TOK_INT)
{
b = atoi(buf);
error = pdf_lex(&tok, file, buf, cap, &len);
if (error)
{
fz_drop_obj(key);
fz_drop_obj(dict);
return fz_rethrow(error, "cannot parse dict");
}
if (tok == PDF_TOK_R)
{
val = fz_new_indirect(a, b, xref);
break;
}
}
fz_drop_obj(key);
fz_drop_obj(dict);
return fz_throw("invalid indirect reference in dict");
default:
fz_drop_obj(key);
fz_drop_obj(dict);
return fz_throw("unknown token in dict");
}
fz_dict_put(dict, key, val);
fz_drop_obj(val);
fz_drop_obj(key);
}
}
pdf_obj *
pdf_parse_ind_obj(fz_context *ctx, pdf_document *doc,
fz_stream *file, pdf_lexbuf *buf,
int *onum, int *ogen, fz_off_t *ostmofs, int *try_repair)
{
pdf_obj *obj = NULL;
int num = 0, gen = 0;
fz_off_t stm_ofs;
pdf_token tok;
fz_off_t a, b;
fz_var(obj);
tok = pdf_lex(ctx, file, buf);
if (tok != PDF_TOK_INT)
{
if (try_repair)
*try_repair = 1;
fz_throw(ctx, FZ_ERROR_GENERIC, "expected object number");
}
num = buf->i;
tok = pdf_lex(ctx, file, buf);
if (tok != PDF_TOK_INT)
{
if (try_repair)
*try_repair = 1;
fz_throw(ctx, FZ_ERROR_GENERIC, "expected generation number (%d ? obj)", num);
}
gen = buf->i;
tok = pdf_lex(ctx, file, buf);
if (tok != PDF_TOK_OBJ)
{
if (try_repair)
*try_repair = 1;
fz_throw(ctx, FZ_ERROR_GENERIC, "expected 'obj' keyword (%d %d ?)", num, gen);
}
tok = pdf_lex(ctx, file, buf);
switch (tok)
{
case PDF_TOK_OPEN_ARRAY:
obj = pdf_parse_array(ctx, doc, file, buf);
break;
case PDF_TOK_OPEN_DICT:
obj = pdf_parse_dict(ctx, doc, file, buf);
break;
case PDF_TOK_NAME: obj = pdf_new_name(ctx, doc, buf->scratch); break;
case PDF_TOK_REAL: obj = pdf_new_real(ctx, doc, buf->f); break;
case PDF_TOK_STRING: obj = pdf_new_string(ctx, doc, buf->scratch, buf->len); break;
case PDF_TOK_TRUE: obj = pdf_new_bool(ctx, doc, 1); break;
case PDF_TOK_FALSE: obj = pdf_new_bool(ctx, doc, 0); break;
case PDF_TOK_NULL: obj = pdf_new_null(ctx, doc); break;
case PDF_TOK_INT:
a = buf->i;
tok = pdf_lex(ctx, file, buf);
if (tok == PDF_TOK_STREAM || tok == PDF_TOK_ENDOBJ)
{
obj = pdf_new_int_offset(ctx, doc, a);
goto skip;
}
if (tok == PDF_TOK_INT)
{
b = buf->i;
tok = pdf_lex(ctx, file, buf);
if (tok == PDF_TOK_R)
{
obj = pdf_new_indirect(ctx, doc, a, b);
break;
}
}
fz_throw(ctx, FZ_ERROR_GENERIC, "expected 'R' keyword (%d %d R)", num, gen);
case PDF_TOK_ENDOBJ:
obj = pdf_new_null(ctx, doc);
goto skip;
default:
fz_throw(ctx, FZ_ERROR_GENERIC, "syntax error in object (%d %d R)", num, gen);
}
fz_try(ctx)
{
tok = pdf_lex(ctx, file, buf);
}
fz_catch(ctx)
{
pdf_drop_obj(ctx, obj);
fz_rethrow(ctx);
}
skip:
if (tok == PDF_TOK_STREAM)
{
int c = fz_read_byte(ctx, file);
while (c == ' ')
c = fz_read_byte(ctx, file);
if (c == '\r')
{
c = fz_peek_byte(ctx, file);
if (c != '\n')
fz_warn(ctx, "line feed missing after stream begin marker (%d %d R)", num, gen);
else
fz_read_byte(ctx, file);
}
stm_ofs = fz_tell(ctx, file);
}
else if (tok == PDF_TOK_ENDOBJ)
{
stm_ofs = 0;
}
else
{
fz_warn(ctx, "expected 'endobj' or 'stream' keyword (%d %d R)", num, gen);
stm_ofs = 0;
}
if (onum) *onum = num;
if (ogen) *ogen = gen;
if (ostmofs) *ostmofs = stm_ofs;
return obj;
}
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);
}
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;
}
fz_error
pdf_parse_ind_obj(fz_obj **op, pdf_xref *xref,
fz_stream *file, char *buf, int cap,
int *onum, int *ogen, int *ostmofs)
{
fz_error error = fz_okay;
fz_obj *obj = NULL;
int num = 0, gen = 0, stm_ofs;
int tok;
int len;
int a, b;
error = pdf_lex(&tok, file, buf, cap, &len);
if (error)
return fz_rethrow(error, "cannot parse indirect object (%d %d R)", num, gen);
if (tok != PDF_TOK_INT)
return fz_throw("expected object number (%d %d R)", num, gen);
num = atoi(buf);
error = pdf_lex(&tok, file, buf, cap, &len);
if (error)
return fz_rethrow(error, "cannot parse indirect object (%d %d R)", num, gen);
if (tok != PDF_TOK_INT)
return fz_throw("expected generation number (%d %d R)", num, gen);
gen = atoi(buf);
error = pdf_lex(&tok, file, buf, cap, &len);
if (error)
return fz_rethrow(error, "cannot parse indirect object (%d %d R)", num, gen);
if (tok != PDF_TOK_OBJ)
return fz_throw("expected 'obj' keyword (%d %d R)", num, gen);
error = pdf_lex(&tok, file, buf, cap, &len);
if (error)
return fz_rethrow(error, "cannot parse indirect object (%d %d R)", num, gen);
switch (tok)
{
case PDF_TOK_OPEN_ARRAY:
error = pdf_parse_array(&obj, xref, file, buf, cap);
if (error)
return fz_rethrow(error, "cannot parse indirect object (%d %d R)", num, gen);
break;
case PDF_TOK_OPEN_DICT:
error = pdf_parse_dict(&obj, xref, file, buf, cap);
if (error)
return fz_rethrow(error, "cannot parse indirect object (%d %d R)", num, gen);
break;
case PDF_TOK_NAME: obj = fz_new_name(buf); break;
case PDF_TOK_REAL: obj = fz_new_real(fz_atof(buf)); break;
case PDF_TOK_STRING: obj = fz_new_string(buf, len); break;
case PDF_TOK_TRUE: obj = fz_new_bool(1); break;
case PDF_TOK_FALSE: obj = fz_new_bool(0); break;
case PDF_TOK_NULL: obj = fz_new_null(); break;
case PDF_TOK_INT:
a = atoi(buf);
error = pdf_lex(&tok, file, buf, cap, &len);
if (error)
return fz_rethrow(error, "cannot parse indirect object (%d %d R)", num, gen);
if (tok == PDF_TOK_STREAM || tok == PDF_TOK_ENDOBJ)
{
obj = fz_new_int(a);
goto skip;
}
if (tok == PDF_TOK_INT)
{
b = atoi(buf);
error = pdf_lex(&tok, file, buf, cap, &len);
if (error)
return fz_rethrow(error, "cannot parse indirect object (%d %d R)", num, gen);
if (tok == PDF_TOK_R)
{
obj = fz_new_indirect(a, b, xref);
break;
}
}
return fz_throw("expected 'R' keyword (%d %d R)", num, gen);
case PDF_TOK_ENDOBJ:
obj = fz_new_null();
goto skip;
default:
return fz_throw("syntax error in object (%d %d R)", num, gen);
}
error = pdf_lex(&tok, file, buf, cap, &len);
if (error)
{
fz_drop_obj(obj);
return fz_rethrow(error, "cannot parse indirect object (%d %d R)", num, gen);
}
skip:
if (tok == PDF_TOK_STREAM)
{
int c = fz_read_byte(file);
while (c == ' ')
c = fz_read_byte(file);
if (c == '\r')
{
c = fz_peek_byte(file);
if (c != '\n')
fz_warn("line feed missing after stream begin marker (%d %d R)", num, gen);
else
fz_read_byte(file);
}
stm_ofs = fz_tell(file);
}
else if (tok == PDF_TOK_ENDOBJ)
{
stm_ofs = 0;
}
else
{
fz_warn("expected 'endobj' or 'stream' keyword (%d %d R)", num, gen);
stm_ofs = 0;
}
if (onum) *onum = num;
if (ogen) *ogen = gen;
if (ostmofs) *ostmofs = stm_ofs;
*op = obj;
return fz_okay;
}
pdf_obj *
pdf_parse_dict(pdf_document *doc, fz_stream *file, pdf_lexbuf *buf)
{
pdf_obj *dict;
pdf_obj *key = NULL;
pdf_obj *val = NULL;
pdf_token tok;
int a, b;
fz_context *ctx = file->ctx;
dict = pdf_new_dict(doc, 8);
fz_var(key);
fz_var(val);
fz_try(ctx)
{
while (1)
{
tok = pdf_lex(file, buf);
skip:
if (tok == PDF_TOK_CLOSE_DICT)
break;
/* for BI .. ID .. EI in content streams */
if (tok == PDF_TOK_KEYWORD && !strcmp(buf->scratch, "ID"))
break;
if (tok != PDF_TOK_NAME)
fz_throw(ctx, FZ_ERROR_GENERIC, "invalid key in dict");
key = pdf_new_name(doc, buf->scratch);
tok = pdf_lex(file, buf);
switch (tok)
{
case PDF_TOK_OPEN_ARRAY:
val = pdf_parse_array(doc, file, buf);
break;
case PDF_TOK_OPEN_DICT:
val = pdf_parse_dict(doc, file, buf);
break;
case PDF_TOK_NAME: val = pdf_new_name(doc, buf->scratch); break;
case PDF_TOK_REAL: val = pdf_new_real(doc, buf->f); break;
case PDF_TOK_STRING: val = pdf_new_string(doc, buf->scratch, buf->len); break;
case PDF_TOK_TRUE: val = pdf_new_bool(doc, 1); break;
case PDF_TOK_FALSE: val = pdf_new_bool(doc, 0); break;
case PDF_TOK_NULL: val = pdf_new_null(doc); break;
case PDF_TOK_INT:
/* 64-bit to allow for numbers > INT_MAX and overflow */
a = buf->i;
tok = pdf_lex(file, buf);
if (tok == PDF_TOK_CLOSE_DICT || tok == PDF_TOK_NAME ||
(tok == PDF_TOK_KEYWORD && !strcmp(buf->scratch, "ID")))
{
val = pdf_new_int(doc, a);
pdf_dict_put(dict, key, val);
pdf_drop_obj(val);
val = NULL;
pdf_drop_obj(key);
key = NULL;
goto skip;
}
if (tok == PDF_TOK_INT)
{
b = buf->i;
tok = pdf_lex(file, buf);
if (tok == PDF_TOK_R)
{
val = pdf_new_indirect(doc, a, b);
break;
}
}
fz_throw(ctx, FZ_ERROR_GENERIC, "invalid indirect reference in dict");
default:
fz_throw(ctx, FZ_ERROR_GENERIC, "unknown token in dict");
}
pdf_dict_put(dict, key, val);
pdf_drop_obj(val);
val = NULL;
pdf_drop_obj(key);
key = NULL;
}
}
fz_catch(ctx)
{
pdf_drop_obj(dict);
pdf_drop_obj(key);
pdf_drop_obj(val);
fz_rethrow_message(ctx, "cannot parse dict");
}
return dict;
}
pdf_obj *
pdf_parse_ind_obj(pdf_document *xref,
fz_stream *file, pdf_lexbuf *buf,
int *onum, int *ogen, int *ostmofs)
{
pdf_obj *obj = NULL;
int num = 0, gen = 0, stm_ofs;
int tok;
int a, b;
fz_context *ctx = file->ctx;
fz_var(obj);
tok = pdf_lex(file, buf);
/* RJW: cannot parse indirect object (%d %d R)", num, gen */
if (tok != PDF_TOK_INT)
fz_throw(ctx, "expected object number (%d %d R)", num, gen);
num = buf->i;
tok = pdf_lex(file, buf);
/* RJW: "cannot parse indirect object (%d %d R)", num, gen */
if (tok != PDF_TOK_INT)
fz_throw(ctx, "expected generation number (%d %d R)", num, gen);
gen = buf->i;
tok = pdf_lex(file, buf);
/* RJW: "cannot parse indirect object (%d %d R)", num, gen */
if (tok != PDF_TOK_OBJ)
fz_throw(ctx, "expected 'obj' keyword (%d %d R)", num, gen);
tok = pdf_lex(file, buf);
/* RJW: "cannot parse indirect object (%d %d R)", num, gen */
switch (tok)
{
case PDF_TOK_OPEN_ARRAY:
obj = pdf_parse_array(xref, file, buf);
/* RJW: "cannot parse indirect object (%d %d R)", num, gen */
break;
case PDF_TOK_OPEN_DICT:
obj = pdf_parse_dict(xref, file, buf);
/* RJW: "cannot parse indirect object (%d %d R)", num, gen */
break;
case PDF_TOK_NAME:
obj = fz_new_name(ctx, buf->scratch);
break;
case PDF_TOK_REAL:
obj = pdf_new_real(ctx, buf->f);
break;
case PDF_TOK_STRING:
obj = pdf_new_string(ctx, buf->scratch, buf->len);
break;
case PDF_TOK_TRUE:
obj = pdf_new_bool(ctx, 1);
break;
case PDF_TOK_FALSE:
obj = pdf_new_bool(ctx, 0);
break;
case PDF_TOK_NULL:
obj = pdf_new_null(ctx);
break;
case PDF_TOK_INT:
a = buf->i;
tok = pdf_lex(file, buf);
/* "cannot parse indirect object (%d %d R)", num, gen */
if (tok == PDF_TOK_STREAM || tok == PDF_TOK_ENDOBJ)
{
obj = pdf_new_int(ctx, a);
goto skip;
}
if (tok == PDF_TOK_INT)
{
b = buf->i;
tok = pdf_lex(file, buf);
/* RJW: "cannot parse indirect object (%d %d R)", num, gen); */
if (tok == PDF_TOK_R)
{
obj = pdf_new_indirect(ctx, a, b, xref);
break;
}
}
fz_throw(ctx, "expected 'R' keyword (%d %d R)", num, gen);
case PDF_TOK_ENDOBJ:
obj = pdf_new_null(ctx);
goto skip;
default:
fz_throw(ctx, "syntax error in object (%d %d R)", num, gen);
}
fz_try(ctx)
{
tok = pdf_lex(file, buf);
}
fz_catch(ctx)
{
pdf_drop_obj(obj);
fz_throw(ctx, "cannot parse indirect object (%d %d R)", num, gen);
}
skip:
if (tok == PDF_TOK_STREAM)
{
int c = fz_read_byte(file);
while (c == ' ')
c = fz_read_byte(file);
if (c == '\r')
{
c = fz_peek_byte(file);
if (c != '\n')
fz_warn(ctx, "line feed missing after stream begin marker (%d %d R)", num, gen);
else
fz_read_byte(file);
}
stm_ofs = fz_tell(file);
}
else if (tok == PDF_TOK_ENDOBJ)
{
stm_ofs = 0;
}
else
{
fz_warn(ctx, "expected 'endobj' or 'stream' keyword (%d %d R)", num, gen);
stm_ofs = 0;
}
if (onum) *onum = num;
if (ogen) *ogen = gen;
if (ostmofs) *ostmofs = stm_ofs;
return obj;
}
fz_error
pdf_parse_array(fz_obj **op, pdf_xref *xref, fz_stream *file, char *buf, int cap)
{
fz_error error = fz_okay;
fz_obj *ary = NULL;
fz_obj *obj = NULL;
int a = 0, b = 0, n = 0;
int tok;
int len;
ary = fz_new_array(4);
while (1)
{
error = pdf_lex(&tok, file, buf, cap, &len);
if (error)
{
fz_drop_obj(ary);
return fz_rethrow(error, "cannot parse array");
}
if (tok != PDF_TOK_INT && tok != PDF_TOK_R)
{
if (n > 0)
{
obj = fz_new_int(a);
fz_array_push(ary, obj);
fz_drop_obj(obj);
}
if (n > 1)
{
obj = fz_new_int(b);
fz_array_push(ary, obj);
fz_drop_obj(obj);
}
n = 0;
}
if (tok == PDF_TOK_INT && n == 2)
{
obj = fz_new_int(a);
fz_array_push(ary, obj);
fz_drop_obj(obj);
a = b;
n --;
}
switch (tok)
{
case PDF_TOK_CLOSE_ARRAY:
*op = ary;
return fz_okay;
case PDF_TOK_INT:
if (n == 0)
a = atoi(buf);
if (n == 1)
b = atoi(buf);
n ++;
break;
case PDF_TOK_R:
if (n != 2)
{
fz_drop_obj(ary);
return fz_throw("cannot parse indirect reference in array");
}
obj = fz_new_indirect(a, b, xref);
fz_array_push(ary, obj);
fz_drop_obj(obj);
n = 0;
break;
case PDF_TOK_OPEN_ARRAY:
error = pdf_parse_array(&obj, xref, file, buf, cap);
if (error)
{
fz_drop_obj(ary);
return fz_rethrow(error, "cannot parse array");
}
fz_array_push(ary, obj);
fz_drop_obj(obj);
break;
case PDF_TOK_OPEN_DICT:
error = pdf_parse_dict(&obj, xref, file, buf, cap);
if (error)
{
fz_drop_obj(ary);
return fz_rethrow(error, "cannot parse array");
}
fz_array_push(ary, obj);
fz_drop_obj(obj);
break;
case PDF_TOK_NAME:
obj = fz_new_name(buf);
fz_array_push(ary, obj);
fz_drop_obj(obj);
break;
case PDF_TOK_REAL:
obj = fz_new_real(fz_atof(buf));
fz_array_push(ary, obj);
fz_drop_obj(obj);
break;
case PDF_TOK_STRING:
obj = fz_new_string(buf, len);
fz_array_push(ary, obj);
fz_drop_obj(obj);
break;
case PDF_TOK_TRUE:
obj = fz_new_bool(1);
fz_array_push(ary, obj);
fz_drop_obj(obj);
break;
case PDF_TOK_FALSE:
obj = fz_new_bool(0);
fz_array_push(ary, obj);
fz_drop_obj(obj);
break;
case PDF_TOK_NULL:
obj = fz_new_null();
fz_array_push(ary, obj);
fz_drop_obj(obj);
break;
default:
fz_drop_obj(ary);
return fz_throw("cannot parse token in array");
}
}
}
