From 534fd269ea1d68160d5e231b5a81f78af3f4784d Mon Sep 17 00:00:00 2001
From: "sumit.ray@baesystems.com" <sumit.ray@baesystems.com>
Date: Mon, 26 Jul 2021 00:23:44 -0400
Subject: [PATCH] Broken -- contains hack to try to pass in stream length to
act_ostm
---
pdf.c | 131 ++++++++++++++++++++++++++++++++++++++++++----------------
1 file changed, 96 insertions(+), 35 deletions(-)
diff --git a/pdf.c b/pdf.c
index 31dab43..419e1b5 100644
--- a/pdf.c
+++ b/pdf.c
@@ -295,6 +295,27 @@ typedef struct {
} Fontinfo_T;
+
+// Used by act_ostm to ensure indices are reasonable
+typedef struct {
+ const Dict *dict;
+ size_t numobjs;
+ size_t stmlength;
+} ostm_spec_T;
+
+
+// Parser for object streams
+HParser *p_objstm__m(HAllocator *, ostm_spec_T *);
+
+// Action for stream continuation
+HParsedToken *act_ks_value(const HParseResult *p, void *u);
+struct streamspec {
+ Dict *dict; /* stream dictionary */
+ HParser *parser; /* data parser */
+};
+
+
+
// ***********************************************************
@@ -2247,6 +2268,7 @@ HParser *ktxtstream(HAllocator *, const HParsedToken *, void *);
HParser *kcontentstream(HAllocator *, const HParsedToken *, void *);
HParser *kbyteostream(HAllocator *, const HParsedToken *, void *);
+
void
init_runlengthdecode_parser(struct Env *aux)
{
@@ -2996,8 +3018,9 @@ depred_png(struct predictor *pred, uint8_t *inp, size_t sz)
return done;
}
+// SR:: Fix Hacky code - N is only useful for object streams
HParseResult *
-FlateDecode(const Dict *parms, HBytes b, HParser *p)
+FlateDecode(const Dict *parms, HBytes b, HParser *p, size_t N)
{
size_t const BUFSIZE = 8 * 1024;
uint8_t *buf;
@@ -3098,10 +3121,15 @@ FlateDecode(const Dict *parms, HBytes b, HParser *p)
// XXX always return NULL on error?
#else
- // DEBUG -- will not always work depending on the font encoding used
-// fprintf (stdout, "FlateDecode:: Inflated string (%lu):\n%.*s\n", pred.nout, (int)pred.nout, pred.out);
- //DEBUG
-
+ // DEBUG --
+ // SR :: Hacky code for Object Streams
+ // If I understood Marcell correctly, I should be able to access the ostrm_spec
+ if (N > 0) { // object streams
+ ostm_spec_T *ospec = (ostm_spec_T *)(p->env + sizeof(HParser*) + sizeof(HAction));
+ ospec->stmlength = pred.nout;
+ fprintf(stdout, "\n\nFlateDecode: %p, N = (%lu, %lu), stmlength = %lu\n",
+ ospec, N, ospec->numobjs, ospec->stmlength);
+ }
res = h_parse(p, pred.out, pred.nout);
free(pred.out);
#endif
@@ -3265,7 +3293,7 @@ int read_lzw_buffer(void)
HParseResult *
-LZWDecode(const Dict *parms, HBytes b, HParser *p)
+LZWDecode(const Dict *parms, HBytes b, HParser *p, size_t numobjs)
{
struct predictor pred = {1, 1, 8, 1};
int (*depredict)(struct predictor *, uint8_t *, size_t);
@@ -3343,7 +3371,7 @@ LZWDecode(const Dict *parms, HBytes b, HParser *p)
}
HParseResult *
-RunLengthDecode(const Dict *parms, HBytes b, HParser *p)
+RunLengthDecode(const Dict *parms, HBytes b, HParser *p, size_t numobjs)
{
HParseResult *res;
@@ -3366,7 +3394,7 @@ RunLengthDecode(const Dict *parms, HBytes b, HParser *p)
* parms should be empty, because the filter has no parameters
*/
HParseResult *
-ASCIIHexDecode(const Dict *parms, HBytes b, HParser *p)
+ASCIIHexDecode(const Dict *parms, HBytes b, HParser *p, size_t numobjs)
{
HParseResult *f_res, *res;
@@ -3397,7 +3425,7 @@ ASCIIHexDecode(const Dict *parms, HBytes b, HParser *p)
* parms should be empty, because the filter has no parameters
*/
HParseResult*
-ASCII85Decode(const Dict *parms, HBytes b, HParser *p)
+ASCII85Decode(const Dict *parms, HBytes b, HParser *p, size_t numobjs)
{
HParseResult *f_res, *res;
@@ -3473,16 +3501,6 @@ p_take__m(HAllocator *mm__, size_t n, struct Env *aux)
}
-// Parser for object streams
-HParser *p_objstm__m(HAllocator *, const Dict *);
-
-// Action for stream continuation
-HParsedToken *act_ks_value(const HParseResult *p, void *u);
-struct streamspec {
- Dict *dict; /* stream dictionary */
- HParser *parser; /* data parser */
-};
-
/*
@@ -3497,7 +3515,7 @@ struct streamspec {
HParseResult *
decode_contentstream(const Dict *d, HBytes b, HParser *p)
{
- HParseResult *(*filter)(const Dict *, HBytes, HParser *);
+ HParseResult *(*filter)(const Dict *, HBytes, HParser *, size_t);
const Dict *parms = NULL;
const HParsedToken *v;
HParseResult *res = NULL;
@@ -3534,7 +3552,7 @@ decode_contentstream(const Dict *d, HBytes b, HParser *p)
else if (bytes_eq(v->bytes, "LZWDecode"))
filter = LZWDecode;
else { /* filter not supported */
- fprintf(stderr, "decode_stream:: Unsupported Filter [%.*s]\n",
+ fprintf(stderr, "decode_contentstream:: Unsupported Filter [%.*s]\n",
(int)v->bytes.len, v->bytes.token);
return NULL; /* Treat the stream as a byte array */
}
@@ -3543,7 +3561,14 @@ decode_contentstream(const Dict *d, HBytes b, HParser *p)
if (v && v->token_type == TT_Dict)
parms = v->user;
- res = filter(parms, b, p);
+ // SR:: Hacky code -- see if we are processing an object stream
+ // if we are, set N
+ size_t N = 0;
+ v = dictentry(d, "N");
+ if ( (v) && (v->token_type == TT_SINT) ) {
+ N = v->sint;
+ }
+ res = filter(parms, b, p, N);
/* Debug */
if (res){
@@ -4036,8 +4061,14 @@ kcontentstream(HAllocator *mm__, const HParsedToken *x, void *env)
v = dictentry(dict, "Type");
if (v == NULL) // XXX -> custom type
spec->parser = p_textstream;
- else if ( (v->token_type == TT_BYTES) && bytes_eq(v->bytes, "ObjStm") )
- spec->parser = p_objstm__m(mm__, dict);
+ else if ( (v->token_type == TT_BYTES) && bytes_eq(v->bytes, "ObjStm") ) {
+ ostm_spec_T *ospec = h_alloc(mm__, sizeof(ostm_spec_T ));
+ ospec->dict = dict;
+ ospec->numobjs = 0;
+ ospec->stmlength = 0;
+
+ spec->parser = p_objstm__m(mm__, ospec);
+ }
else {
fprintf(stdout, "kcontentstream: Not a text or object stream!\n");
return p_fail;
@@ -4868,7 +4899,7 @@ parse_catalog(struct Env *aux, const HParsedToken *root)
HParseResult *
decode_stream(const Dict *d, HBytes b, HParser *p)
{
- HParseResult *(*filter)(const Dict *, HBytes, HParser *);
+ HParseResult *(*filter)(const Dict *, HBytes, HParser *, size_t);
const Dict *parms = NULL;
const HParsedToken *v;
@@ -4908,7 +4939,14 @@ decode_stream(const Dict *d, HBytes b, HParser *p)
if (v && v->token_type == TT_Dict)
parms = v->user;
- return filter(parms, b, p);
+ // SR:: Hacky code -- see if we are processing an object stream
+ // if we are, set N
+ size_t N = 0;
+ v = dictentry(d, "N");
+ if ( (v) && (v->token_type == TT_SINT) ) {
+ N = v->sint;
+ }
+ return filter(parms, b, p, N);
}
@@ -4930,7 +4968,17 @@ p_stream_data__m(HAllocator *mm__, const Dict *dict, struct Env *aux)
if (bytes_eq(v->bytes, "ObjStm")) {
fprintf(stdout, "\np_stream_data__m:: Parsing object stream\n");
- return p_objstm__m(mm__, dict);
+ ostm_spec_T *ospec = h_alloc(mm__, sizeof(ostm_spec_T ));
+ ospec->dict = dict;
+ v = dictentry(dict, "N");
+ if ( (v == NULL) || (v->token_type != TT_SINT)) {
+ fprintf(stderr, "\np_stream_data__m:: Missing # of objects -N- in stream!!\n");
+ return NULL;
+ }
+ ospec->numobjs = v->sint;
+ ospec->stmlength = 0;
+
+ return p_objstm__m(mm__, ospec);
}
if (bytes_eq(v->bytes, "XObject")) {
@@ -5219,20 +5267,28 @@ p_xrefdata__m(HAllocator *mm__, const Dict *dict)
HParsedToken *
act_ostm(const HParseResult *p, void *u)
{
- assert(((HParsedToken *)u)->token_type == TT_SINT);
- size_t N = ((HParsedToken *)u)->sint;
+ ostm_spec_T *spec = (ostm_spec_T *)u;
+ fprintf(stdout, "\nact_ostm:: N objects = %lu, Stream Length = %lu\n",
+ spec->numobjs, spec->stmlength);
Objstm *ostrm = H_ALLOC(Objstm);
- ostrm->numObjs = N;
- ostrm->tok = h_arena_malloc(p->arena, N* sizeof(Objref_T));
+ ostrm->numObjs = spec->numobjs;
+ ostrm->tok = h_arena_malloc(p->arena, ostrm->numObjs* sizeof(Objref_T));
ostrm->arena = p->arena;
for (int i=0; i<ostrm->numObjs; i++) {
const HParsedToken *num = H_FIELD_TOKEN(0, 2*i);
- assert(num->token_type == TT_UINT);
+ const HParsedToken *offt = H_FIELD_TOKEN(0, 2*i+1);
+ assert((num->token_type == TT_UINT) && (offt->token_type == TT_UINT));
ostrm->tok[i].oid.nr = H_CAST_UINT(num);
ostrm->tok[i].oid.gen = 0;
ostrm->tok[i].obj = H_FIELD_TOKEN(1, i);
+ size_t offs = H_CAST_UINT(offt);
+ if (offs < 0) { // need to figure out a way to get the stream
+ fprintf(stdout, "\nact_ostm:: This should be warning and not generate an excption!\n\n");
+ // not sure whether returning NULL will generate an exception
+ return NULL;
+ }
}
// const HCountedArray *indices = H_FIELD_SEQ(0);
@@ -5248,24 +5304,29 @@ act_ostm(const HParseResult *p, void *u)
}
HParser *
-p_objstm__m(HAllocator *mm__, const Dict *dict)
+p_objstm__m(HAllocator *mm__, ostm_spec_T *spec)
{
const HParsedToken *v;
size_t N;
- v = dictentry(dict, "N");
+ if ((!spec) || (!spec->dict)) {
+ fprintf(stdout, "\np_objstm__m: The object stream dictionary missing!\n");
+ return p_fail;
+ }
+ v = dictentry(spec->dict, "N");
if (v == NULL || v->token_type != TT_SINT || v->sint < 0 ||
(uint64_t)v->sint > SIZE_MAX) {
fprintf(stderr, "p_objstm__m: missing /N on object stream\n");
return p_fail;
}
N = v->sint;
+ spec->numobjs = N;
HParser *wel_ws = h_sequence__m(mm__, p_wel, p_ws, NULL);
HParser *idx = p_sepBy_n__m(mm__, p_npair, wel_ws, N);
HParser *p_ostm = h_sequence__m(mm__, p_ws, idx, p_elemr, p_ws, NULL);
- HParser *ostm_p = h_action__m(mm__, p_ostm, act_ostm, (void *)v);
+ HParser *ostm_p = h_action__m(mm__, p_ostm, act_ostm, (void *)spec);
return ostm_p;
// XXX leading and trailing ws OK?
--
GitLab