UV
srl_iterator_stack_info(pTHX_ srl_iterator_t *iter, UV *length_ptr)
{
UV type = 0;
srl_stack_t *stack = iter->pstack;
srl_iterator_stack_ptr stack_ptr = stack->ptr;
SRL_ITER_ASSERT_STACK(iter);
if (length_ptr) *length_ptr = stack_ptr->count;
switch (stack_ptr->tag) {
case SRL_ITER_STACK_ROOT_TAG:
type = SRL_ITERATOR_OBJ_IS_ROOT;
break;
case SRL_HDR_HASH:
CASE_SRL_HDR_HASHREF:
type = SRL_ITERATOR_OBJ_IS_HASH;
break;
case SRL_HDR_ARRAY:
CASE_SRL_HDR_ARRAYREF:
type = SRL_ITERATOR_OBJ_IS_ARRAY;
break;
default:
SRL_RDR_ERROR_UNEXPECTED(iter->pbuf, stack_ptr->tag, "ARRAY or HASH");
}
return type;
}
UV
srl_iterator_object_info(pTHX_ srl_iterator_t *iter, UV *length_ptr)
{
U8 tag;
UV offset, type = 0;
srl_reader_char_ptr orig_pos = iter->buf.pos;
DEBUG_ASSERT_RDR_SANE(iter->pbuf);
if (length_ptr) *length_ptr = 0;
read_again:
SRL_ITER_ASSERT_EOF(iter, "serialized object");
tag = *iter->buf.pos & ~SRL_HDR_TRACK_FLAG;
SRL_ITER_REPORT_TAG(iter, tag);
iter->buf.pos++;
switch (tag) {
case SRL_HDR_PAD:
case SRL_HDR_REFN:
case SRL_HDR_WEAKEN:
/* advanced pointer to the object */
goto read_again;
case SRL_HDR_HASH:
type = SRL_ITERATOR_OBJ_IS_HASH;
if (length_ptr) *length_ptr = srl_read_varint_uv_count(aTHX_ iter->pbuf, " while reading HASH");
break;
CASE_SRL_HDR_HASHREF:
type = SRL_ITERATOR_OBJ_IS_HASH;
if (length_ptr) *length_ptr = SRL_HDR_HASHREF_LEN_FROM_TAG(tag);
break;
case SRL_HDR_ARRAY:
type = SRL_ITERATOR_OBJ_IS_ARRAY;
if (length_ptr) *length_ptr = srl_read_varint_uv_count(aTHX_ iter->pbuf, " while reading ARRAY");
break;
CASE_SRL_HDR_ARRAYREF:
type = SRL_ITERATOR_OBJ_IS_ARRAY;
if (length_ptr) *length_ptr = SRL_HDR_ARRAYREF_LEN_FROM_TAG(tag);
break;
CASE_SRL_HDR_POS:
CASE_SRL_HDR_NEG:
CASE_SRL_HDR_SHORT_BINARY:
case SRL_HDR_BINARY:
case SRL_HDR_STR_UTF8:
case SRL_HDR_VARINT:
case SRL_HDR_ZIGZAG:
case SRL_HDR_FLOAT:
case SRL_HDR_DOUBLE:
case SRL_HDR_LONG_DOUBLE:
case SRL_HDR_TRUE:
case SRL_HDR_FALSE:
case SRL_HDR_UNDEF:
case SRL_HDR_CANONICAL_UNDEF:
type = SRL_ITERATOR_OBJ_IS_SCALAR;
break;
case SRL_HDR_REFP:
case SRL_HDR_COPY:
offset = srl_read_varint_uv_offset(aTHX_ iter->pbuf, " while reading COPY tag");
iter->buf.pos = iter->buf.body_pos + offset;
goto read_again;
default:
iter->buf.pos = orig_pos;
SRL_RDR_ERROR_UNEXPECTED(iter->pbuf, tag, "ARRAY or HASH or SCALAR");
}
iter->buf.pos = orig_pos;
DEBUG_ASSERT_RDR_SANE(iter->pbuf);
return type;
}
IV
srl_iterator_hash_exists(pTHX_ srl_iterator_t *iter, const char *name, STRLEN name_len)
{
U8 tag;
UV length, offset;
const char *key_ptr;
IV stack_depth = iter->stack.depth;
srl_iterator_stack_ptr stack_ptr = iter->stack.ptr;
DEBUG_ASSERT_RDR_SANE(iter->pbuf);
SRL_ITER_ASSERT_EOF(iter, "stringish");
SRL_ITER_ASSERT_STACK(iter);
SRL_ITER_ASSERT_HASH_ON_STACK(iter);
SRL_ITER_TRACE("name=%.*s", (int) name_len, name);
SRL_ITER_REPORT_STACK_STATE(iter);
while (stack_ptr->idx) {
stack_ptr->idx--; // do not make it be part of while clause
SRL_ITER_ASSERT_STACK(iter);
assert(stack_ptr->idx % 2 == 1);
assert(iter->stack.depth == stack_depth);
DEBUG_ASSERT_RDR_SANE(iter->pbuf);
tag = *iter->buf.pos & ~SRL_HDR_TRACK_FLAG;
SRL_ITER_REPORT_TAG(iter, tag);
iter->buf.pos++;
switch (tag) {
CASE_SRL_HDR_SHORT_BINARY:
length = SRL_HDR_SHORT_BINARY_LEN_FROM_TAG(tag);
key_ptr = (const char *) iter->buf.pos;
iter->buf.pos += length;
break;
case SRL_HDR_BINARY:
length = srl_read_varint_uv_length(aTHX_ iter->pbuf, " while reading BINARY");
key_ptr = (const char *) iter->buf.pos;
iter->buf.pos += length;
break;
case SRL_HDR_STR_UTF8:
// TODO deal with UTF8
length = srl_read_varint_uv_length(aTHX_ iter->pbuf, " while reading STR_UTF8");
key_ptr = (const char *) iter->buf.pos;
iter->buf.pos += length;
break;
case SRL_HDR_COPY:
offset = srl_read_varint_uv_offset(aTHX_ iter->pbuf, " while reading COPY tag");
key_ptr = (const char *) iter->buf.body_pos + offset;
tag = *key_ptr & ~SRL_HDR_TRACK_FLAG;
key_ptr++;
/* Note we do NOT validate these items, as we have already read them
* and if they were a problem we would not be here to process them! */
switch (tag) {
CASE_SRL_HDR_SHORT_BINARY:
length = SRL_HDR_SHORT_BINARY_LEN_FROM_TAG(tag);
break;
case SRL_HDR_BINARY:
SET_UV_FROM_VARINT(iter->pbuf, length, key_ptr);
break;
case SRL_HDR_STR_UTF8:
// TODO deal with UTF8
SET_UV_FROM_VARINT(iter->pbuf, length, key_ptr);
break;
default:
SRL_RDR_ERROR_BAD_COPY(iter->pbuf, SRL_HDR_HASH);
}
break;
default:
SRL_RDR_ERROR_UNEXPECTED(iter->pbuf, tag, "stringish");
}
if (expect_false((srl_reader_char_ptr) key_ptr >= iter->buf.end)) {
SRL_RDR_ERROR_EOF(iter->pbuf, "string content");
}
if ( length == name_len
&& memcmp(name, key_ptr, name_len) == 0)
{
SRL_ITER_TRACE("found key '%.*s' at offset %"UVuf,
(int) name_len, name, SRL_RDR_BODY_POS_OFS(iter->pbuf));
return SRL_RDR_BODY_POS_OFS(iter->pbuf);
}
// srl_iterator_next garantee that we remans on current stack
srl_iterator_next(aTHX_ iter, 1);
stack_ptr = iter->stack.ptr;
}
SRL_ITER_TRACE("didn't found key '%.*s'", (int) name_len, name);
return SRL_ITER_NOT_FOUND;
}
const char *
srl_iterator_hash_key(pTHX_ srl_iterator_t *iter, STRLEN *len_out)
{
U8 tag;
UV length, offset;
const char *result = NULL;
srl_reader_char_ptr orig_pos = iter->buf.pos;
*len_out = 0;
DEBUG_ASSERT_RDR_SANE(iter->pbuf);
SRL_ITER_ASSERT_EOF(iter, "stringish");
SRL_ITER_ASSERT_STACK(iter);
SRL_ITER_ASSERT_HASH_ON_STACK(iter);
tag = *iter->buf.pos & ~SRL_HDR_TRACK_FLAG;
SRL_ITER_REPORT_TAG(iter, tag);
iter->buf.pos++;
switch (tag) {
CASE_SRL_HDR_SHORT_BINARY:
length = SRL_HDR_SHORT_BINARY_LEN_FROM_TAG(tag);
break;
case SRL_HDR_BINARY:
length = srl_read_varint_uv_length(aTHX_ iter->pbuf, " while reading BINARY");
break;
case SRL_HDR_STR_UTF8:
// TODO deal with UTF8
length = srl_read_varint_uv_length(aTHX_ iter->pbuf, " while reading STR_UTF8");
break;
case SRL_HDR_COPY:
offset = srl_read_varint_uv_offset(aTHX_ iter->pbuf, " while reading COPY tag");
iter->buf.pos = iter->buf.body_pos + offset;
/* Note we do NOT validate these items, as we have already read them
* and if they were a problem we would not be here to process them! */
tag = *iter->buf.pos & ~SRL_HDR_TRACK_FLAG;
SRL_ITER_REPORT_TAG(iter, tag);
iter->buf.pos++;
switch (tag) {
CASE_SRL_HDR_SHORT_BINARY:
length = SRL_HDR_SHORT_BINARY_LEN_FROM_TAG(tag);
break;
case SRL_HDR_BINARY:
SET_UV_FROM_VARINT(iter->pbuf, length, iter->buf.pos);
break;
case SRL_HDR_STR_UTF8:
// TODO deal with UTF8
SET_UV_FROM_VARINT(iter->pbuf, length, iter->buf.pos);
break;
default:
SRL_RDR_ERROR_BAD_COPY(iter->pbuf, SRL_HDR_HASH);
}
break;
default:
SRL_RDR_ERROR_UNEXPECTED(iter->pbuf, tag, "stringish");
}
if (expect_false(iter->buf.pos + length >= iter->buf.end)) {
SRL_RDR_ERROR_EOF(iter->pbuf, "string content");
}
*len_out = length;
result = (const char *) iter->buf.pos;
iter->buf.pos = orig_pos; // restore original position
DEBUG_ASSERT_RDR_SANE(iter->pbuf);
return result;
}
SRL_STATIC_INLINE void
srl_merge_single_value(pTHX_ srl_merger_t *mrg)
{
U8 tag;
UV length, offset;
ptable_entry_ptr ptable_entry;
read_again:
assert(mrg->recursion_depth >= 0);
DEBUG_ASSERT_RDR_SANE(mrg->pibuf);
DEBUG_ASSERT_BUF_SANE(&mrg->obuf);
if (expect_false(++mrg->recursion_depth > mrg->max_recursion_depth))
SRL_RDR_ERRORf1(mrg->pibuf, "Reached recursion limit (%lu) during merging", mrg->max_recursion_depth);
ptable_entry = NULL;
if (expect_false(SRL_RDR_DONE(mrg->pibuf)))
SRL_RDR_ERROR(mrg->pibuf, "Unexpected termination of input buffer");
tag = *mrg->ibuf.pos & ~SRL_HDR_TRACK_FLAG;
SRL_REPORT_CURRENT_TAG(mrg, tag);
if (mrg->tracked_offsets && !srl_stack_empty(mrg->tracked_offsets)) {
UV itag_offset = SRL_RDR_BODY_POS_OFS(mrg->pibuf);
if (expect_false(itag_offset == srl_stack_peek_nocheck(aTHX_ mrg->tracked_offsets))) {
// trackme case
srl_stack_pop_nocheck(mrg->tracked_offsets);
ptable_entry = srl_store_tracked_offset(aTHX_ mrg, itag_offset, BODY_POS_OFS(&mrg->obuf));
}
}
if (tag <= SRL_HDR_NEG_HIGH) {
srl_buf_cat_tag_nocheck(mrg, tag);
} else if (tag >= SRL_HDR_ARRAYREF_LOW && tag <= SRL_HDR_ARRAYREF_HIGH) {
srl_merge_array(aTHX_ mrg, tag, SRL_HDR_ARRAYREF_LEN_FROM_TAG(tag));
} else if (tag >= SRL_HDR_HASHREF_LOW && tag <= SRL_HDR_HASHREF_HIGH) {
srl_merge_hash(aTHX_ mrg, tag, SRL_HDR_HASHREF_LEN_FROM_TAG(tag));
} else if (tag >= SRL_HDR_SHORT_BINARY_LOW) {
srl_merge_short_binary(aTHX_ mrg, tag, ptable_entry);
} else {
switch (tag) {
case SRL_HDR_VARINT:
case SRL_HDR_ZIGZAG:
srl_buf_cat_tag_nocheck(mrg, tag);
srl_copy_varint(aTHX_ mrg);
break;
case SRL_HDR_FLOAT: srl_buf_copy_content_nocheck(aTHX_ mrg, 5); break;
case SRL_HDR_DOUBLE: srl_buf_copy_content_nocheck(aTHX_ mrg, 9); break;
case SRL_HDR_LONG_DOUBLE: srl_buf_copy_content_nocheck(aTHX_ mrg, 17); break;
case SRL_HDR_TRUE:
case SRL_HDR_FALSE:
case SRL_HDR_UNDEF:
case SRL_HDR_CANONICAL_UNDEF:
srl_buf_cat_tag_nocheck(mrg, tag);
break;
case SRL_HDR_BINARY:
case SRL_HDR_STR_UTF8:
srl_merge_binary_utf8(aTHX_ mrg, ptable_entry);
break;
case SRL_HDR_HASH:
mrg->ibuf.pos++; // skip tag in input buffer
length = srl_read_varint_uv_count(aTHX_ mrg->pibuf, " while reading ARRAY or HASH");
srl_merge_hash(aTHX_ mrg, tag, length);
break;
case SRL_HDR_ARRAY:
mrg->ibuf.pos++; // skip tag in input buffer
length = srl_read_varint_uv_count(aTHX_ mrg->pibuf, " while reading ARRAY or HASH");
srl_merge_array(aTHX_ mrg, tag, length);
break;
default:
switch (tag) {
case SRL_HDR_COPY:
case SRL_HDR_REFP:
case SRL_HDR_ALIAS:
mrg->ibuf.pos++; // skip tag in input buffer
offset = srl_read_varint_uv_offset(aTHX_ mrg->pibuf, " while reading COPY/ALIAS/REFP");
offset = srl_lookup_tracked_offset(aTHX_ mrg, offset); // convert ibuf offset to obuf offset
srl_buf_cat_varint(aTHX_ &mrg->obuf, tag, offset);
if (tag == SRL_HDR_REFP || tag == SRL_HDR_ALIAS) {
SRL_SET_TRACK_FLAG(*(mrg->obuf.body_pos + offset));
}
break;
case SRL_HDR_REFN:
case SRL_HDR_WEAKEN:
case SRL_HDR_EXTEND:
srl_buf_cat_tag_nocheck(mrg, tag);
goto read_again;
case SRL_HDR_OBJECT:
case SRL_HDR_OBJECT_FREEZE:
srl_merge_object(aTHX_ mrg, tag);
break;
case SRL_HDR_REGEXP:
srl_buf_cat_tag_nocheck(mrg, tag);
srl_merge_stringish(aTHX_ mrg);
tag = *mrg->ibuf.pos;
if (expect_false(tag < SRL_HDR_SHORT_BINARY_LOW))
SRL_RDR_ERROR_UNEXPECTED(mrg->pibuf, tag, "SRL_HDR_SHORT_BINARY");
srl_buf_copy_content_nocheck(aTHX_ mrg, SRL_HDR_SHORT_BINARY_LEN_FROM_TAG(tag) + 1);
break;
case SRL_HDR_OBJECTV:
case SRL_HDR_OBJECTV_FREEZE:
mrg->ibuf.pos++; // skip tag in input buffer
offset = srl_read_varint_uv_offset(aTHX_ mrg->pibuf, " while reading OBJECTV/OBJECTV_FREEZE");
offset = srl_lookup_tracked_offset(aTHX_ mrg, offset); // convert ibuf offset to obuf offset
srl_buf_cat_varint(aTHX_ &mrg->obuf, tag, offset);
goto read_again;
case SRL_HDR_PAD:
while (SRL_RDR_NOT_DONE(mrg->pibuf) && *mrg->ibuf.pos == SRL_HDR_PAD) {
srl_buf_cat_tag_nocheck(mrg, SRL_HDR_PAD);
}
goto read_again;
default:
SRL_RDR_ERROR_UNIMPLEMENTED(mrg->pibuf, tag, "");
break;
}
}
}
--mrg->recursion_depth;
DEBUG_ASSERT_RDR_SANE(mrg->pibuf);
DEBUG_ASSERT_BUF_SANE(&mrg->obuf);
}