SRL_STATIC_INLINE void srl_set_input_buffer(pTHX_ srl_merger_t *mrg, SV *src) { STRLEN len; UV header_len; U8 encoding_flags; U8 protocol_version; srl_buffer_char *tmp; IV proto_version_and_encoding_flags_int; SRL_RDR_CLEAR(&mrg->ibuf); tmp = (srl_buffer_char*) SvPV(src, len); mrg->ibuf.start = mrg->ibuf.pos = tmp; mrg->ibuf.end = mrg->ibuf.start + len; proto_version_and_encoding_flags_int = srl_validate_header_version(aTHX_ (srl_reader_char_ptr) mrg->ibuf.start, len); if (proto_version_and_encoding_flags_int < 1) { if (proto_version_and_encoding_flags_int == 0) SRL_RDR_ERROR(mrg->pibuf, "Bad Sereal header: It seems your document was accidentally UTF-8 encoded"); else SRL_RDR_ERROR(mrg->pibuf, "Bad Sereal header: Not a valid Sereal document."); } mrg->ibuf.pos += 5; encoding_flags = (U8) (proto_version_and_encoding_flags_int & SRL_PROTOCOL_ENCODING_MASK); protocol_version = (U8) (proto_version_and_encoding_flags_int & SRL_PROTOCOL_VERSION_MASK); if (expect_false(protocol_version > 3 || protocol_version < 1)) { SRL_RDR_ERRORf1(mrg->pibuf, "Unsupported Sereal protocol version %u", (unsigned int) protocol_version); } // skip header in any case header_len = srl_read_varint_uv_length(aTHX_ mrg->pibuf, " while reading header"); mrg->ibuf.pos += header_len; if (encoding_flags == SRL_PROTOCOL_ENCODING_RAW) { /* no op */ } else if ( encoding_flags == SRL_PROTOCOL_ENCODING_SNAPPY || encoding_flags == SRL_PROTOCOL_ENCODING_SNAPPY_INCREMENTAL) { srl_decompress_body_snappy(aTHX_ mrg->pibuf, encoding_flags, NULL); } else if (encoding_flags == SRL_PROTOCOL_ENCODING_ZLIB) { srl_decompress_body_zlib(aTHX_ mrg->pibuf, NULL); } else { SRL_RDR_ERROR(mrg->pibuf, "Sereal document encoded in an unknown format"); } /* this functions *MUST* be called after srl_decompress_body* */ SRL_RDR_UPDATE_BODY_POS(mrg->pibuf, protocol_version); DEBUG_ASSERT_RDR_SANE(mrg->pibuf); }
void srl_iterator_set(pTHX_ srl_iterator_t *iter, SV *src) { SV *sv; STRLEN len; UV header_len; U8 encoding_flags; U8 protocol_version; srl_reader_char_ptr tmp; IV proto_version_and_encoding_flags_int; srl_iterator_stack_ptr stack_ptr = NULL; if (iter->document) { SvREFCNT_dec(iter->document); iter->document = NULL; } iter->document = src; SvREFCNT_inc(iter->document); tmp = (srl_reader_char_ptr) SvPV(src, len); iter->buf.start = iter->buf.pos = tmp; iter->buf.end = iter->buf.start + len; proto_version_and_encoding_flags_int = srl_validate_header_version(aTHX_ iter->buf.start, len); if (proto_version_and_encoding_flags_int < 1) { if (proto_version_and_encoding_flags_int == 0) SRL_RDR_ERROR(iter->pbuf, "Bad Sereal header: It seems your document was accidentally UTF-8 encoded"); else SRL_RDR_ERROR(iter->pbuf, "Bad Sereal header: Not a valid Sereal document."); } iter->buf.pos += 5; encoding_flags = (U8) (proto_version_and_encoding_flags_int & SRL_PROTOCOL_ENCODING_MASK); protocol_version = (U8) (proto_version_and_encoding_flags_int & SRL_PROTOCOL_VERSION_MASK); if (expect_false(protocol_version > 3 || protocol_version < 1)) { SRL_RDR_ERRORf1(iter->pbuf, "Unsupported Sereal protocol version %u", (unsigned int) protocol_version); } // skip header in any case header_len = srl_read_varint_uv_length(aTHX_ iter->pbuf, " while reading header"); iter->buf.pos += header_len; if (encoding_flags == SRL_PROTOCOL_ENCODING_RAW) { /* no op */ } else if ( encoding_flags == SRL_PROTOCOL_ENCODING_SNAPPY || encoding_flags == SRL_PROTOCOL_ENCODING_SNAPPY_INCREMENTAL) { srl_decompress_body_snappy(aTHX_ iter->pbuf, encoding_flags, &sv); SvREFCNT_dec(iter->document); SvREFCNT_inc(sv); iter->document = sv; } else if (encoding_flags == SRL_PROTOCOL_ENCODING_ZLIB) { srl_decompress_body_zlib(aTHX_ iter->pbuf, &sv); SvREFCNT_dec(iter->document); SvREFCNT_inc(sv); iter->document = sv; } else { SRL_RDR_ERROR(iter->pbuf, "Sereal document encoded in an unknown format"); } /* this function *MUST* be called after calling srl_decompress_body* */ SRL_RDR_UPDATE_BODY_POS(iter->pbuf, protocol_version); DEBUG_ASSERT_RDR_SANE(iter->pbuf); srl_stack_push_and_set(iter, SRL_ITER_STACK_ROOT_TAG, 1, stack_ptr); srl_iterator_reset(aTHX_ iter); }
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); }