Chiebot-Mirror
/
protobuf
mirror of https://github.com/protocolbuffers/protobuf.git
8
0
Fork 0
Code Issues Projects Releases Wiki Activity

Simplified main parsing function.

Browse Source
pull/13171/head
Joshua Haberman 3 years ago
parent 2e7da49bbe
commit 69bb5d1d94
1 changed files with 108 additions and 92 deletions
Whitespace
Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL
Unified View
Diff Options
Show Stats Download Patch File Download Diff File
  1. 200
      upb/decode.c

200
upb/decode.c
Unescape View File

@ -92,6 +92,7 @@ static const unsigned fixed64_ok = (1 << UPB_DTYPE_DOUBLE) |
(1 << UPB_DTYPE_SFIXED64); (1 << UPB_DTYPE_SFIXED64);
/* Op: an action to be performed for a wire-type/field-type combination. */ /* Op: an action to be performed for a wire-type/field-type combination. */
#define OP_UNKNOWN -1
#define OP_SCALAR_LG2(n) (n) /* n in [0, 2, 3] => op in [0, 2, 3] */ #define OP_SCALAR_LG2(n) (n) /* n in [0, 2, 3] => op in [0, 2, 3] */
#define OP_STRING 4 #define OP_STRING 4
#define OP_BYTES 5 #define OP_BYTES 5
@ -101,48 +102,48 @@ static const unsigned fixed64_ok = (1 << UPB_DTYPE_DOUBLE) |
#define OP_VARPCK_LG2(n) (n + 9) /* n in [0, 2, 3] => op in [9, 11, 12] */ #define OP_VARPCK_LG2(n) (n + 9) /* n in [0, 2, 3] => op in [9, 11, 12] */
static const int8_t varint_ops[19] = { static const int8_t varint_ops[19] = {
-1, /* field not found */ OP_UNKNOWN, /* field not found */
-1, /* DOUBLE */ OP_UNKNOWN, /* DOUBLE */
-1, /* FLOAT */ OP_UNKNOWN, /* FLOAT */
OP_SCALAR_LG2(3), /* INT64 */ OP_SCALAR_LG2(3), /* INT64 */
OP_SCALAR_LG2(3), /* UINT64 */ OP_SCALAR_LG2(3), /* UINT64 */
OP_SCALAR_LG2(2), /* INT32 */ OP_SCALAR_LG2(2), /* INT32 */
-1, /* FIXED64 */ OP_UNKNOWN, /* FIXED64 */
-1, /* FIXED32 */ OP_UNKNOWN, /* FIXED32 */
OP_SCALAR_LG2(0), /* BOOL */ OP_SCALAR_LG2(0), /* BOOL */
-1, /* STRING */ OP_UNKNOWN, /* STRING */
-1, /* GROUP */ OP_UNKNOWN, /* GROUP */
-1, /* MESSAGE */ OP_UNKNOWN, /* MESSAGE */
-1, /* BYTES */ OP_UNKNOWN, /* BYTES */
OP_SCALAR_LG2(2), /* UINT32 */ OP_SCALAR_LG2(2), /* UINT32 */
OP_SCALAR_LG2(2), /* ENUM */ OP_SCALAR_LG2(2), /* ENUM */
-1, /* SFIXED32 */ OP_UNKNOWN, /* SFIXED32 */
-1, /* SFIXED64 */ OP_UNKNOWN, /* SFIXED64 */
OP_SCALAR_LG2(2), /* SINT32 */ OP_SCALAR_LG2(2), /* SINT32 */
OP_SCALAR_LG2(3), /* SINT64 */ OP_SCALAR_LG2(3), /* SINT64 */
}; };
static const int8_t delim_ops[37] = { static const int8_t delim_ops[37] = {
/* For non-repeated field type. */ /* For non-repeated field type. */
-1, /* field not found */ OP_UNKNOWN, /* field not found */
-1, /* DOUBLE */ OP_UNKNOWN, /* DOUBLE */
-1, /* FLOAT */ OP_UNKNOWN, /* FLOAT */
-1, /* INT64 */ OP_UNKNOWN, /* INT64 */
-1, /* UINT64 */ OP_UNKNOWN, /* UINT64 */
-1, /* INT32 */ OP_UNKNOWN, /* INT32 */
-1, /* FIXED64 */ OP_UNKNOWN, /* FIXED64 */
-1, /* FIXED32 */ OP_UNKNOWN, /* FIXED32 */
-1, /* BOOL */ OP_UNKNOWN, /* BOOL */
OP_STRING, /* STRING */ OP_STRING, /* STRING */
-1, /* GROUP */ OP_UNKNOWN, /* GROUP */
OP_SUBMSG, /* MESSAGE */ OP_SUBMSG, /* MESSAGE */
OP_BYTES, /* BYTES */ OP_BYTES, /* BYTES */
-1, /* UINT32 */ OP_UNKNOWN, /* UINT32 */
-1, /* ENUM */ OP_UNKNOWN, /* ENUM */
-1, /* SFIXED32 */ OP_UNKNOWN, /* SFIXED32 */
-1, /* SFIXED64 */ OP_UNKNOWN, /* SFIXED64 */
-1, /* SINT32 */ OP_UNKNOWN, /* SINT32 */
-1, /* SINT64 */ OP_UNKNOWN, /* SINT64 */
/* For repeated field type. */ /* For repeated field type. */
OP_FIXPCK_LG2(3), /* REPEATED DOUBLE */ OP_FIXPCK_LG2(3), /* REPEATED DOUBLE */
OP_FIXPCK_LG2(2), /* REPEATED FLOAT */ OP_FIXPCK_LG2(2), /* REPEATED FLOAT */
@ -608,6 +609,76 @@ static bool decode_tryfastdispatch(upb_decstate *d, const char **ptr,
return false; return false;
} }
UPB_FORCEINLINE
static const char *decode_wireval(upb_decstate *d, const char *ptr,
const upb_msglayout_field *field,
int wire_type, wireval *val, int *op) {
switch (wire_type) {
case UPB_WIRE_TYPE_VARINT:
ptr = decode_varint64(d, ptr, &val->uint64_val);
*op = varint_ops[field->descriptortype];
decode_munge(field->descriptortype, val);
return ptr;
case UPB_WIRE_TYPE_32BIT:
memcpy(&val->uint32_val, ptr, 4);
val->uint32_val = _upb_be_swap32(val->uint32_val);
*op = OP_SCALAR_LG2(2);
if (((1 << field->descriptortype) & fixed32_ok) == 0) *op = OP_UNKNOWN;
return ptr + 4;
case UPB_WIRE_TYPE_64BIT:
memcpy(&val->uint64_val, ptr, 8);
val->uint64_val = _upb_be_swap64(val->uint64_val);
*op = OP_SCALAR_LG2(3);
if (((1 << field->descriptortype) & fixed64_ok) == 0) *op = OP_UNKNOWN;
return ptr + 8;
case UPB_WIRE_TYPE_DELIMITED: {
int ndx = field->descriptortype;
uint64_t size;
if (_upb_getmode(field) == _UPB_MODE_ARRAY) ndx += 18;
ptr = decode_varint64(d, ptr, &size);
if (size >= INT32_MAX ||
ptr - d->end + (int32_t)size > d->limit) {
decode_err(d); /* Length overflow. */
}
*op = delim_ops[ndx];
val->size = size;
return ptr;
}
case UPB_WIRE_TYPE_START_GROUP:
val->uint32_val = field->number;
*op = OP_SUBMSG;
if (field->descriptortype != UPB_DTYPE_GROUP) *op = OP_UNKNOWN;
return ptr;
default:
decode_err(d);
}
}
UPB_FORCEINLINE
static const char *decode_skipunknown(upb_decstate *d, const char *ptr,
upb_msg *msg, int field_number,
int wire_type, wireval val,
const char **field_start) {
if (field_number == 0) decode_err(d);
if (wire_type == UPB_WIRE_TYPE_DELIMITED) ptr += val.size;
if (msg) {
if (wire_type == UPB_WIRE_TYPE_START_GROUP) {
d->unknown = *field_start;
d->unknown_msg = msg;
ptr = decode_group(d, ptr, NULL, NULL, field_number);
d->unknown_msg = NULL;
*field_start = d->unknown;
}
if (!_upb_msg_addunknown(msg, *field_start, ptr - *field_start,
&d->arena)) {
decode_err(d);
}
} else if (wire_type == UPB_WIRE_TYPE_START_GROUP) {
ptr = decode_group(d, ptr, NULL, NULL, field_number);
}
return ptr;
}
UPB_NOINLINE UPB_NOINLINE
static const char *decode_msg(upb_decstate *d, const char *ptr, upb_msg *msg, static const char *decode_msg(upb_decstate *d, const char *ptr, upb_msg *msg,
const upb_msglayout *layout) { const upb_msglayout *layout) {
@ -628,51 +699,13 @@ static const char *decode_msg(upb_decstate *d, const char *ptr, upb_msg *msg,
field = upb_find_field(d, layout, field_number, &last_field_index); field = upb_find_field(d, layout, field_number, &last_field_index);
switch (wire_type) { if (wire_type == UPB_WIRE_TYPE_END_GROUP) {
case UPB_WIRE_TYPE_VARINT: d->end_group = field_number;
ptr = decode_varint64(d, ptr, &val.uint64_val); return ptr;
op = varint_ops[field->descriptortype];
decode_munge(field->descriptortype, &val);
break;
case UPB_WIRE_TYPE_32BIT:
memcpy(&val.uint32_val, ptr, 4);
val.uint32_val = _upb_be_swap32(val.uint32_val);
ptr += 4;
op = OP_SCALAR_LG2(2);
if (((1 << field->descriptortype) & fixed32_ok) == 0) goto unknown;
break;
case UPB_WIRE_TYPE_64BIT:
memcpy(&val.uint64_val, ptr, 8);
val.uint64_val = _upb_be_swap64(val.uint64_val);
ptr += 8;
op = OP_SCALAR_LG2(3);
if (((1 << field->descriptortype) & fixed64_ok) == 0) goto unknown;
break;
case UPB_WIRE_TYPE_DELIMITED: {
int ndx = field->descriptortype;
uint64_t size;
if (_upb_getmode(field) == _UPB_MODE_ARRAY) ndx += 18;
ptr = decode_varint64(d, ptr, &size);
if (size >= INT32_MAX ||
ptr - d->end + (int32_t)size > d->limit) {
decode_err(d); /* Length overflow. */
}
op = delim_ops[ndx];
val.size = size;
break;
}
case UPB_WIRE_TYPE_START_GROUP:
val.uint32_val = field_number;
op = OP_SUBMSG;
if (field->descriptortype != UPB_DTYPE_GROUP) goto unknown;
break;
case UPB_WIRE_TYPE_END_GROUP:
d->end_group = field_number;
return ptr;
default:
decode_err(d);
} }
ptr = decode_wireval(d, ptr, field, wire_type, &val, &op);
if (op >= 0) { if (op >= 0) {
/* Known field, possibly an extension. */ /* Known field, possibly an extension. */
upb_msg *field_msg = msg; upb_msg *field_msg = msg;
@ -702,25 +735,8 @@ static const char *decode_msg(upb_decstate *d, const char *ptr, upb_msg *msg,
UPB_UNREACHABLE(); UPB_UNREACHABLE();
} }
} else { } else {
unknown: ptr = decode_skipunknown(d, ptr, msg, field_number, wire_type, val,
/* Skip unknown field. */ &field_start);
if (field_number == 0) decode_err(d);
if (wire_type == UPB_WIRE_TYPE_DELIMITED) ptr += val.size;
if (msg) {
if (wire_type == UPB_WIRE_TYPE_START_GROUP) {
d->unknown = field_start;
d->unknown_msg = msg;
ptr = decode_group(d, ptr, NULL, NULL, field_number);
d->unknown_msg = NULL;
field_start = d->unknown;
}
if (!_upb_msg_addunknown(msg, field_start, ptr - field_start,
&d->arena)) {
decode_err(d);
}
} else if (wire_type == UPB_WIRE_TYPE_START_GROUP) {
ptr = decode_group(d, ptr, NULL, NULL, field_number);
}
} }
if (decode_isdone(d, &ptr)) return ptr; if (decode_isdone(d, &ptr)) return ptr;

Write Preview
Loading…
Cancel
Save