go-ethereum/rlp/typecache.go

220 lines
5.6 KiB
Go
Raw Normal View History

2015-07-07 03:54:22 +03:00
// Copyright 2014 The go-ethereum Authors
// This file is part of the go-ethereum library.
2015-07-07 03:54:22 +03:00
//
// The go-ethereum library is free software: you can redistribute it and/or modify
2015-07-07 03:54:22 +03:00
// it under the terms of the GNU Lesser General Public License as published by
// the Free Software Foundation, either version 3 of the License, or
// (at your option) any later version.
//
// The go-ethereum library is distributed in the hope that it will be useful,
2015-07-07 03:54:22 +03:00
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
2015-07-07 03:54:22 +03:00
// GNU Lesser General Public License for more details.
//
// You should have received a copy of the GNU Lesser General Public License
// along with the go-ethereum library. If not, see <http://www.gnu.org/licenses/>.
2015-07-07 03:54:22 +03:00
package rlp
import (
2015-12-21 22:05:20 +02:00
"fmt"
"reflect"
2015-12-21 22:05:20 +02:00
"strings"
"sync"
)
var (
typeCacheMutex sync.RWMutex
typeCache = make(map[typekey]*typeinfo)
)
type typeinfo struct {
decoder decoder
decoderErr error // error from makeDecoder
writer writer
writerErr error // error from makeWriter
}
rlp: improve nil pointer handling (#20064) * rlp: improve nil pointer handling In both encoder and decoder, the rules for encoding nil pointers were a bit hard to understand, and didn't leave much choice. Since RLP allows two empty values (empty list, empty string), any protocol built on RLP must choose either of these values to represent the null value in a certain context. This change adds choice in the form of two new struct tags, "nilString" and "nilList". These can be used to specify how a nil pointer value is encoded. The "nil" tag still exists, but its implementation is now explicit and defines exactly how nil pointers are handled in a single place. Another important change in this commit is how nil pointers and the Encoder interface interact. The EncodeRLP method was previously called even on nil values, which was supposed to give users a choice of how their value would be handled when nil. It turns out this is a stupid idea. If you create a network protocol containing an object defined in another package, it's better to be able to say that the object should be a list or string when nil in the definition of the protocol message rather than defining the encoding of nil on the object itself. As of this commit, the encoding rules for pointers now take precedence over the Encoder interface rule. I think the "nil" tag will work fine for most cases. For special kinds of objects which are a struct in Go but strings in RLP, code using the object can specify the desired encoding of nil using the "nilString" and "nilList" tags. * rlp: propagate struct field type errors If a struct contained fields of undecodable type, the encoder and decoder would panic instead of returning an error. Fix this by propagating type errors in makeStruct{Writer,Decoder} and add a test.
2019-09-13 12:10:57 +03:00
// tags represents struct tags.
type tags struct {
2015-12-21 22:05:20 +02:00
// rlp:"nil" controls whether empty input results in a nil pointer.
nilOK bool
rlp: improve nil pointer handling (#20064) * rlp: improve nil pointer handling In both encoder and decoder, the rules for encoding nil pointers were a bit hard to understand, and didn't leave much choice. Since RLP allows two empty values (empty list, empty string), any protocol built on RLP must choose either of these values to represent the null value in a certain context. This change adds choice in the form of two new struct tags, "nilString" and "nilList". These can be used to specify how a nil pointer value is encoded. The "nil" tag still exists, but its implementation is now explicit and defines exactly how nil pointers are handled in a single place. Another important change in this commit is how nil pointers and the Encoder interface interact. The EncodeRLP method was previously called even on nil values, which was supposed to give users a choice of how their value would be handled when nil. It turns out this is a stupid idea. If you create a network protocol containing an object defined in another package, it's better to be able to say that the object should be a list or string when nil in the definition of the protocol message rather than defining the encoding of nil on the object itself. As of this commit, the encoding rules for pointers now take precedence over the Encoder interface rule. I think the "nil" tag will work fine for most cases. For special kinds of objects which are a struct in Go but strings in RLP, code using the object can specify the desired encoding of nil using the "nilString" and "nilList" tags. * rlp: propagate struct field type errors If a struct contained fields of undecodable type, the encoder and decoder would panic instead of returning an error. Fix this by propagating type errors in makeStruct{Writer,Decoder} and add a test.
2019-09-13 12:10:57 +03:00
// This controls whether nil pointers are encoded/decoded as empty strings
// or empty lists.
nilKind Kind
2015-12-21 22:05:20 +02:00
// rlp:"tail" controls whether this field swallows additional list
// elements. It can only be set for the last field, which must be
// of slice type.
tail bool
rlp: improve nil pointer handling (#20064) * rlp: improve nil pointer handling In both encoder and decoder, the rules for encoding nil pointers were a bit hard to understand, and didn't leave much choice. Since RLP allows two empty values (empty list, empty string), any protocol built on RLP must choose either of these values to represent the null value in a certain context. This change adds choice in the form of two new struct tags, "nilString" and "nilList". These can be used to specify how a nil pointer value is encoded. The "nil" tag still exists, but its implementation is now explicit and defines exactly how nil pointers are handled in a single place. Another important change in this commit is how nil pointers and the Encoder interface interact. The EncodeRLP method was previously called even on nil values, which was supposed to give users a choice of how their value would be handled when nil. It turns out this is a stupid idea. If you create a network protocol containing an object defined in another package, it's better to be able to say that the object should be a list or string when nil in the definition of the protocol message rather than defining the encoding of nil on the object itself. As of this commit, the encoding rules for pointers now take precedence over the Encoder interface rule. I think the "nil" tag will work fine for most cases. For special kinds of objects which are a struct in Go but strings in RLP, code using the object can specify the desired encoding of nil using the "nilString" and "nilList" tags. * rlp: propagate struct field type errors If a struct contained fields of undecodable type, the encoder and decoder would panic instead of returning an error. Fix this by propagating type errors in makeStruct{Writer,Decoder} and add a test.
2019-09-13 12:10:57 +03:00
2017-03-07 14:37:53 +03:00
// rlp:"-" ignores fields.
ignored bool
}
rlp: improve nil pointer handling (#20064) * rlp: improve nil pointer handling In both encoder and decoder, the rules for encoding nil pointers were a bit hard to understand, and didn't leave much choice. Since RLP allows two empty values (empty list, empty string), any protocol built on RLP must choose either of these values to represent the null value in a certain context. This change adds choice in the form of two new struct tags, "nilString" and "nilList". These can be used to specify how a nil pointer value is encoded. The "nil" tag still exists, but its implementation is now explicit and defines exactly how nil pointers are handled in a single place. Another important change in this commit is how nil pointers and the Encoder interface interact. The EncodeRLP method was previously called even on nil values, which was supposed to give users a choice of how their value would be handled when nil. It turns out this is a stupid idea. If you create a network protocol containing an object defined in another package, it's better to be able to say that the object should be a list or string when nil in the definition of the protocol message rather than defining the encoding of nil on the object itself. As of this commit, the encoding rules for pointers now take precedence over the Encoder interface rule. I think the "nil" tag will work fine for most cases. For special kinds of objects which are a struct in Go but strings in RLP, code using the object can specify the desired encoding of nil using the "nilString" and "nilList" tags. * rlp: propagate struct field type errors If a struct contained fields of undecodable type, the encoder and decoder would panic instead of returning an error. Fix this by propagating type errors in makeStruct{Writer,Decoder} and add a test.
2019-09-13 12:10:57 +03:00
// typekey is the key of a type in typeCache. It includes the struct tags because
// they might generate a different decoder.
type typekey struct {
reflect.Type
tags
}
type decoder func(*Stream, reflect.Value) error
type writer func(reflect.Value, *encbuf) error
func cachedDecoder(typ reflect.Type) (decoder, error) {
info := cachedTypeInfo(typ, tags{})
return info.decoder, info.decoderErr
}
func cachedWriter(typ reflect.Type) (writer, error) {
info := cachedTypeInfo(typ, tags{})
return info.writer, info.writerErr
}
func cachedTypeInfo(typ reflect.Type, tags tags) *typeinfo {
typeCacheMutex.RLock()
info := typeCache[typekey{typ, tags}]
typeCacheMutex.RUnlock()
if info != nil {
return info
}
// not in the cache, need to generate info for this type.
typeCacheMutex.Lock()
defer typeCacheMutex.Unlock()
return cachedTypeInfo1(typ, tags)
}
func cachedTypeInfo1(typ reflect.Type, tags tags) *typeinfo {
key := typekey{typ, tags}
info := typeCache[key]
if info != nil {
// another goroutine got the write lock first
return info
}
2018-09-11 18:05:28 +03:00
// put a dummy value into the cache before generating.
// if the generator tries to lookup itself, it will get
// the dummy value and won't call itself recursively.
info = new(typeinfo)
typeCache[key] = info
info.generate(typ, tags)
return info
}
type field struct {
index int
info *typeinfo
}
func structFields(typ reflect.Type) (fields []field, err error) {
lastPublic := lastPublicField(typ)
for i := 0; i < typ.NumField(); i++ {
if f := typ.Field(i); f.PkgPath == "" { // exported
tags, err := parseStructTag(typ, i, lastPublic)
2015-12-21 22:05:20 +02:00
if err != nil {
return nil, err
}
2017-03-07 14:37:53 +03:00
if tags.ignored {
continue
}
info := cachedTypeInfo1(f.Type, tags)
fields = append(fields, field{i, info})
}
}
return fields, nil
}
rlp: improve nil pointer handling (#20064) * rlp: improve nil pointer handling In both encoder and decoder, the rules for encoding nil pointers were a bit hard to understand, and didn't leave much choice. Since RLP allows two empty values (empty list, empty string), any protocol built on RLP must choose either of these values to represent the null value in a certain context. This change adds choice in the form of two new struct tags, "nilString" and "nilList". These can be used to specify how a nil pointer value is encoded. The "nil" tag still exists, but its implementation is now explicit and defines exactly how nil pointers are handled in a single place. Another important change in this commit is how nil pointers and the Encoder interface interact. The EncodeRLP method was previously called even on nil values, which was supposed to give users a choice of how their value would be handled when nil. It turns out this is a stupid idea. If you create a network protocol containing an object defined in another package, it's better to be able to say that the object should be a list or string when nil in the definition of the protocol message rather than defining the encoding of nil on the object itself. As of this commit, the encoding rules for pointers now take precedence over the Encoder interface rule. I think the "nil" tag will work fine for most cases. For special kinds of objects which are a struct in Go but strings in RLP, code using the object can specify the desired encoding of nil using the "nilString" and "nilList" tags. * rlp: propagate struct field type errors If a struct contained fields of undecodable type, the encoder and decoder would panic instead of returning an error. Fix this by propagating type errors in makeStruct{Writer,Decoder} and add a test.
2019-09-13 12:10:57 +03:00
type structFieldError struct {
typ reflect.Type
field int
err error
}
func (e structFieldError) Error() string {
return fmt.Sprintf("%v (struct field %v.%s)", e.err, e.typ, e.typ.Field(e.field).Name)
}
type structTagError struct {
typ reflect.Type
field, tag, err string
}
func (e structTagError) Error() string {
return fmt.Sprintf("rlp: invalid struct tag %q for %v.%s (%s)", e.tag, e.typ, e.field, e.err)
}
func parseStructTag(typ reflect.Type, fi, lastPublic int) (tags, error) {
2015-12-21 22:05:20 +02:00
f := typ.Field(fi)
var ts tags
for _, t := range strings.Split(f.Tag.Get("rlp"), ",") {
switch t = strings.TrimSpace(t); t {
case "":
2017-03-07 14:37:53 +03:00
case "-":
ts.ignored = true
rlp: improve nil pointer handling (#20064) * rlp: improve nil pointer handling In both encoder and decoder, the rules for encoding nil pointers were a bit hard to understand, and didn't leave much choice. Since RLP allows two empty values (empty list, empty string), any protocol built on RLP must choose either of these values to represent the null value in a certain context. This change adds choice in the form of two new struct tags, "nilString" and "nilList". These can be used to specify how a nil pointer value is encoded. The "nil" tag still exists, but its implementation is now explicit and defines exactly how nil pointers are handled in a single place. Another important change in this commit is how nil pointers and the Encoder interface interact. The EncodeRLP method was previously called even on nil values, which was supposed to give users a choice of how their value would be handled when nil. It turns out this is a stupid idea. If you create a network protocol containing an object defined in another package, it's better to be able to say that the object should be a list or string when nil in the definition of the protocol message rather than defining the encoding of nil on the object itself. As of this commit, the encoding rules for pointers now take precedence over the Encoder interface rule. I think the "nil" tag will work fine for most cases. For special kinds of objects which are a struct in Go but strings in RLP, code using the object can specify the desired encoding of nil using the "nilString" and "nilList" tags. * rlp: propagate struct field type errors If a struct contained fields of undecodable type, the encoder and decoder would panic instead of returning an error. Fix this by propagating type errors in makeStruct{Writer,Decoder} and add a test.
2019-09-13 12:10:57 +03:00
case "nil", "nilString", "nilList":
2015-12-21 22:05:20 +02:00
ts.nilOK = true
rlp: improve nil pointer handling (#20064) * rlp: improve nil pointer handling In both encoder and decoder, the rules for encoding nil pointers were a bit hard to understand, and didn't leave much choice. Since RLP allows two empty values (empty list, empty string), any protocol built on RLP must choose either of these values to represent the null value in a certain context. This change adds choice in the form of two new struct tags, "nilString" and "nilList". These can be used to specify how a nil pointer value is encoded. The "nil" tag still exists, but its implementation is now explicit and defines exactly how nil pointers are handled in a single place. Another important change in this commit is how nil pointers and the Encoder interface interact. The EncodeRLP method was previously called even on nil values, which was supposed to give users a choice of how their value would be handled when nil. It turns out this is a stupid idea. If you create a network protocol containing an object defined in another package, it's better to be able to say that the object should be a list or string when nil in the definition of the protocol message rather than defining the encoding of nil on the object itself. As of this commit, the encoding rules for pointers now take precedence over the Encoder interface rule. I think the "nil" tag will work fine for most cases. For special kinds of objects which are a struct in Go but strings in RLP, code using the object can specify the desired encoding of nil using the "nilString" and "nilList" tags. * rlp: propagate struct field type errors If a struct contained fields of undecodable type, the encoder and decoder would panic instead of returning an error. Fix this by propagating type errors in makeStruct{Writer,Decoder} and add a test.
2019-09-13 12:10:57 +03:00
if f.Type.Kind() != reflect.Ptr {
return ts, structTagError{typ, f.Name, t, "field is not a pointer"}
}
switch t {
case "nil":
ts.nilKind = defaultNilKind(f.Type.Elem())
case "nilString":
ts.nilKind = String
case "nilList":
ts.nilKind = List
}
2015-12-21 22:05:20 +02:00
case "tail":
ts.tail = true
if fi != lastPublic {
rlp: improve nil pointer handling (#20064) * rlp: improve nil pointer handling In both encoder and decoder, the rules for encoding nil pointers were a bit hard to understand, and didn't leave much choice. Since RLP allows two empty values (empty list, empty string), any protocol built on RLP must choose either of these values to represent the null value in a certain context. This change adds choice in the form of two new struct tags, "nilString" and "nilList". These can be used to specify how a nil pointer value is encoded. The "nil" tag still exists, but its implementation is now explicit and defines exactly how nil pointers are handled in a single place. Another important change in this commit is how nil pointers and the Encoder interface interact. The EncodeRLP method was previously called even on nil values, which was supposed to give users a choice of how their value would be handled when nil. It turns out this is a stupid idea. If you create a network protocol containing an object defined in another package, it's better to be able to say that the object should be a list or string when nil in the definition of the protocol message rather than defining the encoding of nil on the object itself. As of this commit, the encoding rules for pointers now take precedence over the Encoder interface rule. I think the "nil" tag will work fine for most cases. For special kinds of objects which are a struct in Go but strings in RLP, code using the object can specify the desired encoding of nil using the "nilString" and "nilList" tags. * rlp: propagate struct field type errors If a struct contained fields of undecodable type, the encoder and decoder would panic instead of returning an error. Fix this by propagating type errors in makeStruct{Writer,Decoder} and add a test.
2019-09-13 12:10:57 +03:00
return ts, structTagError{typ, f.Name, t, "must be on last field"}
2015-12-21 22:05:20 +02:00
}
if f.Type.Kind() != reflect.Slice {
rlp: improve nil pointer handling (#20064) * rlp: improve nil pointer handling In both encoder and decoder, the rules for encoding nil pointers were a bit hard to understand, and didn't leave much choice. Since RLP allows two empty values (empty list, empty string), any protocol built on RLP must choose either of these values to represent the null value in a certain context. This change adds choice in the form of two new struct tags, "nilString" and "nilList". These can be used to specify how a nil pointer value is encoded. The "nil" tag still exists, but its implementation is now explicit and defines exactly how nil pointers are handled in a single place. Another important change in this commit is how nil pointers and the Encoder interface interact. The EncodeRLP method was previously called even on nil values, which was supposed to give users a choice of how their value would be handled when nil. It turns out this is a stupid idea. If you create a network protocol containing an object defined in another package, it's better to be able to say that the object should be a list or string when nil in the definition of the protocol message rather than defining the encoding of nil on the object itself. As of this commit, the encoding rules for pointers now take precedence over the Encoder interface rule. I think the "nil" tag will work fine for most cases. For special kinds of objects which are a struct in Go but strings in RLP, code using the object can specify the desired encoding of nil using the "nilString" and "nilList" tags. * rlp: propagate struct field type errors If a struct contained fields of undecodable type, the encoder and decoder would panic instead of returning an error. Fix this by propagating type errors in makeStruct{Writer,Decoder} and add a test.
2019-09-13 12:10:57 +03:00
return ts, structTagError{typ, f.Name, t, "field type is not slice"}
2015-12-21 22:05:20 +02:00
}
default:
return ts, fmt.Errorf("rlp: unknown struct tag %q on %v.%s", t, typ, f.Name)
}
}
return ts, nil
}
func lastPublicField(typ reflect.Type) int {
last := 0
for i := 0; i < typ.NumField(); i++ {
if typ.Field(i).PkgPath == "" {
last = i
}
}
return last
}
func (i *typeinfo) generate(typ reflect.Type, tags tags) {
i.decoder, i.decoderErr = makeDecoder(typ, tags)
i.writer, i.writerErr = makeWriter(typ, tags)
}
rlp: improve nil pointer handling (#20064) * rlp: improve nil pointer handling In both encoder and decoder, the rules for encoding nil pointers were a bit hard to understand, and didn't leave much choice. Since RLP allows two empty values (empty list, empty string), any protocol built on RLP must choose either of these values to represent the null value in a certain context. This change adds choice in the form of two new struct tags, "nilString" and "nilList". These can be used to specify how a nil pointer value is encoded. The "nil" tag still exists, but its implementation is now explicit and defines exactly how nil pointers are handled in a single place. Another important change in this commit is how nil pointers and the Encoder interface interact. The EncodeRLP method was previously called even on nil values, which was supposed to give users a choice of how their value would be handled when nil. It turns out this is a stupid idea. If you create a network protocol containing an object defined in another package, it's better to be able to say that the object should be a list or string when nil in the definition of the protocol message rather than defining the encoding of nil on the object itself. As of this commit, the encoding rules for pointers now take precedence over the Encoder interface rule. I think the "nil" tag will work fine for most cases. For special kinds of objects which are a struct in Go but strings in RLP, code using the object can specify the desired encoding of nil using the "nilString" and "nilList" tags. * rlp: propagate struct field type errors If a struct contained fields of undecodable type, the encoder and decoder would panic instead of returning an error. Fix this by propagating type errors in makeStruct{Writer,Decoder} and add a test.
2019-09-13 12:10:57 +03:00
// defaultNilKind determines whether a nil pointer to typ encodes/decodes
// as an empty string or empty list.
func defaultNilKind(typ reflect.Type) Kind {
k := typ.Kind()
if isUint(k) || k == reflect.String || k == reflect.Bool || isByteArray(typ) {
return String
}
return List
}
func isUint(k reflect.Kind) bool {
return k >= reflect.Uint && k <= reflect.Uintptr
}
rlp: improve nil pointer handling (#20064) * rlp: improve nil pointer handling In both encoder and decoder, the rules for encoding nil pointers were a bit hard to understand, and didn't leave much choice. Since RLP allows two empty values (empty list, empty string), any protocol built on RLP must choose either of these values to represent the null value in a certain context. This change adds choice in the form of two new struct tags, "nilString" and "nilList". These can be used to specify how a nil pointer value is encoded. The "nil" tag still exists, but its implementation is now explicit and defines exactly how nil pointers are handled in a single place. Another important change in this commit is how nil pointers and the Encoder interface interact. The EncodeRLP method was previously called even on nil values, which was supposed to give users a choice of how their value would be handled when nil. It turns out this is a stupid idea. If you create a network protocol containing an object defined in another package, it's better to be able to say that the object should be a list or string when nil in the definition of the protocol message rather than defining the encoding of nil on the object itself. As of this commit, the encoding rules for pointers now take precedence over the Encoder interface rule. I think the "nil" tag will work fine for most cases. For special kinds of objects which are a struct in Go but strings in RLP, code using the object can specify the desired encoding of nil using the "nilString" and "nilList" tags. * rlp: propagate struct field type errors If a struct contained fields of undecodable type, the encoder and decoder would panic instead of returning an error. Fix this by propagating type errors in makeStruct{Writer,Decoder} and add a test.
2019-09-13 12:10:57 +03:00
func isByte(typ reflect.Type) bool {
return typ.Kind() == reflect.Uint8 && !typ.Implements(encoderInterface)
}
rlp: improve nil pointer handling (#20064) * rlp: improve nil pointer handling In both encoder and decoder, the rules for encoding nil pointers were a bit hard to understand, and didn't leave much choice. Since RLP allows two empty values (empty list, empty string), any protocol built on RLP must choose either of these values to represent the null value in a certain context. This change adds choice in the form of two new struct tags, "nilString" and "nilList". These can be used to specify how a nil pointer value is encoded. The "nil" tag still exists, but its implementation is now explicit and defines exactly how nil pointers are handled in a single place. Another important change in this commit is how nil pointers and the Encoder interface interact. The EncodeRLP method was previously called even on nil values, which was supposed to give users a choice of how their value would be handled when nil. It turns out this is a stupid idea. If you create a network protocol containing an object defined in another package, it's better to be able to say that the object should be a list or string when nil in the definition of the protocol message rather than defining the encoding of nil on the object itself. As of this commit, the encoding rules for pointers now take precedence over the Encoder interface rule. I think the "nil" tag will work fine for most cases. For special kinds of objects which are a struct in Go but strings in RLP, code using the object can specify the desired encoding of nil using the "nilString" and "nilList" tags. * rlp: propagate struct field type errors If a struct contained fields of undecodable type, the encoder and decoder would panic instead of returning an error. Fix this by propagating type errors in makeStruct{Writer,Decoder} and add a test.
2019-09-13 12:10:57 +03:00
func isByteArray(typ reflect.Type) bool {
return (typ.Kind() == reflect.Slice || typ.Kind() == reflect.Array) && isByte(typ.Elem())
}