2024-01-25 23:45:22 +01:00
|
|
|
// Copyright 2023 Google Inc. All rights reserved.
|
|
|
|
//
|
|
|
|
// Licensed under the Apache License, Version 2.0 (the "License");
|
|
|
|
// you may not use this file except in compliance with the License.
|
|
|
|
// You may obtain a copy of the License at
|
|
|
|
//
|
|
|
|
// http://www.apache.org/licenses/LICENSE-2.0
|
|
|
|
//
|
|
|
|
// Unless required by applicable law or agreed to in writing, software
|
|
|
|
// distributed under the License is distributed on an "AS IS" BASIS,
|
|
|
|
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
|
|
// See the License for the specific language governing permissions and
|
|
|
|
// limitations under the License.
|
|
|
|
|
|
|
|
package proptools
|
|
|
|
|
|
|
|
import (
|
|
|
|
"cmp"
|
|
|
|
"encoding/binary"
|
|
|
|
"fmt"
|
|
|
|
"hash/maphash"
|
|
|
|
"math"
|
|
|
|
"reflect"
|
|
|
|
"sort"
|
|
|
|
)
|
|
|
|
|
|
|
|
var seed maphash.Seed = maphash.MakeSeed()
|
|
|
|
|
|
|
|
// byte to insert between elements of lists, fields of structs/maps, etc in order
|
|
|
|
// to try and make sure the hash is different when values are moved around between
|
|
|
|
// elements. 36 is arbitrary, but it's the ascii code for a record separator
|
|
|
|
var recordSeparator []byte = []byte{36}
|
|
|
|
|
|
|
|
func HashProvider(provider interface{}) (uint64, error) {
|
|
|
|
hasher := maphash.Hash{}
|
|
|
|
hasher.SetSeed(seed)
|
|
|
|
ptrs := make(map[uintptr]bool)
|
|
|
|
v := reflect.ValueOf(provider)
|
|
|
|
var err error
|
|
|
|
if v.IsValid() {
|
|
|
|
err = hashProviderInternal(&hasher, v, ptrs)
|
|
|
|
}
|
|
|
|
return hasher.Sum64(), err
|
|
|
|
}
|
|
|
|
|
2024-02-01 21:34:01 +01:00
|
|
|
func hashProviderInternal(hasher *maphash.Hash, v reflect.Value, ptrs map[uintptr]bool) error {
|
2024-01-25 23:45:22 +01:00
|
|
|
var int64Array [8]byte
|
|
|
|
int64Buf := int64Array[:]
|
|
|
|
binary.LittleEndian.PutUint64(int64Buf, uint64(v.Kind()))
|
|
|
|
hasher.Write(int64Buf)
|
|
|
|
v.IsValid()
|
|
|
|
switch v.Kind() {
|
|
|
|
case reflect.Struct:
|
|
|
|
binary.LittleEndian.PutUint64(int64Buf, uint64(v.NumField()))
|
|
|
|
hasher.Write(int64Buf)
|
|
|
|
for i := 0; i < v.NumField(); i++ {
|
|
|
|
hasher.Write(recordSeparator)
|
|
|
|
err := hashProviderInternal(hasher, v.Field(i), ptrs)
|
|
|
|
if err != nil {
|
|
|
|
return fmt.Errorf("in field %s: %s", v.Type().Field(i).Name, err.Error())
|
|
|
|
}
|
|
|
|
}
|
|
|
|
case reflect.Map:
|
|
|
|
binary.LittleEndian.PutUint64(int64Buf, uint64(v.Len()))
|
|
|
|
hasher.Write(int64Buf)
|
|
|
|
indexes := make([]int, v.Len())
|
|
|
|
keys := make([]reflect.Value, v.Len())
|
|
|
|
values := make([]reflect.Value, v.Len())
|
|
|
|
iter := v.MapRange()
|
|
|
|
for i := 0; iter.Next(); i++ {
|
|
|
|
indexes[i] = i
|
|
|
|
keys[i] = iter.Key()
|
|
|
|
values[i] = iter.Value()
|
|
|
|
}
|
|
|
|
sort.SliceStable(indexes, func(i, j int) bool {
|
|
|
|
return compare_values(keys[indexes[i]], keys[indexes[j]]) < 0
|
|
|
|
})
|
|
|
|
for i := 0; i < v.Len(); i++ {
|
|
|
|
hasher.Write(recordSeparator)
|
|
|
|
err := hashProviderInternal(hasher, keys[indexes[i]], ptrs)
|
|
|
|
if err != nil {
|
|
|
|
return fmt.Errorf("in map: %s", err.Error())
|
|
|
|
}
|
|
|
|
hasher.Write(recordSeparator)
|
|
|
|
err = hashProviderInternal(hasher, keys[indexes[i]], ptrs)
|
|
|
|
if err != nil {
|
|
|
|
return fmt.Errorf("in map: %s", err.Error())
|
|
|
|
}
|
|
|
|
}
|
|
|
|
case reflect.Slice, reflect.Array:
|
|
|
|
binary.LittleEndian.PutUint64(int64Buf, uint64(v.Len()))
|
|
|
|
hasher.Write(int64Buf)
|
|
|
|
for i := 0; i < v.Len(); i++ {
|
|
|
|
hasher.Write(recordSeparator)
|
|
|
|
err := hashProviderInternal(hasher, v.Index(i), ptrs)
|
|
|
|
if err != nil {
|
|
|
|
return fmt.Errorf("in %s at index %d: %s", v.Kind().String(), i, err.Error())
|
|
|
|
}
|
|
|
|
}
|
|
|
|
case reflect.Pointer:
|
|
|
|
if v.IsNil() {
|
|
|
|
int64Buf[0] = 0
|
|
|
|
hasher.Write(int64Buf[:1])
|
|
|
|
return nil
|
|
|
|
}
|
|
|
|
addr := v.Pointer()
|
|
|
|
binary.LittleEndian.PutUint64(int64Buf, uint64(addr))
|
|
|
|
hasher.Write(int64Buf)
|
|
|
|
if _, ok := ptrs[addr]; ok {
|
|
|
|
// We could make this an error if we want to disallow pointer cycles in the future
|
|
|
|
return nil
|
|
|
|
}
|
|
|
|
ptrs[addr] = true
|
|
|
|
err := hashProviderInternal(hasher, v.Elem(), ptrs)
|
|
|
|
if err != nil {
|
|
|
|
return fmt.Errorf("in pointer: %s", err.Error())
|
|
|
|
}
|
|
|
|
case reflect.Interface:
|
|
|
|
if v.IsNil() {
|
|
|
|
int64Buf[0] = 0
|
|
|
|
hasher.Write(int64Buf[:1])
|
|
|
|
} else {
|
|
|
|
// The only way get the pointer out of an interface to hash it or check for cycles
|
|
|
|
// would be InterfaceData(), but that's deprecated and seems like it has undefined behavior.
|
|
|
|
err := hashProviderInternal(hasher, v.Elem(), ptrs)
|
|
|
|
if err != nil {
|
|
|
|
return fmt.Errorf("in interface: %s", err.Error())
|
|
|
|
}
|
|
|
|
}
|
|
|
|
case reflect.String:
|
2024-02-01 21:34:01 +01:00
|
|
|
hasher.WriteString(v.String())
|
2024-01-25 23:45:22 +01:00
|
|
|
case reflect.Bool:
|
|
|
|
if v.Bool() {
|
|
|
|
int64Buf[0] = 1
|
|
|
|
} else {
|
|
|
|
int64Buf[0] = 0
|
|
|
|
}
|
|
|
|
hasher.Write(int64Buf[:1])
|
|
|
|
case reflect.Uint, reflect.Uint8, reflect.Uint16, reflect.Uint32, reflect.Uint64, reflect.Uintptr:
|
|
|
|
binary.LittleEndian.PutUint64(int64Buf, v.Uint())
|
|
|
|
hasher.Write(int64Buf)
|
|
|
|
case reflect.Int, reflect.Int8, reflect.Int16, reflect.Int32, reflect.Int64:
|
|
|
|
binary.LittleEndian.PutUint64(int64Buf, uint64(v.Int()))
|
|
|
|
hasher.Write(int64Buf)
|
|
|
|
case reflect.Float32, reflect.Float64:
|
|
|
|
binary.LittleEndian.PutUint64(int64Buf, math.Float64bits(v.Float()))
|
|
|
|
hasher.Write(int64Buf)
|
|
|
|
default:
|
|
|
|
return fmt.Errorf("providers may only contain primitives, strings, arrays, slices, structs, maps, and pointers, found: %s", v.Kind().String())
|
|
|
|
}
|
|
|
|
return nil
|
|
|
|
}
|
|
|
|
|
|
|
|
func compare_values(x, y reflect.Value) int {
|
|
|
|
if x.Type() != y.Type() {
|
|
|
|
panic("Expected equal types")
|
|
|
|
}
|
|
|
|
|
|
|
|
switch x.Kind() {
|
|
|
|
case reflect.Uint, reflect.Uint8, reflect.Uint16, reflect.Uint32, reflect.Uint64, reflect.Uintptr:
|
|
|
|
return cmp.Compare(x.Uint(), y.Uint())
|
|
|
|
case reflect.Int, reflect.Int8, reflect.Int16, reflect.Int32, reflect.Int64:
|
|
|
|
return cmp.Compare(x.Int(), y.Int())
|
|
|
|
case reflect.Float32, reflect.Float64:
|
|
|
|
return cmp.Compare(x.Float(), y.Float())
|
|
|
|
case reflect.String:
|
|
|
|
return cmp.Compare(x.String(), y.String())
|
|
|
|
case reflect.Bool:
|
|
|
|
if x.Bool() == y.Bool() {
|
|
|
|
return 0
|
|
|
|
} else if x.Bool() {
|
|
|
|
return 1
|
|
|
|
} else {
|
|
|
|
return -1
|
|
|
|
}
|
|
|
|
case reflect.Pointer:
|
|
|
|
return cmp.Compare(x.Pointer(), y.Pointer())
|
|
|
|
case reflect.Array:
|
|
|
|
for i := 0; i < x.Len(); i++ {
|
|
|
|
if result := compare_values(x.Index(i), y.Index(i)); result != 0 {
|
|
|
|
return result
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return 0
|
|
|
|
case reflect.Struct:
|
|
|
|
for i := 0; i < x.NumField(); i++ {
|
|
|
|
if result := compare_values(x.Field(i), y.Field(i)); result != 0 {
|
|
|
|
return result
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return 0
|
|
|
|
case reflect.Interface:
|
|
|
|
if x.IsNil() && y.IsNil() {
|
|
|
|
return 0
|
|
|
|
} else if x.IsNil() {
|
|
|
|
return 1
|
|
|
|
} else if y.IsNil() {
|
|
|
|
return -1
|
|
|
|
}
|
|
|
|
return compare_values(x.Elem(), y.Elem())
|
|
|
|
default:
|
|
|
|
panic(fmt.Sprintf("Could not compare types %s and %s", x.Type().String(), y.Type().String()))
|
|
|
|
}
|
|
|
|
}
|