474 lines
11 KiB
Go
474 lines
11 KiB
Go
// Copyright 2017 The OPA Authors. All rights reserved.
|
|
// Use of this source code is governed by an Apache2
|
|
// license that can be found in the LICENSE file.
|
|
|
|
package topdown
|
|
|
|
import (
|
|
"fmt"
|
|
"strconv"
|
|
"strings"
|
|
|
|
"github.com/open-policy-agent/opa/v1/ast"
|
|
)
|
|
|
|
type undo struct {
|
|
k *ast.Term
|
|
u *bindings
|
|
}
|
|
|
|
func (u *undo) Undo() {
|
|
if u == nil {
|
|
// Allow call on zero value of Undo for ease-of-use.
|
|
return
|
|
}
|
|
if u.u == nil {
|
|
// Call on empty unifier undos a no-op unify operation.
|
|
return
|
|
}
|
|
u.u.delete(u.k)
|
|
}
|
|
|
|
type bindings struct {
|
|
id uint64
|
|
values bindingsArrayHashmap
|
|
instr *Instrumentation
|
|
}
|
|
|
|
func newBindings(id uint64, instr *Instrumentation) *bindings {
|
|
values := newBindingsArrayHashmap()
|
|
return &bindings{id, values, instr}
|
|
}
|
|
|
|
// newBindingsWithSize creates bindings pre-sized for the expected number of entries.
|
|
// This avoids over-allocation when the binding count is known in advance (e.g., function arguments).
|
|
// For sizeHint <= maxLinearScan, it uses array mode; for larger hints, it pre-allocates a map.
|
|
func newBindingsWithSize(id uint64, instr *Instrumentation, sizeHint int) *bindings {
|
|
values := newBindingsArrayHashmapWithSize(sizeHint)
|
|
return &bindings{id, values, instr}
|
|
}
|
|
|
|
func (u *bindings) Iter(caller *bindings, iter func(*ast.Term, *ast.Term) error) error {
|
|
|
|
var err error
|
|
|
|
u.values.Iter(func(k *ast.Term, _ value) bool {
|
|
if err != nil {
|
|
return true
|
|
}
|
|
err = iter(k, u.PlugNamespaced(k, caller))
|
|
|
|
return false
|
|
})
|
|
|
|
return err
|
|
}
|
|
|
|
func (u *bindings) Namespace(x ast.Node, caller *bindings) {
|
|
vis := namespacingVisitor{
|
|
b: u,
|
|
caller: caller,
|
|
}
|
|
ast.NewGenericVisitor(vis.Visit).Walk(x)
|
|
}
|
|
|
|
func (u *bindings) Plug(a *ast.Term) *ast.Term {
|
|
return u.PlugNamespaced(a, nil)
|
|
}
|
|
|
|
func (u *bindings) PlugNamespaced(a *ast.Term, caller *bindings) *ast.Term {
|
|
if u != nil && u.instr != nil {
|
|
u.instr.startTimer(evalOpPlug)
|
|
t := u.plugNamespaced(a, caller)
|
|
u.instr.stopTimer(evalOpPlug)
|
|
return t
|
|
}
|
|
|
|
return u.plugNamespaced(a, caller)
|
|
}
|
|
|
|
func (u *bindings) plugNamespaced(a *ast.Term, caller *bindings) *ast.Term {
|
|
switch v := a.Value.(type) {
|
|
case ast.Var:
|
|
b, next := u.apply(a)
|
|
if a != b || u != next {
|
|
return next.plugNamespaced(b, caller)
|
|
}
|
|
return u.namespaceVar(b, caller)
|
|
case *ast.Array:
|
|
if a.IsGround() {
|
|
return a
|
|
}
|
|
cpy := *a
|
|
arr := make([]*ast.Term, v.Len())
|
|
for i := range arr {
|
|
arr[i] = u.plugNamespaced(v.Elem(i), caller)
|
|
}
|
|
cpy.Value = ast.NewArray(arr...)
|
|
return &cpy
|
|
case ast.Object:
|
|
if a.IsGround() {
|
|
return a
|
|
}
|
|
cpy := *a
|
|
cpy.Value, _ = v.Map(func(k, v *ast.Term) (*ast.Term, *ast.Term, error) {
|
|
return u.plugNamespaced(k, caller), u.plugNamespaced(v, caller), nil
|
|
})
|
|
return &cpy
|
|
case ast.Set:
|
|
if a.IsGround() {
|
|
return a
|
|
}
|
|
cpy := *a
|
|
cpy.Value, _ = v.Map(func(x *ast.Term) (*ast.Term, error) {
|
|
return u.plugNamespaced(x, caller), nil
|
|
})
|
|
return &cpy
|
|
case ast.Ref:
|
|
cpy := *a
|
|
ref := make(ast.Ref, len(v))
|
|
for i := range ref {
|
|
ref[i] = u.plugNamespaced(v[i], caller)
|
|
}
|
|
cpy.Value = ref
|
|
return &cpy
|
|
}
|
|
return a
|
|
}
|
|
|
|
func (u *bindings) bind(a *ast.Term, b *ast.Term, other *bindings, und *undo) {
|
|
u.values.Put(a, value{
|
|
u: other,
|
|
v: b,
|
|
})
|
|
und.k = a
|
|
und.u = u
|
|
}
|
|
|
|
func (u *bindings) apply(a *ast.Term) (*ast.Term, *bindings) {
|
|
// Early exit for non-var terms. Only vars are bound in the binding list,
|
|
// so the lookup below will always fail for non-var terms. In some cases,
|
|
// the lookup may be expensive as it has to hash the term (which for large
|
|
// inputs can be costly).
|
|
_, ok := a.Value.(ast.Var)
|
|
if !ok {
|
|
return a, u
|
|
}
|
|
val, ok := u.get(a)
|
|
if !ok {
|
|
return a, u
|
|
}
|
|
return val.u.apply(val.v)
|
|
}
|
|
|
|
func (u *bindings) delete(v *ast.Term) {
|
|
u.values.Delete(v)
|
|
}
|
|
|
|
func (u *bindings) get(v *ast.Term) (value, bool) {
|
|
if u == nil {
|
|
return value{}, false
|
|
}
|
|
return u.values.Get(v)
|
|
}
|
|
|
|
func (u *bindings) String() string {
|
|
if u == nil {
|
|
return "()"
|
|
}
|
|
var buf []string
|
|
u.values.Iter(func(a *ast.Term, b value) bool {
|
|
buf = append(buf, fmt.Sprintf("%v: %v", a, b))
|
|
return false
|
|
})
|
|
return fmt.Sprintf("({%v}, %v)", strings.Join(buf, ", "), u.id)
|
|
}
|
|
|
|
func (u *bindings) namespaceVar(v *ast.Term, caller *bindings) *ast.Term {
|
|
name, ok := v.Value.(ast.Var)
|
|
if !ok {
|
|
panic("illegal value")
|
|
}
|
|
if caller != nil && caller != u {
|
|
// Root documents (i.e., data, input) should never be namespaced because they
|
|
// are globally unique.
|
|
if !ast.RootDocumentNames.Contains(v) {
|
|
return ast.VarTerm(string(name) + strconv.FormatUint(u.id, 10))
|
|
}
|
|
}
|
|
return v
|
|
}
|
|
|
|
type value struct {
|
|
u *bindings
|
|
v *ast.Term
|
|
}
|
|
|
|
func (v value) String() string {
|
|
return fmt.Sprintf("(%v, %d)", v.v, v.u.id)
|
|
}
|
|
|
|
func (v value) equal(other *value) bool {
|
|
if v.u == other.u {
|
|
return v.v.Equal(other.v)
|
|
}
|
|
return false
|
|
}
|
|
|
|
type namespacingVisitor struct {
|
|
b *bindings
|
|
caller *bindings
|
|
}
|
|
|
|
func (vis namespacingVisitor) Visit(x any) bool {
|
|
switch x := x.(type) {
|
|
case *ast.ArrayComprehension:
|
|
x.Term = vis.namespaceTerm(x.Term)
|
|
vis := ast.NewGenericVisitor(vis.Visit)
|
|
for _, expr := range x.Body {
|
|
vis.Walk(expr)
|
|
}
|
|
return true
|
|
case *ast.SetComprehension:
|
|
x.Term = vis.namespaceTerm(x.Term)
|
|
vis := ast.NewGenericVisitor(vis.Visit)
|
|
for _, expr := range x.Body {
|
|
vis.Walk(expr)
|
|
}
|
|
return true
|
|
case *ast.ObjectComprehension:
|
|
x.Key = vis.namespaceTerm(x.Key)
|
|
x.Value = vis.namespaceTerm(x.Value)
|
|
vis := ast.NewGenericVisitor(vis.Visit)
|
|
for _, expr := range x.Body {
|
|
vis.Walk(expr)
|
|
}
|
|
return true
|
|
case *ast.Expr:
|
|
switch terms := x.Terms.(type) {
|
|
case []*ast.Term:
|
|
for i := 1; i < len(terms); i++ {
|
|
terms[i] = vis.namespaceTerm(terms[i])
|
|
}
|
|
case *ast.Term:
|
|
x.Terms = vis.namespaceTerm(terms)
|
|
}
|
|
for _, w := range x.With {
|
|
w.Target = vis.namespaceTerm(w.Target)
|
|
w.Value = vis.namespaceTerm(w.Value)
|
|
}
|
|
}
|
|
return false
|
|
}
|
|
|
|
func (vis namespacingVisitor) namespaceTerm(a *ast.Term) *ast.Term {
|
|
switch v := a.Value.(type) {
|
|
case ast.Var:
|
|
return vis.b.namespaceVar(a, vis.caller)
|
|
case *ast.Array:
|
|
if a.IsGround() {
|
|
return a
|
|
}
|
|
cpy := *a
|
|
arr := make([]*ast.Term, v.Len())
|
|
for i := range arr {
|
|
arr[i] = vis.namespaceTerm(v.Elem(i))
|
|
}
|
|
cpy.Value = ast.NewArray(arr...)
|
|
return &cpy
|
|
case ast.Object:
|
|
if a.IsGround() {
|
|
return a
|
|
}
|
|
cpy := *a
|
|
cpy.Value, _ = v.Map(func(k, v *ast.Term) (*ast.Term, *ast.Term, error) {
|
|
return vis.namespaceTerm(k), vis.namespaceTerm(v), nil
|
|
})
|
|
return &cpy
|
|
case ast.Set:
|
|
if a.IsGround() {
|
|
return a
|
|
}
|
|
cpy := *a
|
|
cpy.Value, _ = v.Map(func(x *ast.Term) (*ast.Term, error) {
|
|
return vis.namespaceTerm(x), nil
|
|
})
|
|
return &cpy
|
|
case ast.Ref:
|
|
cpy := *a
|
|
ref := make(ast.Ref, len(v))
|
|
for i := range ref {
|
|
ref[i] = vis.namespaceTerm(v[i])
|
|
}
|
|
cpy.Value = ref
|
|
return &cpy
|
|
}
|
|
return a
|
|
}
|
|
|
|
const maxLinearScan = 16
|
|
|
|
// bindingsArrayHashMap uses a dynamically growing slice with linear scan instead
|
|
// of a hash map for smaller # of entries. Hash maps start to
|
|
// show off their performance advantage only after 16 keys.
|
|
//
|
|
// Memory optimization: The slice grows incrementally (2 -> 4 -> 8 -> 16) to avoid
|
|
// wasting memory when only a few bindings are used. This is critical for scenarios
|
|
// like comprehensions and functions with few arguments that are called thousands of times.
|
|
type bindingsArrayHashmap struct {
|
|
n int // Entries in the slice.
|
|
a []bindingArrayKeyValue
|
|
m map[ast.Var]bindingArrayKeyValue
|
|
}
|
|
|
|
type bindingArrayKeyValue struct {
|
|
key *ast.Term
|
|
value value
|
|
}
|
|
|
|
func newBindingsArrayHashmap() bindingsArrayHashmap {
|
|
return bindingsArrayHashmap{}
|
|
}
|
|
|
|
// newBindingsArrayHashmapWithSize creates a bindingsArrayHashmap pre-sized for the expected number of entries.
|
|
// This optimization reduces memory waste when the binding count is known in advance.
|
|
//
|
|
// Size selection strategy:
|
|
// - sizeHint == 0: lazy allocation (no pre-allocation)
|
|
// - sizeHint <= maxLinearScan: pre-allocate slice with exact capacity to avoid reallocation
|
|
// - sizeHint > maxLinearScan: pre-allocate map with exact capacity
|
|
//
|
|
// Memory impact example:
|
|
// - Without hint: dynamic growth 0 -> 2 -> 4 -> 8 -> 16 (saves memory for small counts)
|
|
// - With hint=2: pre-allocates slice with capacity 2 (exact fit, no waste)
|
|
// - With hint=20: pre-allocates map with capacity 20 (saves array allocation + reallocation)
|
|
func newBindingsArrayHashmapWithSize(sizeHint int) bindingsArrayHashmap {
|
|
if sizeHint <= 0 {
|
|
// For unknown sizes, use default lazy allocation with dynamic growth.
|
|
return bindingsArrayHashmap{}
|
|
}
|
|
|
|
if sizeHint <= maxLinearScan {
|
|
// For small known sizes, pre-allocate slice with exact capacity to avoid growth overhead.
|
|
return bindingsArrayHashmap{
|
|
a: make([]bindingArrayKeyValue, 0, sizeHint),
|
|
}
|
|
}
|
|
|
|
// For larger sizes, pre-allocate map to avoid array allocation + transition cost.
|
|
return bindingsArrayHashmap{
|
|
m: make(map[ast.Var]bindingArrayKeyValue, sizeHint),
|
|
}
|
|
}
|
|
|
|
func (b *bindingsArrayHashmap) Put(key *ast.Term, value value) {
|
|
if b.m == nil {
|
|
// Check if key already exists and update value
|
|
if i := b.find(key); i >= 0 {
|
|
b.a[i].value = value
|
|
return
|
|
}
|
|
|
|
// Still room in slice mode (< maxLinearScan)
|
|
if b.n < maxLinearScan {
|
|
// Grow slice if needed using exponential growth strategy
|
|
if b.n == cap(b.a) {
|
|
newCap := cap(b.a) * 2
|
|
if newCap == 0 {
|
|
newCap = 2 // Start with 2 elements
|
|
}
|
|
if newCap > maxLinearScan {
|
|
newCap = maxLinearScan
|
|
}
|
|
newA := make([]bindingArrayKeyValue, b.n, newCap)
|
|
copy(newA, b.a)
|
|
b.a = newA
|
|
}
|
|
b.a = append(b.a, bindingArrayKeyValue{key, value})
|
|
b.n++
|
|
return
|
|
}
|
|
|
|
// Slice is full (reached maxLinearScan), transition to map mode.
|
|
b.m = make(map[ast.Var]bindingArrayKeyValue, maxLinearScan+1)
|
|
for _, kv := range b.a {
|
|
b.m[kv.key.Value.(ast.Var)] = bindingArrayKeyValue{kv.key, kv.value}
|
|
}
|
|
b.m[key.Value.(ast.Var)] = bindingArrayKeyValue{key, value}
|
|
|
|
// Clear slice to allow GC
|
|
b.a = nil
|
|
b.n = 0
|
|
return
|
|
}
|
|
|
|
b.m[key.Value.(ast.Var)] = bindingArrayKeyValue{key, value}
|
|
}
|
|
|
|
func (b *bindingsArrayHashmap) Get(key *ast.Term) (value, bool) {
|
|
if b.m == nil {
|
|
if i := b.find(key); i >= 0 {
|
|
return b.a[i].value, true
|
|
}
|
|
|
|
return value{}, false
|
|
}
|
|
|
|
v, ok := b.m[key.Value.(ast.Var)]
|
|
if ok {
|
|
return v.value, true
|
|
}
|
|
|
|
return value{}, false
|
|
}
|
|
|
|
func (b *bindingsArrayHashmap) Delete(key *ast.Term) {
|
|
if b.m == nil {
|
|
if i := b.find(key); i >= 0 {
|
|
n := b.n - 1
|
|
if i < n {
|
|
b.a[i] = b.a[n]
|
|
}
|
|
// Shrink slice to reflect deletion
|
|
b.a = b.a[:n]
|
|
b.n = n
|
|
}
|
|
return
|
|
}
|
|
|
|
delete(b.m, key.Value.(ast.Var))
|
|
}
|
|
|
|
func (b *bindingsArrayHashmap) Iter(f func(k *ast.Term, v value) bool) {
|
|
if b.m == nil {
|
|
if b.a != nil {
|
|
for i := range b.n {
|
|
if f(b.a[i].key, b.a[i].value) {
|
|
return
|
|
}
|
|
}
|
|
}
|
|
return
|
|
}
|
|
|
|
for _, v := range b.m {
|
|
if f(v.key, v.value) {
|
|
return
|
|
}
|
|
}
|
|
}
|
|
|
|
func (b *bindingsArrayHashmap) find(key *ast.Term) int {
|
|
if b.a == nil || b.n == 0 {
|
|
return -1
|
|
}
|
|
v := key.Value.(ast.Var)
|
|
for i := range b.n {
|
|
if b.a[i].key.Value.(ast.Var) == v {
|
|
return i
|
|
}
|
|
}
|
|
|
|
return -1
|
|
}
|