Files
mgmt/lang/funcs/simple/simple.go
James Shubin 37bb67dffd lang: Improve graph shape with speculative execution
Most of the time, we don't need to have a dynamic call sub graph, since
the actual function call could be represented statically as it
originally was before lambda functions were implemented. Simplifying the
graph shape has important performance benefits in terms of both keep the
graph smaller (memory, etc) and in avoiding the need to run transactions
at runtime (speed) to reshape the graph.

Co-authored-by: Samuel Gélineau <gelisam@gmail.com>
2025-04-27 22:14:51 -04:00

306 lines
11 KiB
Go

// Mgmt
// Copyright (C) James Shubin and the project contributors
// Written by James Shubin <james@shubin.ca> and the project contributors
//
// This program is free software: you can redistribute it and/or modify
// it under the terms of the GNU General Public License as published by
// the Free Software Foundation, either version 3 of the License, or
// (at your option) any later version.
//
// This program is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
// GNU General Public License for more details.
//
// You should have received a copy of the GNU General Public License
// along with this program. If not, see <https://www.gnu.org/licenses/>.
//
// Additional permission under GNU GPL version 3 section 7
//
// If you modify this program, or any covered work, by linking or combining it
// with embedded mcl code and modules (and that the embedded mcl code and
// modules which link with this program, contain a copy of their source code in
// the authoritative form) containing parts covered by the terms of any other
// license, the licensors of this program grant you additional permission to
// convey the resulting work. Furthermore, the licensors of this program grant
// the original author, James Shubin, additional permission to update this
// additional permission if he deems it necessary to achieve the goals of this
// additional permission.
package simple
import (
"context"
"fmt"
"reflect"
"strings"
docsUtil "github.com/purpleidea/mgmt/docs/util"
"github.com/purpleidea/mgmt/lang/funcs"
"github.com/purpleidea/mgmt/lang/funcs/wrapped"
"github.com/purpleidea/mgmt/lang/interfaces"
"github.com/purpleidea/mgmt/lang/types"
unificationUtil "github.com/purpleidea/mgmt/lang/unification/util"
"github.com/purpleidea/mgmt/util/errwrap"
)
const (
// DirectInterface specifies whether we should use the direct function
// API or not. If we don't use it, then these simple functions are
// wrapped with the struct below.
DirectInterface = false // XXX: fix any bugs and set to true!
)
// RegisteredFuncs maps a function name to the corresponding function scaffold.
var RegisteredFuncs = make(map[string]*Scaffold) // must initialize
// Info holds some information about this function.
type Info struct {
Pure bool // is the function pure? (can it be memoized?)
Memo bool // should the function be memoized? (false if too much output)
Fast bool // is the function slow? (avoid speculative execution)
Spec bool // can we speculatively execute it? (true for most)
}
// Scaffold holds the necessary data to build a (possibly polymorphic) function
// with this API.
type Scaffold struct {
// I is some general info about the function.
I *Info
// T is the type of the function. It can include unification variables.
// At a minimum, this must be a `func(?1) ?2` as a naked `?1` is not
// allowed. (TODO: Because of ArgGen.)
T *types.Type
// C is a check function to run after type unification. It will get
// passed the solved type of this function. It should error if this is
// not an acceptable option. This function can be omitted.
C func(typ *types.Type) error
// F is the implementation of the function. The input type can be
// determined by inspecting the values. Of note, this API does not tell
// the implementation what the correct return type should be. If it
// can't be determined from the input types, then a different function
// API needs to be used. XXX: Should we extend this here?
F interfaces.FuncSig
// D is the documentation handle for this function. We look on that
// struct or function for the doc string instead of the F field if this
// is specified. (This is used for facts.)
D interface{}
}
// Register registers a simple, static, pure, polymorphic function. It is easier
// to use than the raw function API. It allows you to build and check a function
// based on a type signature that contains unification variables. You may only
// specify a single type signature with the API, so some complex patterns are
// not possible with this API. Implementing a function like `printf` would not
// be possible. Implementing a function which counts the number of elements in a
// list would be.
func Register(name string, scaffold *Scaffold) {
if _, exists := RegisteredFuncs[name]; exists {
panic(fmt.Sprintf("a simple func named %s is already registered", name))
}
if scaffold == nil {
panic("no scaffold specified for simple func")
}
if scaffold.T == nil {
panic("no type specified for simple func")
}
if scaffold.T.Kind != types.KindFunc {
panic("type must be a func")
}
if scaffold.T.HasVariant() {
panic("func contains a variant type signature")
}
// It's okay if scaffold.C is nil.
if scaffold.F == nil {
panic("no implementation specified for simple func")
}
RegisteredFuncs[name] = scaffold // store a copy for ourselves
// TODO: Do we need to special case either of these?
//if strings.HasPrefix(name, "embedded/") {}
//if strings.HasPrefix(name, "golang/") {}
var f interface{} = scaffold.F
if scaffold.D != nil { // override the doc lookup location if specified
f = scaffold.D
}
metadata, err := funcs.GetFunctionMetadata(f)
if err != nil {
panic(fmt.Sprintf("could not locate function filename for %s", name))
}
funcInfo := &wrapped.Info{}
if scaffold.I != nil {
funcInfo = &wrapped.Info{
Pure: scaffold.I.Pure,
Memo: scaffold.I.Memo,
Fast: scaffold.I.Fast,
Spec: scaffold.I.Spec,
}
}
// register a copy in the main function database
funcs.Register(name, func() interfaces.Func {
return &Func{
Metadata: metadata,
WrappedFunc: &wrapped.Func{
Name: name,
FuncInfo: funcInfo,
// NOTE: It might be more correct to Copy here,
// but we do the copy inside of ExprFunc.Copy()
// instead, so that the same type can be unified
// in more than one way. Doing it here wouldn't
// be harmful, but it's an extra copy we don't
// need to do AFAICT.
Type: scaffold.T, // .Copy(),
},
Check: scaffold.C,
Func: scaffold.F,
}
})
}
// ModuleRegister is exactly like Register, except that it registers within a
// named module. This is a helper function.
func ModuleRegister(module, name string, scaffold *Scaffold) {
Register(module+funcs.ModuleSep+name, scaffold)
}
// WrappedFunc is a type alias so that we can embed `wrapped.Func` inside our
// struct, since the Func name collides with our Func field name.
type WrappedFunc = wrapped.Func
var _ interfaces.BuildableFunc = &Func{} // ensure it meets this expectation
// Func is a scaffolding function struct which fulfills the boiler-plate for the
// function API, but that can run a very simple, static, pure, polymorphic
// function.
type Func struct {
*docsUtil.Metadata
*WrappedFunc // *wrapped.Func as a type alias to pull in the base impl.
// Check is a check function to run after type unification. It will get
// passed the solved type of this function. It should error if this is
// not an acceptable option. This function can be omitted.
Check func(typ *types.Type) error
// Func is the implementation of the function. The input type can be
// determined by inspecting the values. Of note, this API does not tell
// the implementation what the correct return type should be. If it
// can't be determined from the input types, then a different function
// API needs to be used. XXX: Should we extend this here?
Func interfaces.FuncSig
}
// Build is run to turn the maybe polymorphic, undetermined function, into the
// specific statically typed version. It is usually run after unification
// completes, and must be run before Info() and any of the other Func interface
// methods are used. For this function API, it just runs the Check function to
// make sure that the type found during unification is one of the valid ones.
func (obj *Func) Build(typ *types.Type) (*types.Type, error) {
// typ is the KindFunc signature we're trying to build...
if obj.Check != nil {
if err := obj.Check(typ); err != nil {
return nil, errwrap.Wrapf(err, "can't build %s with %s", obj.Name, typ)
}
}
fn := &types.FuncValue{
T: typ,
V: obj.Func, // implementation
}
obj.Fn = fn
return obj.Fn.T, nil
}
// TypeMatch accepts a list of possible type signatures that we want to check
// against after type unification. This helper function returns a function which
// is suitable for use in the scaffold check function field.
func TypeMatch(typeList []string) func(*types.Type) error {
return func(typ *types.Type) error {
for _, s := range typeList {
t := types.NewType(s)
if t == nil {
// TODO: should we panic?
continue // skip
}
if unificationUtil.UnifyCmp(typ, t) == nil {
return nil
}
}
return fmt.Errorf("did not match")
}
}
// StructRegister takes an CLI args struct with optional struct tags, and
// generates simple functions from the contained fields in the specified
// namespace. If no struct field named `func` is included, then a default
// function name which is the lower case representation of the field name will
// be used, otherwise the struct tag contents are used. If the struct tag
// contains the `-` character, then the field will be skipped.
// TODO: An alternative version of this might choose to return all of the values
// as a single giant struct.
func StructRegister(moduleName string, args interface{}) error {
if args == nil {
// programming error
return fmt.Errorf("could not convert/access our struct")
}
//fmt.Printf("A: %+v\n", args)
val := reflect.ValueOf(args)
if val.Kind() == reflect.Ptr { // max one de-referencing
val = val.Elem()
}
typ := val.Type()
for i := 0; i < typ.NumField(); i++ {
v := val.Field(i) // value of the field
t := typ.Field(i) // struct type, get real type with .Type
name := strings.ToLower(t.Name) // default
if alias, ok := t.Tag.Lookup("func"); ok {
if alias == "-" { // skip
continue
}
name = alias
}
//fmt.Printf("N: %+v\n", name) // debug
if len(strings.Trim(name, "abcdefghijklmnopqrstuvwxyz_")) > 0 {
return fmt.Errorf("struct field index(%d) has invalid char(s) in function name", i)
}
typed, err := types.TypeOf(t.Type) // reflect.Type -> (*types.Type, error)
if err != nil {
return err
}
//fmt.Printf("T: %+v\n", typed.String()) // debug
ModuleRegister(moduleName, name, &Scaffold{
T: types.NewType(fmt.Sprintf("func() %s", typed.String())),
F: func(ctx context.Context, input []types.Value) (types.Value, error) {
//if args == nil {
// // programming error
// return nil, fmt.Errorf("could not convert/access our struct")
//}
value, err := types.ValueOf(v) // reflect.Value -> (types.Value, error)
if err != nil {
return nil, errwrap.Wrapf(err, "func `%s.%s()` has nil value", moduleName, name)
}
//fmt.Printf("V: %+v\n", value) // debug
return value, nil
},
})
}
return nil
}