mirror of https://github.com/pulumi/pulumi.git
304 lines
9.8 KiB
Go
304 lines
9.8 KiB
Go
// Copyright 2016-2020, Pulumi Corporation.
|
|
//
|
|
// 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 pcl
|
|
|
|
import (
|
|
"io"
|
|
"sort"
|
|
"strings"
|
|
"unicode"
|
|
"unicode/utf8"
|
|
|
|
"github.com/pulumi/pulumi/sdk/v3/go/common/util/contract"
|
|
|
|
"github.com/hashicorp/hcl/v2"
|
|
"github.com/pulumi/pulumi/pkg/v3/codegen"
|
|
"github.com/pulumi/pulumi/pkg/v3/codegen/hcl2/model"
|
|
"github.com/pulumi/pulumi/pkg/v3/codegen/hcl2/model/format"
|
|
"github.com/pulumi/pulumi/pkg/v3/codegen/schema"
|
|
"github.com/pulumi/pulumi/sdk/v3/go/common/slice"
|
|
)
|
|
|
|
// titleCase replaces the first character in the given string with its upper-case equivalent.
|
|
func titleCase(s string) string {
|
|
c, sz := utf8.DecodeRuneInString(s)
|
|
if sz == 0 || unicode.IsUpper(c) {
|
|
return s
|
|
}
|
|
return string([]rune{unicode.ToUpper(c)}) + s[sz:]
|
|
}
|
|
|
|
func SourceOrderNodes(nodes []Node) []Node {
|
|
sort.Slice(nodes, func(i, j int) bool {
|
|
return model.SourceOrderLess(nodes[i].SyntaxNode().Range(), nodes[j].SyntaxNode().Range())
|
|
})
|
|
return nodes
|
|
}
|
|
|
|
func DecomposeToken(tok string, sourceRange hcl.Range) (string, string, string, hcl.Diagnostics) {
|
|
components := strings.Split(tok, ":")
|
|
if len(components) != 3 {
|
|
// If we don't have a valid type token, return the invalid token as the type name.
|
|
return "", "", tok, hcl.Diagnostics{malformedToken(tok, sourceRange)}
|
|
}
|
|
return components[0], components[1], components[2], nil
|
|
}
|
|
|
|
func linearizeNode(n Node, done codegen.Set, list *[]Node) {
|
|
if !done.Has(n) {
|
|
for _, d := range n.getDependencies() {
|
|
linearizeNode(d, done, list)
|
|
}
|
|
|
|
*list = append(*list, n)
|
|
done.Add(n)
|
|
}
|
|
}
|
|
|
|
// Linearize performs a topological sort of the nodes in the program so that they can be processed by tools that need
|
|
// to see all of a node's dependencies before the node itself (e.g. a code generator for a programming language that
|
|
// requires variables to be defined before they can be referenced). The sort is stable, and nodes are kept in source
|
|
// order as much as possible.
|
|
func Linearize(p *Program) []Node {
|
|
type file struct {
|
|
name string // The name of the HCL source file.
|
|
nodes []Node // The list of nodes defined by the source file.
|
|
}
|
|
|
|
// First, collect nodes into files. Ignore config and outputs, as these are sources and sinks, respectively.
|
|
files := map[string]*file{}
|
|
for _, n := range p.Nodes {
|
|
filename := n.SyntaxNode().Range().Filename
|
|
f, ok := files[filename]
|
|
if !ok {
|
|
f = &file{name: filename}
|
|
files[filename] = f
|
|
}
|
|
f.nodes = append(f.nodes, n)
|
|
}
|
|
|
|
// Now build a worklist out of the set of files, sorting the nodes in each file in source order as we go.
|
|
worklist := slice.Prealloc[*file](len(files))
|
|
for _, f := range files {
|
|
SourceOrderNodes(f.nodes)
|
|
worklist = append(worklist, f)
|
|
}
|
|
|
|
// While the worklist is not empty, add the nodes in the file with the fewest unsatisfied dependencies on nodes in
|
|
// other files.
|
|
doneNodes, nodes := codegen.Set{}, slice.Prealloc[Node](len(p.Nodes))
|
|
for len(worklist) > 0 {
|
|
// Recalculate file weights and find the file with the lowest weight.
|
|
var next *file
|
|
var nextIndex, nextWeight int
|
|
for i, f := range worklist {
|
|
weight, processed := 0, codegen.Set{}
|
|
for _, n := range f.nodes {
|
|
for _, d := range n.getDependencies() {
|
|
// We don't count nodes that we've already counted or nodes that have already been ordered.
|
|
if processed.Has(d) || doneNodes.Has(d) {
|
|
continue
|
|
}
|
|
|
|
// If this dependency resides in a different file, increment the current file's weight and mark the
|
|
// depdendency as processed.
|
|
depFilename := d.SyntaxNode().Range().Filename
|
|
if depFilename != f.name {
|
|
weight++
|
|
}
|
|
processed.Add(d)
|
|
}
|
|
}
|
|
|
|
// If we haven't yet chosen a file to generate or if this file has fewer unsatisfied dependencies than the
|
|
// current choice, choose this file. Ties are broken by the lexical order of the filenames.
|
|
if next == nil || weight < nextWeight || weight == nextWeight && f.name < next.name {
|
|
next, nextIndex, nextWeight = f, i, weight
|
|
}
|
|
}
|
|
|
|
// Swap the chosen file with the tail of the list, then trim the worklist by one.
|
|
worklist[len(worklist)-1], worklist[nextIndex] = worklist[nextIndex], worklist[len(worklist)-1]
|
|
worklist = worklist[:len(worklist)-1]
|
|
|
|
// Now generate the nodes in the chosen file and mark the file as done.
|
|
for _, n := range next.nodes {
|
|
linearizeNode(n, doneNodes, &nodes)
|
|
}
|
|
}
|
|
|
|
return nodes
|
|
}
|
|
|
|
// Remaps the "pulumi:providers:$Package" token style to "$Package:index:Provider", consistent with code generation.
|
|
// This mapping is consistent with how provider resources are projected into the schema and removes special casing logic
|
|
// to generate registering explicit providers.
|
|
//
|
|
// The resultant program should be a shallow copy of the source with only the modified resource nodes copied.
|
|
func MapProvidersAsResources(p *Program) {
|
|
for _, n := range p.Nodes {
|
|
if r, ok := n.(*Resource); ok && r.Schema != nil {
|
|
pkg, mod, name, _ := r.DecomposeToken()
|
|
if r.Schema.IsProvider && pkg == "pulumi" && mod == "providers" {
|
|
// the binder emits tokens like this when the module is "index"
|
|
r.Token = name + "::Provider"
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
func FixupPulumiPackageTokens(r *Resource) {
|
|
pkg, mod, name, _ := r.DecomposeToken()
|
|
if pkg == "pulumi" && mod == "pulumi" {
|
|
r.Token = "pulumi::" + name
|
|
}
|
|
}
|
|
|
|
// SortedFunctionParameters returns a list of properties of the input type from the schema
|
|
// for an invoke function call which has multi argument inputs. We assume here
|
|
// that the expression is an invoke which has it's args (2nd parameter) annotated
|
|
// with the original schema type. The original schema type has properties sorted.
|
|
// This is important because model.ObjectType has no guarantee of property order.
|
|
func SortedFunctionParameters(expr *model.FunctionCallExpression) []*schema.Property {
|
|
if !expr.Signature.MultiArgumentInputs {
|
|
return []*schema.Property{}
|
|
}
|
|
|
|
switch args := expr.Signature.Parameters[1].Type.(type) {
|
|
case *model.ObjectType:
|
|
if len(args.Annotations) == 0 {
|
|
return []*schema.Property{}
|
|
}
|
|
|
|
originalSchemaType, ok := args.Annotations[0].(*schema.ObjectType)
|
|
if !ok {
|
|
return []*schema.Property{}
|
|
}
|
|
|
|
return originalSchemaType.Properties
|
|
default:
|
|
return []*schema.Property{}
|
|
}
|
|
}
|
|
|
|
// GenerateMultiArguments takes the input bag (object) of a function invoke and spreads the values of that object
|
|
// into multi-argument function call.
|
|
// For example, { a: 1, b: 2 } with multiInputArguments: ["a", "b"] would become: 1, 2
|
|
//
|
|
// However, when optional parameters are omitted, then <undefinedLiteral> is used where they should be.
|
|
// Take for example { a: 1, c: 3 } with multiInputArguments: ["a", "b", "c"], it becomes 1, <undefinedLiteral>, 3
|
|
// because b was omitted and c was provided so b had to be the provided <undefinedLiteral>
|
|
func GenerateMultiArguments(
|
|
f *format.Formatter,
|
|
w io.Writer,
|
|
undefinedLiteral string,
|
|
expr *model.ObjectConsExpression,
|
|
multiArguments []*schema.Property,
|
|
) {
|
|
items := make(map[string]model.Expression)
|
|
for _, item := range expr.Items {
|
|
lit := item.Key.(*model.LiteralValueExpression)
|
|
propertyKey := lit.Value.AsString()
|
|
items[propertyKey] = item.Value
|
|
}
|
|
|
|
hasMoreArgs := func(index int) bool {
|
|
for _, arg := range multiArguments[index:] {
|
|
if _, ok := items[arg.Name]; ok {
|
|
return true
|
|
}
|
|
}
|
|
|
|
return false
|
|
}
|
|
|
|
for index, arg := range multiArguments {
|
|
value, ok := items[arg.Name]
|
|
if ok {
|
|
f.Fgenf(w, "%.v", value)
|
|
} else if hasMoreArgs(index) {
|
|
// a positional argument was not provided in the input bag
|
|
// assume it is optional
|
|
f.Fgen(w, undefinedLiteral)
|
|
}
|
|
|
|
if hasMoreArgs(index + 1) {
|
|
f.Fgen(w, ", ")
|
|
}
|
|
}
|
|
}
|
|
|
|
func SortedStringKeys[V any](m map[string]V) []string {
|
|
keys := make([]string, 0)
|
|
for propertyName := range m {
|
|
keys = append(keys, propertyName)
|
|
}
|
|
|
|
sort.Strings(keys)
|
|
return keys
|
|
}
|
|
|
|
// UnwrapOption returns type T if the input is an Option(T)
|
|
func UnwrapOption(exprType model.Type) model.Type {
|
|
switch exprType := exprType.(type) {
|
|
case *model.UnionType:
|
|
if len(exprType.ElementTypes) == 2 && exprType.ElementTypes[0] == model.NoneType {
|
|
return exprType.ElementTypes[1]
|
|
} else if len(exprType.ElementTypes) == 2 && exprType.ElementTypes[1] == model.NoneType {
|
|
return exprType.ElementTypes[0]
|
|
} else {
|
|
return exprType
|
|
}
|
|
default:
|
|
return exprType
|
|
}
|
|
}
|
|
|
|
// VariableAccessed returns whether the given variable name is accessed in the given expression.
|
|
func VariableAccessed(variableName string, expr model.Expression) bool {
|
|
accessed := false
|
|
visitor := func(subExpr model.Expression) (model.Expression, hcl.Diagnostics) {
|
|
if traversal, ok := subExpr.(*model.ScopeTraversalExpression); ok {
|
|
if traversal.RootName == variableName {
|
|
accessed = true
|
|
}
|
|
}
|
|
return subExpr, nil
|
|
}
|
|
|
|
_, diags := model.VisitExpression(expr, model.IdentityVisitor, visitor)
|
|
contract.Assertf(len(diags) == 0, "expected no diagnostics from VisitExpression")
|
|
return accessed
|
|
}
|
|
|
|
// LiteralValueString evaluates the given expression and returns the string value if it is a literal value expression
|
|
// otherwise returns an empty string for anything else.
|
|
func LiteralValueString(x model.Expression) string {
|
|
switch x := x.(type) {
|
|
case *model.LiteralValueExpression:
|
|
if model.StringType.AssignableFrom(x.Type()) {
|
|
return x.Value.AsString()
|
|
}
|
|
case *model.TemplateExpression:
|
|
if len(x.Parts) == 1 {
|
|
if lit, ok := x.Parts[0].(*model.LiteralValueExpression); ok && model.StringType.AssignableFrom(lit.Type()) {
|
|
return lit.Value.AsString()
|
|
}
|
|
}
|
|
}
|
|
|
|
return ""
|
|
}
|