1
0
mirror of https://github.com/golang/go synced 2024-11-18 19:24:39 -07:00
go/internal/lsp/source/util.go

650 lines
16 KiB
Go
Raw Normal View History

// Copyright 2019 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
package source
import (
"context"
"encoding/json"
"fmt"
"go/ast"
"go/printer"
"go/token"
"go/types"
"path/filepath"
"regexp"
"sort"
"strings"
"golang.org/x/tools/internal/lsp/protocol"
"golang.org/x/tools/internal/span"
errors "golang.org/x/xerrors"
)
type mappedRange struct {
spanRange span.Range
m *protocol.ColumnMapper
// protocolRange is the result of converting the spanRange using the mapper.
// It is computed on-demand.
protocolRange *protocol.Range
}
func newMappedRange(fset *token.FileSet, m *protocol.ColumnMapper, start, end token.Pos) mappedRange {
return mappedRange{
spanRange: span.Range{
FileSet: fset,
Start: start,
End: end,
Converter: m.Converter,
},
m: m,
}
}
func (s mappedRange) Range() (protocol.Range, error) {
if s.protocolRange == nil {
spn, err := s.spanRange.Span()
if err != nil {
return protocol.Range{}, err
}
prng, err := s.m.Range(spn)
if err != nil {
return protocol.Range{}, err
}
s.protocolRange = &prng
}
return *s.protocolRange, nil
}
func (s mappedRange) Span() (span.Span, error) {
return s.spanRange.Span()
}
func (s mappedRange) URI() span.URI {
return s.m.URI
}
// getParsedFile is a convenience function that extracts the Package and ParsedGoFile for a File in a Snapshot.
// selectPackage is typically Narrowest/WidestPackageHandle below.
internal/lsp: replace ParseGoHandle with concrete data ParseGoHandles serve two purposes: they pin cache entries so that redundant calculations are cached, and they allow users to obtain the actual parsed AST. The former is an implementation detail, and the latter turns out to just be an annoyance. Parsed Go files are obtained from two places. By far the most common is from a type checked package. But a type checked package must by definition have already parsed all the files it contains, so the PGH is already computed and cannot have failed. Type checked packages can simply return the parsed file without requiring a separate Check operation. We do want to pin the cache entries in this case, which I've done by holding on to the PGH in cache.pkg. There are some cases where we directly parse a file, such as for the FoldingRange LSP call, which doesn't need type information. Those parses can actually fail, so we do need an error check. But we don't need the PGH; in all cases we are immediately using and discarding it. So it turns out we don't actually need the PGH type at all, at least not in the public API. Instead, we can pass around a concrete struct that has the various pieces of data directly available. This uncovered a bug in typeCheck: it should fail if it encounters any real errors. Change-Id: I203bf2dd79d5d65c01392d69c2cf4f7744fde7fc Reviewed-on: https://go-review.googlesource.com/c/tools/+/244021 Run-TryBot: Heschi Kreinick <heschi@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Rebecca Stambler <rstambler@golang.org>
2020-07-21 13:15:06 -06:00
func getParsedFile(ctx context.Context, snapshot Snapshot, fh FileHandle, selectPackage PackagePolicy) (Package, *ParsedGoFile, error) {
phs, err := snapshot.PackagesForFile(ctx, fh.URI(), TypecheckWorkspace)
if err != nil {
return nil, nil, err
}
pkg, err := selectPackage(phs)
if err != nil {
return nil, nil, err
}
pgh, err := pkg.File(fh.URI())
return pkg, pgh, err
}
type PackagePolicy func([]Package) (Package, error)
// NarrowestPackage picks the "narrowest" package for a given file.
//
// By "narrowest" package, we mean the package with the fewest number of files
// that includes the given file. This solves the problem of test variants,
// as the test will have more files than the non-test package.
func NarrowestPackage(pkgs []Package) (Package, error) {
if len(pkgs) < 1 {
return nil, errors.Errorf("no packages")
}
result := pkgs[0]
for _, handle := range pkgs[1:] {
if result == nil || len(handle.CompiledGoFiles()) < len(result.CompiledGoFiles()) {
result = handle
}
}
if result == nil {
return nil, errors.Errorf("no packages in input")
}
return result, nil
}
// WidestPackage returns the Package containing the most files.
//
// This is useful for something like diagnostics, where we'd prefer to offer diagnostics
// for as many files as possible.
func WidestPackage(pkgs []Package) (Package, error) {
if len(pkgs) < 1 {
return nil, errors.Errorf("no packages")
}
result := pkgs[0]
for _, handle := range pkgs[1:] {
if result == nil || len(handle.CompiledGoFiles()) > len(result.CompiledGoFiles()) {
result = handle
}
}
if result == nil {
return nil, errors.Errorf("no packages in input")
}
return result, nil
}
func IsGenerated(ctx context.Context, snapshot Snapshot, uri span.URI) bool {
fh, err := snapshot.GetFile(ctx, uri)
if err != nil {
return false
}
internal/lsp: replace ParseGoHandle with concrete data ParseGoHandles serve two purposes: they pin cache entries so that redundant calculations are cached, and they allow users to obtain the actual parsed AST. The former is an implementation detail, and the latter turns out to just be an annoyance. Parsed Go files are obtained from two places. By far the most common is from a type checked package. But a type checked package must by definition have already parsed all the files it contains, so the PGH is already computed and cannot have failed. Type checked packages can simply return the parsed file without requiring a separate Check operation. We do want to pin the cache entries in this case, which I've done by holding on to the PGH in cache.pkg. There are some cases where we directly parse a file, such as for the FoldingRange LSP call, which doesn't need type information. Those parses can actually fail, so we do need an error check. But we don't need the PGH; in all cases we are immediately using and discarding it. So it turns out we don't actually need the PGH type at all, at least not in the public API. Instead, we can pass around a concrete struct that has the various pieces of data directly available. This uncovered a bug in typeCheck: it should fail if it encounters any real errors. Change-Id: I203bf2dd79d5d65c01392d69c2cf4f7744fde7fc Reviewed-on: https://go-review.googlesource.com/c/tools/+/244021 Run-TryBot: Heschi Kreinick <heschi@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Rebecca Stambler <rstambler@golang.org>
2020-07-21 13:15:06 -06:00
pgf, err := snapshot.ParseGo(ctx, fh, ParseHeader)
if err != nil {
return false
}
tok := snapshot.FileSet().File(pgf.File.Pos())
if tok == nil {
return false
}
internal/lsp: replace ParseGoHandle with concrete data ParseGoHandles serve two purposes: they pin cache entries so that redundant calculations are cached, and they allow users to obtain the actual parsed AST. The former is an implementation detail, and the latter turns out to just be an annoyance. Parsed Go files are obtained from two places. By far the most common is from a type checked package. But a type checked package must by definition have already parsed all the files it contains, so the PGH is already computed and cannot have failed. Type checked packages can simply return the parsed file without requiring a separate Check operation. We do want to pin the cache entries in this case, which I've done by holding on to the PGH in cache.pkg. There are some cases where we directly parse a file, such as for the FoldingRange LSP call, which doesn't need type information. Those parses can actually fail, so we do need an error check. But we don't need the PGH; in all cases we are immediately using and discarding it. So it turns out we don't actually need the PGH type at all, at least not in the public API. Instead, we can pass around a concrete struct that has the various pieces of data directly available. This uncovered a bug in typeCheck: it should fail if it encounters any real errors. Change-Id: I203bf2dd79d5d65c01392d69c2cf4f7744fde7fc Reviewed-on: https://go-review.googlesource.com/c/tools/+/244021 Run-TryBot: Heschi Kreinick <heschi@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Rebecca Stambler <rstambler@golang.org>
2020-07-21 13:15:06 -06:00
for _, commentGroup := range pgf.File.Comments {
for _, comment := range commentGroup.List {
if matched := generatedRx.MatchString(comment.Text); matched {
// Check if comment is at the beginning of the line in source.
if pos := tok.Position(comment.Slash); pos.Column == 1 {
return true
}
}
}
}
return false
}
func nodeToProtocolRange(snapshot Snapshot, pkg Package, n ast.Node) (protocol.Range, error) {
mrng, err := posToMappedRange(snapshot, pkg, n.Pos(), n.End())
if err != nil {
return protocol.Range{}, err
}
return mrng.Range()
}
func objToMappedRange(snapshot Snapshot, pkg Package, obj types.Object) (mappedRange, error) {
if pkgName, ok := obj.(*types.PkgName); ok {
// An imported Go package has a package-local, unqualified name.
// When the name matches the imported package name, there is no
// identifier in the import spec with the local package name.
//
// For example:
// import "go/ast" // name "ast" matches package name
// import a "go/ast" // name "a" does not match package name
//
// When the identifier does not appear in the source, have the range
// of the object be the import path, including quotes.
if pkgName.Imported().Name() == pkgName.Name() {
return posToMappedRange(snapshot, pkg, obj.Pos(), obj.Pos()+token.Pos(len(pkgName.Imported().Path())+2))
}
}
return nameToMappedRange(snapshot, pkg, obj.Pos(), obj.Name())
}
func nameToMappedRange(snapshot Snapshot, pkg Package, pos token.Pos, name string) (mappedRange, error) {
return posToMappedRange(snapshot, pkg, pos, pos+token.Pos(len(name)))
}
func posToMappedRange(snapshot Snapshot, pkg Package, pos, end token.Pos) (mappedRange, error) {
logicalFilename := snapshot.FileSet().File(pos).Position(pos).Filename
internal/lsp: replace ParseGoHandle with concrete data ParseGoHandles serve two purposes: they pin cache entries so that redundant calculations are cached, and they allow users to obtain the actual parsed AST. The former is an implementation detail, and the latter turns out to just be an annoyance. Parsed Go files are obtained from two places. By far the most common is from a type checked package. But a type checked package must by definition have already parsed all the files it contains, so the PGH is already computed and cannot have failed. Type checked packages can simply return the parsed file without requiring a separate Check operation. We do want to pin the cache entries in this case, which I've done by holding on to the PGH in cache.pkg. There are some cases where we directly parse a file, such as for the FoldingRange LSP call, which doesn't need type information. Those parses can actually fail, so we do need an error check. But we don't need the PGH; in all cases we are immediately using and discarding it. So it turns out we don't actually need the PGH type at all, at least not in the public API. Instead, we can pass around a concrete struct that has the various pieces of data directly available. This uncovered a bug in typeCheck: it should fail if it encounters any real errors. Change-Id: I203bf2dd79d5d65c01392d69c2cf4f7744fde7fc Reviewed-on: https://go-review.googlesource.com/c/tools/+/244021 Run-TryBot: Heschi Kreinick <heschi@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Rebecca Stambler <rstambler@golang.org>
2020-07-21 13:15:06 -06:00
pgf, _, err := findFileInDeps(pkg, span.URIFromPath(logicalFilename))
if err != nil {
return mappedRange{}, err
}
if !pos.IsValid() {
return mappedRange{}, errors.Errorf("invalid position for %v", pos)
}
if !end.IsValid() {
return mappedRange{}, errors.Errorf("invalid position for %v", end)
}
return newMappedRange(snapshot.FileSet(), pgf.Mapper, pos, end), nil
}
// Matches cgo generated comment as well as the proposed standard:
// https://golang.org/s/generatedcode
var generatedRx = regexp.MustCompile(`// .*DO NOT EDIT\.?`)
func DetectLanguage(langID, filename string) FileKind {
switch langID {
case "go":
return Go
case "go.mod":
return Mod
case "go.sum":
return Sum
}
// Fallback to detecting the language based on the file extension.
switch filepath.Ext(filename) {
case ".mod":
return Mod
case ".sum":
return Sum
default: // fallback to Go
return Go
}
}
func (k FileKind) String() string {
switch k {
case Mod:
return "go.mod"
case Sum:
return "go.sum"
default:
return "go"
}
}
// Returns the index and the node whose position is contained inside the node list.
func nodeAtPos(nodes []ast.Node, pos token.Pos) (ast.Node, int) {
if nodes == nil {
return nil, -1
}
for i, node := range nodes {
if node.Pos() <= pos && pos <= node.End() {
return node, i
}
}
return nil, -1
}
// indexExprAtPos returns the index of the expression containing pos.
func exprAtPos(pos token.Pos, args []ast.Expr) int {
for i, expr := range args {
if expr.Pos() <= pos && pos <= expr.End() {
return i
}
}
return len(args)
}
internal/lsp/source: optimize enumeration of a type's fields When searching for deep completions, we can end up enumerating struct types' fields a lot. Optimize fieldSelections to reduce work: - Wait until we see an embedded field before we create the "seen" map. - Use a callback style to iterate over the struct's fields rather than returning a slice of fields. - Change "seen" checking strategy back to track struct types rather than each individual field. Struct with 5 non-embedded fields: name old time/op new time/op delta Fields-16 293ns ± 1% 20ns ± 2% -93.13% (p=0.008 n=5+5) name old alloc/op new alloc/op delta Fields-16 120B ± 0% 0B -100.00% (p=0.008 n=5+5) name old allocs/op new allocs/op delta Fields-16 4.00 ± 0% 0.00 -100.00% (p=0.008 n=5+5) Same struct but add an embedded struct with 2 fields: name old time/op new time/op delta Fields-16 389ns ± 1% 142ns ± 1% -63.53% (p=0.008 n=5+5) name old alloc/op new alloc/op delta Fields-16 120B ± 0% 144B ± 0% +20.00% (p=0.008 n=5+5) name old allocs/op new allocs/op delta Fields-16 4.00 ± 0% 2.00 ± 0% -50.00% (p=0.008 n=5+5) I think the alloc/op went up because the "seen" map is no longer allocated on the stack. There is more room for more optimization, but it's probably not worth making things more complicated. Change-Id: I6f9f2124334a8594ef9d6f9b5ac4b3a8aead5f49 Reviewed-on: https://go-review.googlesource.com/c/tools/+/223419 Run-TryBot: Muir Manders <muir@mnd.rs> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Rebecca Stambler <rstambler@golang.org>
2020-03-13 13:10:42 -06:00
// eachField invokes fn for each field that can be selected from a
// value of type T.
func eachField(T types.Type, fn func(*types.Var)) {
// TODO(adonovan): this algorithm doesn't exclude ambiguous
// selections that match more than one field/method.
// types.NewSelectionSet should do that for us.
internal/lsp/source: optimize enumeration of a type's fields When searching for deep completions, we can end up enumerating struct types' fields a lot. Optimize fieldSelections to reduce work: - Wait until we see an embedded field before we create the "seen" map. - Use a callback style to iterate over the struct's fields rather than returning a slice of fields. - Change "seen" checking strategy back to track struct types rather than each individual field. Struct with 5 non-embedded fields: name old time/op new time/op delta Fields-16 293ns ± 1% 20ns ± 2% -93.13% (p=0.008 n=5+5) name old alloc/op new alloc/op delta Fields-16 120B ± 0% 0B -100.00% (p=0.008 n=5+5) name old allocs/op new allocs/op delta Fields-16 4.00 ± 0% 0.00 -100.00% (p=0.008 n=5+5) Same struct but add an embedded struct with 2 fields: name old time/op new time/op delta Fields-16 389ns ± 1% 142ns ± 1% -63.53% (p=0.008 n=5+5) name old alloc/op new alloc/op delta Fields-16 120B ± 0% 144B ± 0% +20.00% (p=0.008 n=5+5) name old allocs/op new allocs/op delta Fields-16 4.00 ± 0% 2.00 ± 0% -50.00% (p=0.008 n=5+5) I think the alloc/op went up because the "seen" map is no longer allocated on the stack. There is more room for more optimization, but it's probably not worth making things more complicated. Change-Id: I6f9f2124334a8594ef9d6f9b5ac4b3a8aead5f49 Reviewed-on: https://go-review.googlesource.com/c/tools/+/223419 Run-TryBot: Muir Manders <muir@mnd.rs> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Rebecca Stambler <rstambler@golang.org>
2020-03-13 13:10:42 -06:00
// for termination on recursive types
var seen map[*types.Struct]bool
var visit func(T types.Type)
visit = func(T types.Type) {
if T, ok := deref(T).Underlying().(*types.Struct); ok {
internal/lsp/source: optimize enumeration of a type's fields When searching for deep completions, we can end up enumerating struct types' fields a lot. Optimize fieldSelections to reduce work: - Wait until we see an embedded field before we create the "seen" map. - Use a callback style to iterate over the struct's fields rather than returning a slice of fields. - Change "seen" checking strategy back to track struct types rather than each individual field. Struct with 5 non-embedded fields: name old time/op new time/op delta Fields-16 293ns ± 1% 20ns ± 2% -93.13% (p=0.008 n=5+5) name old alloc/op new alloc/op delta Fields-16 120B ± 0% 0B -100.00% (p=0.008 n=5+5) name old allocs/op new allocs/op delta Fields-16 4.00 ± 0% 0.00 -100.00% (p=0.008 n=5+5) Same struct but add an embedded struct with 2 fields: name old time/op new time/op delta Fields-16 389ns ± 1% 142ns ± 1% -63.53% (p=0.008 n=5+5) name old alloc/op new alloc/op delta Fields-16 120B ± 0% 144B ± 0% +20.00% (p=0.008 n=5+5) name old allocs/op new allocs/op delta Fields-16 4.00 ± 0% 2.00 ± 0% -50.00% (p=0.008 n=5+5) I think the alloc/op went up because the "seen" map is no longer allocated on the stack. There is more room for more optimization, but it's probably not worth making things more complicated. Change-Id: I6f9f2124334a8594ef9d6f9b5ac4b3a8aead5f49 Reviewed-on: https://go-review.googlesource.com/c/tools/+/223419 Run-TryBot: Muir Manders <muir@mnd.rs> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Rebecca Stambler <rstambler@golang.org>
2020-03-13 13:10:42 -06:00
if seen[T] {
return
}
for i := 0; i < T.NumFields(); i++ {
f := T.Field(i)
internal/lsp/source: optimize enumeration of a type's fields When searching for deep completions, we can end up enumerating struct types' fields a lot. Optimize fieldSelections to reduce work: - Wait until we see an embedded field before we create the "seen" map. - Use a callback style to iterate over the struct's fields rather than returning a slice of fields. - Change "seen" checking strategy back to track struct types rather than each individual field. Struct with 5 non-embedded fields: name old time/op new time/op delta Fields-16 293ns ± 1% 20ns ± 2% -93.13% (p=0.008 n=5+5) name old alloc/op new alloc/op delta Fields-16 120B ± 0% 0B -100.00% (p=0.008 n=5+5) name old allocs/op new allocs/op delta Fields-16 4.00 ± 0% 0.00 -100.00% (p=0.008 n=5+5) Same struct but add an embedded struct with 2 fields: name old time/op new time/op delta Fields-16 389ns ± 1% 142ns ± 1% -63.53% (p=0.008 n=5+5) name old alloc/op new alloc/op delta Fields-16 120B ± 0% 144B ± 0% +20.00% (p=0.008 n=5+5) name old allocs/op new allocs/op delta Fields-16 4.00 ± 0% 2.00 ± 0% -50.00% (p=0.008 n=5+5) I think the alloc/op went up because the "seen" map is no longer allocated on the stack. There is more room for more optimization, but it's probably not worth making things more complicated. Change-Id: I6f9f2124334a8594ef9d6f9b5ac4b3a8aead5f49 Reviewed-on: https://go-review.googlesource.com/c/tools/+/223419 Run-TryBot: Muir Manders <muir@mnd.rs> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Rebecca Stambler <rstambler@golang.org>
2020-03-13 13:10:42 -06:00
fn(f)
if f.Anonymous() {
internal/lsp/source: optimize enumeration of a type's fields When searching for deep completions, we can end up enumerating struct types' fields a lot. Optimize fieldSelections to reduce work: - Wait until we see an embedded field before we create the "seen" map. - Use a callback style to iterate over the struct's fields rather than returning a slice of fields. - Change "seen" checking strategy back to track struct types rather than each individual field. Struct with 5 non-embedded fields: name old time/op new time/op delta Fields-16 293ns ± 1% 20ns ± 2% -93.13% (p=0.008 n=5+5) name old alloc/op new alloc/op delta Fields-16 120B ± 0% 0B -100.00% (p=0.008 n=5+5) name old allocs/op new allocs/op delta Fields-16 4.00 ± 0% 0.00 -100.00% (p=0.008 n=5+5) Same struct but add an embedded struct with 2 fields: name old time/op new time/op delta Fields-16 389ns ± 1% 142ns ± 1% -63.53% (p=0.008 n=5+5) name old alloc/op new alloc/op delta Fields-16 120B ± 0% 144B ± 0% +20.00% (p=0.008 n=5+5) name old allocs/op new allocs/op delta Fields-16 4.00 ± 0% 2.00 ± 0% -50.00% (p=0.008 n=5+5) I think the alloc/op went up because the "seen" map is no longer allocated on the stack. There is more room for more optimization, but it's probably not worth making things more complicated. Change-Id: I6f9f2124334a8594ef9d6f9b5ac4b3a8aead5f49 Reviewed-on: https://go-review.googlesource.com/c/tools/+/223419 Run-TryBot: Muir Manders <muir@mnd.rs> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Rebecca Stambler <rstambler@golang.org>
2020-03-13 13:10:42 -06:00
if seen == nil {
// Lazily create "seen" since it is only needed for
// embedded structs.
seen = make(map[*types.Struct]bool)
}
seen[T] = true
visit(f.Type())
}
}
}
}
visit(T)
}
// typeIsValid reports whether typ doesn't contain any Invalid types.
func typeIsValid(typ types.Type) bool {
// Check named types separately, because we don't want
// to call Underlying() on them to avoid problems with recursive types.
if _, ok := typ.(*types.Named); ok {
return true
}
switch typ := typ.Underlying().(type) {
case *types.Basic:
return typ.Kind() != types.Invalid
case *types.Array:
return typeIsValid(typ.Elem())
case *types.Slice:
return typeIsValid(typ.Elem())
case *types.Pointer:
return typeIsValid(typ.Elem())
case *types.Map:
return typeIsValid(typ.Key()) && typeIsValid(typ.Elem())
case *types.Chan:
return typeIsValid(typ.Elem())
case *types.Signature:
return typeIsValid(typ.Params()) && typeIsValid(typ.Results())
case *types.Tuple:
for i := 0; i < typ.Len(); i++ {
if !typeIsValid(typ.At(i).Type()) {
return false
}
}
return true
case *types.Struct, *types.Interface:
// Don't bother checking structs, interfaces for validity.
return true
default:
return false
}
}
// resolveInvalid traverses the node of the AST that defines the scope
// containing the declaration of obj, and attempts to find a user-friendly
// name for its invalid type. The resulting Object and its Type are fake.
func resolveInvalid(fset *token.FileSet, obj types.Object, node ast.Node, info *types.Info) types.Object {
var resultExpr ast.Expr
ast.Inspect(node, func(node ast.Node) bool {
switch n := node.(type) {
case *ast.ValueSpec:
for _, name := range n.Names {
if info.Defs[name] == obj {
resultExpr = n.Type
}
}
return false
case *ast.Field: // This case handles parameters and results of a FuncDecl or FuncLit.
for _, name := range n.Names {
if info.Defs[name] == obj {
resultExpr = n.Type
}
}
return false
default:
return true
}
})
// Construct a fake type for the object and return a fake object with this type.
typename := formatNode(fset, resultExpr)
typ := types.NewNamed(types.NewTypeName(token.NoPos, obj.Pkg(), typename, nil), types.Typ[types.Invalid], nil)
return types.NewVar(obj.Pos(), obj.Pkg(), obj.Name(), typ)
}
func formatNode(fset *token.FileSet, n ast.Node) string {
var buf strings.Builder
if err := printer.Fprint(&buf, fset, n); err != nil {
return ""
}
return buf.String()
}
func isPointer(T types.Type) bool {
_, ok := T.(*types.Pointer)
return ok
}
func isVar(obj types.Object) bool {
_, ok := obj.(*types.Var)
return ok
}
// deref returns a pointer's element type, traversing as many levels as needed.
// Otherwise it returns typ.
func deref(typ types.Type) types.Type {
for {
p, ok := typ.Underlying().(*types.Pointer)
if !ok {
return typ
}
typ = p.Elem()
}
}
func isTypeName(obj types.Object) bool {
_, ok := obj.(*types.TypeName)
return ok
}
func isFunc(obj types.Object) bool {
_, ok := obj.(*types.Func)
return ok
}
func isEmptyInterface(T types.Type) bool {
intf, _ := T.(*types.Interface)
return intf != nil && intf.NumMethods() == 0
}
func isUntyped(T types.Type) bool {
if basic, ok := T.(*types.Basic); ok {
return basic.Info()&types.IsUntyped > 0
}
return false
}
func isPkgName(obj types.Object) bool {
_, ok := obj.(*types.PkgName)
return ok
}
func isASTFile(n ast.Node) bool {
_, ok := n.(*ast.File)
return ok
}
internal/lsp/source: untangle completion type comparison The code to check if a candidate object matches our candidate inference had become complicated, messy, and in some cases incorrect. The main source of the complexity is the "derived" expected and candidate types. When considering a candidate object "foo", we also consider "&foo", "foo()", and "*foo", as appropriate. On the expected side of things, when completing the a variadic function parameter we expect either the variadic slice type and the scalar element type. The code had grown organically to handle the expanding concerns, but that resulted in confused code that didn't handle the interplay between the various facets of candidate inference. For example, we were inappropriately invoking func candidates when completing variadic args: func foo(...func()) func bar() {} foo(bar<>) // oops - expanded to "bar()" and we weren't type matching functions properly as builtin args: func myMap() map[string]int { ... } delete(myM<>) // we weren't preferring (or invoking) "myMap()" We also had methods like "typeMatches" which took both a "candidate" object and a "candType" type, which doesn't make sense because the candidate contains the type already. Now instead we explicitly iterate over all the derived candidate and expected types so they are treated the same. There are still some warts left but I think this is a step in the right direction. Change-Id: If84a84b34a8fb771a32231f7ab64ca192f611b3d Reviewed-on: https://go-review.googlesource.com/c/tools/+/218877 Run-TryBot: Muir Manders <muir@mnd.rs> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Robert Findley <rfindley@google.com>
2020-02-08 20:59:28 -07:00
func deslice(T types.Type) types.Type {
if slice, ok := T.Underlying().(*types.Slice); ok {
return slice.Elem()
}
return nil
}
// isSelector returns the enclosing *ast.SelectorExpr when pos is in the
// selector.
func enclosingSelector(path []ast.Node, pos token.Pos) *ast.SelectorExpr {
if len(path) == 0 {
return nil
}
if sel, ok := path[0].(*ast.SelectorExpr); ok {
return sel
}
if _, ok := path[0].(*ast.Ident); ok && len(path) > 1 {
if sel, ok := path[1].(*ast.SelectorExpr); ok && pos >= sel.Sel.Pos() {
return sel
}
}
return nil
}
func enclosingValueSpec(path []ast.Node) *ast.ValueSpec {
for _, n := range path {
if vs, ok := n.(*ast.ValueSpec); ok {
return vs
}
}
return nil
}
// exprObj returns the types.Object associated with the *ast.Ident or
// *ast.SelectorExpr e.
func exprObj(info *types.Info, e ast.Expr) types.Object {
var ident *ast.Ident
switch expr := e.(type) {
case *ast.Ident:
ident = expr
case *ast.SelectorExpr:
ident = expr.Sel
default:
return nil
}
return info.ObjectOf(ident)
}
// typeConversion returns the type being converted to if call is a type
// conversion expression.
func typeConversion(call *ast.CallExpr, info *types.Info) types.Type {
// Type conversion (e.g. "float64(foo)").
if fun, _ := exprObj(info, call.Fun).(*types.TypeName); fun != nil {
return fun.Type()
}
return nil
}
// fieldsAccessible returns whether s has at least one field accessible by p.
func fieldsAccessible(s *types.Struct, p *types.Package) bool {
for i := 0; i < s.NumFields(); i++ {
f := s.Field(i)
if f.Exported() || f.Pkg() == p {
return true
}
}
return false
}
func SortDiagnostics(d []*Diagnostic) {
sort.Slice(d, func(i int, j int) bool {
return CompareDiagnostic(d[i], d[j]) < 0
})
}
func CompareDiagnostic(a, b *Diagnostic) int {
if r := protocol.CompareRange(a.Range, b.Range); r != 0 {
return r
}
if a.Source < b.Source {
return -1
}
if a.Message < b.Message {
return -1
}
if a.Message == b.Message {
return 0
}
return 1
}
func findPosInPackage(snapshot Snapshot, searchpkg Package, pos token.Pos) (*ParsedGoFile, Package, error) {
tok := snapshot.FileSet().File(pos)
if tok == nil {
return nil, nil, errors.Errorf("no file for pos in package %s", searchpkg.ID())
}
uri := span.URIFromPath(tok.Name())
internal/lsp: replace ParseGoHandle with concrete data ParseGoHandles serve two purposes: they pin cache entries so that redundant calculations are cached, and they allow users to obtain the actual parsed AST. The former is an implementation detail, and the latter turns out to just be an annoyance. Parsed Go files are obtained from two places. By far the most common is from a type checked package. But a type checked package must by definition have already parsed all the files it contains, so the PGH is already computed and cannot have failed. Type checked packages can simply return the parsed file without requiring a separate Check operation. We do want to pin the cache entries in this case, which I've done by holding on to the PGH in cache.pkg. There are some cases where we directly parse a file, such as for the FoldingRange LSP call, which doesn't need type information. Those parses can actually fail, so we do need an error check. But we don't need the PGH; in all cases we are immediately using and discarding it. So it turns out we don't actually need the PGH type at all, at least not in the public API. Instead, we can pass around a concrete struct that has the various pieces of data directly available. This uncovered a bug in typeCheck: it should fail if it encounters any real errors. Change-Id: I203bf2dd79d5d65c01392d69c2cf4f7744fde7fc Reviewed-on: https://go-review.googlesource.com/c/tools/+/244021 Run-TryBot: Heschi Kreinick <heschi@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Rebecca Stambler <rstambler@golang.org>
2020-07-21 13:15:06 -06:00
pgf, pkg, err := findFileInDeps(searchpkg, uri)
if err != nil {
return nil, nil, err
}
internal/lsp: replace ParseGoHandle with concrete data ParseGoHandles serve two purposes: they pin cache entries so that redundant calculations are cached, and they allow users to obtain the actual parsed AST. The former is an implementation detail, and the latter turns out to just be an annoyance. Parsed Go files are obtained from two places. By far the most common is from a type checked package. But a type checked package must by definition have already parsed all the files it contains, so the PGH is already computed and cannot have failed. Type checked packages can simply return the parsed file without requiring a separate Check operation. We do want to pin the cache entries in this case, which I've done by holding on to the PGH in cache.pkg. There are some cases where we directly parse a file, such as for the FoldingRange LSP call, which doesn't need type information. Those parses can actually fail, so we do need an error check. But we don't need the PGH; in all cases we are immediately using and discarding it. So it turns out we don't actually need the PGH type at all, at least not in the public API. Instead, we can pass around a concrete struct that has the various pieces of data directly available. This uncovered a bug in typeCheck: it should fail if it encounters any real errors. Change-Id: I203bf2dd79d5d65c01392d69c2cf4f7744fde7fc Reviewed-on: https://go-review.googlesource.com/c/tools/+/244021 Run-TryBot: Heschi Kreinick <heschi@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Rebecca Stambler <rstambler@golang.org>
2020-07-21 13:15:06 -06:00
return pgf, pkg, nil
}
internal/lsp: replace ParseGoHandle with concrete data ParseGoHandles serve two purposes: they pin cache entries so that redundant calculations are cached, and they allow users to obtain the actual parsed AST. The former is an implementation detail, and the latter turns out to just be an annoyance. Parsed Go files are obtained from two places. By far the most common is from a type checked package. But a type checked package must by definition have already parsed all the files it contains, so the PGH is already computed and cannot have failed. Type checked packages can simply return the parsed file without requiring a separate Check operation. We do want to pin the cache entries in this case, which I've done by holding on to the PGH in cache.pkg. There are some cases where we directly parse a file, such as for the FoldingRange LSP call, which doesn't need type information. Those parses can actually fail, so we do need an error check. But we don't need the PGH; in all cases we are immediately using and discarding it. So it turns out we don't actually need the PGH type at all, at least not in the public API. Instead, we can pass around a concrete struct that has the various pieces of data directly available. This uncovered a bug in typeCheck: it should fail if it encounters any real errors. Change-Id: I203bf2dd79d5d65c01392d69c2cf4f7744fde7fc Reviewed-on: https://go-review.googlesource.com/c/tools/+/244021 Run-TryBot: Heschi Kreinick <heschi@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Rebecca Stambler <rstambler@golang.org>
2020-07-21 13:15:06 -06:00
// findFileInDeps finds uri in pkg or its dependencies.
func findFileInDeps(pkg Package, uri span.URI) (*ParsedGoFile, Package, error) {
queue := []Package{pkg}
seen := make(map[string]bool)
for len(queue) > 0 {
pkg := queue[0]
queue = queue[1:]
seen[pkg.ID()] = true
internal/lsp: replace ParseGoHandle with concrete data ParseGoHandles serve two purposes: they pin cache entries so that redundant calculations are cached, and they allow users to obtain the actual parsed AST. The former is an implementation detail, and the latter turns out to just be an annoyance. Parsed Go files are obtained from two places. By far the most common is from a type checked package. But a type checked package must by definition have already parsed all the files it contains, so the PGH is already computed and cannot have failed. Type checked packages can simply return the parsed file without requiring a separate Check operation. We do want to pin the cache entries in this case, which I've done by holding on to the PGH in cache.pkg. There are some cases where we directly parse a file, such as for the FoldingRange LSP call, which doesn't need type information. Those parses can actually fail, so we do need an error check. But we don't need the PGH; in all cases we are immediately using and discarding it. So it turns out we don't actually need the PGH type at all, at least not in the public API. Instead, we can pass around a concrete struct that has the various pieces of data directly available. This uncovered a bug in typeCheck: it should fail if it encounters any real errors. Change-Id: I203bf2dd79d5d65c01392d69c2cf4f7744fde7fc Reviewed-on: https://go-review.googlesource.com/c/tools/+/244021 Run-TryBot: Heschi Kreinick <heschi@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Rebecca Stambler <rstambler@golang.org>
2020-07-21 13:15:06 -06:00
if pgf, err := pkg.File(uri); err == nil {
return pgf, pkg, nil
}
for _, dep := range pkg.Imports() {
if !seen[dep.ID()] {
queue = append(queue, dep)
}
}
}
return nil, nil, errors.Errorf("no file for %s in package %s", uri, pkg.ID())
}
// prevStmt returns the statement that precedes the statement containing pos.
// For example:
//
// foo := 1
// bar(1 + 2<>)
//
// If "<>" is pos, prevStmt returns "foo := 1"
func prevStmt(pos token.Pos, path []ast.Node) ast.Stmt {
var blockLines []ast.Stmt
for i := 0; i < len(path) && blockLines == nil; i++ {
switch n := path[i].(type) {
case *ast.BlockStmt:
blockLines = n.List
case *ast.CommClause:
blockLines = n.Body
case *ast.CaseClause:
blockLines = n.Body
}
}
for i := len(blockLines) - 1; i >= 0; i-- {
if blockLines[i].End() < pos {
return blockLines[i]
}
}
return nil
}
// formatZeroValue produces Go code representing the zero value of T. It
// returns the empty string if T is invalid.
func formatZeroValue(T types.Type, qf types.Qualifier) string {
switch u := T.Underlying().(type) {
case *types.Basic:
switch {
case u.Info()&types.IsNumeric > 0:
return "0"
case u.Info()&types.IsString > 0:
return `""`
case u.Info()&types.IsBoolean > 0:
return "false"
default:
return ""
}
case *types.Pointer, *types.Interface, *types.Chan, *types.Map, *types.Slice, *types.Signature:
return "nil"
default:
return types.TypeString(T, qf) + "{}"
}
}
// MarshalArgs encodes the given arguments to json.RawMessages. This function
// is used to construct arguments to a protocol.Command.
//
// Example usage:
//
// jsonArgs, err := EncodeArgs(1, "hello", true, StructuredArg{42, 12.6})
//
func MarshalArgs(args ...interface{}) ([]json.RawMessage, error) {
var out []json.RawMessage
for _, arg := range args {
argJSON, err := json.Marshal(arg)
if err != nil {
return nil, err
}
out = append(out, argJSON)
}
return out, nil
}
// UnmarshalArgs decodes the given json.RawMessages to the variables provided
// by args. Each element of args should be a pointer.
//
// Example usage:
//
// var (
// num int
// str string
// bul bool
// structured StructuredArg
// )
// err := UnmarshalArgs(args, &num, &str, &bul, &structured)
//
func UnmarshalArgs(jsonArgs []json.RawMessage, args ...interface{}) error {
if len(args) != len(jsonArgs) {
return fmt.Errorf("DecodeArgs: expected %d input arguments, got %d JSON arguments", len(args), len(jsonArgs))
}
for i, arg := range args {
if err := json.Unmarshal(jsonArgs[i], arg); err != nil {
return err
}
}
return nil
}