imports

package
v0.0.0-...-b73abcb Latest Latest
Warning

This package is not in the latest version of its module.

Go to latest
Published: Oct 21, 2024 License: MIT Imports: 34 Imported by: 0

Documentation

Overview

Package imports implements a Go pretty-printer (like package "go/format") that also adds or removes import statements as necessary.

Index

Constants

View Source
const MaxRelevance = 7.0

MaxRelevance is the highest relevance, used for the standard library. Chosen arbitrarily to match pre-existing gopls code.

Variables

This section is empty.

Functions

func ApplyFixes

func ApplyFixes(fixes []*ImportFix, filename string, src []byte, opt *Options, extraMode parser.Mode) (formatted []byte, err error)

ApplyFixes applies all of the fixes to the file and formats it. extraMode is added in when parsing the file. src and opts must be specified, but no env is needed.

func GetAllCandidates

func GetAllCandidates(ctx context.Context, wrapped func(ImportFix), searchPrefix, filename, filePkg string, env *ProcessEnv) error

GetAllCandidates calls wrapped for each package whose name starts with searchPrefix, and can be imported from filename with the package name filePkg.

Beware that the wrapped function may be called multiple times concurrently. TODO(adonovan): encapsulate the concurrency.

func GetImportPaths

func GetImportPaths(ctx context.Context, wrapped func(ImportFix), searchPrefix, filename, filePkg string, env *ProcessEnv) error

GetImportPaths calls wrapped for each package whose import path starts with searchPrefix, and can be imported from filename with the package name filePkg.

func GetPackageExports

func GetPackageExports(ctx context.Context, wrapped func(PackageExport), searchPkg, filename, filePkg string, env *ProcessEnv) error

GetPackageExports returns all known packages with name pkg and their exports.

func ImportPathToAssumedName

func ImportPathToAssumedName(importPath string) string

ImportPathToAssumedName returns the assumed package name of an import path. It does this using only string parsing of the import path. It picks the last element of the path that does not look like a major version, and then picks the valid identifier off the start of that element. It is used to determine if a local rename should be added to an import for clarity. This function could be moved to a standard package and exported if we want for use in other tools.

func PrimeCache

func PrimeCache(ctx context.Context, resolver Resolver) error

func Process

func Process(filename string, src []byte, opt *Options) (formatted []byte, err error)

Process implements golang.org/x/tools/imports.Process with explicit context in opt.Env.

func ScanModuleCache

func ScanModuleCache(dir string, cache *DirInfoCache, logf func(string, ...any))

ScanModuleCache walks the given directory, which must be a GOMODCACHE value, for directory package information, storing the results in cache.

func ScoreImportPaths

func ScoreImportPaths(ctx context.Context, env *ProcessEnv, paths []string) (map[string]float64, error)

func VendorlessPath

func VendorlessPath(ipath string) string

VendorlessPath returns the devendorized version of the import path ipath. For example, VendorlessPath("foo/bar/vendor/a/b") returns "a/b".

Types

type DirInfoCache

type DirInfoCache struct {
	// contains filtered or unexported fields
}

DirInfoCache is a concurrency-safe map for storing information about directories that may contain packages.

The information in this cache is built incrementally. Entries are initialized in scan. No new keys should be added in any other functions, as all directories containing packages are identified in scan.

Other functions, including loadExports and findPackage, may update entries in this cache as they discover new things about the directory.

The information in the cache is not expected to change for the cache's lifetime, so there is no protection against competing writes. Users should take care not to hold the cache across changes to the underlying files.

func NewDirInfoCache

func NewDirInfoCache() *DirInfoCache

func (*DirInfoCache) CacheExports

func (d *DirInfoCache) CacheExports(ctx context.Context, env *ProcessEnv, info directoryPackageInfo) (string, []stdlib.Symbol, error)

func (*DirInfoCache) CachePackageName

func (d *DirInfoCache) CachePackageName(info directoryPackageInfo) (string, error)

func (*DirInfoCache) Keys

func (d *DirInfoCache) Keys() (keys []string)

Keys returns the keys currently present in d.

func (*DirInfoCache) Load

func (d *DirInfoCache) Load(dir string) (directoryPackageInfo, bool)

Load returns a copy of the directoryPackageInfo for absolute directory dir.

func (*DirInfoCache) ScanAndListen

func (d *DirInfoCache) ScanAndListen(ctx context.Context, listener cacheListener) func()

ScanAndListen calls listener on all the items in the cache, and on anything newly added. The returned stop function waits for all in-flight callbacks to finish and blocks new ones.

func (*DirInfoCache) Store

func (d *DirInfoCache) Store(dir string, info directoryPackageInfo)

Store stores the package info for dir.

type ImportFix

type ImportFix struct {
	// StmtInfo represents the import statement this fix will add, remove, or change.
	StmtInfo ImportInfo
	// IdentName is the identifier that this fix will add or remove.
	IdentName string
	// FixType is the type of fix this is (AddImport, DeleteImport, SetImportName).
	FixType   ImportFixType
	Relevance float64 // see pkg
}

func FixImports

func FixImports(ctx context.Context, filename string, src []byte, opt *Options) (fixes []*ImportFix, err error)

FixImports returns a list of fixes to the imports that, when applied, will leave the imports in the same state as Process. src and opt must be specified.

Note that filename's directory influences which imports can be chosen, so it is important that filename be accurate.

type ImportFixType

type ImportFixType int
const (
	AddImport ImportFixType = iota
	DeleteImport
	SetImportName
)

type ImportInfo

type ImportInfo struct {
	ImportPath string // import path, e.g. "crypto/rand".
	Name       string // import name, e.g. "crand", or "" if none.
}

An ImportInfo represents a single import statement.

type ModuleResolver

type ModuleResolver struct {
	// contains filtered or unexported fields
}

ModuleResolver implements the Resolver interface for a workspace using modules.

A goal of the ModuleResolver is to invoke the Go command as little as possible. To this end, it runs the Go command only for listing module information (i.e. `go list -m -e -json ...`). Package scanning, the process of loading package information for the modules, is implemented internally via the scan method.

It has two types of state: the state derived from the go command, which is populated by init, and the state derived from scans, which is populated via scan. A root is considered scanned if it has been walked to discover directories. However, if the scan did not require additional information from the directory (such as package name or exports), the directory information itself may be partially populated. It will be lazily filled in as needed by scans, using the scanCallback.

func (*ModuleResolver) ClearForNewScan

func (r *ModuleResolver) ClearForNewScan() Resolver

ClearForNewScan invalidates the last scan.

It preserves the set of roots, but forgets about the set of directories. Though it forgets the set of module cache directories, it remembers their contents, since they are assumed to be immutable.

type Options

type Options struct {
	Env *ProcessEnv // The environment to use. Note: this contains the cached module and filesystem state.

	// LocalPrefix is a comma-separated string of import path prefixes, which, if
	// set, instructs Process to sort the import paths with the given prefixes
	// into another group after 3rd-party packages.
	LocalPrefix string

	Fragment  bool // Accept fragment of a source file (no package statement)
	AllErrors bool // Report all errors (not just the first 10 on different lines)

	Comments  bool // Print comments (true if nil *Options provided)
	TabIndent bool // Use tabs for indent (true if nil *Options provided)
	TabWidth  int  // Tab width (8 if nil *Options provided)

	FormatOnly bool // Disable the insertion and deletion of imports
}

Options is golang.org/x/tools/imports.Options with extra internal-only options.

type PackageExport

type PackageExport struct {
	Fix     *ImportFix
	Exports []stdlib.Symbol
}

A PackageExport is a package and its exports.

type ProcessEnv

type ProcessEnv struct {
	GocmdRunner *gocommand.Runner

	BuildFlags []string
	ModFlag    string

	// SkipPathInScan returns true if the path should be skipped from scans of
	// the RootCurrentModule root type. The function argument is a clean,
	// absolute path.
	SkipPathInScan func(string) bool

	// Env overrides the OS environment, and can be used to specify
	// GOPROXY, GO111MODULE, etc. PATH cannot be set here, because
	// exec.Command will not honor it.
	// Specifying all of requiredGoEnvVars avoids a call to `go env`.
	Env map[string]string

	WorkingDir string

	// If Logf is non-nil, debug logging is enabled through this function.
	Logf func(format string, args ...interface{})

	// If set, ModCache holds a shared cache of directory info to use across
	// multiple ProcessEnvs.
	ModCache *DirInfoCache
	// contains filtered or unexported fields
}

ProcessEnv contains environment variables and settings that affect the use of the go command, the go/build package, etc.

...a ProcessEnv *also* overwrites its Env along with derived state in the form of the resolver. And because it is lazily initialized, an env may just be broken and unusable, but there is no way for the caller to detect that: all queries will just fail.

TODO(rfindley): refactor this package so that this type (perhaps renamed to just Env or Config) is an immutable configuration struct, to be exchanged for an initialized object via a constructor that returns an error. Perhaps the signature should be `func NewResolver(*Env) (*Resolver, error)`, where resolver is a concrete type used for resolving imports. Via this refactoring, we can avoid the need to call ProcessEnv.init and ProcessEnv.GoEnv everywhere, and implicitly fix all the places where this these are misused. Also, we'd delegate the caller the decision of how to handle a broken environment.

func (*ProcessEnv) ClearModuleInfo

func (e *ProcessEnv) ClearModuleInfo()

ClearModuleInfo invalidates resolver state that depends on go.mod file contents (essentially, the output of go list -m -json ...).

Notably, it does not forget directory contents, which are reset asynchronously via ClearForNewScan.

If the ProcessEnv is a GOPATH environment, ClearModuleInfo is a no op.

TODO(rfindley): move this to a new env.go, consolidating ProcessEnv methods.

func (*ProcessEnv) CopyConfig

func (e *ProcessEnv) CopyConfig() *ProcessEnv

CopyConfig copies the env's configuration into a new env.

func (*ProcessEnv) GetResolver

func (e *ProcessEnv) GetResolver() (Resolver, error)

func (*ProcessEnv) UpdateResolver

func (e *ProcessEnv) UpdateResolver(r Resolver)

UpdateResolver sets the resolver for the ProcessEnv to use in imports operations. Only for use with the result of [Resolver.ClearForNewScan].

TODO(rfindley): this awkward API is a result of the (arguably) inverted relationship between configuration and state described in the doc comment for ProcessEnv.

type Resolver

type Resolver interface {

	// ClearForNewScan returns a new Resolver based on the receiver that has
	// cleared its internal caches of directory contents.
	//
	// The new resolver should be primed and then set via
	// [ProcessEnv.UpdateResolver].
	ClearForNewScan() Resolver
	// contains filtered or unexported methods
}

A Resolver does the build-system-specific parts of goimports.

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL