1
0
mirror of https://github.com/golang/go synced 2024-11-19 03:54:42 -07:00
go/imports/fix_test.go

1676 lines
32 KiB
Go
Raw Normal View History

// Copyright 2013 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 imports
import (
"bytes"
"flag"
"go/build"
"io/ioutil"
"os"
"path/filepath"
"runtime"
"strings"
"sync"
"testing"
)
var only = flag.String("only", "", "If non-empty, the fix test to run")
var tests = []struct {
name string
formatOnly bool
in, out string
}{
// Adding an import to an existing parenthesized import
{
name: "factored_imports_add",
in: `package foo
import (
"fmt"
)
func bar() {
var b bytes.Buffer
fmt.Println(b.String())
}
`,
out: `package foo
import (
"bytes"
"fmt"
)
func bar() {
var b bytes.Buffer
fmt.Println(b.String())
}
`,
},
// Adding an import to an existing parenthesized import,
// verifying it goes into the first section.
{
name: "factored_imports_add_first_sec",
in: `package foo
import (
"fmt"
"appengine"
)
func bar() {
var b bytes.Buffer
_ = appengine.IsDevServer
fmt.Println(b.String())
}
`,
out: `package foo
import (
"bytes"
"fmt"
"appengine"
)
func bar() {
var b bytes.Buffer
_ = appengine.IsDevServer
fmt.Println(b.String())
}
`,
},
// Adding an import to an existing parenthesized import,
// verifying it goes into the first section. (test 2)
{
name: "factored_imports_add_first_sec_2",
in: `package foo
import (
"fmt"
"appengine"
)
func bar() {
_ = math.NaN
_ = fmt.Sprintf
_ = appengine.IsDevServer
}
`,
out: `package foo
import (
"fmt"
"math"
"appengine"
)
func bar() {
_ = math.NaN
_ = fmt.Sprintf
_ = appengine.IsDevServer
}
`,
},
// Adding a new import line, without parens
{
name: "add_import_section",
in: `package foo
func bar() {
var b bytes.Buffer
}
`,
out: `package foo
import "bytes"
func bar() {
var b bytes.Buffer
}
`,
},
// Adding two new imports, which should make a parenthesized import decl.
{
name: "add_import_paren_section",
in: `package foo
func bar() {
_, _ := bytes.Buffer, zip.NewReader
}
`,
out: `package foo
import (
"archive/zip"
"bytes"
)
func bar() {
_, _ := bytes.Buffer, zip.NewReader
}
`,
},
// Make sure we don't add things twice
{
name: "no_double_add",
in: `package foo
func bar() {
_, _ := bytes.Buffer, bytes.NewReader
}
`,
out: `package foo
import "bytes"
func bar() {
_, _ := bytes.Buffer, bytes.NewReader
}
`,
},
// Remove unused imports, 1 of a factored block
{
name: "remove_unused_1_of_2",
in: `package foo
import (
"bytes"
"fmt"
)
func bar() {
_, _ := bytes.Buffer, bytes.NewReader
}
`,
out: `package foo
import (
"bytes"
)
func bar() {
_, _ := bytes.Buffer, bytes.NewReader
}
`,
},
// Remove unused imports, 2 of 2
{
name: "remove_unused_2_of_2",
in: `package foo
import (
"bytes"
"fmt"
)
func bar() {
}
`,
out: `package foo
func bar() {
}
`,
},
// Remove unused imports, 1 of 1
{
name: "remove_unused_1_of_1",
in: `package foo
import "fmt"
func bar() {
}
`,
out: `package foo
func bar() {
}
`,
},
// Don't remove empty imports.
{
name: "dont_remove_empty_imports",
in: `package foo
import (
_ "image/png"
_ "image/jpeg"
)
`,
out: `package foo
import (
_ "image/jpeg"
_ "image/png"
)
`,
},
// Don't remove dot imports.
{
name: "dont_remove_dot_imports",
in: `package foo
import (
. "foo"
. "bar"
)
`,
out: `package foo
import (
. "bar"
. "foo"
)
`,
},
// Skip refs the parser can resolve.
{
name: "skip_resolved_refs",
in: `package foo
func f() {
type t struct{ Println func(string) }
fmt := t{Println: func(string) {}}
fmt.Println("foo")
}
`,
out: `package foo
func f() {
type t struct{ Println func(string) }
fmt := t{Println: func(string) {}}
fmt.Println("foo")
}
`,
},
// Do not add a package we already have a resolution for.
{
name: "skip_template",
in: `package foo
import "html/template"
func f() { t = template.New("sometemplate") }
`,
out: `package foo
import "html/template"
func f() { t = template.New("sometemplate") }
`,
},
// Don't touch cgo
{
name: "cgo",
in: `package foo
/*
#include <foo.h>
*/
import "C"
`,
out: `package foo
/*
#include <foo.h>
*/
import "C"
`,
},
// Put some things in their own section
{
name: "make_sections",
in: `package foo
import (
"os"
)
func foo () {
_, _ = os.Args, fmt.Println
_, _ = appengine.FooSomething, user.Current
}
`,
out: `package foo
import (
"fmt"
"os"
"appengine"
"appengine/user"
)
func foo() {
_, _ = os.Args, fmt.Println
_, _ = appengine.FooSomething, user.Current
}
`,
},
// Delete existing empty import block
{
name: "delete_empty_import_block",
in: `package foo
import ()
`,
out: `package foo
`,
},
// Use existing empty import block
{
name: "use_empty_import_block",
in: `package foo
import ()
func f() {
_ = fmt.Println
}
`,
out: `package foo
import "fmt"
func f() {
_ = fmt.Println
}
`,
},
// Blank line before adding new section.
{
name: "blank_line_before_new_group",
in: `package foo
import (
"fmt"
"net"
)
func f() {
_ = net.Dial
_ = fmt.Printf
_ = snappy.Foo
}
`,
out: `package foo
import (
"fmt"
"net"
"code.google.com/p/snappy-go/snappy"
)
func f() {
_ = net.Dial
_ = fmt.Printf
_ = snappy.Foo
}
`,
},
// Blank line between standard library and third-party stuff.
{
name: "blank_line_separating_std_and_third_party",
in: `package foo
import (
"code.google.com/p/snappy-go/snappy"
"fmt"
"net"
)
func f() {
_ = net.Dial
_ = fmt.Printf
_ = snappy.Foo
}
`,
out: `package foo
import (
"fmt"
"net"
"code.google.com/p/snappy-go/snappy"
)
func f() {
_ = net.Dial
_ = fmt.Printf
_ = snappy.Foo
}
`,
},
// golang.org/issue/6884
{
name: "issue 6884",
in: `package main
// A comment
func main() {
fmt.Println("Hello, world")
}
`,
out: `package main
import "fmt"
// A comment
func main() {
fmt.Println("Hello, world")
}
`,
},
// golang.org/issue/7132
{
name: "issue 7132",
in: `package main
import (
"fmt"
"gu"
"github.com/foo/bar"
)
var (
a = bar.a
b = gu.a
c = fmt.Printf
)
`,
out: `package main
import (
"fmt"
"gu"
"github.com/foo/bar"
)
var (
a = bar.a
b = gu.a
c = fmt.Printf
)
`,
},
{
name: "renamed package",
in: `package main
var _ = str.HasPrefix
`,
out: `package main
import str "strings"
var _ = str.HasPrefix
`,
},
{
name: "fragment with main",
in: `func main(){fmt.Println("Hello, world")}`,
out: `package main
import "fmt"
func main() { fmt.Println("Hello, world") }
`,
},
{
name: "fragment without main",
in: `func notmain(){fmt.Println("Hello, world")}`,
out: `import "fmt"
func notmain() { fmt.Println("Hello, world") }`,
},
go.tools/astutil: fix edge case in DeleteImport causing merging of import sections. The issue occurs only when deleting an import that has a blank line immediately preceding, and other imports before that. Currently, DeleteImport assumes there's a blank line-sized hole left behind where the import was, and always deletes it. That blank line-sized hole is there in all cases except the above edge case. This fix checks for that edge case, and does not remove the blank line-sized hole. The CL also adds a previously failing test case that catches this scenario. After the change to DeleteImport, the new test passes (along with all other tests). Fixes golang/go#7679. Note that there is no attempt to ensure the result *ast.File and *token.FileSet are perfectly matching to what you would get if you printed the AST and parsed it back. This is how the rest of the package and the current tests work (i.e., they only check that printing the AST gives the correct output). Changing that is very hard, if not impossible, at least not without resorting to manipulating AST via printing, text manipulation and parsing. This is okay for most usages, but it does create potential problems. For example, astutil.Imports() currently only works correctly on freshly parsed AST. If that AST is manipulated via astutil funcs, then Imports() may not always generate correct output. However, thas is a separate issue and should be treated as such. LGTM=bradfitz R=golang-codereviews, gobot, adonovan, bradfitz CC=golang-codereviews https://golang.org/cl/92250045
2014-05-19 15:04:30 -06:00
// Remove first import within in a 2nd/3rd/4th/etc. section.
// golang.org/issue/7679
{
name: "issue 7679",
in: `package main
import (
"fmt"
"github.com/foo/bar"
"github.com/foo/qux"
)
func main() {
var _ = fmt.Println
//var _ = bar.A
var _ = qux.B
}
`,
out: `package main
import (
"fmt"
"github.com/foo/qux"
)
func main() {
var _ = fmt.Println
//var _ = bar.A
var _ = qux.B
}
`,
},
// Blank line can be added before all types of import declarations.
// golang.org/issue/7866
{
name: "issue 7866",
in: `package main
import (
"fmt"
renamed_bar "github.com/foo/bar"
. "github.com/foo/baz"
"io"
_ "github.com/foo/qux"
"strings"
)
func main() {
_, _, _, _, _ = fmt.Errorf, io.Copy, strings.Contains, renamed_bar.A, B
}
`,
out: `package main
import (
"fmt"
renamed_bar "github.com/foo/bar"
"io"
. "github.com/foo/baz"
"strings"
_ "github.com/foo/qux"
)
func main() {
_, _, _, _, _ = fmt.Errorf, io.Copy, strings.Contains, renamed_bar.A, B
}
`,
},
// Non-idempotent comment formatting
// golang.org/issue/8035
{
name: "issue 8035",
in: `package main
import (
"fmt" // A
"go/ast" // B
_ "launchpad.net/gocheck" // C
)
func main() { _, _ = fmt.Print, ast.Walk }
`,
out: `package main
import (
"fmt" // A
"go/ast" // B
_ "launchpad.net/gocheck" // C
)
func main() { _, _ = fmt.Print, ast.Walk }
`,
},
// Failure to delete all duplicate imports
// golang.org/issue/8459
{
name: "issue 8459",
in: `package main
import (
"fmt"
"log"
"log"
"math"
)
func main() { fmt.Println("pi:", math.Pi) }
`,
out: `package main
import (
"fmt"
"math"
)
func main() { fmt.Println("pi:", math.Pi) }
`,
},
// Too aggressive prefix matching
// golang.org/issue/9961
{
name: "issue 9961",
in: `package p
import (
"zip"
"rsc.io/p"
)
var (
_ = fmt.Print
_ = zip.Store
_ p.P
_ = regexp.Compile
)
`,
out: `package p
import (
"fmt"
"regexp"
"zip"
"rsc.io/p"
)
var (
_ = fmt.Print
_ = zip.Store
_ p.P
_ = regexp.Compile
)
`,
},
// Unused named import is mistaken for unnamed import
// golang.org/issue/8149
{
name: "issue 8149",
in: `package main
import foo "fmt"
func main() { fmt.Println() }
`,
out: `package main
import "fmt"
func main() { fmt.Println() }
`,
},
// Unused named import is mistaken for unnamed import
// golang.org/issue/8149
{
name: "issue 8149",
in: `package main
import (
"fmt"
x "fmt"
)
func main() { fmt.Println() }
`,
out: `package main
import (
"fmt"
)
func main() { fmt.Println() }
`,
},
// FormatOnly
{
name: "format only",
formatOnly: true,
in: `package main
import (
"fmt"
"golang.org/x/foo"
)
func main() {}
`,
out: `package main
import (
"fmt"
"golang.org/x/foo"
)
func main() {}
`,
},
{
name: "do not make grouped imports non-grouped",
in: `package p
import (
"bytes"
"fmt"
)
var _ = fmt.Sprintf
`,
out: `package p
import (
"fmt"
)
var _ = fmt.Sprintf
go.tools/astutil: fix edge case in DeleteImport causing merging of import sections. The issue occurs only when deleting an import that has a blank line immediately preceding, and other imports before that. Currently, DeleteImport assumes there's a blank line-sized hole left behind where the import was, and always deletes it. That blank line-sized hole is there in all cases except the above edge case. This fix checks for that edge case, and does not remove the blank line-sized hole. The CL also adds a previously failing test case that catches this scenario. After the change to DeleteImport, the new test passes (along with all other tests). Fixes golang/go#7679. Note that there is no attempt to ensure the result *ast.File and *token.FileSet are perfectly matching to what you would get if you printed the AST and parsed it back. This is how the rest of the package and the current tests work (i.e., they only check that printing the AST gives the correct output). Changing that is very hard, if not impossible, at least not without resorting to manipulating AST via printing, text manipulation and parsing. This is okay for most usages, but it does create potential problems. For example, astutil.Imports() currently only works correctly on freshly parsed AST. If that AST is manipulated via astutil funcs, then Imports() may not always generate correct output. However, thas is a separate issue and should be treated as such. LGTM=bradfitz R=golang-codereviews, gobot, adonovan, bradfitz CC=golang-codereviews https://golang.org/cl/92250045
2014-05-19 15:04:30 -06:00
`,
},
}
func TestFixImports(t *testing.T) {
simplePkgs := map[string]string{
"appengine": "appengine",
"bytes": "bytes",
"fmt": "fmt",
"math": "math",
"os": "os",
"p": "rsc.io/p",
"regexp": "regexp",
"snappy": "code.google.com/p/snappy-go/snappy",
"str": "strings",
"user": "appengine/user",
"zip": "archive/zip",
}
old := findImport
defer func() {
findImport = old
}()
findImport = func(pkgName string, symbols map[string]bool, filename string) (string, bool, error) {
return simplePkgs[pkgName], pkgName == "str", nil
}
options := &Options{
TabWidth: 8,
TabIndent: true,
Comments: true,
Fragment: true,
}
for _, tt := range tests {
options.FormatOnly = tt.formatOnly
if *only != "" && tt.name != *only {
continue
}
buf, err := Process(tt.name+".go", []byte(tt.in), options)
if err != nil {
t.Errorf("error on %q: %v", tt.name, err)
continue
}
if got := string(buf); got != tt.out {
t.Errorf("results diff on %q\nGOT:\n%s\nWANT:\n%s\n", tt.name, got, tt.out)
}
}
}
// Test support for packages in GOPATH that are actually symlinks.
// Also test that a symlink loop does not block the process.
func TestImportSymlinks(t *testing.T) {
switch runtime.GOOS {
case "windows", "plan9":
t.Skipf("skipping test on %q as there are no symlinks", runtime.GOOS)
}
newGoPath, err := ioutil.TempDir("", "symlinktest")
if err != nil {
t.Fatal(err)
}
defer os.RemoveAll(newGoPath)
targetPath := newGoPath + "/target"
if err := os.MkdirAll(targetPath, 0755); err != nil {
t.Fatal(err)
}
if err := ioutil.WriteFile(targetPath+"/f.go", []byte("package mypkg\nvar Foo = 123\n"), 0666); err != nil {
t.Fatal(err)
}
symlinkPath := newGoPath + "/src/x/mypkg"
if err := os.MkdirAll(filepath.Dir(symlinkPath), 0755); err != nil {
t.Fatal(err)
}
if err := os.Symlink(targetPath, symlinkPath); err != nil {
t.Fatal(err)
}
// Add a symlink loop.
if err := os.Symlink(newGoPath+"/src/x", newGoPath+"/src/x/apkg"); err != nil {
t.Fatal(err)
}
cmd/goimports, imports: make goimports great again I felt the burn of my laptop on my legs, spinning away while processing goimports, and felt that it was time to make goimports great again. Over the past few years goimports fell into a slow state of disrepair with too many feature additions and no attention to the performance death by a thousand cuts. This was particularly terrible on OS X with its lackluster filesystem buffering. This CL makes goimports stronger, together with various optimizations and more visibility into what goimports is doing. * adds more internal documentation * avoids scanning $GOPATH for answers when running goimports on a file under $GOROOT (for Go core hackers) * don't read all $GOROOT & $GOPATH directories' Go code looking for their package names until much later. Require the package name of missing imports to be present in the last two directory path components. Then only try importing them in order from best to worst (shortest to longest, as before), so we can stop early. * when adding imports, add names to imports when the imported package name doesn't match the baes of its import path. For example: import foo "example.net/foo/v1" * don't read all *.go files in a package directory once the first file in a directory has revealed itself to be a package we're not looking for. For example, if we're looking for the right "client" for "client.Foo", we used to consider a directory "bar/client" as a candidate and read all 50 of its *.go files instead of stopping after its first *.go file had a "package main" line. * add some fast paths to remove allocations * add some fast paths to remove disk I/O when looking up the base package name of a standard library import (of existing imports in a file, which are very common) * adds a special case for import "C", to avoid some disk I/O. * add a -verbose flag to goimports for debugging On my Mac laptop with a huge $GOPATH, with a test file like: package foo import ( "fmt" "net/http" ) /* */ import "C" var _ = cloudbilling.New var _ = http.NewRequest var _ = client.New ... this took like 10 seconds before, and now 1.3 seconds. (Still slow; disk-based caching can come later) Updates golang/go#16367 (goimports is slow) Updates golang/go#16384 (refactor TestRename is broken on Windows) Change-Id: I97e85d3016afc9f2ad5501f97babad30c7989183 Reviewed-on: https://go-review.googlesource.com/24941 Reviewed-by: Andrew Gerrand <adg@golang.org> Reviewed-by: Rob Pike <r@golang.org> Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-14 18:08:27 -06:00
withEmptyGoPath(func() {
build.Default.GOPATH = newGoPath
cmd/goimports, imports: make goimports great again I felt the burn of my laptop on my legs, spinning away while processing goimports, and felt that it was time to make goimports great again. Over the past few years goimports fell into a slow state of disrepair with too many feature additions and no attention to the performance death by a thousand cuts. This was particularly terrible on OS X with its lackluster filesystem buffering. This CL makes goimports stronger, together with various optimizations and more visibility into what goimports is doing. * adds more internal documentation * avoids scanning $GOPATH for answers when running goimports on a file under $GOROOT (for Go core hackers) * don't read all $GOROOT & $GOPATH directories' Go code looking for their package names until much later. Require the package name of missing imports to be present in the last two directory path components. Then only try importing them in order from best to worst (shortest to longest, as before), so we can stop early. * when adding imports, add names to imports when the imported package name doesn't match the baes of its import path. For example: import foo "example.net/foo/v1" * don't read all *.go files in a package directory once the first file in a directory has revealed itself to be a package we're not looking for. For example, if we're looking for the right "client" for "client.Foo", we used to consider a directory "bar/client" as a candidate and read all 50 of its *.go files instead of stopping after its first *.go file had a "package main" line. * add some fast paths to remove allocations * add some fast paths to remove disk I/O when looking up the base package name of a standard library import (of existing imports in a file, which are very common) * adds a special case for import "C", to avoid some disk I/O. * add a -verbose flag to goimports for debugging On my Mac laptop with a huge $GOPATH, with a test file like: package foo import ( "fmt" "net/http" ) /* */ import "C" var _ = cloudbilling.New var _ = http.NewRequest var _ = client.New ... this took like 10 seconds before, and now 1.3 seconds. (Still slow; disk-based caching can come later) Updates golang/go#16367 (goimports is slow) Updates golang/go#16384 (refactor TestRename is broken on Windows) Change-Id: I97e85d3016afc9f2ad5501f97babad30c7989183 Reviewed-on: https://go-review.googlesource.com/24941 Reviewed-by: Andrew Gerrand <adg@golang.org> Reviewed-by: Rob Pike <r@golang.org> Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-14 18:08:27 -06:00
input := `package p
var (
_ = fmt.Print
_ = mypkg.Foo
)
`
cmd/goimports, imports: make goimports great again I felt the burn of my laptop on my legs, spinning away while processing goimports, and felt that it was time to make goimports great again. Over the past few years goimports fell into a slow state of disrepair with too many feature additions and no attention to the performance death by a thousand cuts. This was particularly terrible on OS X with its lackluster filesystem buffering. This CL makes goimports stronger, together with various optimizations and more visibility into what goimports is doing. * adds more internal documentation * avoids scanning $GOPATH for answers when running goimports on a file under $GOROOT (for Go core hackers) * don't read all $GOROOT & $GOPATH directories' Go code looking for their package names until much later. Require the package name of missing imports to be present in the last two directory path components. Then only try importing them in order from best to worst (shortest to longest, as before), so we can stop early. * when adding imports, add names to imports when the imported package name doesn't match the baes of its import path. For example: import foo "example.net/foo/v1" * don't read all *.go files in a package directory once the first file in a directory has revealed itself to be a package we're not looking for. For example, if we're looking for the right "client" for "client.Foo", we used to consider a directory "bar/client" as a candidate and read all 50 of its *.go files instead of stopping after its first *.go file had a "package main" line. * add some fast paths to remove allocations * add some fast paths to remove disk I/O when looking up the base package name of a standard library import (of existing imports in a file, which are very common) * adds a special case for import "C", to avoid some disk I/O. * add a -verbose flag to goimports for debugging On my Mac laptop with a huge $GOPATH, with a test file like: package foo import ( "fmt" "net/http" ) /* */ import "C" var _ = cloudbilling.New var _ = http.NewRequest var _ = client.New ... this took like 10 seconds before, and now 1.3 seconds. (Still slow; disk-based caching can come later) Updates golang/go#16367 (goimports is slow) Updates golang/go#16384 (refactor TestRename is broken on Windows) Change-Id: I97e85d3016afc9f2ad5501f97babad30c7989183 Reviewed-on: https://go-review.googlesource.com/24941 Reviewed-by: Andrew Gerrand <adg@golang.org> Reviewed-by: Rob Pike <r@golang.org> Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-14 18:08:27 -06:00
output := `package p
import (
"fmt"
"x/mypkg"
)
var (
_ = fmt.Print
_ = mypkg.Foo
)
`
cmd/goimports, imports: make goimports great again I felt the burn of my laptop on my legs, spinning away while processing goimports, and felt that it was time to make goimports great again. Over the past few years goimports fell into a slow state of disrepair with too many feature additions and no attention to the performance death by a thousand cuts. This was particularly terrible on OS X with its lackluster filesystem buffering. This CL makes goimports stronger, together with various optimizations and more visibility into what goimports is doing. * adds more internal documentation * avoids scanning $GOPATH for answers when running goimports on a file under $GOROOT (for Go core hackers) * don't read all $GOROOT & $GOPATH directories' Go code looking for their package names until much later. Require the package name of missing imports to be present in the last two directory path components. Then only try importing them in order from best to worst (shortest to longest, as before), so we can stop early. * when adding imports, add names to imports when the imported package name doesn't match the baes of its import path. For example: import foo "example.net/foo/v1" * don't read all *.go files in a package directory once the first file in a directory has revealed itself to be a package we're not looking for. For example, if we're looking for the right "client" for "client.Foo", we used to consider a directory "bar/client" as a candidate and read all 50 of its *.go files instead of stopping after its first *.go file had a "package main" line. * add some fast paths to remove allocations * add some fast paths to remove disk I/O when looking up the base package name of a standard library import (of existing imports in a file, which are very common) * adds a special case for import "C", to avoid some disk I/O. * add a -verbose flag to goimports for debugging On my Mac laptop with a huge $GOPATH, with a test file like: package foo import ( "fmt" "net/http" ) /* */ import "C" var _ = cloudbilling.New var _ = http.NewRequest var _ = client.New ... this took like 10 seconds before, and now 1.3 seconds. (Still slow; disk-based caching can come later) Updates golang/go#16367 (goimports is slow) Updates golang/go#16384 (refactor TestRename is broken on Windows) Change-Id: I97e85d3016afc9f2ad5501f97babad30c7989183 Reviewed-on: https://go-review.googlesource.com/24941 Reviewed-by: Andrew Gerrand <adg@golang.org> Reviewed-by: Rob Pike <r@golang.org> Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-14 18:08:27 -06:00
buf, err := Process(newGoPath+"/src/myotherpkg/toformat.go", []byte(input), &Options{})
if err != nil {
t.Fatal(err)
}
if got := string(buf); got != output {
t.Fatalf("results differ\nGOT:\n%s\nWANT:\n%s\n", got, output)
}
})
// Add a .goimportsignore and ensure it is respected.
if err := ioutil.WriteFile(newGoPath+"/src/.goimportsignore", []byte("x/mypkg\n"), 0666); err != nil {
t.Fatal(err)
}
withEmptyGoPath(func() {
build.Default.GOPATH = newGoPath
input := `package p
var (
_ = fmt.Print
_ = mypkg.Foo
)
`
output := `package p
import "fmt"
var (
_ = fmt.Print
_ = mypkg.Foo
)
`
buf, err := Process(newGoPath+"/src/myotherpkg/toformat.go", []byte(input), &Options{})
if err != nil {
t.Fatal(err)
}
if got := string(buf); got != output {
t.Fatalf("ignored results differ\nGOT:\n%s\nWANT:\n%s\n", got, output)
}
})
}
// Test for correctly identifying the name of a vendored package when it
// differs from its directory name. In this test, the import line
// "mypkg.com/mypkg.v1" would be removed if goimports wasn't able to detect
// that the package name is "mypkg".
func TestFixImportsVendorPackage(t *testing.T) {
// Skip this test on go versions with no vendor support.
if _, err := os.Stat(filepath.Join(runtime.GOROOT(), "src/vendor")); err != nil {
t.Skip(err)
}
cmd/goimports, imports: make goimports great again I felt the burn of my laptop on my legs, spinning away while processing goimports, and felt that it was time to make goimports great again. Over the past few years goimports fell into a slow state of disrepair with too many feature additions and no attention to the performance death by a thousand cuts. This was particularly terrible on OS X with its lackluster filesystem buffering. This CL makes goimports stronger, together with various optimizations and more visibility into what goimports is doing. * adds more internal documentation * avoids scanning $GOPATH for answers when running goimports on a file under $GOROOT (for Go core hackers) * don't read all $GOROOT & $GOPATH directories' Go code looking for their package names until much later. Require the package name of missing imports to be present in the last two directory path components. Then only try importing them in order from best to worst (shortest to longest, as before), so we can stop early. * when adding imports, add names to imports when the imported package name doesn't match the baes of its import path. For example: import foo "example.net/foo/v1" * don't read all *.go files in a package directory once the first file in a directory has revealed itself to be a package we're not looking for. For example, if we're looking for the right "client" for "client.Foo", we used to consider a directory "bar/client" as a candidate and read all 50 of its *.go files instead of stopping after its first *.go file had a "package main" line. * add some fast paths to remove allocations * add some fast paths to remove disk I/O when looking up the base package name of a standard library import (of existing imports in a file, which are very common) * adds a special case for import "C", to avoid some disk I/O. * add a -verbose flag to goimports for debugging On my Mac laptop with a huge $GOPATH, with a test file like: package foo import ( "fmt" "net/http" ) /* */ import "C" var _ = cloudbilling.New var _ = http.NewRequest var _ = client.New ... this took like 10 seconds before, and now 1.3 seconds. (Still slow; disk-based caching can come later) Updates golang/go#16367 (goimports is slow) Updates golang/go#16384 (refactor TestRename is broken on Windows) Change-Id: I97e85d3016afc9f2ad5501f97babad30c7989183 Reviewed-on: https://go-review.googlesource.com/24941 Reviewed-by: Andrew Gerrand <adg@golang.org> Reviewed-by: Rob Pike <r@golang.org> Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-14 18:08:27 -06:00
testConfig{
gopathFiles: map[string]string{
"mypkg.com/outpkg/vendor/mypkg.com/mypkg.v1/f.go": "package mypkg\nvar Foo = 123\n",
},
}.test(t, func(t *goimportTest) {
input := `package p
import (
"fmt"
"mypkg.com/mypkg.v1"
)
var (
_ = fmt.Print
_ = mypkg.Foo
)
`
cmd/goimports, imports: make goimports great again I felt the burn of my laptop on my legs, spinning away while processing goimports, and felt that it was time to make goimports great again. Over the past few years goimports fell into a slow state of disrepair with too many feature additions and no attention to the performance death by a thousand cuts. This was particularly terrible on OS X with its lackluster filesystem buffering. This CL makes goimports stronger, together with various optimizations and more visibility into what goimports is doing. * adds more internal documentation * avoids scanning $GOPATH for answers when running goimports on a file under $GOROOT (for Go core hackers) * don't read all $GOROOT & $GOPATH directories' Go code looking for their package names until much later. Require the package name of missing imports to be present in the last two directory path components. Then only try importing them in order from best to worst (shortest to longest, as before), so we can stop early. * when adding imports, add names to imports when the imported package name doesn't match the baes of its import path. For example: import foo "example.net/foo/v1" * don't read all *.go files in a package directory once the first file in a directory has revealed itself to be a package we're not looking for. For example, if we're looking for the right "client" for "client.Foo", we used to consider a directory "bar/client" as a candidate and read all 50 of its *.go files instead of stopping after its first *.go file had a "package main" line. * add some fast paths to remove allocations * add some fast paths to remove disk I/O when looking up the base package name of a standard library import (of existing imports in a file, which are very common) * adds a special case for import "C", to avoid some disk I/O. * add a -verbose flag to goimports for debugging On my Mac laptop with a huge $GOPATH, with a test file like: package foo import ( "fmt" "net/http" ) /* */ import "C" var _ = cloudbilling.New var _ = http.NewRequest var _ = client.New ... this took like 10 seconds before, and now 1.3 seconds. (Still slow; disk-based caching can come later) Updates golang/go#16367 (goimports is slow) Updates golang/go#16384 (refactor TestRename is broken on Windows) Change-Id: I97e85d3016afc9f2ad5501f97babad30c7989183 Reviewed-on: https://go-review.googlesource.com/24941 Reviewed-by: Andrew Gerrand <adg@golang.org> Reviewed-by: Rob Pike <r@golang.org> Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-14 18:08:27 -06:00
buf, err := Process(filepath.Join(t.gopath, "src/mypkg.com/outpkg/toformat.go"), []byte(input), &Options{})
if err != nil {
t.Fatal(err)
}
if got := string(buf); got != input {
t.Fatalf("results differ\nGOT:\n%s\nWANT:\n%s\n", got, input)
}
})
}
func TestFindImportGoPath(t *testing.T) {
goroot, err := ioutil.TempDir("", "goimports-")
if err != nil {
t.Fatal(err)
}
defer os.RemoveAll(goroot)
origStdlib := stdlib
defer func() {
stdlib = origStdlib
}()
stdlib = nil
cmd/goimports, imports: make goimports great again I felt the burn of my laptop on my legs, spinning away while processing goimports, and felt that it was time to make goimports great again. Over the past few years goimports fell into a slow state of disrepair with too many feature additions and no attention to the performance death by a thousand cuts. This was particularly terrible on OS X with its lackluster filesystem buffering. This CL makes goimports stronger, together with various optimizations and more visibility into what goimports is doing. * adds more internal documentation * avoids scanning $GOPATH for answers when running goimports on a file under $GOROOT (for Go core hackers) * don't read all $GOROOT & $GOPATH directories' Go code looking for their package names until much later. Require the package name of missing imports to be present in the last two directory path components. Then only try importing them in order from best to worst (shortest to longest, as before), so we can stop early. * when adding imports, add names to imports when the imported package name doesn't match the baes of its import path. For example: import foo "example.net/foo/v1" * don't read all *.go files in a package directory once the first file in a directory has revealed itself to be a package we're not looking for. For example, if we're looking for the right "client" for "client.Foo", we used to consider a directory "bar/client" as a candidate and read all 50 of its *.go files instead of stopping after its first *.go file had a "package main" line. * add some fast paths to remove allocations * add some fast paths to remove disk I/O when looking up the base package name of a standard library import (of existing imports in a file, which are very common) * adds a special case for import "C", to avoid some disk I/O. * add a -verbose flag to goimports for debugging On my Mac laptop with a huge $GOPATH, with a test file like: package foo import ( "fmt" "net/http" ) /* */ import "C" var _ = cloudbilling.New var _ = http.NewRequest var _ = client.New ... this took like 10 seconds before, and now 1.3 seconds. (Still slow; disk-based caching can come later) Updates golang/go#16367 (goimports is slow) Updates golang/go#16384 (refactor TestRename is broken on Windows) Change-Id: I97e85d3016afc9f2ad5501f97babad30c7989183 Reviewed-on: https://go-review.googlesource.com/24941 Reviewed-by: Andrew Gerrand <adg@golang.org> Reviewed-by: Rob Pike <r@golang.org> Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-14 18:08:27 -06:00
withEmptyGoPath(func() {
// Test against imaginary bits/bytes package in std lib
bytesDir := filepath.Join(goroot, "src", "pkg", "bits", "bytes")
for _, tag := range build.Default.ReleaseTags {
// Go 1.4 rearranged the GOROOT tree to remove the "pkg" path component.
if tag == "go1.4" {
bytesDir = filepath.Join(goroot, "src", "bits", "bytes")
}
}
cmd/goimports, imports: make goimports great again I felt the burn of my laptop on my legs, spinning away while processing goimports, and felt that it was time to make goimports great again. Over the past few years goimports fell into a slow state of disrepair with too many feature additions and no attention to the performance death by a thousand cuts. This was particularly terrible on OS X with its lackluster filesystem buffering. This CL makes goimports stronger, together with various optimizations and more visibility into what goimports is doing. * adds more internal documentation * avoids scanning $GOPATH for answers when running goimports on a file under $GOROOT (for Go core hackers) * don't read all $GOROOT & $GOPATH directories' Go code looking for their package names until much later. Require the package name of missing imports to be present in the last two directory path components. Then only try importing them in order from best to worst (shortest to longest, as before), so we can stop early. * when adding imports, add names to imports when the imported package name doesn't match the baes of its import path. For example: import foo "example.net/foo/v1" * don't read all *.go files in a package directory once the first file in a directory has revealed itself to be a package we're not looking for. For example, if we're looking for the right "client" for "client.Foo", we used to consider a directory "bar/client" as a candidate and read all 50 of its *.go files instead of stopping after its first *.go file had a "package main" line. * add some fast paths to remove allocations * add some fast paths to remove disk I/O when looking up the base package name of a standard library import (of existing imports in a file, which are very common) * adds a special case for import "C", to avoid some disk I/O. * add a -verbose flag to goimports for debugging On my Mac laptop with a huge $GOPATH, with a test file like: package foo import ( "fmt" "net/http" ) /* */ import "C" var _ = cloudbilling.New var _ = http.NewRequest var _ = client.New ... this took like 10 seconds before, and now 1.3 seconds. (Still slow; disk-based caching can come later) Updates golang/go#16367 (goimports is slow) Updates golang/go#16384 (refactor TestRename is broken on Windows) Change-Id: I97e85d3016afc9f2ad5501f97babad30c7989183 Reviewed-on: https://go-review.googlesource.com/24941 Reviewed-by: Andrew Gerrand <adg@golang.org> Reviewed-by: Rob Pike <r@golang.org> Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-14 18:08:27 -06:00
if err := os.MkdirAll(bytesDir, 0755); err != nil {
t.Fatal(err)
}
bytesSrcPath := filepath.Join(bytesDir, "bytes.go")
bytesPkgPath := "bits/bytes"
bytesSrc := []byte(`package bytes
type Buffer2 struct {}
`)
cmd/goimports, imports: make goimports great again I felt the burn of my laptop on my legs, spinning away while processing goimports, and felt that it was time to make goimports great again. Over the past few years goimports fell into a slow state of disrepair with too many feature additions and no attention to the performance death by a thousand cuts. This was particularly terrible on OS X with its lackluster filesystem buffering. This CL makes goimports stronger, together with various optimizations and more visibility into what goimports is doing. * adds more internal documentation * avoids scanning $GOPATH for answers when running goimports on a file under $GOROOT (for Go core hackers) * don't read all $GOROOT & $GOPATH directories' Go code looking for their package names until much later. Require the package name of missing imports to be present in the last two directory path components. Then only try importing them in order from best to worst (shortest to longest, as before), so we can stop early. * when adding imports, add names to imports when the imported package name doesn't match the baes of its import path. For example: import foo "example.net/foo/v1" * don't read all *.go files in a package directory once the first file in a directory has revealed itself to be a package we're not looking for. For example, if we're looking for the right "client" for "client.Foo", we used to consider a directory "bar/client" as a candidate and read all 50 of its *.go files instead of stopping after its first *.go file had a "package main" line. * add some fast paths to remove allocations * add some fast paths to remove disk I/O when looking up the base package name of a standard library import (of existing imports in a file, which are very common) * adds a special case for import "C", to avoid some disk I/O. * add a -verbose flag to goimports for debugging On my Mac laptop with a huge $GOPATH, with a test file like: package foo import ( "fmt" "net/http" ) /* */ import "C" var _ = cloudbilling.New var _ = http.NewRequest var _ = client.New ... this took like 10 seconds before, and now 1.3 seconds. (Still slow; disk-based caching can come later) Updates golang/go#16367 (goimports is slow) Updates golang/go#16384 (refactor TestRename is broken on Windows) Change-Id: I97e85d3016afc9f2ad5501f97babad30c7989183 Reviewed-on: https://go-review.googlesource.com/24941 Reviewed-by: Andrew Gerrand <adg@golang.org> Reviewed-by: Rob Pike <r@golang.org> Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-14 18:08:27 -06:00
if err := ioutil.WriteFile(bytesSrcPath, bytesSrc, 0775); err != nil {
t.Fatal(err)
}
build.Default.GOROOT = goroot
cmd/goimports, imports: make goimports great again I felt the burn of my laptop on my legs, spinning away while processing goimports, and felt that it was time to make goimports great again. Over the past few years goimports fell into a slow state of disrepair with too many feature additions and no attention to the performance death by a thousand cuts. This was particularly terrible on OS X with its lackluster filesystem buffering. This CL makes goimports stronger, together with various optimizations and more visibility into what goimports is doing. * adds more internal documentation * avoids scanning $GOPATH for answers when running goimports on a file under $GOROOT (for Go core hackers) * don't read all $GOROOT & $GOPATH directories' Go code looking for their package names until much later. Require the package name of missing imports to be present in the last two directory path components. Then only try importing them in order from best to worst (shortest to longest, as before), so we can stop early. * when adding imports, add names to imports when the imported package name doesn't match the baes of its import path. For example: import foo "example.net/foo/v1" * don't read all *.go files in a package directory once the first file in a directory has revealed itself to be a package we're not looking for. For example, if we're looking for the right "client" for "client.Foo", we used to consider a directory "bar/client" as a candidate and read all 50 of its *.go files instead of stopping after its first *.go file had a "package main" line. * add some fast paths to remove allocations * add some fast paths to remove disk I/O when looking up the base package name of a standard library import (of existing imports in a file, which are very common) * adds a special case for import "C", to avoid some disk I/O. * add a -verbose flag to goimports for debugging On my Mac laptop with a huge $GOPATH, with a test file like: package foo import ( "fmt" "net/http" ) /* */ import "C" var _ = cloudbilling.New var _ = http.NewRequest var _ = client.New ... this took like 10 seconds before, and now 1.3 seconds. (Still slow; disk-based caching can come later) Updates golang/go#16367 (goimports is slow) Updates golang/go#16384 (refactor TestRename is broken on Windows) Change-Id: I97e85d3016afc9f2ad5501f97babad30c7989183 Reviewed-on: https://go-review.googlesource.com/24941 Reviewed-by: Andrew Gerrand <adg@golang.org> Reviewed-by: Rob Pike <r@golang.org> Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-14 18:08:27 -06:00
got, rename, err := findImportGoPath("bytes", map[string]bool{"Buffer2": true}, "x.go")
if err != nil {
t.Fatal(err)
}
if got != bytesPkgPath || rename {
t.Errorf(`findImportGoPath("bytes", Buffer2 ...)=%q, %t, want "%s", false`, got, rename, bytesPkgPath)
}
cmd/goimports, imports: make goimports great again I felt the burn of my laptop on my legs, spinning away while processing goimports, and felt that it was time to make goimports great again. Over the past few years goimports fell into a slow state of disrepair with too many feature additions and no attention to the performance death by a thousand cuts. This was particularly terrible on OS X with its lackluster filesystem buffering. This CL makes goimports stronger, together with various optimizations and more visibility into what goimports is doing. * adds more internal documentation * avoids scanning $GOPATH for answers when running goimports on a file under $GOROOT (for Go core hackers) * don't read all $GOROOT & $GOPATH directories' Go code looking for their package names until much later. Require the package name of missing imports to be present in the last two directory path components. Then only try importing them in order from best to worst (shortest to longest, as before), so we can stop early. * when adding imports, add names to imports when the imported package name doesn't match the baes of its import path. For example: import foo "example.net/foo/v1" * don't read all *.go files in a package directory once the first file in a directory has revealed itself to be a package we're not looking for. For example, if we're looking for the right "client" for "client.Foo", we used to consider a directory "bar/client" as a candidate and read all 50 of its *.go files instead of stopping after its first *.go file had a "package main" line. * add some fast paths to remove allocations * add some fast paths to remove disk I/O when looking up the base package name of a standard library import (of existing imports in a file, which are very common) * adds a special case for import "C", to avoid some disk I/O. * add a -verbose flag to goimports for debugging On my Mac laptop with a huge $GOPATH, with a test file like: package foo import ( "fmt" "net/http" ) /* */ import "C" var _ = cloudbilling.New var _ = http.NewRequest var _ = client.New ... this took like 10 seconds before, and now 1.3 seconds. (Still slow; disk-based caching can come later) Updates golang/go#16367 (goimports is slow) Updates golang/go#16384 (refactor TestRename is broken on Windows) Change-Id: I97e85d3016afc9f2ad5501f97babad30c7989183 Reviewed-on: https://go-review.googlesource.com/24941 Reviewed-by: Andrew Gerrand <adg@golang.org> Reviewed-by: Rob Pike <r@golang.org> Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-14 18:08:27 -06:00
got, rename, err = findImportGoPath("bytes", map[string]bool{"Missing": true}, "x.go")
if err != nil {
t.Fatal(err)
}
if got != "" || rename {
t.Errorf(`findImportGoPath("bytes", Missing ...)=%q, %t, want "", false`, got, rename)
}
})
}
func init() {
inTests = true
}
func withEmptyGoPath(fn func()) {
testMu.Lock()
cmd/goimports, imports: make goimports great again I felt the burn of my laptop on my legs, spinning away while processing goimports, and felt that it was time to make goimports great again. Over the past few years goimports fell into a slow state of disrepair with too many feature additions and no attention to the performance death by a thousand cuts. This was particularly terrible on OS X with its lackluster filesystem buffering. This CL makes goimports stronger, together with various optimizations and more visibility into what goimports is doing. * adds more internal documentation * avoids scanning $GOPATH for answers when running goimports on a file under $GOROOT (for Go core hackers) * don't read all $GOROOT & $GOPATH directories' Go code looking for their package names until much later. Require the package name of missing imports to be present in the last two directory path components. Then only try importing them in order from best to worst (shortest to longest, as before), so we can stop early. * when adding imports, add names to imports when the imported package name doesn't match the baes of its import path. For example: import foo "example.net/foo/v1" * don't read all *.go files in a package directory once the first file in a directory has revealed itself to be a package we're not looking for. For example, if we're looking for the right "client" for "client.Foo", we used to consider a directory "bar/client" as a candidate and read all 50 of its *.go files instead of stopping after its first *.go file had a "package main" line. * add some fast paths to remove allocations * add some fast paths to remove disk I/O when looking up the base package name of a standard library import (of existing imports in a file, which are very common) * adds a special case for import "C", to avoid some disk I/O. * add a -verbose flag to goimports for debugging On my Mac laptop with a huge $GOPATH, with a test file like: package foo import ( "fmt" "net/http" ) /* */ import "C" var _ = cloudbilling.New var _ = http.NewRequest var _ = client.New ... this took like 10 seconds before, and now 1.3 seconds. (Still slow; disk-based caching can come later) Updates golang/go#16367 (goimports is slow) Updates golang/go#16384 (refactor TestRename is broken on Windows) Change-Id: I97e85d3016afc9f2ad5501f97babad30c7989183 Reviewed-on: https://go-review.googlesource.com/24941 Reviewed-by: Andrew Gerrand <adg@golang.org> Reviewed-by: Rob Pike <r@golang.org> Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-14 18:08:27 -06:00
dirScanMu.Lock()
populateIgnoreOnce = sync.Once{}
scanGoRootOnce = sync.Once{}
scanGoPathOnce = sync.Once{}
cmd/goimports, imports: make goimports great again I felt the burn of my laptop on my legs, spinning away while processing goimports, and felt that it was time to make goimports great again. Over the past few years goimports fell into a slow state of disrepair with too many feature additions and no attention to the performance death by a thousand cuts. This was particularly terrible on OS X with its lackluster filesystem buffering. This CL makes goimports stronger, together with various optimizations and more visibility into what goimports is doing. * adds more internal documentation * avoids scanning $GOPATH for answers when running goimports on a file under $GOROOT (for Go core hackers) * don't read all $GOROOT & $GOPATH directories' Go code looking for their package names until much later. Require the package name of missing imports to be present in the last two directory path components. Then only try importing them in order from best to worst (shortest to longest, as before), so we can stop early. * when adding imports, add names to imports when the imported package name doesn't match the baes of its import path. For example: import foo "example.net/foo/v1" * don't read all *.go files in a package directory once the first file in a directory has revealed itself to be a package we're not looking for. For example, if we're looking for the right "client" for "client.Foo", we used to consider a directory "bar/client" as a candidate and read all 50 of its *.go files instead of stopping after its first *.go file had a "package main" line. * add some fast paths to remove allocations * add some fast paths to remove disk I/O when looking up the base package name of a standard library import (of existing imports in a file, which are very common) * adds a special case for import "C", to avoid some disk I/O. * add a -verbose flag to goimports for debugging On my Mac laptop with a huge $GOPATH, with a test file like: package foo import ( "fmt" "net/http" ) /* */ import "C" var _ = cloudbilling.New var _ = http.NewRequest var _ = client.New ... this took like 10 seconds before, and now 1.3 seconds. (Still slow; disk-based caching can come later) Updates golang/go#16367 (goimports is slow) Updates golang/go#16384 (refactor TestRename is broken on Windows) Change-Id: I97e85d3016afc9f2ad5501f97babad30c7989183 Reviewed-on: https://go-review.googlesource.com/24941 Reviewed-by: Andrew Gerrand <adg@golang.org> Reviewed-by: Rob Pike <r@golang.org> Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-14 18:08:27 -06:00
dirScan = nil
ignoredDirs = nil
scanGoRootDone = make(chan struct{})
cmd/goimports, imports: make goimports great again I felt the burn of my laptop on my legs, spinning away while processing goimports, and felt that it was time to make goimports great again. Over the past few years goimports fell into a slow state of disrepair with too many feature additions and no attention to the performance death by a thousand cuts. This was particularly terrible on OS X with its lackluster filesystem buffering. This CL makes goimports stronger, together with various optimizations and more visibility into what goimports is doing. * adds more internal documentation * avoids scanning $GOPATH for answers when running goimports on a file under $GOROOT (for Go core hackers) * don't read all $GOROOT & $GOPATH directories' Go code looking for their package names until much later. Require the package name of missing imports to be present in the last two directory path components. Then only try importing them in order from best to worst (shortest to longest, as before), so we can stop early. * when adding imports, add names to imports when the imported package name doesn't match the baes of its import path. For example: import foo "example.net/foo/v1" * don't read all *.go files in a package directory once the first file in a directory has revealed itself to be a package we're not looking for. For example, if we're looking for the right "client" for "client.Foo", we used to consider a directory "bar/client" as a candidate and read all 50 of its *.go files instead of stopping after its first *.go file had a "package main" line. * add some fast paths to remove allocations * add some fast paths to remove disk I/O when looking up the base package name of a standard library import (of existing imports in a file, which are very common) * adds a special case for import "C", to avoid some disk I/O. * add a -verbose flag to goimports for debugging On my Mac laptop with a huge $GOPATH, with a test file like: package foo import ( "fmt" "net/http" ) /* */ import "C" var _ = cloudbilling.New var _ = http.NewRequest var _ = client.New ... this took like 10 seconds before, and now 1.3 seconds. (Still slow; disk-based caching can come later) Updates golang/go#16367 (goimports is slow) Updates golang/go#16384 (refactor TestRename is broken on Windows) Change-Id: I97e85d3016afc9f2ad5501f97babad30c7989183 Reviewed-on: https://go-review.googlesource.com/24941 Reviewed-by: Andrew Gerrand <adg@golang.org> Reviewed-by: Rob Pike <r@golang.org> Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-14 18:08:27 -06:00
dirScanMu.Unlock()
oldGOPATH := build.Default.GOPATH
cmd/goimports, imports: make goimports great again I felt the burn of my laptop on my legs, spinning away while processing goimports, and felt that it was time to make goimports great again. Over the past few years goimports fell into a slow state of disrepair with too many feature additions and no attention to the performance death by a thousand cuts. This was particularly terrible on OS X with its lackluster filesystem buffering. This CL makes goimports stronger, together with various optimizations and more visibility into what goimports is doing. * adds more internal documentation * avoids scanning $GOPATH for answers when running goimports on a file under $GOROOT (for Go core hackers) * don't read all $GOROOT & $GOPATH directories' Go code looking for their package names until much later. Require the package name of missing imports to be present in the last two directory path components. Then only try importing them in order from best to worst (shortest to longest, as before), so we can stop early. * when adding imports, add names to imports when the imported package name doesn't match the baes of its import path. For example: import foo "example.net/foo/v1" * don't read all *.go files in a package directory once the first file in a directory has revealed itself to be a package we're not looking for. For example, if we're looking for the right "client" for "client.Foo", we used to consider a directory "bar/client" as a candidate and read all 50 of its *.go files instead of stopping after its first *.go file had a "package main" line. * add some fast paths to remove allocations * add some fast paths to remove disk I/O when looking up the base package name of a standard library import (of existing imports in a file, which are very common) * adds a special case for import "C", to avoid some disk I/O. * add a -verbose flag to goimports for debugging On my Mac laptop with a huge $GOPATH, with a test file like: package foo import ( "fmt" "net/http" ) /* */ import "C" var _ = cloudbilling.New var _ = http.NewRequest var _ = client.New ... this took like 10 seconds before, and now 1.3 seconds. (Still slow; disk-based caching can come later) Updates golang/go#16367 (goimports is slow) Updates golang/go#16384 (refactor TestRename is broken on Windows) Change-Id: I97e85d3016afc9f2ad5501f97babad30c7989183 Reviewed-on: https://go-review.googlesource.com/24941 Reviewed-by: Andrew Gerrand <adg@golang.org> Reviewed-by: Rob Pike <r@golang.org> Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-14 18:08:27 -06:00
oldGOROOT := build.Default.GOROOT
build.Default.GOPATH = ""
cmd/goimports, imports: make goimports great again I felt the burn of my laptop on my legs, spinning away while processing goimports, and felt that it was time to make goimports great again. Over the past few years goimports fell into a slow state of disrepair with too many feature additions and no attention to the performance death by a thousand cuts. This was particularly terrible on OS X with its lackluster filesystem buffering. This CL makes goimports stronger, together with various optimizations and more visibility into what goimports is doing. * adds more internal documentation * avoids scanning $GOPATH for answers when running goimports on a file under $GOROOT (for Go core hackers) * don't read all $GOROOT & $GOPATH directories' Go code looking for their package names until much later. Require the package name of missing imports to be present in the last two directory path components. Then only try importing them in order from best to worst (shortest to longest, as before), so we can stop early. * when adding imports, add names to imports when the imported package name doesn't match the baes of its import path. For example: import foo "example.net/foo/v1" * don't read all *.go files in a package directory once the first file in a directory has revealed itself to be a package we're not looking for. For example, if we're looking for the right "client" for "client.Foo", we used to consider a directory "bar/client" as a candidate and read all 50 of its *.go files instead of stopping after its first *.go file had a "package main" line. * add some fast paths to remove allocations * add some fast paths to remove disk I/O when looking up the base package name of a standard library import (of existing imports in a file, which are very common) * adds a special case for import "C", to avoid some disk I/O. * add a -verbose flag to goimports for debugging On my Mac laptop with a huge $GOPATH, with a test file like: package foo import ( "fmt" "net/http" ) /* */ import "C" var _ = cloudbilling.New var _ = http.NewRequest var _ = client.New ... this took like 10 seconds before, and now 1.3 seconds. (Still slow; disk-based caching can come later) Updates golang/go#16367 (goimports is slow) Updates golang/go#16384 (refactor TestRename is broken on Windows) Change-Id: I97e85d3016afc9f2ad5501f97babad30c7989183 Reviewed-on: https://go-review.googlesource.com/24941 Reviewed-by: Andrew Gerrand <adg@golang.org> Reviewed-by: Rob Pike <r@golang.org> Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-14 18:08:27 -06:00
visitedSymlinks.m = nil
testHookScanDir = func(string) {}
testMu.Unlock()
defer func() {
testMu.Lock()
cmd/goimports, imports: make goimports great again I felt the burn of my laptop on my legs, spinning away while processing goimports, and felt that it was time to make goimports great again. Over the past few years goimports fell into a slow state of disrepair with too many feature additions and no attention to the performance death by a thousand cuts. This was particularly terrible on OS X with its lackluster filesystem buffering. This CL makes goimports stronger, together with various optimizations and more visibility into what goimports is doing. * adds more internal documentation * avoids scanning $GOPATH for answers when running goimports on a file under $GOROOT (for Go core hackers) * don't read all $GOROOT & $GOPATH directories' Go code looking for their package names until much later. Require the package name of missing imports to be present in the last two directory path components. Then only try importing them in order from best to worst (shortest to longest, as before), so we can stop early. * when adding imports, add names to imports when the imported package name doesn't match the baes of its import path. For example: import foo "example.net/foo/v1" * don't read all *.go files in a package directory once the first file in a directory has revealed itself to be a package we're not looking for. For example, if we're looking for the right "client" for "client.Foo", we used to consider a directory "bar/client" as a candidate and read all 50 of its *.go files instead of stopping after its first *.go file had a "package main" line. * add some fast paths to remove allocations * add some fast paths to remove disk I/O when looking up the base package name of a standard library import (of existing imports in a file, which are very common) * adds a special case for import "C", to avoid some disk I/O. * add a -verbose flag to goimports for debugging On my Mac laptop with a huge $GOPATH, with a test file like: package foo import ( "fmt" "net/http" ) /* */ import "C" var _ = cloudbilling.New var _ = http.NewRequest var _ = client.New ... this took like 10 seconds before, and now 1.3 seconds. (Still slow; disk-based caching can come later) Updates golang/go#16367 (goimports is slow) Updates golang/go#16384 (refactor TestRename is broken on Windows) Change-Id: I97e85d3016afc9f2ad5501f97babad30c7989183 Reviewed-on: https://go-review.googlesource.com/24941 Reviewed-by: Andrew Gerrand <adg@golang.org> Reviewed-by: Rob Pike <r@golang.org> Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-14 18:08:27 -06:00
testHookScanDir = func(string) {}
build.Default.GOPATH = oldGOPATH
cmd/goimports, imports: make goimports great again I felt the burn of my laptop on my legs, spinning away while processing goimports, and felt that it was time to make goimports great again. Over the past few years goimports fell into a slow state of disrepair with too many feature additions and no attention to the performance death by a thousand cuts. This was particularly terrible on OS X with its lackluster filesystem buffering. This CL makes goimports stronger, together with various optimizations and more visibility into what goimports is doing. * adds more internal documentation * avoids scanning $GOPATH for answers when running goimports on a file under $GOROOT (for Go core hackers) * don't read all $GOROOT & $GOPATH directories' Go code looking for their package names until much later. Require the package name of missing imports to be present in the last two directory path components. Then only try importing them in order from best to worst (shortest to longest, as before), so we can stop early. * when adding imports, add names to imports when the imported package name doesn't match the baes of its import path. For example: import foo "example.net/foo/v1" * don't read all *.go files in a package directory once the first file in a directory has revealed itself to be a package we're not looking for. For example, if we're looking for the right "client" for "client.Foo", we used to consider a directory "bar/client" as a candidate and read all 50 of its *.go files instead of stopping after its first *.go file had a "package main" line. * add some fast paths to remove allocations * add some fast paths to remove disk I/O when looking up the base package name of a standard library import (of existing imports in a file, which are very common) * adds a special case for import "C", to avoid some disk I/O. * add a -verbose flag to goimports for debugging On my Mac laptop with a huge $GOPATH, with a test file like: package foo import ( "fmt" "net/http" ) /* */ import "C" var _ = cloudbilling.New var _ = http.NewRequest var _ = client.New ... this took like 10 seconds before, and now 1.3 seconds. (Still slow; disk-based caching can come later) Updates golang/go#16367 (goimports is slow) Updates golang/go#16384 (refactor TestRename is broken on Windows) Change-Id: I97e85d3016afc9f2ad5501f97babad30c7989183 Reviewed-on: https://go-review.googlesource.com/24941 Reviewed-by: Andrew Gerrand <adg@golang.org> Reviewed-by: Rob Pike <r@golang.org> Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-14 18:08:27 -06:00
build.Default.GOROOT = oldGOROOT
testMu.Unlock()
}()
fn()
}
func TestFindImportInternal(t *testing.T) {
withEmptyGoPath(func() {
// Check for src/internal/race, not just src/internal,
// so that we can run this test also against go1.5
// (which doesn't contain that file).
_, err := os.Stat(filepath.Join(runtime.GOROOT(), "src/internal/race"))
if err != nil {
t.Skip(err)
}
got, rename, err := findImportGoPath("race", map[string]bool{"Acquire": true}, filepath.Join(runtime.GOROOT(), "src/math/x.go"))
if err != nil {
t.Fatal(err)
}
if got != "internal/race" || rename {
cmd/goimports, imports: make goimports great again I felt the burn of my laptop on my legs, spinning away while processing goimports, and felt that it was time to make goimports great again. Over the past few years goimports fell into a slow state of disrepair with too many feature additions and no attention to the performance death by a thousand cuts. This was particularly terrible on OS X with its lackluster filesystem buffering. This CL makes goimports stronger, together with various optimizations and more visibility into what goimports is doing. * adds more internal documentation * avoids scanning $GOPATH for answers when running goimports on a file under $GOROOT (for Go core hackers) * don't read all $GOROOT & $GOPATH directories' Go code looking for their package names until much later. Require the package name of missing imports to be present in the last two directory path components. Then only try importing them in order from best to worst (shortest to longest, as before), so we can stop early. * when adding imports, add names to imports when the imported package name doesn't match the baes of its import path. For example: import foo "example.net/foo/v1" * don't read all *.go files in a package directory once the first file in a directory has revealed itself to be a package we're not looking for. For example, if we're looking for the right "client" for "client.Foo", we used to consider a directory "bar/client" as a candidate and read all 50 of its *.go files instead of stopping after its first *.go file had a "package main" line. * add some fast paths to remove allocations * add some fast paths to remove disk I/O when looking up the base package name of a standard library import (of existing imports in a file, which are very common) * adds a special case for import "C", to avoid some disk I/O. * add a -verbose flag to goimports for debugging On my Mac laptop with a huge $GOPATH, with a test file like: package foo import ( "fmt" "net/http" ) /* */ import "C" var _ = cloudbilling.New var _ = http.NewRequest var _ = client.New ... this took like 10 seconds before, and now 1.3 seconds. (Still slow; disk-based caching can come later) Updates golang/go#16367 (goimports is slow) Updates golang/go#16384 (refactor TestRename is broken on Windows) Change-Id: I97e85d3016afc9f2ad5501f97babad30c7989183 Reviewed-on: https://go-review.googlesource.com/24941 Reviewed-by: Andrew Gerrand <adg@golang.org> Reviewed-by: Rob Pike <r@golang.org> Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-14 18:08:27 -06:00
t.Errorf(`findImportGoPath("race", Acquire ...) = %q, %t; want "internal/race", false`, got, rename)
}
// should not be able to use internal from outside that tree
got, rename, err = findImportGoPath("race", map[string]bool{"Acquire": true}, filepath.Join(runtime.GOROOT(), "x.go"))
if err != nil {
t.Fatal(err)
}
if got != "" || rename {
t.Errorf(`findImportGoPath("race", Acquire ...)=%q, %t, want "", false`, got, rename)
}
})
}
// rand.Read should prefer crypto/rand.Read, not math/rand.Read.
func TestFindImportRandRead(t *testing.T) {
withEmptyGoPath(func() {
file := filepath.Join(runtime.GOROOT(), "src/foo/x.go") // dummy
tests := []struct {
syms []string
want string
}{
{
syms: []string{"Read"},
want: "crypto/rand",
},
{
syms: []string{"Read", "NewZipf"},
want: "math/rand",
},
{
syms: []string{"NewZipf"},
want: "math/rand",
},
{
syms: []string{"Read", "Prime"},
want: "crypto/rand",
},
}
for _, tt := range tests {
m := map[string]bool{}
for _, sym := range tt.syms {
m[sym] = true
}
got, _, err := findImportGoPath("rand", m, file)
if err != nil {
t.Errorf("for %q: %v", tt.syms, err)
continue
}
if got != tt.want {
t.Errorf("for %q, findImportGoPath = %q; want %q", tt.syms, got, tt.want)
}
}
})
}
func TestFindImportVendor(t *testing.T) {
cmd/goimports, imports: make goimports great again I felt the burn of my laptop on my legs, spinning away while processing goimports, and felt that it was time to make goimports great again. Over the past few years goimports fell into a slow state of disrepair with too many feature additions and no attention to the performance death by a thousand cuts. This was particularly terrible on OS X with its lackluster filesystem buffering. This CL makes goimports stronger, together with various optimizations and more visibility into what goimports is doing. * adds more internal documentation * avoids scanning $GOPATH for answers when running goimports on a file under $GOROOT (for Go core hackers) * don't read all $GOROOT & $GOPATH directories' Go code looking for their package names until much later. Require the package name of missing imports to be present in the last two directory path components. Then only try importing them in order from best to worst (shortest to longest, as before), so we can stop early. * when adding imports, add names to imports when the imported package name doesn't match the baes of its import path. For example: import foo "example.net/foo/v1" * don't read all *.go files in a package directory once the first file in a directory has revealed itself to be a package we're not looking for. For example, if we're looking for the right "client" for "client.Foo", we used to consider a directory "bar/client" as a candidate and read all 50 of its *.go files instead of stopping after its first *.go file had a "package main" line. * add some fast paths to remove allocations * add some fast paths to remove disk I/O when looking up the base package name of a standard library import (of existing imports in a file, which are very common) * adds a special case for import "C", to avoid some disk I/O. * add a -verbose flag to goimports for debugging On my Mac laptop with a huge $GOPATH, with a test file like: package foo import ( "fmt" "net/http" ) /* */ import "C" var _ = cloudbilling.New var _ = http.NewRequest var _ = client.New ... this took like 10 seconds before, and now 1.3 seconds. (Still slow; disk-based caching can come later) Updates golang/go#16367 (goimports is slow) Updates golang/go#16384 (refactor TestRename is broken on Windows) Change-Id: I97e85d3016afc9f2ad5501f97babad30c7989183 Reviewed-on: https://go-review.googlesource.com/24941 Reviewed-by: Andrew Gerrand <adg@golang.org> Reviewed-by: Rob Pike <r@golang.org> Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-14 18:08:27 -06:00
testConfig{
gorootFiles: map[string]string{
"vendor/golang.org/x/net/http2/hpack/huffman.go": "package hpack\nfunc HuffmanDecode() { }\n",
},
}.test(t, func(t *goimportTest) {
got, rename, err := findImportGoPath("hpack", map[string]bool{"HuffmanDecode": true}, filepath.Join(t.goroot, "src/math/x.go"))
if err != nil {
t.Fatal(err)
}
want := "golang.org/x/net/http2/hpack"
if got != want || rename {
t.Errorf(`findImportGoPath("hpack", HuffmanDecode ...) = %q, %t; want %q, false`, got, rename, want)
}
})
}
func TestProcessVendor(t *testing.T) {
cmd/goimports, imports: make goimports great again I felt the burn of my laptop on my legs, spinning away while processing goimports, and felt that it was time to make goimports great again. Over the past few years goimports fell into a slow state of disrepair with too many feature additions and no attention to the performance death by a thousand cuts. This was particularly terrible on OS X with its lackluster filesystem buffering. This CL makes goimports stronger, together with various optimizations and more visibility into what goimports is doing. * adds more internal documentation * avoids scanning $GOPATH for answers when running goimports on a file under $GOROOT (for Go core hackers) * don't read all $GOROOT & $GOPATH directories' Go code looking for their package names until much later. Require the package name of missing imports to be present in the last two directory path components. Then only try importing them in order from best to worst (shortest to longest, as before), so we can stop early. * when adding imports, add names to imports when the imported package name doesn't match the baes of its import path. For example: import foo "example.net/foo/v1" * don't read all *.go files in a package directory once the first file in a directory has revealed itself to be a package we're not looking for. For example, if we're looking for the right "client" for "client.Foo", we used to consider a directory "bar/client" as a candidate and read all 50 of its *.go files instead of stopping after its first *.go file had a "package main" line. * add some fast paths to remove allocations * add some fast paths to remove disk I/O when looking up the base package name of a standard library import (of existing imports in a file, which are very common) * adds a special case for import "C", to avoid some disk I/O. * add a -verbose flag to goimports for debugging On my Mac laptop with a huge $GOPATH, with a test file like: package foo import ( "fmt" "net/http" ) /* */ import "C" var _ = cloudbilling.New var _ = http.NewRequest var _ = client.New ... this took like 10 seconds before, and now 1.3 seconds. (Still slow; disk-based caching can come later) Updates golang/go#16367 (goimports is slow) Updates golang/go#16384 (refactor TestRename is broken on Windows) Change-Id: I97e85d3016afc9f2ad5501f97babad30c7989183 Reviewed-on: https://go-review.googlesource.com/24941 Reviewed-by: Andrew Gerrand <adg@golang.org> Reviewed-by: Rob Pike <r@golang.org> Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-14 18:08:27 -06:00
withEmptyGoPath(func() {
_, err := os.Stat(filepath.Join(runtime.GOROOT(), "src/vendor"))
if err != nil {
t.Skip(err)
}
cmd/goimports, imports: make goimports great again I felt the burn of my laptop on my legs, spinning away while processing goimports, and felt that it was time to make goimports great again. Over the past few years goimports fell into a slow state of disrepair with too many feature additions and no attention to the performance death by a thousand cuts. This was particularly terrible on OS X with its lackluster filesystem buffering. This CL makes goimports stronger, together with various optimizations and more visibility into what goimports is doing. * adds more internal documentation * avoids scanning $GOPATH for answers when running goimports on a file under $GOROOT (for Go core hackers) * don't read all $GOROOT & $GOPATH directories' Go code looking for their package names until much later. Require the package name of missing imports to be present in the last two directory path components. Then only try importing them in order from best to worst (shortest to longest, as before), so we can stop early. * when adding imports, add names to imports when the imported package name doesn't match the baes of its import path. For example: import foo "example.net/foo/v1" * don't read all *.go files in a package directory once the first file in a directory has revealed itself to be a package we're not looking for. For example, if we're looking for the right "client" for "client.Foo", we used to consider a directory "bar/client" as a candidate and read all 50 of its *.go files instead of stopping after its first *.go file had a "package main" line. * add some fast paths to remove allocations * add some fast paths to remove disk I/O when looking up the base package name of a standard library import (of existing imports in a file, which are very common) * adds a special case for import "C", to avoid some disk I/O. * add a -verbose flag to goimports for debugging On my Mac laptop with a huge $GOPATH, with a test file like: package foo import ( "fmt" "net/http" ) /* */ import "C" var _ = cloudbilling.New var _ = http.NewRequest var _ = client.New ... this took like 10 seconds before, and now 1.3 seconds. (Still slow; disk-based caching can come later) Updates golang/go#16367 (goimports is slow) Updates golang/go#16384 (refactor TestRename is broken on Windows) Change-Id: I97e85d3016afc9f2ad5501f97babad30c7989183 Reviewed-on: https://go-review.googlesource.com/24941 Reviewed-by: Andrew Gerrand <adg@golang.org> Reviewed-by: Rob Pike <r@golang.org> Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-14 18:08:27 -06:00
target := filepath.Join(runtime.GOROOT(), "src/math/x.go")
out, err := Process(target, []byte("package http\nimport \"bytes\"\nfunc f() { strings.NewReader(); hpack.HuffmanDecode() }\n"), nil)
cmd/goimports, imports: make goimports great again I felt the burn of my laptop on my legs, spinning away while processing goimports, and felt that it was time to make goimports great again. Over the past few years goimports fell into a slow state of disrepair with too many feature additions and no attention to the performance death by a thousand cuts. This was particularly terrible on OS X with its lackluster filesystem buffering. This CL makes goimports stronger, together with various optimizations and more visibility into what goimports is doing. * adds more internal documentation * avoids scanning $GOPATH for answers when running goimports on a file under $GOROOT (for Go core hackers) * don't read all $GOROOT & $GOPATH directories' Go code looking for their package names until much later. Require the package name of missing imports to be present in the last two directory path components. Then only try importing them in order from best to worst (shortest to longest, as before), so we can stop early. * when adding imports, add names to imports when the imported package name doesn't match the baes of its import path. For example: import foo "example.net/foo/v1" * don't read all *.go files in a package directory once the first file in a directory has revealed itself to be a package we're not looking for. For example, if we're looking for the right "client" for "client.Foo", we used to consider a directory "bar/client" as a candidate and read all 50 of its *.go files instead of stopping after its first *.go file had a "package main" line. * add some fast paths to remove allocations * add some fast paths to remove disk I/O when looking up the base package name of a standard library import (of existing imports in a file, which are very common) * adds a special case for import "C", to avoid some disk I/O. * add a -verbose flag to goimports for debugging On my Mac laptop with a huge $GOPATH, with a test file like: package foo import ( "fmt" "net/http" ) /* */ import "C" var _ = cloudbilling.New var _ = http.NewRequest var _ = client.New ... this took like 10 seconds before, and now 1.3 seconds. (Still slow; disk-based caching can come later) Updates golang/go#16367 (goimports is slow) Updates golang/go#16384 (refactor TestRename is broken on Windows) Change-Id: I97e85d3016afc9f2ad5501f97babad30c7989183 Reviewed-on: https://go-review.googlesource.com/24941 Reviewed-by: Andrew Gerrand <adg@golang.org> Reviewed-by: Rob Pike <r@golang.org> Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-14 18:08:27 -06:00
if err != nil {
t.Fatal(err)
}
cmd/goimports, imports: make goimports great again I felt the burn of my laptop on my legs, spinning away while processing goimports, and felt that it was time to make goimports great again. Over the past few years goimports fell into a slow state of disrepair with too many feature additions and no attention to the performance death by a thousand cuts. This was particularly terrible on OS X with its lackluster filesystem buffering. This CL makes goimports stronger, together with various optimizations and more visibility into what goimports is doing. * adds more internal documentation * avoids scanning $GOPATH for answers when running goimports on a file under $GOROOT (for Go core hackers) * don't read all $GOROOT & $GOPATH directories' Go code looking for their package names until much later. Require the package name of missing imports to be present in the last two directory path components. Then only try importing them in order from best to worst (shortest to longest, as before), so we can stop early. * when adding imports, add names to imports when the imported package name doesn't match the baes of its import path. For example: import foo "example.net/foo/v1" * don't read all *.go files in a package directory once the first file in a directory has revealed itself to be a package we're not looking for. For example, if we're looking for the right "client" for "client.Foo", we used to consider a directory "bar/client" as a candidate and read all 50 of its *.go files instead of stopping after its first *.go file had a "package main" line. * add some fast paths to remove allocations * add some fast paths to remove disk I/O when looking up the base package name of a standard library import (of existing imports in a file, which are very common) * adds a special case for import "C", to avoid some disk I/O. * add a -verbose flag to goimports for debugging On my Mac laptop with a huge $GOPATH, with a test file like: package foo import ( "fmt" "net/http" ) /* */ import "C" var _ = cloudbilling.New var _ = http.NewRequest var _ = client.New ... this took like 10 seconds before, and now 1.3 seconds. (Still slow; disk-based caching can come later) Updates golang/go#16367 (goimports is slow) Updates golang/go#16384 (refactor TestRename is broken on Windows) Change-Id: I97e85d3016afc9f2ad5501f97babad30c7989183 Reviewed-on: https://go-review.googlesource.com/24941 Reviewed-by: Andrew Gerrand <adg@golang.org> Reviewed-by: Rob Pike <r@golang.org> Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-14 18:08:27 -06:00
want := "golang_org/x/net/http2/hpack"
if _, err := os.Stat(filepath.Join(runtime.GOROOT(), "src/vendor", want)); os.IsNotExist(err) {
want = "golang.org/x/net/http2/hpack"
}
if !bytes.Contains(out, []byte(want)) {
t.Fatalf("Process(%q) did not add expected hpack import %q; got:\n%s", target, want, out)
}
})
}
func TestFindImportStdlib(t *testing.T) {
tests := []struct {
pkg string
symbols []string
want string
}{
{"http", []string{"Get"}, "net/http"},
{"http", []string{"Get", "Post"}, "net/http"},
{"http", []string{"Get", "Foo"}, ""},
{"bytes", []string{"Buffer"}, "bytes"},
{"ioutil", []string{"Discard"}, "io/ioutil"},
}
for _, tt := range tests {
got, rename, ok := findImportStdlib(tt.pkg, strSet(tt.symbols))
if (got != "") != ok {
t.Error("findImportStdlib return value inconsistent")
}
if got != tt.want || rename {
t.Errorf("findImportStdlib(%q, %q) = %q, %t; want %q, false", tt.pkg, tt.symbols, got, rename, tt.want)
}
}
}
cmd/goimports, imports: make goimports great again I felt the burn of my laptop on my legs, spinning away while processing goimports, and felt that it was time to make goimports great again. Over the past few years goimports fell into a slow state of disrepair with too many feature additions and no attention to the performance death by a thousand cuts. This was particularly terrible on OS X with its lackluster filesystem buffering. This CL makes goimports stronger, together with various optimizations and more visibility into what goimports is doing. * adds more internal documentation * avoids scanning $GOPATH for answers when running goimports on a file under $GOROOT (for Go core hackers) * don't read all $GOROOT & $GOPATH directories' Go code looking for their package names until much later. Require the package name of missing imports to be present in the last two directory path components. Then only try importing them in order from best to worst (shortest to longest, as before), so we can stop early. * when adding imports, add names to imports when the imported package name doesn't match the baes of its import path. For example: import foo "example.net/foo/v1" * don't read all *.go files in a package directory once the first file in a directory has revealed itself to be a package we're not looking for. For example, if we're looking for the right "client" for "client.Foo", we used to consider a directory "bar/client" as a candidate and read all 50 of its *.go files instead of stopping after its first *.go file had a "package main" line. * add some fast paths to remove allocations * add some fast paths to remove disk I/O when looking up the base package name of a standard library import (of existing imports in a file, which are very common) * adds a special case for import "C", to avoid some disk I/O. * add a -verbose flag to goimports for debugging On my Mac laptop with a huge $GOPATH, with a test file like: package foo import ( "fmt" "net/http" ) /* */ import "C" var _ = cloudbilling.New var _ = http.NewRequest var _ = client.New ... this took like 10 seconds before, and now 1.3 seconds. (Still slow; disk-based caching can come later) Updates golang/go#16367 (goimports is slow) Updates golang/go#16384 (refactor TestRename is broken on Windows) Change-Id: I97e85d3016afc9f2ad5501f97babad30c7989183 Reviewed-on: https://go-review.googlesource.com/24941 Reviewed-by: Andrew Gerrand <adg@golang.org> Reviewed-by: Rob Pike <r@golang.org> Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-14 18:08:27 -06:00
type testConfig struct {
// goroot and gopath optionally specifies the path on disk
// to use for the GOROOT and GOPATH. If empty, a temp directory
// is made if needed.
goroot, gopath string
cmd/goimports, imports: make goimports great again I felt the burn of my laptop on my legs, spinning away while processing goimports, and felt that it was time to make goimports great again. Over the past few years goimports fell into a slow state of disrepair with too many feature additions and no attention to the performance death by a thousand cuts. This was particularly terrible on OS X with its lackluster filesystem buffering. This CL makes goimports stronger, together with various optimizations and more visibility into what goimports is doing. * adds more internal documentation * avoids scanning $GOPATH for answers when running goimports on a file under $GOROOT (for Go core hackers) * don't read all $GOROOT & $GOPATH directories' Go code looking for their package names until much later. Require the package name of missing imports to be present in the last two directory path components. Then only try importing them in order from best to worst (shortest to longest, as before), so we can stop early. * when adding imports, add names to imports when the imported package name doesn't match the baes of its import path. For example: import foo "example.net/foo/v1" * don't read all *.go files in a package directory once the first file in a directory has revealed itself to be a package we're not looking for. For example, if we're looking for the right "client" for "client.Foo", we used to consider a directory "bar/client" as a candidate and read all 50 of its *.go files instead of stopping after its first *.go file had a "package main" line. * add some fast paths to remove allocations * add some fast paths to remove disk I/O when looking up the base package name of a standard library import (of existing imports in a file, which are very common) * adds a special case for import "C", to avoid some disk I/O. * add a -verbose flag to goimports for debugging On my Mac laptop with a huge $GOPATH, with a test file like: package foo import ( "fmt" "net/http" ) /* */ import "C" var _ = cloudbilling.New var _ = http.NewRequest var _ = client.New ... this took like 10 seconds before, and now 1.3 seconds. (Still slow; disk-based caching can come later) Updates golang/go#16367 (goimports is slow) Updates golang/go#16384 (refactor TestRename is broken on Windows) Change-Id: I97e85d3016afc9f2ad5501f97babad30c7989183 Reviewed-on: https://go-review.googlesource.com/24941 Reviewed-by: Andrew Gerrand <adg@golang.org> Reviewed-by: Rob Pike <r@golang.org> Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-14 18:08:27 -06:00
// gorootFiles optionally specifies the complete contents of GOROOT to use,
// If nil, the normal current $GOROOT is used.
gorootFiles map[string]string // paths relative to $GOROOT/src to contents
// gopathFiles is like gorootFiles, but for $GOPATH.
// If nil, there is no GOPATH, though.
gopathFiles map[string]string // paths relative to $GOPATH/src to contents
}
func mustTempDir(t *testing.T, prefix string) string {
dir, err := ioutil.TempDir("", prefix)
if err != nil {
t.Fatal(err)
}
return dir
}
func mapToDir(destDir string, files map[string]string) error {
for path, contents := range files {
file := filepath.Join(destDir, "src", path)
if err := os.MkdirAll(filepath.Dir(file), 0755); err != nil {
return err
}
var err error
if strings.HasPrefix(contents, "LINK:") {
err = os.Symlink(strings.TrimPrefix(contents, "LINK:"), file)
} else {
err = ioutil.WriteFile(file, []byte(contents), 0644)
}
if err != nil {
cmd/goimports, imports: make goimports great again I felt the burn of my laptop on my legs, spinning away while processing goimports, and felt that it was time to make goimports great again. Over the past few years goimports fell into a slow state of disrepair with too many feature additions and no attention to the performance death by a thousand cuts. This was particularly terrible on OS X with its lackluster filesystem buffering. This CL makes goimports stronger, together with various optimizations and more visibility into what goimports is doing. * adds more internal documentation * avoids scanning $GOPATH for answers when running goimports on a file under $GOROOT (for Go core hackers) * don't read all $GOROOT & $GOPATH directories' Go code looking for their package names until much later. Require the package name of missing imports to be present in the last two directory path components. Then only try importing them in order from best to worst (shortest to longest, as before), so we can stop early. * when adding imports, add names to imports when the imported package name doesn't match the baes of its import path. For example: import foo "example.net/foo/v1" * don't read all *.go files in a package directory once the first file in a directory has revealed itself to be a package we're not looking for. For example, if we're looking for the right "client" for "client.Foo", we used to consider a directory "bar/client" as a candidate and read all 50 of its *.go files instead of stopping after its first *.go file had a "package main" line. * add some fast paths to remove allocations * add some fast paths to remove disk I/O when looking up the base package name of a standard library import (of existing imports in a file, which are very common) * adds a special case for import "C", to avoid some disk I/O. * add a -verbose flag to goimports for debugging On my Mac laptop with a huge $GOPATH, with a test file like: package foo import ( "fmt" "net/http" ) /* */ import "C" var _ = cloudbilling.New var _ = http.NewRequest var _ = client.New ... this took like 10 seconds before, and now 1.3 seconds. (Still slow; disk-based caching can come later) Updates golang/go#16367 (goimports is slow) Updates golang/go#16384 (refactor TestRename is broken on Windows) Change-Id: I97e85d3016afc9f2ad5501f97babad30c7989183 Reviewed-on: https://go-review.googlesource.com/24941 Reviewed-by: Andrew Gerrand <adg@golang.org> Reviewed-by: Rob Pike <r@golang.org> Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-14 18:08:27 -06:00
return err
}
}
return nil
}
func (c testConfig) test(t *testing.T, fn func(*goimportTest)) {
goroot := c.goroot
gopath := c.gopath
cmd/goimports, imports: make goimports great again I felt the burn of my laptop on my legs, spinning away while processing goimports, and felt that it was time to make goimports great again. Over the past few years goimports fell into a slow state of disrepair with too many feature additions and no attention to the performance death by a thousand cuts. This was particularly terrible on OS X with its lackluster filesystem buffering. This CL makes goimports stronger, together with various optimizations and more visibility into what goimports is doing. * adds more internal documentation * avoids scanning $GOPATH for answers when running goimports on a file under $GOROOT (for Go core hackers) * don't read all $GOROOT & $GOPATH directories' Go code looking for their package names until much later. Require the package name of missing imports to be present in the last two directory path components. Then only try importing them in order from best to worst (shortest to longest, as before), so we can stop early. * when adding imports, add names to imports when the imported package name doesn't match the baes of its import path. For example: import foo "example.net/foo/v1" * don't read all *.go files in a package directory once the first file in a directory has revealed itself to be a package we're not looking for. For example, if we're looking for the right "client" for "client.Foo", we used to consider a directory "bar/client" as a candidate and read all 50 of its *.go files instead of stopping after its first *.go file had a "package main" line. * add some fast paths to remove allocations * add some fast paths to remove disk I/O when looking up the base package name of a standard library import (of existing imports in a file, which are very common) * adds a special case for import "C", to avoid some disk I/O. * add a -verbose flag to goimports for debugging On my Mac laptop with a huge $GOPATH, with a test file like: package foo import ( "fmt" "net/http" ) /* */ import "C" var _ = cloudbilling.New var _ = http.NewRequest var _ = client.New ... this took like 10 seconds before, and now 1.3 seconds. (Still slow; disk-based caching can come later) Updates golang/go#16367 (goimports is slow) Updates golang/go#16384 (refactor TestRename is broken on Windows) Change-Id: I97e85d3016afc9f2ad5501f97babad30c7989183 Reviewed-on: https://go-review.googlesource.com/24941 Reviewed-by: Andrew Gerrand <adg@golang.org> Reviewed-by: Rob Pike <r@golang.org> Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-14 18:08:27 -06:00
if c.gorootFiles != nil && goroot == "" {
cmd/goimports, imports: make goimports great again I felt the burn of my laptop on my legs, spinning away while processing goimports, and felt that it was time to make goimports great again. Over the past few years goimports fell into a slow state of disrepair with too many feature additions and no attention to the performance death by a thousand cuts. This was particularly terrible on OS X with its lackluster filesystem buffering. This CL makes goimports stronger, together with various optimizations and more visibility into what goimports is doing. * adds more internal documentation * avoids scanning $GOPATH for answers when running goimports on a file under $GOROOT (for Go core hackers) * don't read all $GOROOT & $GOPATH directories' Go code looking for their package names until much later. Require the package name of missing imports to be present in the last two directory path components. Then only try importing them in order from best to worst (shortest to longest, as before), so we can stop early. * when adding imports, add names to imports when the imported package name doesn't match the baes of its import path. For example: import foo "example.net/foo/v1" * don't read all *.go files in a package directory once the first file in a directory has revealed itself to be a package we're not looking for. For example, if we're looking for the right "client" for "client.Foo", we used to consider a directory "bar/client" as a candidate and read all 50 of its *.go files instead of stopping after its first *.go file had a "package main" line. * add some fast paths to remove allocations * add some fast paths to remove disk I/O when looking up the base package name of a standard library import (of existing imports in a file, which are very common) * adds a special case for import "C", to avoid some disk I/O. * add a -verbose flag to goimports for debugging On my Mac laptop with a huge $GOPATH, with a test file like: package foo import ( "fmt" "net/http" ) /* */ import "C" var _ = cloudbilling.New var _ = http.NewRequest var _ = client.New ... this took like 10 seconds before, and now 1.3 seconds. (Still slow; disk-based caching can come later) Updates golang/go#16367 (goimports is slow) Updates golang/go#16384 (refactor TestRename is broken on Windows) Change-Id: I97e85d3016afc9f2ad5501f97babad30c7989183 Reviewed-on: https://go-review.googlesource.com/24941 Reviewed-by: Andrew Gerrand <adg@golang.org> Reviewed-by: Rob Pike <r@golang.org> Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-14 18:08:27 -06:00
goroot = mustTempDir(t, "goroot-")
defer os.RemoveAll(goroot)
}
if err := mapToDir(goroot, c.gorootFiles); err != nil {
t.Fatal(err)
}
if c.gopathFiles != nil && gopath == "" {
cmd/goimports, imports: make goimports great again I felt the burn of my laptop on my legs, spinning away while processing goimports, and felt that it was time to make goimports great again. Over the past few years goimports fell into a slow state of disrepair with too many feature additions and no attention to the performance death by a thousand cuts. This was particularly terrible on OS X with its lackluster filesystem buffering. This CL makes goimports stronger, together with various optimizations and more visibility into what goimports is doing. * adds more internal documentation * avoids scanning $GOPATH for answers when running goimports on a file under $GOROOT (for Go core hackers) * don't read all $GOROOT & $GOPATH directories' Go code looking for their package names until much later. Require the package name of missing imports to be present in the last two directory path components. Then only try importing them in order from best to worst (shortest to longest, as before), so we can stop early. * when adding imports, add names to imports when the imported package name doesn't match the baes of its import path. For example: import foo "example.net/foo/v1" * don't read all *.go files in a package directory once the first file in a directory has revealed itself to be a package we're not looking for. For example, if we're looking for the right "client" for "client.Foo", we used to consider a directory "bar/client" as a candidate and read all 50 of its *.go files instead of stopping after its first *.go file had a "package main" line. * add some fast paths to remove allocations * add some fast paths to remove disk I/O when looking up the base package name of a standard library import (of existing imports in a file, which are very common) * adds a special case for import "C", to avoid some disk I/O. * add a -verbose flag to goimports for debugging On my Mac laptop with a huge $GOPATH, with a test file like: package foo import ( "fmt" "net/http" ) /* */ import "C" var _ = cloudbilling.New var _ = http.NewRequest var _ = client.New ... this took like 10 seconds before, and now 1.3 seconds. (Still slow; disk-based caching can come later) Updates golang/go#16367 (goimports is slow) Updates golang/go#16384 (refactor TestRename is broken on Windows) Change-Id: I97e85d3016afc9f2ad5501f97babad30c7989183 Reviewed-on: https://go-review.googlesource.com/24941 Reviewed-by: Andrew Gerrand <adg@golang.org> Reviewed-by: Rob Pike <r@golang.org> Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-14 18:08:27 -06:00
gopath = mustTempDir(t, "gopath-")
defer os.RemoveAll(gopath)
}
if err := mapToDir(gopath, c.gopathFiles); err != nil {
t.Fatal(err)
cmd/goimports, imports: make goimports great again I felt the burn of my laptop on my legs, spinning away while processing goimports, and felt that it was time to make goimports great again. Over the past few years goimports fell into a slow state of disrepair with too many feature additions and no attention to the performance death by a thousand cuts. This was particularly terrible on OS X with its lackluster filesystem buffering. This CL makes goimports stronger, together with various optimizations and more visibility into what goimports is doing. * adds more internal documentation * avoids scanning $GOPATH for answers when running goimports on a file under $GOROOT (for Go core hackers) * don't read all $GOROOT & $GOPATH directories' Go code looking for their package names until much later. Require the package name of missing imports to be present in the last two directory path components. Then only try importing them in order from best to worst (shortest to longest, as before), so we can stop early. * when adding imports, add names to imports when the imported package name doesn't match the baes of its import path. For example: import foo "example.net/foo/v1" * don't read all *.go files in a package directory once the first file in a directory has revealed itself to be a package we're not looking for. For example, if we're looking for the right "client" for "client.Foo", we used to consider a directory "bar/client" as a candidate and read all 50 of its *.go files instead of stopping after its first *.go file had a "package main" line. * add some fast paths to remove allocations * add some fast paths to remove disk I/O when looking up the base package name of a standard library import (of existing imports in a file, which are very common) * adds a special case for import "C", to avoid some disk I/O. * add a -verbose flag to goimports for debugging On my Mac laptop with a huge $GOPATH, with a test file like: package foo import ( "fmt" "net/http" ) /* */ import "C" var _ = cloudbilling.New var _ = http.NewRequest var _ = client.New ... this took like 10 seconds before, and now 1.3 seconds. (Still slow; disk-based caching can come later) Updates golang/go#16367 (goimports is slow) Updates golang/go#16384 (refactor TestRename is broken on Windows) Change-Id: I97e85d3016afc9f2ad5501f97babad30c7989183 Reviewed-on: https://go-review.googlesource.com/24941 Reviewed-by: Andrew Gerrand <adg@golang.org> Reviewed-by: Rob Pike <r@golang.org> Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-14 18:08:27 -06:00
}
withEmptyGoPath(func() {
if goroot != "" {
build.Default.GOROOT = goroot
}
build.Default.GOPATH = gopath
it := &goimportTest{
T: t,
goroot: build.Default.GOROOT,
gopath: gopath,
ctx: &build.Default,
}
fn(it)
})
}
type goimportTest struct {
*testing.T
ctx *build.Context
goroot string
gopath string
}
// Tests that added imports are renamed when the import path's base doesn't
// match its package name. For example, we want to generate:
//
// import cloudbilling "google.golang.org/api/cloudbilling/v1"
func TestRenameWhenPackageNameMismatch(t *testing.T) {
testConfig{
gopathFiles: map[string]string{
"foo/bar/v1/x.go": "package bar \n const X = 1",
},
}.test(t, func(t *goimportTest) {
buf, err := Process(t.gopath+"/src/test/t.go", []byte("package main \n const Y = bar.X"), &Options{})
if err != nil {
t.Fatal(err)
}
const want = `package main
import bar "foo/bar/v1"
const Y = bar.X
`
if string(buf) != want {
t.Errorf("Got:\n%s\nWant:\n%s", buf, want)
}
})
}
// Tests that the LocalPrefix option causes imports
// to be added into a later group (num=3).
func TestLocalPrefix(t *testing.T) {
defer func(s string) { LocalPrefix = s }(LocalPrefix)
LocalPrefix = "foo/"
testConfig{
gopathFiles: map[string]string{
"foo/bar/bar.go": "package bar \n const X = 1",
},
}.test(t, func(t *goimportTest) {
buf, err := Process(t.gopath+"/src/test/t.go", []byte("package main \n const Y = bar.X \n const _ = runtime.GOOS"), &Options{})
if err != nil {
t.Fatal(err)
}
const want = `package main
import (
"runtime"
"foo/bar"
)
const Y = bar.X
const _ = runtime.GOOS
`
if string(buf) != want {
t.Errorf("Got:\n%s\nWant:\n%s", buf, want)
}
})
}
cmd/goimports, imports: make goimports great again I felt the burn of my laptop on my legs, spinning away while processing goimports, and felt that it was time to make goimports great again. Over the past few years goimports fell into a slow state of disrepair with too many feature additions and no attention to the performance death by a thousand cuts. This was particularly terrible on OS X with its lackluster filesystem buffering. This CL makes goimports stronger, together with various optimizations and more visibility into what goimports is doing. * adds more internal documentation * avoids scanning $GOPATH for answers when running goimports on a file under $GOROOT (for Go core hackers) * don't read all $GOROOT & $GOPATH directories' Go code looking for their package names until much later. Require the package name of missing imports to be present in the last two directory path components. Then only try importing them in order from best to worst (shortest to longest, as before), so we can stop early. * when adding imports, add names to imports when the imported package name doesn't match the baes of its import path. For example: import foo "example.net/foo/v1" * don't read all *.go files in a package directory once the first file in a directory has revealed itself to be a package we're not looking for. For example, if we're looking for the right "client" for "client.Foo", we used to consider a directory "bar/client" as a candidate and read all 50 of its *.go files instead of stopping after its first *.go file had a "package main" line. * add some fast paths to remove allocations * add some fast paths to remove disk I/O when looking up the base package name of a standard library import (of existing imports in a file, which are very common) * adds a special case for import "C", to avoid some disk I/O. * add a -verbose flag to goimports for debugging On my Mac laptop with a huge $GOPATH, with a test file like: package foo import ( "fmt" "net/http" ) /* */ import "C" var _ = cloudbilling.New var _ = http.NewRequest var _ = client.New ... this took like 10 seconds before, and now 1.3 seconds. (Still slow; disk-based caching can come later) Updates golang/go#16367 (goimports is slow) Updates golang/go#16384 (refactor TestRename is broken on Windows) Change-Id: I97e85d3016afc9f2ad5501f97babad30c7989183 Reviewed-on: https://go-review.googlesource.com/24941 Reviewed-by: Andrew Gerrand <adg@golang.org> Reviewed-by: Rob Pike <r@golang.org> Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org>
2016-07-14 18:08:27 -06:00
// Tests that running goimport on files in GOROOT (for people hacking
// on Go itself) don't cause the GOPATH to be scanned (which might be
// much bigger).
func TestOptimizationWhenInGoroot(t *testing.T) {
testConfig{
gopathFiles: map[string]string{
"foo/foo.go": "package foo\nconst X = 1\n",
},
}.test(t, func(t *goimportTest) {
testHookScanDir = func(dir string) {
if dir != filepath.Join(build.Default.GOROOT, "src") {
t.Errorf("unexpected dir scan of %s", dir)
}
}
const in = "package foo\n\nconst Y = bar.X\n"
buf, err := Process(t.goroot+"/src/foo/foo.go", []byte(in), nil)
if err != nil {
t.Fatal(err)
}
if string(buf) != in {
t.Errorf("got:\n%q\nwant unchanged:\n%q\n", in, buf)
}
})
}
// Tests that "package documentation" files are ignored.
func TestIgnoreDocumentationPackage(t *testing.T) {
testConfig{
gopathFiles: map[string]string{
"foo/foo.go": "package foo\nconst X = 1\n",
"foo/doc.go": "package documentation \n // just to confuse things\n",
},
}.test(t, func(t *goimportTest) {
const in = "package x\n\nconst Y = foo.X\n"
const want = "package x\n\nimport \"foo\"\n\nconst Y = foo.X\n"
buf, err := Process(t.gopath+"/src/x/x.go", []byte(in), nil)
if err != nil {
t.Fatal(err)
}
if string(buf) != want {
t.Errorf("wrong output.\ngot:\n%q\nwant:\n%q\n", in, want)
}
})
}
// Tests importPathToNameGoPathParse and in particular that it stops
// after finding the first non-documentation package name, not
// reporting an error on inconsistent package names (since it should
// never make it that far).
func TestImportPathToNameGoPathParse(t *testing.T) {
testConfig{
gopathFiles: map[string]string{
"example.net/pkg/doc.go": "package documentation\n", // ignored
"example.net/pkg/gen.go": "package main\n", // also ignored
"example.net/pkg/pkg.go": "package the_pkg_name_to_find\n and this syntax error is ignored because of parser.PackageClauseOnly",
"example.net/pkg/z.go": "package inconsistent\n", // inconsistent but ignored
},
}.test(t, func(t *goimportTest) {
got, err := importPathToNameGoPathParse("example.net/pkg", filepath.Join(t.gopath, "src", "other.net"))
if err != nil {
t.Fatal(err)
}
const want = "the_pkg_name_to_find"
if got != want {
t.Errorf("importPathToNameGoPathParse(..) = %q; want %q", got, want)
}
})
}
func TestIgnoreConfiguration(t *testing.T) {
testConfig{
gopathFiles: map[string]string{
".goimportsignore": "# comment line\n\n example.net", // tests comment, blank line, whitespace trimming
"example.net/pkg/pkg.go": "package pkg\nconst X = 1",
"otherwise-longer-so-worse.example.net/foo/pkg/pkg.go": "package pkg\nconst X = 1",
},
}.test(t, func(t *goimportTest) {
const in = "package x\n\nconst _ = pkg.X\n"
const want = "package x\n\nimport \"otherwise-longer-so-worse.example.net/foo/pkg\"\n\nconst _ = pkg.X\n"
buf, err := Process(t.gopath+"/src/x/x.go", []byte(in), nil)
if err != nil {
t.Fatal(err)
}
if string(buf) != want {
t.Errorf("wrong output.\ngot:\n%q\nwant:\n%q\n", buf, want)
}
})
}
// Skip "node_modules" directory.
func TestSkipNodeModules(t *testing.T) {
testConfig{
gopathFiles: map[string]string{
"example.net/node_modules/pkg/a.go": "package pkg\nconst X = 1",
"otherwise-longer.net/not_modules/pkg/a.go": "package pkg\nconst X = 1",
},
}.test(t, func(t *goimportTest) {
const in = "package x\n\nconst _ = pkg.X\n"
const want = "package x\n\nimport \"otherwise-longer.net/not_modules/pkg\"\n\nconst _ = pkg.X\n"
buf, err := Process(t.gopath+"/src/x/x.go", []byte(in), nil)
if err != nil {
t.Fatal(err)
}
if string(buf) != want {
t.Errorf("wrong output.\ngot:\n%q\nwant:\n%q\n", buf, want)
}
})
}
// golang.org/issue/16458 -- if GOROOT is a prefix of GOPATH, GOPATH is ignored.
func TestGoRootPrefixOfGoPath(t *testing.T) {
dir := mustTempDir(t, "importstest")
defer os.RemoveAll(dir)
testConfig{
goroot: filepath.Join(dir, "go"),
gopath: filepath.Join(dir, "gopath"),
gopathFiles: map[string]string{
"example.com/foo/pkg.go": "package foo\nconst X = 1",
},
}.test(t, func(t *goimportTest) {
const in = "package x\n\nconst _ = foo.X\n"
const want = "package x\n\nimport \"example.com/foo\"\n\nconst _ = foo.X\n"
buf, err := Process(t.gopath+"/src/x/x.go", []byte(in), nil)
if err != nil {
t.Fatal(err)
}
if string(buf) != want {
t.Errorf("wrong output.\ngot:\n%q\nwant:\n%q\n", buf, want)
}
})
}
const testGlobalImportsUsesGlobal = `package globalimporttest
func doSomething() {
t := time.Now()
}
`
const testGlobalImportsGlobalDecl = `package globalimporttest
type Time struct{}
func (t Time) Now() Time {
return Time{}
}
var time Time
`
// Tests that package global variables with the same name and function name as
// a function in a separate package do not result in an import which masks
// the global variable
func TestGlobalImports(t *testing.T) {
const pkg = "globalimporttest"
const usesGlobalFile = pkg + "/uses_global.go"
testConfig{
gopathFiles: map[string]string{
usesGlobalFile: testGlobalImportsUsesGlobal,
pkg + "/global.go": testGlobalImportsGlobalDecl,
},
}.test(t, func(t *goimportTest) {
buf, err := Process(
t.gopath+"/src/"+usesGlobalFile, []byte(testGlobalImportsUsesGlobal), nil)
if err != nil {
t.Fatal(err)
}
if string(buf) != testGlobalImportsUsesGlobal {
t.Errorf("wrong output.\ngot:\n%q\nwant:\n%q\n", buf, testGlobalImportsUsesGlobal)
}
})
}
func strSet(ss []string) map[string]bool {
m := make(map[string]bool)
for _, s := range ss {
m[s] = true
}
return m
}
func TestPkgIsCandidate(t *testing.T) {
tests := [...]struct {
filename string
pkgIdent string
pkg *pkg
want bool
}{
// normal match
0: {
filename: "/gopath/src/my/pkg/pkg.go",
pkgIdent: "client",
pkg: &pkg{
dir: "/gopath/src/client",
importPath: "client",
importPathShort: "client",
},
want: true,
},
// not a match
1: {
filename: "/gopath/src/my/pkg/pkg.go",
pkgIdent: "zzz",
pkg: &pkg{
dir: "/gopath/src/client",
importPath: "client",
importPathShort: "client",
},
want: false,
},
// would be a match, but "client" appears too deep.
2: {
filename: "/gopath/src/my/pkg/pkg.go",
pkgIdent: "client",
pkg: &pkg{
dir: "/gopath/src/client/foo/foo/foo",
importPath: "client/foo/foo",
importPathShort: "client/foo/foo",
},
want: false,
},
// not an exact match, but substring is good enough.
3: {
filename: "/gopath/src/my/pkg/pkg.go",
pkgIdent: "client",
pkg: &pkg{
dir: "/gopath/src/foo/go-client",
importPath: "foo/go-client",
importPathShort: "foo/go-client",
},
want: true,
},
// "internal" package, and not visible
4: {
filename: "/gopath/src/my/pkg/pkg.go",
pkgIdent: "client",
pkg: &pkg{
dir: "/gopath/src/foo/internal/client",
importPath: "foo/internal/client",
importPathShort: "foo/internal/client",
},
want: false,
},
// "internal" package but visible
5: {
filename: "/gopath/src/foo/bar.go",
pkgIdent: "client",
pkg: &pkg{
dir: "/gopath/src/foo/internal/client",
importPath: "foo/internal/client",
importPathShort: "foo/internal/client",
},
want: true,
},
// "vendor" package not visible
6: {
filename: "/gopath/src/foo/bar.go",
pkgIdent: "client",
pkg: &pkg{
dir: "/gopath/src/other/vendor/client",
importPath: "other/vendor/client",
importPathShort: "client",
},
want: false,
},
// "vendor" package, visible
7: {
filename: "/gopath/src/foo/bar.go",
pkgIdent: "client",
pkg: &pkg{
dir: "/gopath/src/foo/vendor/client",
importPath: "other/foo/client",
importPathShort: "client",
},
want: true,
},
// Ignore hyphens.
8: {
filename: "/gopath/src/foo/bar.go",
pkgIdent: "socketio",
pkg: &pkg{
dir: "/gopath/src/foo/socket-io",
importPath: "foo/socket-io",
importPathShort: "foo/socket-io",
},
want: true,
},
// Ignore case.
9: {
filename: "/gopath/src/foo/bar.go",
pkgIdent: "fooprod",
pkg: &pkg{
dir: "/gopath/src/foo/FooPROD",
importPath: "foo/FooPROD",
importPathShort: "foo/FooPROD",
},
want: true,
},
// Ignoring both hyphens and case together.
10: {
filename: "/gopath/src/foo/bar.go",
pkgIdent: "fooprod",
pkg: &pkg{
dir: "/gopath/src/foo/Foo-PROD",
importPath: "foo/Foo-PROD",
importPathShort: "foo/Foo-PROD",
},
want: true,
},
}
for i, tt := range tests {
got := pkgIsCandidate(tt.filename, tt.pkgIdent, tt.pkg)
if got != tt.want {
t.Errorf("test %d. pkgIsCandidate(%q, %q, %+v) = %v; want %v",
i, tt.filename, tt.pkgIdent, *tt.pkg, got, tt.want)
}
}
}