clean: switch -test to a regex -whitelist flag

README.md

@@ -3,7 +3,15 @@
 [![Build Status](https://travis-ci.org/LK4D4/vndr.svg?branch=master)](https://travis-ci.org/LK4D4/vndr)
 
 Vndr is simple vendoring tool, which is inspired by Docker vendor script.
-Vndr has only one option: `-verbose`.
+Vndr has only two options: `-verbose` and `-whitelist`, both of which do
+exactly what they say on the tin.
+
+* `-verbose` adds additional output, helpful for debugging issues.
+* `-whitelist` allows you to specify several regular expressions for paths
+  which will *not* be cleaned in the final stage of vendoring -- this is useful
+  for running tests in a vendored project or otherwise ensuring that some
+  important files are retained after `vndr` is done cleaning unused files from
+  your `vendor/` directory.
 
 ## vendor.conf
 
@@ -20,7 +28,7 @@ This config format is also accepted by [trash](https://github.com/rancher/trash)
 ## Initialization
 
 You can initiate your project with vendor directory and `vendor.conf` using command
-`vndr init`. This will populate your vendor directory with latest versions of 
+`vndr init`. This will populate your vendor directory with latest versions of
 all dependecies and also write `vendor.conf` config which you can use for changing
 versions later.
 

clean.go

@@ -72,14 +72,14 @@ func cleanVendor(vendorDir string, realDeps []*build.Package) error {
 			return nil
 		}
 
-		// When preserving tests don't delete testdata, let alone _test.go files.
-		if preserveTest {
-			if i.Name() == "testdata" {
-				return nil
-			}
-			if strings.HasSuffix(path, "_test.go") {
-				return nil
-			}
+		// Make sure we don't delete anything that matches the whitelist. The
+		// whitelist is relative to the vendor directory.
+		relPath, err := filepath.Rel(vendorDir, path)
+		if err != nil {
+			return err
+		}
+		if cleanWhitelist.matchString(relPath) {
+			return nil
 		}
 
 		if strings.HasPrefix(i.Name(), ".") || strings.HasPrefix(i.Name(), "_") {

main.go

@@ -8,6 +8,7 @@ import (
 	"log"
 	"os"
 	"path/filepath"
+	"regexp"
 	"sort"
 	"strings"
 	"time"
@@ -22,18 +23,49 @@ const (
 )
 
 var (
-	verbose      bool
-	preserveTest bool
+	verbose        bool
+	cleanWhitelist regexpSlice
 )
 
+type regexpSlice []*regexp.Regexp
+
+var _ flag.Value = new(regexpSlice)
+
+func (rs *regexpSlice) Set(exp string) error {
+	regex, err := regexp.Compile(exp)
+	if err != nil {
+		return err
+	}
+
+	*rs = append(*rs, regex)
+	return nil
+}
+
+func (rs *regexpSlice) String() string {
+	exps := []string{}
+	for _, regex := range *rs {
+		exps = append(exps, fmt.Sprintf("%q", regex.String()))
+	}
+	return fmt.Sprintf("%v", exps)
+}
+
+func (rs *regexpSlice) matchString(str string) bool {
+	for _, regex := range *rs {
+		if regex.MatchString(str) {
+			return true
+		}
+	}
+	return false
+}
+
 func init() {
 	flag.Usage = func() {
 		fmt.Fprintf(os.Stderr, "Usage of %s:\n", os.Args[0])
 		fmt.Fprintf(os.Stderr, "%s [[import path] [revision]] [repository]\n%s init\n", os.Args[0], os.Args[0])
 		flag.PrintDefaults()
 	}
 	flag.BoolVar(&verbose, "verbose", false, "shows all warnings")
-	flag.BoolVar(&preserveTest, "test", false, "preserve _test.go files in vendor'd projects")
+	flag.Var(&cleanWhitelist, "whitelist", "regular expressions to whitelist for cleaning phase of vendoring, relative to the vendor/ directory")
 }
 
 func validateArgs() {
@@ -137,22 +169,22 @@ func main() {
 			log.Fatal("There must not be vendor dir and vendor.conf file for initialization")
 		}
 	}
+
 	wd, err := os.Getwd()
 	if err != nil {
 		log.Fatalf("Error getting working directory: %v", err)
 	}
-
 	wd, err = filepath.EvalSymlinks(wd)
 	if err != nil {
 		log.Fatalf("Error getting working directory after evalsymlinks: %v", err)
 	}
+	vd := filepath.Join(wd, vendorDir)
 
 	log.Println("Collecting initial packages")
 	initPkgs, err := collectPkgs(wd)
 	if err != nil {
 		log.Fatalf("Error collecting initial packages: %v", err)
 	}
-	vd := filepath.Join(wd, vendorDir)
 	// variables for init
 	var dlFunc func(string) (*build.Package, error)
 	var deps []depEntry
@@ -194,6 +226,9 @@ func main() {
 		log.Fatalf("Error on collecting all dependencies: %v", err)
 	}
 	log.Println("Clean vendor dir from unused packages")
+	for _, regex := range cleanWhitelist {
+		log.Printf("\tIgnoring paths matching %q", regex.String())
+	}
 	if err := cleanVendor(vd, pkgs); err != nil {
 		log.Fatal(err)
 	}

test/vndr_test.go

@@ -2,6 +2,7 @@ package vndrtest
 
 import (
 	"bytes"
+	"fmt"
 	"io/ioutil"
 	"os"
 	"os/exec"
@@ -186,3 +187,87 @@ github.com/docker/swarmkit branch
 		t.Fatalf("libcompose should not be reported: %s", out)
 	}
 }
+
+func TestCleanWhitelist(t *testing.T) {
+	vndrBin, err := exec.LookPath("vndr")
+	if err != nil {
+		t.Fatal(err)
+	}
+	tmp, err := ioutil.TempDir("", "test-vndr-")
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer os.RemoveAll(tmp)
+	repoDir := filepath.Join(tmp, "src", testRepo)
+	if err := os.MkdirAll(repoDir, 0700); err != nil {
+		t.Fatal(err)
+	}
+	content := []byte(`github.com/containers/image master
+github.com/projectatomic/skopeo master`)
+	vendorConf := filepath.Join(repoDir, "vendor.conf")
+	if err := ioutil.WriteFile(vendorConf, content, 0666); err != nil {
+		t.Fatal(err)
+	}
+	vndrCmd := exec.Command(vndrBin,
+		"-whitelist", `github\.com/containers/image/MAINTAINERS`,
+		"-whitelist", `github\.com/projectatomic/skopeo/integration/.*`)
+	vndrCmd.Dir = repoDir
+	setGopath(vndrCmd, tmp)
+
+	out, err := vndrCmd.CombinedOutput()
+	if err != nil {
+		t.Logf("output: %v", string(out))
+		t.Fatalf("error was not expected: %v", err)
+	}
+
+	if !bytes.Contains(out, []byte(fmt.Sprintf(`Ignoring paths matching %q`, `github\.com/containers/image/MAINTAINERS`))) {
+		t.Logf("output: %v", string(out))
+		t.Errorf(`output missing regular expression "github\.com/containers/image/MAINTAINERS"`)
+	}
+	if !bytes.Contains(out, []byte(fmt.Sprintf(`Ignoring paths matching %q`, `github\.com/projectatomic/skopeo/integration/.*`))) {
+		t.Logf("output: %v", string(out))
+		t.Errorf(`output missing regular expression "github\.com/projectatomic/skopeo/integration/.*"`)
+	}
+
+	// Make sure that the files were not "cleaned".
+	for _, path := range []string{
+		"github.com/containers/image/MAINTAINERS",
+		"github.com/projectatomic/skopeo/integration",
+	} {
+		path = filepath.Join(repoDir, "vendor", path)
+		if _, err := os.Lstat(path); err != nil {
+			t.Errorf("%s was cleaned but shouldn't have been", path)
+		}
+	}
+
+	// Run again to make sure the above will be cleaned.
+	vndrCmd = exec.Command(vndrBin)
+	vndrCmd.Dir = repoDir
+	setGopath(vndrCmd, tmp)
+
+	out, err = vndrCmd.CombinedOutput()
+	if err != nil {
+		t.Logf("output: %v", string(out))
+		t.Fatalf("[no -whitelist] error was not expected: %v", err)
+	}
+
+	if bytes.Contains(out, []byte(fmt.Sprintf(`Ignoring paths matching %q`, `github\.com/containers/image/MAINTAINERS`))) {
+		t.Logf("output: %v", string(out))
+		t.Errorf(`[no -whitelist] output missing regular expression "github\.com/containers/image/MAINTAINERS"`)
+	}
+	if bytes.Contains(out, []byte(fmt.Sprintf(`Ignoring paths matching %q`, `github\.com/projectatomic/skopeo/integration/.*`))) {
+		t.Logf("output: %v", string(out))
+		t.Errorf(`[no -whitelist] output missing regular expression "github\.com/projectatomic/skopeo/integration/.*"`)
+	}
+
+	// Make sure that the files were "cleaned".
+	for _, path := range []string{
+		"github.com/containers/image/MAINTAINERS",
+		"github.com/projectatomic/skopeo/integration",
+	} {
+		path = filepath.Join(repoDir, "vendor", path)
+		if _, err := os.Lstat(path); err == nil {
+			t.Errorf("[no -whitelist] %s was NOT cleaned but should have been", path)
+		}
+	}
+}