Calculate localized path for remote file (#4878)

* Calculate localized path for remote file

* Add documentation

api/internal/localizer/util.go

@@ -5,6 +5,7 @@ package localizer
 
 import (
 	"log"
+	"net/url"
 	"path/filepath"
 	"strings"
 
@@ -120,8 +121,25 @@ func cleanFilePath(fSys filesys.FileSystem, root filesys.ConfirmedDir, file stri
 	return locPath
 }
 
-// locFilePath returns the relative localized path of validated file url fileURL
-// TODO(annasong): implement
-func locFilePath(_ string) string {
-	return filepath.Join(LocalizeDir, "")
+// locFilePath converts a URL to its localized form, e.g.
+// https://raw.githubusercontent.com/kubernetes-sigs/kustomize/master/api/krusty/testdata/localize/simple/service.yaml ->
+// localized-files/raw.githubusercontent.com/kubernetes-sigs/kustomize/master/api/krusty/testdata/localize/simple/service.yaml.
+//
+// fileURL must be a validated file URL.
+func locFilePath(fileURL string) string {
+	// File urls must have http or https scheme, so it is safe to use url.Parse.
+	u, err := url.Parse(fileURL)
+	if err != nil {
+		log.Fatalf("cannot parse validated file url %q: %s", fileURL, err.Error())
+	}
+
+	// Percent-encodings should be preserved in case sub-delims have special meaning.
+	// Extraneous '..' parent directory dot-segments should be removed.
+	path := filepath.Join(string(filepath.Separator), filepath.FromSlash(u.EscapedPath()))
+
+	// The host should not include userinfo or port.
+	// Raw github urls are the only type of file urls kustomize officially accepts.
+	// In this case, the path already consists of org, repo, version, and path in repo, in order,
+	// so we can use it as is.
+	return filepath.Join(LocalizeDir, u.Hostname(), path)
 }

api/internal/localizer/util_test.go

@@ -4,6 +4,8 @@
 package localizer //nolint:testpackage
 
 import (
+	"path/filepath"
+	"strings"
 	"testing"
 
 	"github.com/stretchr/testify/require"
@@ -16,3 +18,49 @@ func TestUrlBase(t *testing.T) {
 func TestUrlBaseTrailingSlash(t *testing.T) {
 	require.Equal(t, "repo", urlBase("github.com/org/repo//"))
 }
+
+// simpleJoin is filepath.Join() without the side effects of filepath.Clean()
+func simpleJoin(t *testing.T, elems ...string) string {
+	t.Helper()
+
+	return strings.Join(elems, string(filepath.Separator))
+}
+
+func TestLocFilePath(t *testing.T) {
+	for name, tUnit := range map[string]struct {
+		url, path string
+	}{
+		"official": {
+			url:  "https://raw.githubusercontent.com/org/repo/ref/path/to/file.yaml",
+			path: simpleJoin(t, "raw.githubusercontent.com", "org", "repo", "ref", "path", "to", "file.yaml"),
+		},
+		"empty_path": {
+			url:  "https://host",
+			path: "host",
+		},
+		"empty_path_segment": {
+			url:  "https://host//",
+			path: "host",
+		},
+		"extraneous_components": {
+			url:  "http://userinfo@host:1234/path/file?query",
+			path: simpleJoin(t, "host", "path", "file"),
+		},
+		"percent-encoding": {
+			url:  "https://host/file%2Eyaml",
+			path: simpleJoin(t, "host", "file%2Eyaml"),
+		},
+		"dot-segments": {
+			url:  "https://host/path/blah/../to/foo/bar/../../file/./",
+			path: simpleJoin(t, "host", "path", "to", "file"),
+		},
+		"extraneous_dot-segments": {
+			url:  "https://host/foo/bar/baz/../../../../file",
+			path: simpleJoin(t, "host", "file"),
+		},
+	} {
+		t.Run(name, func(t *testing.T) {
+			require.Equal(t, simpleJoin(t, LocalizeDir, tUnit.path), locFilePath(tUnit.url))
+		})
+	}
+}