Harden line break detection

Handle multi-line breaks use in the single file app format, as users from different OSes (windows/unix) can use different line breaks inadvertently, and this is still valid YAML

Signed-off-by: Silvin Lubecki <[email protected]>

Author: silvin-lubecki

loader/loader.go

@@ -7,26 +7,64 @@ import (
 	"os"
 	"path/filepath"
 
-	"github.com/docker/cli/cli/compose/loader"
-
-	"github.com/docker/app/specification"
-	"github.com/docker/cli/cli/compose/schema"
-
 	"github.com/docker/app/internal"
+	"github.com/docker/app/specification"
 	"github.com/docker/app/types"
+	"github.com/docker/cli/cli/compose/loader"
+	"github.com/docker/cli/cli/compose/schema"
 	"github.com/docker/docker/pkg/archive"
 	"github.com/pkg/errors"
 )
 
+var (
+	crlf       = []byte{'\r', '\n'}
+	lf         = []byte{'\n'}
+	delimiters = [][]byte{
+		[]byte("\r\n---\r\n"),
+		[]byte("\n---\r\n"),
+		[]byte("\r\n---\n"),
+		[]byte("\n---\n"),
+	}
+)
+
+// useCRLF detects which line break should be used
+func useCRLF(data []byte) bool {
+	nbCrlf := bytes.Count(data, crlf)
+	nbLf := bytes.Count(data, lf)
+	switch {
+	case nbCrlf == nbLf:
+		// document contains only CRLF
+		return true
+	case nbCrlf == 0:
+		// document does not contain any CRLF
+		return false
+	default:
+		// document contains mixed line breaks, so use the OS default
+		return bytes.Equal(defaultLineBreak, crlf)
+	}
+}
+
+// splitSingleFile split a multidocument using all possible document delimiters
+func splitSingleFile(data []byte) [][]byte {
+	parts := [][]byte{data}
+	for _, delimiter := range delimiters {
+		var intermediate [][]byte
+		for _, part := range parts {
+			intermediate = append(intermediate, bytes.Split(part, delimiter)...)
+		}
+		parts = intermediate
+	}
+	return parts
+}
+
 // LoadFromSingleFile loads a docker app from a single-file format (as a reader)
 func LoadFromSingleFile(path string, r io.Reader, ops ...func(*types.App) error) (*types.App, error) {
 	data, err := ioutil.ReadAll(r)
 	if err != nil {
 		return nil, errors.Wrap(err, "error reading single-file")
 	}
-	hasCRLF := bytes.Contains(data, []byte{'\r', '\n'})
 
-	parts := bytes.Split(data, types.YamlSingleFileSeparator(hasCRLF))
+	parts := splitSingleFile(data)
 	if len(parts) != 3 {
 		return nil, errors.Errorf("malformed single-file application: expected 3 documents, got %d", len(parts))
 	}
@@ -51,11 +89,12 @@ func LoadFromSingleFile(path string, r io.Reader, ops ...func(*types.App) error)
 			return nil, errors.New("malformed single-file application")
 		}
 	}
+
 	appOps := append([]func(*types.App) error{
 		types.WithComposes(compose),
 		types.WithParameters(params),
 		types.Metadata(metadata),
-		types.WithCRLF(hasCRLF),
+		types.WithCRLF(useCRLF(data)),
 	}, ops...)
 	return types.NewApp(path, appOps...)
 }

loader/loader_test.go

@@ -20,10 +20,10 @@ const (
 	metadata = `name: my-app
 version: 1.0.0
 `
-	compose = `services:
+	compose = `version: "3.1"
+services:
   web:
     image: nginx
-version: "3.1"
 `
 	params = `foo: bar
 `
@@ -42,6 +42,10 @@ func TestLoadFromSingleFile(t *testing.T) {
 			name: "carriage-return-line-feed",
 			file: fmt.Sprintf("%s\r\n---\r\n%s\r\n---\r\n%s", metadata, compose, params),
 		},
+		{
+			name: "mixed-carriage-return-line-feed",
+			file: fmt.Sprintf("%s\r\n---\r\n%s\r\n---\n%s", metadata, compose, params),
+		},
 		{
 			name: "unordered-documents",
 			file: fmt.Sprintf("%s\n---\n%s\n---\n%s", params, metadata, compose),

loader/loader_unix.go

@@ -0,0 +1,7 @@
+// +build !windows
+
+package loader
+
+var (
+	defaultLineBreak = []byte{'\n'}
+)

loader/loader_windows.go

@@ -0,0 +1,5 @@
+package loader
+
+var (
+	defaultLineBreak = []byte{'\r', '\n'}
+)