fix(misconf): handle heredocs in dockerfile instructions (#8284)

Signed-off-by: nikpivkin <nikita.pivkin@smartforce.io>

Author: nikpivkin

pkg/iac/scanners/dockerfile/parser/parser.go

@@ -30,7 +30,7 @@ func Parse(_ context.Context, r io.Reader, path string) (any, error) {
 
 		instr, err := instructions.ParseInstruction(child)
 		if err != nil {
-			return nil, fmt.Errorf("process dockerfile instructions: %w", err)
+			return nil, fmt.Errorf("parse dockerfile instruction: %w", err)
 		}
 
 		if _, ok := instr.(*instructions.Stage); ok {
@@ -56,14 +56,27 @@ func Parse(_ context.Context, r io.Reader, path string) (any, error) {
 			EndLine:   child.EndLine,
 		}
 
+		// processing statement with sub-statement
+		// example: ONBUILD RUN foo bar
+		// https://github.com/moby/buildkit/blob/master/frontend/dockerfile/docs/reference.md#onbuild
 		if child.Next != nil && len(child.Next.Children) > 0 {
 			cmd.SubCmd = child.Next.Children[0].Value
 			child = child.Next.Children[0]
 		}
 
+		// mark if the instruction is in exec form
+		// https://github.com/moby/buildkit/blob/master/frontend/dockerfile/docs/reference.md#exec-form
 		cmd.JSON = child.Attributes["json"]
-		for n := child.Next; n != nil; n = n.Next {
-			cmd.Value = append(cmd.Value, n.Value)
+
+		// heredoc may contain a script that will be executed in the shell, so we need to process it
+		// https://github.com/moby/buildkit/blob/master/frontend/dockerfile/docs/reference.md#here-documents
+		if len(child.Heredocs) > 0 && child.Next != nil {
+			cmd.Original = originalFromHeredoc(child)
+			cmd.Value = []string{processHeredoc(child)}
+		} else {
+			for n := child.Next; n != nil; n = n.Next {
+				cmd.Value = append(cmd.Value, n.Value)
+			}
 		}
 
 		stage.Commands = append(stage.Commands, cmd)
@@ -75,3 +88,44 @@ func Parse(_ context.Context, r io.Reader, path string) (any, error) {
 
 	return &parsedFile, nil
 }
+
+func originalFromHeredoc(node *parser.Node) string {
+	var sb strings.Builder
+	sb.WriteString(node.Original)
+	sb.WriteRune('\n')
+	for i, heredoc := range node.Heredocs {
+		sb.WriteString(heredoc.Content)
+		sb.WriteString(heredoc.Name)
+		if i != len(node.Heredocs)-1 {
+			sb.WriteRune('\n')
+		}
+	}
+
+	return sb.String()
+}
+
+// heredoc processing taken from here
+// https://github.com/moby/buildkit/blob/9a39e2c112b7c98353c27e64602bc08f31fe356e/frontend/dockerfile/dockerfile2llb/convert.go#L1200
+func processHeredoc(node *parser.Node) string {
+	if parser.MustParseHeredoc(node.Next.Value) == nil || strings.HasPrefix(node.Heredocs[0].Content, "#!") {
+		// more complex heredoc is passed to the shell as is
+		var sb strings.Builder
+		sb.WriteString(node.Next.Value)
+		for _, heredoc := range node.Heredocs {
+			sb.WriteRune('\n')
+			sb.WriteString(heredoc.Content)
+			sb.WriteString(heredoc.Name)
+		}
+		return sb.String()
+	}
+
+	// simple heredoc and the content is run in a shell
+	content := node.Heredocs[0].Content
+	if node.Heredocs[0].Chomp {
+		content = parser.ChompHeredocContent(content)
+	}
+
+	content = strings.ReplaceAll(content, "\r\n", "\n")
+	cmds := strings.Split(strings.TrimSuffix(content, "\n"), "\n")
+	return strings.Join(cmds, " ; ")
+}

pkg/iac/scanners/dockerfile/parser/parser_test.go

@@ -59,3 +59,60 @@ CMD python /app/app.py
 	assert.Equal(t, 4, commands[3].StartLine)
 	assert.Equal(t, 4, commands[3].EndLine)
 }
+
+func Test_ParseHeredocs(t *testing.T) {
+	tests := []struct {
+		name     string
+		src      string
+		expected string
+	}{
+		{
+			name: "multi-line script",
+			src: `RUN <<EOF
+apk add curl
+apk add git
+EOF`,
+			expected: "apk add curl ; apk add git",
+		},
+		{
+			name: "file redirection and chained command",
+			src: `RUN cat <<EOF > /tmp/output && echo 'done'
+hello
+mr
+potato
+EOF`,
+			expected: "cat <<EOF > /tmp/output && echo 'done'\nhello\nmr\npotato\nEOF",
+		},
+		{
+			name: "redirect to file",
+			src: `RUN <<EOF > /etc/config.yaml
+key1: value1
+key2: value2
+EOF`,
+			expected: "<<EOF > /etc/config.yaml\nkey1: value1\nkey2: value2\nEOF",
+		},
+		{
+			name: "with a shebang",
+			src: `RUN <<EOF
+#!/usr/bin/env python
+print("hello world")
+EOF`,
+			expected: "<<EOF\n#!/usr/bin/env python\nprint(\"hello world\")\nEOF",
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			res, err := parser.Parse(context.TODO(), strings.NewReader(tt.src), "Dockerfile")
+			require.NoError(t, err)
+
+			df, ok := res.(*dockerfile.Dockerfile)
+			require.True(t, ok)
+
+			cmd := df.Stages[0].Commands[0]
+
+			assert.Equal(t, tt.src, cmd.Original)
+			assert.Equal(t, []string{tt.expected}, cmd.Value)
+		})
+	}
+}