account for non-import shapes (#3997)

Signed-off-by: Alex Goodman <wagoodman@users.noreply.github.com>

syft/format/syftjson/model/document.go

@@ -28,7 +28,7 @@ func (d *Document) UnmarshalJSON(data []byte) error {
 	if d.Schema.Version == "1.0.0" && d.Descriptor.Name == "anchorectl" {
 		// convert all file modes from decimal to octal
 		for i := range d.Files {
-			d.Files[i].Metadata.Mode = convertFileModeToBase8(d.Files[i].Metadata.Mode)
+			d.Files[i].Metadata.Mode = convertBase10ToBase8(d.Files[i].Metadata.Mode)
 		}
 	}
 

syft/format/syftjson/model/file.go

@@ -31,14 +31,21 @@ type FileMetadataEntry struct {
 	Size            int64  `json:"size"`
 }
 
-func (f *FileMetadataEntry) UnmarshalJSON(data []byte) error {
-	type Alias FileMetadataEntry
-	aux := (*Alias)(f)
+type auxFileMetadataEntry FileMetadataEntry
+type fileMetadataEntryWithLegacyHint struct {
+	*auxFileMetadataEntry `json:",inline"`
+	LegacyHint            any `json:"FileInfo"`
+}
 
-	if err := json.Unmarshal(data, aux); err == nil {
-		// we should have at least one field set to a non-zero value... otherwise this is a legacy entry
-		if f.Mode != 0 || f.Type != "" || f.LinkDestination != "" ||
-			f.UserID != 0 || f.GroupID != 0 || f.MIMEType != "" || f.Size != 0 {
+func (f *FileMetadataEntry) UnmarshalJSON(data []byte) error {
+	aux := fileMetadataEntryWithLegacyHint{
+		auxFileMetadataEntry: (*auxFileMetadataEntry)(f),
+	}
+	if err := json.Unmarshal(data, &aux); err == nil {
+		fieldsSpecified := f.Mode != 0 || f.Type != "" || f.LinkDestination != "" ||
+			f.UserID != 0 || f.GroupID != 0 || f.MIMEType != "" || f.Size != 0
+		if aux.LegacyHint == nil && fieldsSpecified {
+			// we should have at least one field set to a non-zero value... (this is not a legacy shape)
 			return nil
 		}
 	}
@@ -48,8 +55,14 @@ func (f *FileMetadataEntry) UnmarshalJSON(data []byte) error {
 		return err
 	}
 
+	if !legacy.Type.WasInt {
+		// this occurs for document shapes from a non-import path and indicates that the mode has already been converted to octal.
+		// That being said, we want to handle all legacy shapes the same, so we will convert this to base 10 for consistency.
+		legacy.Mode = convertBase8ToBase10(legacy.Mode)
+	}
+
 	f.Mode = legacy.Mode
-	f.Type = string(legacy.Type)
+	f.Type = legacy.Type.Value
 	f.LinkDestination = legacy.LinkDestination
 	f.UserID = legacy.UserID
 	f.GroupID = legacy.GroupID
@@ -82,12 +95,15 @@ type FileLicenseEvidence struct {
 	Extent     int `json:"extent"`
 }
 
-type intOrStringFileType string
+type intOrStringFileType struct {
+	Value  string
+	WasInt bool
+}
 
 func (lt *intOrStringFileType) UnmarshalJSON(data []byte) error {
 	var str string
 	if err := json.Unmarshal(data, &str); err == nil {
-		*lt = intOrStringFileType(str)
+		lt.Value = str
 		return nil
 	}
 
@@ -96,13 +112,21 @@ func (lt *intOrStringFileType) UnmarshalJSON(data []byte) error {
 		return fmt.Errorf("file.Type must be either string or int, got: %s", string(data))
 	}
 
-	*lt = intOrStringFileType(num.String())
+	lt.Value = num.String()
+	lt.WasInt = true
 	return nil
 }
 
-func convertFileModeToBase8(rawMode int) int {
+func convertBase10ToBase8(rawMode int) int {
 	octalStr := fmt.Sprintf("%o", rawMode)
 	// we don't need to check that this is a valid octal string since the input is always an integer
 	result, _ := strconv.Atoi(octalStr)
 	return result
 }
+
+func convertBase8ToBase10(octalMode int) int {
+	octalStr := strconv.Itoa(octalMode)
+	result, _ := strconv.ParseInt(octalStr, 8, 64)
+
+	return int(result)
+}

syft/format/syftjson/model/file_test.go

@@ -36,8 +36,9 @@ func Test_FileMetadataEntry_UnmarshalJSON(t *testing.T) {
 			},
 		},
 		{
-			name: "unmarshal legacy sbom import format",
+			name: "unmarshal legacy image add internal document format",
 			jsonData: []byte(`{
+             "FileInfo": {},
              "Mode": 644,
              "Type": "RegularFile",
              "LinkDestination": "/usr/bin/python3",
@@ -47,7 +48,7 @@ func Test_FileMetadataEntry_UnmarshalJSON(t *testing.T) {
              "Size": 10174
           }`),
 			expected: FileMetadataEntry{
-				Mode:            644,
+				Mode:            420, // important! we convert this to base 10 so that all documents are consistent
 				Type:            "RegularFile",
 				LinkDestination: "/usr/bin/python3",
 				UserID:          1000,
@@ -57,8 +58,9 @@ func Test_FileMetadataEntry_UnmarshalJSON(t *testing.T) {
 			},
 		},
 		{
-			name: "unmarshal legacy sbom import format - integer type",
+			name: "unmarshal legacy sbom import format",
 			jsonData: []byte(`{
+             "FileInfo": {},
              "Mode": 644,
              "Type": 0,
              "LinkDestination": "/usr/bin/python3",
@@ -93,6 +95,7 @@ func Test_FileMetadataEntry_UnmarshalJSON(t *testing.T) {
 		{
 			name: "unmarshal minimal legacy format",
 			jsonData: []byte(`{
+             "FileInfo": {},
              "Mode": 0,
              "Type": "RegularFile",
              "UserID": 0,
@@ -120,10 +123,11 @@ func Test_FileMetadataEntry_UnmarshalJSON(t *testing.T) {
 
 func Test_intOrStringFileType_UnmarshalJSON(t *testing.T) {
 	tests := []struct {
-		name     string
-		jsonData []byte
-		expected string
-		wantErr  require.ErrorAssertionFunc
+		name           string
+		jsonData       []byte
+		expected       string
+		expectedWasInt bool
+		wantErr        require.ErrorAssertionFunc
 	}{
 		// string inputs - should pass through unchanged
 		{
@@ -148,59 +152,70 @@ func Test_intOrStringFileType_UnmarshalJSON(t *testing.T) {
 		},
 		// integer inputs - should convert to string representation
 		{
-			name:     "int 0 (TypeRegular)",
-			jsonData: []byte(`0`),
-			expected: "RegularFile",
+			name:           "int 0 (TypeRegular)",
+			jsonData:       []byte(`0`),
+			expected:       "RegularFile",
+			expectedWasInt: true,
 		},
 		{
-			name:     "int 1 (TypeHardLink)",
-			jsonData: []byte(`1`),
-			expected: "HardLink",
+			name:           "int 1 (TypeHardLink)",
+			jsonData:       []byte(`1`),
+			expected:       "HardLink",
+			expectedWasInt: true,
 		},
 		{
-			name:     "int 2 (TypeSymLink)",
-			jsonData: []byte(`2`),
-			expected: "SymbolicLink",
+			name:           "int 2 (TypeSymLink)",
+			jsonData:       []byte(`2`),
+			expected:       "SymbolicLink",
+			expectedWasInt: true,
 		},
 		{
-			name:     "int 3 (TypeCharacterDevice)",
-			jsonData: []byte(`3`),
-			expected: "CharacterDevice",
+			name:           "int 3 (TypeCharacterDevice)",
+			jsonData:       []byte(`3`),
+			expected:       "CharacterDevice",
+			expectedWasInt: true,
 		},
 		{
-			name:     "int 4 (TypeBlockDevice)",
-			jsonData: []byte(`4`),
-			expected: "BlockDevice",
+			name:           "int 4 (TypeBlockDevice)",
+			jsonData:       []byte(`4`),
+			expected:       "BlockDevice",
+			expectedWasInt: true,
 		},
 		{
-			name:     "int 5 (TypeDirectory)",
-			jsonData: []byte(`5`),
-			expected: "Directory",
+			name:           "int 5 (TypeDirectory)",
+			jsonData:       []byte(`5`),
+			expected:       "Directory",
+			expectedWasInt: true,
 		},
 		{
-			name:     "int 6 (TypeFIFO)",
-			jsonData: []byte(`6`),
-			expected: "FIFONode",
+			name:           "int 6 (TypeFIFO)",
+			jsonData:       []byte(`6`),
+			expected:       "FIFONode",
+			expectedWasInt: true,
 		},
 		{
-			name:     "int 7 (TypeSocket)",
-			jsonData: []byte(`7`),
-			expected: "Socket",
+			name:           "int 7 (TypeSocket)",
+			jsonData:       []byte(`7`),
+			expected:       "Socket",
+			expectedWasInt: true,
 		},
 		{
-			name:     "int 8 (TypeIrregular)",
-			jsonData: []byte(`8`),
-			expected: "IrregularFile",
+			name:           "int 8 (TypeIrregular)",
+			jsonData:       []byte(`8`),
+			expected:       "IrregularFile",
+			expectedWasInt: true,
 		},
 		{
-			name:     "unknown int",
-			jsonData: []byte(`99`),
-			expected: "Unknown",
+			name:           "unknown int",
+			jsonData:       []byte(`99`),
+			expected:       "Unknown",
+			expectedWasInt: true,
 		},
 		{
-			name:     "negative int",
-			jsonData: []byte(`-1`),
-			expected: "Unknown",
+			name:           "negative int",
+			jsonData:       []byte(`-1`),
+			expected:       "Unknown",
+			expectedWasInt: true,
 		},
 		{
 			name:     "null value",
@@ -244,12 +259,13 @@ func Test_intOrStringFileType_UnmarshalJSON(t *testing.T) {
 			if err != nil {
 				return
 			}
-			assert.Equal(t, test.expected, string(ft))
+			assert.Equal(t, test.expected, ft.Value)
+			assert.Equal(t, test.expectedWasInt, ft.WasInt)
 		})
 	}
 }
 
-func Test_convertFileModeToBase8(t *testing.T) {
+func Test_convertBase10ToBase8(t *testing.T) {
 	tests := []struct {
 		name     string
 		input    int
@@ -269,7 +285,34 @@ func Test_convertFileModeToBase8(t *testing.T) {
 
 	for _, tt := range tests {
 		t.Run(tt.name, func(t *testing.T) {
-			actual := convertFileModeToBase8(tt.input)
+			actual := convertBase10ToBase8(tt.input)
+
+			require.Equal(t, tt.expected, actual)
+		})
+	}
+}
+
+func Test_convertBase8ToBase10(t *testing.T) {
+	tests := []struct {
+		name     string
+		input    int
+		expected int
+	}{
+		{
+			name:     "no permissions",
+			input:    0,
+			expected: 0,
+		},
+		{
+			name:     "symlink + rwxrwxrwx",
+			input:    1000000777,
+			expected: 134218239,
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			actual := convertBase8ToBase10(tt.input)
 
 			require.Equal(t, tt.expected, actual)
 		})