Allow Multipart Declaration for File and Package

 - Files and Packages can now be declared in several parts in different parent tags

Signed-off-by: Rishabh Bhatnagar <bhatnagarrishabh4@gmail.com>

5 files changed, 110 insertions, 8 deletions

diff --git a/rdfloader/parser2v2/parse_file.go b/rdfloader/parser2v2/parse_file.go
index 9fbf3b4..38a95ea 100644
--- a/rdfloader/parser2v2/parse_file.go
+++ b/rdfloader/parser2v2/parse_file.go
@@ -18,6 +18,10 @@ func (parser *rdfParser2_2) getFileFromNode(fileNode *gordfParser.Node) (file *s
 		return nil, err
 	}
 
+	if existingFile := parser.files[file.FileSPDXIdentifier]; existingFile != nil {
+		file = existingFile
+	}
+
 	for _, subTriple := range parser.nodeToTriples(fileNode) {
 		switch subTriple.Predicate.ID {
 		case SPDX_FILE_NAME: // 4.1
diff --git a/rdfloader/parser2v2/parse_file_test.go b/rdfloader/parser2v2/parse_file_test.go
index 27c8fb4..48db28f 100644
--- a/rdfloader/parser2v2/parse_file_test.go
+++ b/rdfloader/parser2v2/parse_file_test.go
@@ -441,7 +441,60 @@ func Test_rdfParser2_2_getFileFromNode(t *testing.T) {
 		t.Error("should've raised an error stating invalid licenseInfoInFile for a file")
 	}
 
-	// TestCase 10: all valid attribute and it's values.
+	// TestCase 10: Splitting of File definition into parents of different tags mustn't create new file objects.
+	fileDefinitions := []string{
+		`<spdx:Package rdf:about="#SPDXRef-Package1">
+			<spdx:hasFile>
+				<spdx:File rdf:about="http://anupam-VirtualBox/repo/SPDX2_time-1.9.tar.gz_1535120734-spdx.rdf#SPDXRef-item177">
+					<spdx:fileName>time-1.9/ChangeLog</spdx:fileName>
+					<spdx:fileType rdf:resource="http://spdx.org/rdf/terms#fileType_source"/>
+				</spdx:File>
+			</spdx:hasFile>
+		</spdx:Package>`,
+		`<spdx:Package rdf:about="#SPDXRef-Package2">
+			<spdx:hasFile>
+				<spdx:File rdf:about="http://anupam-VirtualBox/repo/SPDX2_time-1.9.tar.gz_1535120734-spdx.rdf#SPDXRef-item177">
+					<spdx:licenseConcluded rdf:resource="http://spdx.org/rdf/terms#noassertion" />
+					<spdx:licenseInfoInFile rdf:resource="http://spdx.org/rdf/terms#NOASSERTION" />
+				</spdx:File>
+			</spdx:hasFile>
+		</spdx:Package>`,
+	}
+	parser, _ = parserFromBodyContent(strings.Join(fileDefinitions, ""))
+
+	var file *spdx.File2_2
+	packageTypeTriples := gordfWriter.FilterTriples(parser.gordfParserObj.Triples, nil, &RDF_TYPE, &SPDX_PACKAGE)
+	for _, typeTriple := range packageTypeTriples {
+		pkg, err := parser.getPackageFromNode(typeTriple.Subject)
+		if err != nil {
+			t.Errorf("unexpected error parsing a valid package: %v", err)
+		}
+		if n := len(pkg.Files); n != 1 {
+			t.Errorf("expected package to contain exactly 1 file. Found %d files", n)
+		}
+		for _, file = range pkg.Files {
+		}
+	}
+
+	// checking if all the attributes that spanned over a several tags are set in the same variable.
+	expectedFileName := "time-1.9/ChangeLog"
+	if file.FileName != expectedFileName {
+		t.Errorf("expected %s, found %s", expectedFileName, file.FileName)
+	}
+	expectedLicenseConcluded := "NOASSERTION"
+	if file.LicenseConcluded != expectedLicenseConcluded {
+		t.Errorf("expected %s, found %s", expectedLicenseConcluded, file.LicenseConcluded)
+	}
+	expectedFileType := "source"
+	if file.FileType[0] != expectedFileType {
+		t.Errorf("expected %s, found %s", expectedFileType, file.FileType)
+	}
+	expectedLicenseInfoInFile := "NOASSERTION"
+	if file.LicenseInfoInFile[0] != expectedLicenseInfoInFile {
+		t.Errorf("expected %s, found %s", expectedLicenseInfoInFile, file.LicenseInfoInFile[0])
+	}
+
+	// TestCase 11: all valid attribute and it's values.
 	parser, _ = parserFromBodyContent(`
 		<spdx:File rdf:about="http://anupam-VirtualBox/repo/SPDX2_time-1.9.tar.gz_1535120734-spdx.rdf#SPDXRef-item177">
 			<spdx:fileName>time-1.9/ChangeLog</spdx:fileName>
@@ -485,14 +538,14 @@ func Test_rdfParser2_2_getFileFromNode(t *testing.T) {
 		</spdx:File>
 	`)
 	fileNode = gordfWriter.FilterTriples(parser.gordfParserObj.Triples, nil, &RDF_TYPE, &SPDX_FILE)[0].Subject
-	file, err := parser.getFileFromNode(fileNode)
+	file, err = parser.getFileFromNode(fileNode)
 	if err != nil {
 		t.Errorf("unexpected error parsing a valid file: %v", err)
 	}
 
 	// checking each and every attribute of the obtained file.
 
-	expectedFileName := "time-1.9/ChangeLog"
+	expectedFileName = "time-1.9/ChangeLog"
 	if file.FileName != expectedFileName {
 		t.Errorf("expected %s, found %s", expectedFileName, file.FileName)
 	}
@@ -500,7 +553,7 @@ func Test_rdfParser2_2_getFileFromNode(t *testing.T) {
 	if len(file.FileType) != 1 {
 		t.Errorf("given file should have 1 fileType attribute. found %d", len(file.FileType))
 	}
-	expectedFileType := "source"
+	expectedFileType = "source"
 	if file.FileType[0] != expectedFileType {
 		t.Errorf("expected %s, found %s", expectedFileType, file.FileType)
 	}
@@ -510,7 +563,7 @@ func Test_rdfParser2_2_getFileFromNode(t *testing.T) {
 		t.Errorf("expected %s, found %s", expectedChecksum, file.FileChecksumSHA1)
 	}
 
-	expectedLicenseConcluded := "NOASSERTION"
+	expectedLicenseConcluded = "NOASSERTION"
 	if file.LicenseConcluded != expectedLicenseConcluded {
 		t.Errorf("expected %s, found %s", expectedLicenseConcluded, file.LicenseConcluded)
 	}
@@ -518,7 +571,7 @@ func Test_rdfParser2_2_getFileFromNode(t *testing.T) {
 	if len(file.LicenseInfoInFile) != 1 {
 		t.Errorf("given file should have 1 licenseInfoInFile attribute. found %d", len(file.LicenseInfoInFile))
 	}
-	expectedLicenseInfoInFile := "NOASSERTION"
+	expectedLicenseInfoInFile = "NOASSERTION"
 	if file.LicenseInfoInFile[0] != expectedLicenseInfoInFile {
 		t.Errorf("expected %s, found %s", expectedLicenseInfoInFile, file.LicenseInfoInFile[0])
 	}
diff --git a/rdfloader/parser2v2/parse_package.go b/rdfloader/parser2v2/parse_package.go
index 20047a6..5be9631 100644
--- a/rdfloader/parser2v2/parse_package.go
+++ b/rdfloader/parser2v2/parse_package.go
@@ -19,6 +19,10 @@ func (parser *rdfParser2_2) getPackageFromNode(packageNode *gordfParser.Node) (p
 	}
 	pkg.PackageSPDXIdentifier = eId // 3.2
 
+	if existingPkg := parser.doc.Packages[eId]; existingPkg != nil {
+		pkg = existingPkg
+	}
+
 	// iterate over all the triples associated with the provided package packageNode.
 	for _, subTriple := range parser.nodeToTriples(packageNode) {
 		switch subTriple.Predicate.ID {
@@ -125,6 +129,7 @@ func (parser *rdfParser2_2) getPackageFromNode(packageNode *gordfParser.Node) (p
 		}
 	}
 
+	parser.doc.Packages[pkg.PackageSPDXIdentifier] = pkg
 	return pkg, nil
 }
 
diff --git a/rdfloader/parser2v2/parse_package_test.go b/rdfloader/parser2v2/parse_package_test.go
index a5f77a8..35a67dc 100644
--- a/rdfloader/parser2v2/parse_package_test.go
+++ b/rdfloader/parser2v2/parse_package_test.go
@@ -418,7 +418,48 @@ func Test_rdfParser2_2_getPackageFromNode(t *testing.T) {
 		t.Errorf("expected an error(invalid homepage uri), found %v", err)
 	}
 
-	// TestCase 7: everything valid
+	// TestCase 7: Package tag declared more than once should be parsed into a single object's definition
+	parser, _ = parserFromBodyContent(`
+		<spdx:Package rdf:about="http://anupam-VirtualBox/repo/SPDX2_time-1.9#SPDXRef-upload2">
+			<spdx:name>Test Package</spdx:name>
+		</spdx:Package>
+	`)
+	node = parser.gordfParserObj.Triples[0].Subject
+	_, err = parser.getPackageFromNode(node)
+	if err != nil {
+		t.Errorf("error parsing a valid package: %v", err)
+	}
+	yetAnotherPkgTriple := gordfParser.Triple{
+		Subject: node,
+		Predicate: &gordfParser.Node{
+			NodeType: gordfParser.IRI,
+			ID:       SPDX_PACKAGE_FILE_NAME,
+		},
+		Object: &gordfParser.Node{
+			NodeType: gordfParser.LITERAL,
+			ID:       "packageFileName",
+		},
+	}
+	parser.nodeStringToTriples[node.String()] = append(parser.nodeStringToTriples[node.String()], &yetAnotherPkgTriple)
+	pkg, err := parser.getPackageFromNode(node)
+	if err != nil {
+		t.Errorf("error parsing a valid package: %v", err)
+	}
+	// validating if all the attributes that spanned over two tags are included in the parsed package.
+	expectedID := "upload2"
+	if string(pkg.PackageSPDXIdentifier) != expectedID {
+		t.Errorf("expected package id: %s, got %s", expectedID, pkg.PackageSPDXIdentifier)
+	}
+	expectedPkgFileName := "packageFileName"
+	if expectedPkgFileName != pkg.PackageFileName {
+		t.Errorf("expected package file name: %s, got %s", expectedPkgFileName, pkg.PackageFileName)
+	}
+	expectedName := "Test Package"
+	if pkg.PackageName != expectedName {
+		t.Errorf("expected package name: %s, got %s", expectedPkgFileName, pkg.PackageName)
+	}
+
+	// TestCase 8: everything valid
 	parser, _ = parserFromBodyContent(`
 		<spdx:Package rdf:about="http://anupam-VirtualBox/repo/SPDX2_time-1.9#SPDXRef-upload2">
 			<spdx:name>Test Package</spdx:name>
diff --git a/rdfloader/parser2v2/parse_relationship.go b/rdfloader/parser2v2/parse_relationship.go
index 3ef9c61..beba838 100644
--- a/rdfloader/parser2v2/parse_relationship.go
+++ b/rdfloader/parser2v2/parse_relationship.go
@@ -88,7 +88,6 @@ func (parser *rdfParser2_2) parseRelatedElementFromTriple(reln *spdx.Relationshi
 			DocumentRefID: "",
 			ElementRefID:  pkg.PackageSPDXIdentifier,
 		}
-		parser.doc.Packages[pkg.PackageSPDXIdentifier] = pkg
 
 	case SPDX_SPDX_ELEMENT:
 		// it shouldn't be associated with any other triple.