Fixed toolkit's handling of RPMs with epoch values in their name

toolkit/tools/graphpkgfetcher/graphpkgfetcher.go

@@ -552,7 +552,8 @@ func assignRPMPath(node *pkggraph.PkgNode, outDir string, resolvedPackages []str
 }
 
 func rpmPackageToRPMPath(rpmPackage, outDir string) string {
-	// Construct the rpm path of the cloned package.
+	// Construct the RPM path of the cloned package.
+	rpmPackage = rpm.StripEpochFromPackageFullQualifiedName(rpmPackage)
 	rpmName := fmt.Sprintf("%s.rpm", rpmPackage)
 	return filepath.Join(outDir, rpmName)
 }

toolkit/tools/internal/rpm/rpm.go

@@ -63,11 +63,25 @@ const (
 )
 
 const (
-	installedRPMRegexRPMIndex        = 1
-	installedRPMRegexVersionIndex    = 2
-	installedRPMRegexArchIndex       = 3
-	installedRPMRegexExpectedMatches = 4
+	packageFQNRegexMatchSubString  = iota
+	packageFQNRegexNameIndex       = iota
+	packageFQNRegexEpochIndex      = iota
+	packageFQNRegexVersionIndex    = iota
+	packageFQNRegexReleaseIndex    = iota
+	packageFQNRegexArchIndex       = iota
+	packageFQNRegexExtensionIndex  = iota
+	packageFQNRegexExpectedMatches = iota
+)
+
+const (
+	installedRPMRegexMatchSubString  = iota
+	installedRPMRegexRPMIndex        = iota
+	installedRPMRegexVersionIndex    = iota
+	installedRPMRegexArchIndex       = iota
+	installedRPMRegexExpectedMatches = iota
+)
 
+const (
 	rpmProgram      = "rpm"
 	rpmSpecProgram  = "rpmspec"
 	rpmBuildProgram = "rpmbuild"
@@ -83,6 +97,25 @@ var (
 	// It works multi-line strings containing the whole file content, thus the need for the 'm' flag.
 	checkSectionRegex = regexp.MustCompile(`(?m)^\s*%check`)
 
+	// A full qualified RPM name contains the package name, epoch, version, release, architecture, and extension.
+	// Optional fields:
+	// 	- epoch,
+	// 	- architecture.
+	//	- "rpm" extension.
+	//
+	// Sample match:
+	//
+	//	pkg-name-0:1.2.3-4.azl3.x86_64.rpm
+	//
+	// Groups can be used to split it into:
+	//   - name:			pkg-name
+	//   - epoch:			0
+	//   - version:			1.2.3
+	//   - release:			4.azl3
+	//   - architecture:	x86_64
+	//   - extension:		rpm
+	packageFQNRegex = regexp.MustCompile(`^\s*(\S+[^-])-(?:(\d+):)?(\d[^-:_]*)-(\d+(?:[^-\s]*?))(?:\.(noarch|x86_64|aarch64|src))?(?:\.(rpm))?\s*$`)
+
 	// Output from 'rpm' prints installed RPMs in a line with the following format:
 	//
 	//	D: ========== +++ [name]-([epoch]:)[version]-[release].[distribution] [architecture]-linux [hex_value]
@@ -187,19 +220,15 @@ func getMacroDirWithFallback(allowDefault bool) (macroDir string, err error) {
 func ExtractNameFromRPMPath(rpmFilePath string) (packageName string, err error) {
 	baseName := filepath.Base(rpmFilePath)
 
+	matches := packageFQNRegex.FindStringSubmatch(baseName)
+
 	// If the path is invalid, return empty string. We consider any string that has at least 1 '-' characters valid.
-	if !strings.Contains(baseName, "-") {
+	if matches == nil {
 		err = fmt.Errorf("invalid RPM file path (%s), can't extract name", rpmFilePath)
 		return
 	}
 
-	rpmFileSplit := strings.Split(baseName, "-")
-	packageName = strings.Join(rpmFileSplit[:len(rpmFileSplit)-2], "-")
-	if packageName == "" {
-		err = fmt.Errorf("invalid RPM file path (%s), can't extract name", rpmFilePath)
-		return
-	}
-	return
+	return matches[packageFQNRegexNameIndex], nil
 }
 
 // getCommonBuildArgs will generate arguments to pass to 'rpmbuild'.
@@ -526,10 +555,6 @@ func extractCompetingPackageInfoFromLine(line string) (match bool, pkgName strin
 		pkgName := matches[installedRPMRegexRPMIndex]
 		version := matches[installedRPMRegexVersionIndex]
 		arch := matches[installedRPMRegexArchIndex]
-		// Names should not contain the epoch, strip everything before the ":"" in the string. "Version": "0:1.2-3", becomes "1.2-3"
-		if strings.Contains(version, ":") {
-			version = strings.Split(version, ":")[1]
-		}
 
 		return true, fmt.Sprintf("%s-%s.%s", pkgName, version, arch)
 	}
@@ -636,6 +661,39 @@ func BuildCompatibleSpecsList(baseDir string, inputSpecPaths []string, defines m
 	return filterCompatibleSpecs(specPaths, defines)
 }
 
+// StripEpochFromPackageFullQualifiedName removes the epoch from a package full qualified name if it is present.
+// Example:
+//
+//	"pkg-name-0:1.2.3-4.azl3.x86_64" -> "pkg-name-1.2.3-4.azl3.x86_64"
+func StripEpochFromPackageFullQualifiedName(packageFQN string) string {
+	var packageFQNBuilder strings.Builder
+
+	matches := packageFQNRegex.FindStringSubmatch(packageFQN)
+	if matches == nil {
+		return packageFQN
+	}
+
+	packageFQNBuilder.WriteString(matches[packageFQNRegexNameIndex])
+	packageFQNBuilder.WriteString("-")
+
+	packageFQNBuilder.WriteString(matches[packageFQNRegexVersionIndex])
+	packageFQNBuilder.WriteString("-")
+
+	packageFQNBuilder.WriteString(matches[packageFQNRegexReleaseIndex])
+
+	if matches[packageFQNRegexArchIndex] != "" {
+		packageFQNBuilder.WriteString(".")
+		packageFQNBuilder.WriteString(matches[packageFQNRegexArchIndex])
+	}
+
+	if matches[packageFQNRegexExtensionIndex] != "" {
+		packageFQNBuilder.WriteString(".")
+		packageFQNBuilder.WriteString(matches[packageFQNRegexExtensionIndex])
+	}
+
+	return packageFQNBuilder.String()
+}
+
 // TestRPMFromSRPM builds an RPM from the given SRPM and runs its '%check' section SRPM file
 // but it does not generate any RPM packages.
 func TestRPMFromSRPM(srpmFile, outArch string, defines map[string]string) (err error) {

toolkit/tools/internal/rpm/rpm_test.go

@@ -470,7 +470,7 @@ func TestConflictingPackageRegex(t *testing.T) {
 			name:           "perl with epoch",
 			inputLine:      "D: ========== +++ perl-4:5.34.1-489.cm2 x86_64-linux 0x0",
 			expectedMatch:  true,
-			expectedOutput: "perl-5.34.1-489.cm2.x86_64",
+			expectedOutput: "perl-4:5.34.1-489.cm2.x86_64",
 		},
 		{
 			name:           "systemd no epoch",
@@ -494,3 +494,234 @@ func TestConflictingPackageRegex(t *testing.T) {
 		})
 	}
 }
+
+func TestPackageFQNRegexWithValidInput(t *testing.T) {
+	tests := []struct {
+		name           string
+		input          string
+		expectedGroups []string
+	}{
+		{
+			name:           "package with epoch and architecture",
+			input:          "pkg-name-0:1.2.3-4.azl3.x86_64.rpm",
+			expectedGroups: []string{"pkg-name", "0", "1.2.3", "4.azl3", "x86_64", "rpm"},
+		},
+		{
+			name:           "package with epoch and architecture but no '.rpm' suffix",
+			input:          "pkg-name-0:1.2.3-4.azl3.x86_64",
+			expectedGroups: []string{"pkg-name", "0", "1.2.3", "4.azl3", "x86_64", ""},
+		},
+		{
+			name:           "package without epoch, and architecture",
+			input:          "pkg-name-1.2.3-4.azl3.rpm",
+			expectedGroups: []string{"pkg-name", "", "1.2.3", "4.azl3", "", "rpm"},
+		},
+		{
+			name:           "package with architecture but no epoch",
+			input:          "pkg-name-1.2.3-4.azl3.aarch64",
+			expectedGroups: []string{"pkg-name", "", "1.2.3", "4.azl3", "aarch64", ""},
+		},
+		{
+			name:           "package with epoch but no architecture",
+			input:          "pkg-name-0:1.2.3-4.azl3",
+			expectedGroups: []string{"pkg-name", "0", "1.2.3", "4.azl3", "", ""},
+		},
+		{
+			name:           "package without '.rpm' suffix",
+			input:          "pkg-name-1.2.3-4.azl3.x86_64",
+			expectedGroups: []string{"pkg-name", "", "1.2.3", "4.azl3", "x86_64", ""},
+		},
+		{
+			name:           "package with version containing the '+' character",
+			input:          "pkg-name-1.2.3+4-4.azl3.x86_64.rpm",
+			expectedGroups: []string{"pkg-name", "", "1.2.3+4", "4.azl3", "x86_64", "rpm"},
+		},
+		{
+			name:           "package with version containing the '~' character",
+			input:          "pkg-name-1.2.3~4-4.azl3.x86_64.rpm",
+			expectedGroups: []string{"pkg-name", "", "1.2.3~4", "4.azl3", "x86_64", "rpm"},
+		},
+		{
+			name:           "package with release containing two '.' characters",
+			input:          "pkg-name-1.2.3-4.5.azl3.x86_64.rpm",
+			expectedGroups: []string{"pkg-name", "", "1.2.3", "4.5.azl3", "x86_64", "rpm"},
+		},
+		{
+			name:           "package with release containing the '_' character",
+			input:          "pkg-name-1.2.3-45.az_l3.x86_64.rpm",
+			expectedGroups: []string{"pkg-name", "", "1.2.3", "45.az_l3", "x86_64", "rpm"},
+		},
+		{
+			name:           "package with release containing the `~` character",
+			input:          "pkg-name-1.2.3-45.azl3~2.x86_64.rpm",
+			expectedGroups: []string{"pkg-name", "", "1.2.3", "45.azl3~2", "x86_64", "rpm"},
+		},
+		{
+			name:           "package with double dash in name",
+			input:          "nvidia-container-toolkit-1.15.0-1.azl3.x86_64.rpm",
+			expectedGroups: []string{"nvidia-container-toolkit", "", "1.15.0", "1.azl3", "x86_64", "rpm"},
+		},
+		{
+			name:           "package with underscore in release",
+			input:          "nvidia-container-toolkit-550.54.15-2_5.15.162.2.1.azl3.x86_64.rpm",
+			expectedGroups: []string{"nvidia-container-toolkit", "", "550.54.15", "2_5.15.162.2.1.azl3", "x86_64", "rpm"},
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			matches := packageFQNRegex.FindStringSubmatch(tt.input)
+			assert.NotNil(t, matches)
+			assert.Equal(t, tt.expectedGroups, matches[1:])
+		})
+	}
+}
+
+func TestPackageFQNRegexWithInvalidInput(t *testing.T) {
+	tests := []struct {
+		name  string
+		input string
+	}{
+		{
+			name:  "package with missing version",
+			input: "pkg-name--4.azl3.x86_64.rpm",
+		},
+		{
+			name:  "package with missing release",
+			input: "pkg-name-1.2.3-.azl3.x86_64.rpm",
+		},
+		{
+			name:  "package with missing name",
+			input: "-1.2.3-4.azl3.x86_64.rpm",
+		},
+		{
+			name:  "package with only hyphen",
+			input: "-",
+		},
+		{
+			name:  "package with version not beginning with a digit",
+			input: "pkg-name-0:a1.2.3-4.azl3.x86_64.rpm",
+		},
+		{
+			name:  "package with release not beginning with a digit",
+			input: "pkg-name-0:1.2.3-D4.azl3.x86_64.rpm",
+		},
+		{
+			name:  "package with epoch not beginning with a digit",
+			input: "pkg-name-0:1.2.3-D4.azl3.x86_64.rpm",
+		},
+		{
+			name:  "package with epoch unsupported architecture",
+			input: "pkg-name-0:1.2.3-D4.azl3.other_arch.rpm",
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			matches := packageFQNRegex.FindStringSubmatch(tt.input)
+			assert.Nil(t, matches)
+		})
+	}
+}
+
+func TestStripEpochFromPackageFullQualifiedNameWithValidInput(t *testing.T) {
+	tests := []struct {
+		name     string
+		input    string
+		expected string
+	}{
+		{
+			name:     "package with epoch and architecture",
+			input:    "pkg-name-0:1.2.3-4.azl3.x86_64.rpm",
+			expected: "pkg-name-1.2.3-4.azl3.x86_64.rpm",
+		},
+		{
+			name:     "package with epoch and architecture but no '.rpm' suffix",
+			input:    "pkg-name-0:1.2.3-4.azl3.x86_64",
+			expected: "pkg-name-1.2.3-4.azl3.x86_64",
+		},
+		{
+			name:     "package with epoch but no architecture",
+			input:    "pkg-name-0:1.2.3-4.azl3",
+			expected: "pkg-name-1.2.3-4.azl3",
+		},
+		{
+			name:     "package with architecture but no epoch",
+			input:    "pkg-name-1.2.3-4.azl3.aarch64",
+			expected: "pkg-name-1.2.3-4.azl3.aarch64",
+		},
+		{
+			name:     "package without epoch, and architecture",
+			input:    "pkg-name-1.2.3-4.azl3.rpm",
+			expected: "pkg-name-1.2.3-4.azl3.rpm",
+		},
+		{
+			name:     "package with version containing the '+' character",
+			input:    "pkg-name-1.2.3+4-4.azl3.x86_64.rpm",
+			expected: "pkg-name-1.2.3+4-4.azl3.x86_64.rpm",
+		},
+		{
+			name:     "package with version containing the '~' character",
+			input:    "pkg-name-1.2.3~4-4.azl3.x86_64.rpm",
+			expected: "pkg-name-1.2.3~4-4.azl3.x86_64.rpm",
+		},
+		{
+			name:     "package with release containing two '.' characters",
+			input:    "pkg-name-1.2.3-4.5.azl3.x86_64.rpm",
+			expected: "pkg-name-1.2.3-4.5.azl3.x86_64.rpm",
+		},
+		{
+			name:     "package with release containing the '_' character",
+			input:    "pkg-name-1.2.3-4_5.azl3.x86_64.rpm",
+			expected: "pkg-name-1.2.3-4_5.azl3.x86_64.rpm",
+		},
+		{
+			name:     "package with release containing the `~` character",
+			input:    "pkg-name-1.2.3-4~5.azl3.x86_64.rpm",
+			expected: "pkg-name-1.2.3-4~5.azl3.x86_64.rpm",
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			actual := StripEpochFromPackageFullQualifiedName(tt.input)
+			assert.Equal(t, tt.expected, actual)
+		})
+	}
+}
+
+func TestStripEpochFromPackageFullQualifiedNameWithInvalidInput(t *testing.T) {
+	tests := []struct {
+		name     string
+		input    string
+		expected string
+	}{
+		{
+			name:     "invalid package name",
+			input:    "invalid-package-name",
+			expected: "invalid-package-name",
+		},
+		{
+			name:     "empty package name",
+			input:    "",
+			expected: "",
+		},
+		{
+			name:     "package name with only hyphens",
+			input:    "----",
+			expected: "----",
+		},
+		{
+			name:     "package name with spaces",
+			input:    "pkg name-1.2.3-4.azl3.x86_64.rpm",
+			expected: "pkg name-1.2.3-4.azl3.x86_64.rpm",
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			actual := StripEpochFromPackageFullQualifiedName(tt.input)
+			assert.Equal(t, tt.expected, actual)
+		})
+	}
+}