Add support for older CentOS versions (6 & 5) by checking additional release files for information

Signed-off-by: Samuel Dacanay <sam.dacanay@anchore.com>

syft/distro/identify.go

@@ -35,6 +35,16 @@ var identityFiles = []parseEntry{
 		path: "/bin/busybox",
 		fn:   parseBusyBox,
 	},
+	{
+		// check for centos:6
+		path: "/etc/system-release-cpe",
+		fn:   parseSystemReleaseCPE,
+	},
+	{
+		// last ditch effort for determining older centos version distro information
+		path: "/etc/redhat-release",
+		fn:   parseRedhatRelease,
+	},
 }
 
 // Identify parses distro-specific files to determine distro metadata like version and release.
@@ -145,3 +155,48 @@ func parseBusyBox(contents string) *Distro {
 	}
 	return nil
 }
+
+// TODO: we should update parseSystemReleaseCPE to use the CPE struct, pkg.CPE, which requires a refactor to avoid a circular import:
+// TODO: pkg depends on distro to support pURLs. To avoid the circular import, either try to make pkg to not depend on distro (medium lift-ish)
+// TODO: or migrate the cpe code out of the pkg package (small lift).
+// example CPE: cpe:/o:centos:linux:6:GA
+var systemReleaseCpeMatcher = regexp.MustCompile(`cpe:\/o:(.*?):.*?:(.*?):.*?$`)
+
+// parseSystemReleaseCPE parses the older centos (6) file to determine distro metadata
+func parseSystemReleaseCPE(contents string) *Distro {
+	matches := systemReleaseCpeMatcher.FindAllStringSubmatch(contents, -1)
+	for _, match := range matches {
+		if len(match) < 3 {
+			log.Warnf("system release cpe does not match expected format")
+			return nil
+		}
+		// note: in SubMatches (capture groups), the 0th index is the full match string
+		// see https://pkg.go.dev/regexp#pkg-overview for more info
+		distro := assemble(match[1], match[2], "")
+		if distro != nil {
+			return distro
+		}
+	}
+	return nil
+}
+
+// example: "CentOS release 6.10 (Final)"
+var redhatReleaseMatcher = regexp.MustCompile(`(.*?)\srelease\s(\d\.\d+)`)
+
+// parseRedhatRelease is a fallback parsing method for determining distro information in older redhat versions
+func parseRedhatRelease(contents string) *Distro {
+	matches := redhatReleaseMatcher.FindAllStringSubmatch(contents, -1)
+	for _, match := range matches {
+		if len(match) < 3 {
+			log.Warnf("failed to parse redhat-release file, unexpected format")
+			return nil
+		}
+		// note: in SubMatches (capture groups), the 0th index is the full match string
+		// see https://pkg.go.dev/regexp#pkg-overview for more info
+		distro := assemble(strings.ToLower(match[1]), match[2], "")
+		if distro != nil {
+			return distro
+		}
+	}
+	return nil
+}

syft/distro/identify_test.go

@@ -2,6 +2,7 @@ package distro
 
 import (
 	"fmt"
+	hashiVer "github.com/hashicorp/go-version"
 	"io/ioutil"
 	"os"
 	"testing"
@@ -103,6 +104,16 @@ func TestIdentifyDistro(t *testing.T) {
 			fixture: "test-fixtures/partial-fields/missing-version",
 			Type:    UnknownDistroType,
 		},
+		{
+			fixture: "test-fixtures/os/centos6",
+			Type:    CentOS,
+			Version: "6.0.0",
+		},
+		{
+			fixture: "test-fixtures/os/centos5",
+			Type:    CentOS,
+			Version: "5.7.0",
+		},
 	}
 
 	observedDistros := internal.NewStringSet()
@@ -205,18 +216,7 @@ func TestParseOsRelease(t *testing.T) {
 
 	for _, test := range tests {
 		name := fmt.Sprintf("%s:%s", test.name, test.RawVersion)
-		fixture, err := os.Open(test.fixture)
-		if err != nil {
-			t.Fatalf("could not open test fixture=%s: %+v", test.fixture, err)
-		}
-		defer fixture.Close()
-
-		b, err := ioutil.ReadAll(fixture)
-		if err != nil {
-			t.Fatalf("unable to read fixture file: %+v", err)
-		}
-
-		contents := string(b)
+		contents := retrieveFixtureContentsAsString(test.fixture, t)
 
 		t.Run(name, func(t *testing.T) {
 			distro := parseOsRelease(contents)
@@ -245,18 +245,7 @@ func TestParseOsReleaseFailures(t *testing.T) {
 
 	for _, test := range tests {
 		name := fmt.Sprintf("%s:%s", test.name, test.fixture)
-		fixture, err := os.Open(test.fixture)
-		if err != nil {
-			t.Fatalf("could not open test fixture=%s: %+v", test.fixture, err)
-		}
-		defer fixture.Close()
-
-		b, err := ioutil.ReadAll(fixture)
-		if err != nil {
-			t.Fatalf("unable to read fixture file: %+v", err)
-		}
-
-		contents := string(b)
+		contents := retrieveFixtureContentsAsString(test.fixture, t)
 
 		t.Run(name, func(t *testing.T) {
 			distro := parseOsRelease(contents)
@@ -265,5 +254,101 @@ func TestParseOsReleaseFailures(t *testing.T) {
 			}
 		})
 	}
-
+}
+
+func TestParseSystemReleaseCPE(t *testing.T) {
+	centos6Version, _ := hashiVer.NewVersion("6")
+	tests := []struct {
+		fixture  string
+		name     string
+		expected *Distro
+	}{
+		{
+			fixture: "test-fixtures/os/centos6/etc/system-release-cpe",
+			name:    "Centos 6",
+			expected: &Distro{
+				Type:       CentOS,
+				Version:    centos6Version,
+				RawVersion: "6",
+			},
+		},
+		{
+			fixture:  "test-fixtures/bad-system-release-cpe",
+			name:     "Centos 6 Bad CPE",
+			expected: nil,
+		},
+	}
+
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			contents := retrieveFixtureContentsAsString(test.fixture, t)
+			actual := parseSystemReleaseCPE(contents)
+
+			if test.expected == nil {
+				assert.Nil(t, actual)
+				return
+			}
+
+			// not comparing the full distro object because the hashiVer is a pointer
+			assert.Equal(t, test.expected.Type, actual.Type)
+			assert.Equal(t, &test.expected.Version, &actual.Version)
+			assert.Equal(t, test.expected.RawVersion, actual.RawVersion)
+		})
+	}
+}
+
+func TestParseRedhatRelease(t *testing.T) {
+	centos5Version, _ := hashiVer.NewVersion("5.7")
+	tests := []struct {
+		fixture  string
+		name     string
+		expected *Distro
+	}{
+		{
+			fixture: "test-fixtures/os/centos5/etc/redhat-release",
+			name:    "Centos 5",
+			expected: &Distro{
+				Type:       CentOS,
+				Version:    centos5Version,
+				RawVersion: "5.7",
+			},
+		},
+		{
+			fixture:  "test-fixtures/bad-redhat-release",
+			name:     "Centos 5 Bad Redhat Release",
+			expected: nil,
+		},
+	}
+
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			contents := retrieveFixtureContentsAsString(test.fixture, t)
+			actual := parseRedhatRelease(contents)
+
+			if test.expected == nil {
+				assert.Nil(t, actual)
+				return
+			}
+
+			// not comparing the full distro object because the hashiVer is a pointer
+			assert.Equal(t, test.expected.Type, actual.Type)
+			assert.Equal(t, &test.expected.Version, &actual.Version)
+			assert.Equal(t, test.expected.RawVersion, actual.RawVersion)
+		})
+	}
+}
+
+func retrieveFixtureContentsAsString(fixturePath string, t *testing.T) string {
+	fixture, err := os.Open(fixturePath)
+	if err != nil {
+		t.Fatalf("could not open test fixture=%s: %+v", fixturePath, err)
+	}
+	defer fixture.Close()
+
+	b, err := ioutil.ReadAll(fixture)
+	if err != nil {
+		t.Fatalf("unable to read fixture file: %+v", err)
+	}
+
+	return string(b)
 }

syft/distro/test-fixtures/bad-redhat-release

@@ -0,0 +1 @@
+CentOS release 5 (Final)

syft/distro/test-fixtures/bad-system-release-cpe

@@ -0,0 +1 @@
+cpe:/o:centos:6:GA

syft/distro/test-fixtures/os/centos5/etc/redhat-release

@@ -0,0 +1 @@
+CentOS release 5.7 (Final)

syft/distro/test-fixtures/os/centos6/etc/system-release-cpe

@@ -0,0 +1 @@
+cpe:/o:centos:linux:6:GA