add options for validating licenses that limits what is considered valid

cmd/doc.go

@@ -10,6 +10,7 @@ spdxexp/license.go file.
 
 Command to run all extractions (run command from the /cmd directory):
 
+	cd cmd
 	go run . extract -l -e
 
 Usage options:

cmd/exceptions.go

@@ -3,7 +3,9 @@ package main
 import (
 	"encoding/json"
 	"fmt"
+	"go/format"
 	"os"
+	"strings"
 )
 
 type ExceptionData struct {
@@ -54,6 +56,27 @@ func extractExceptionLicenseIDs() error {
 // Code generated by go-spdx cmd/exceptions.go. DO NOT EDIT.
 // Source: https://github.com/spdx/license-list-data specifies official SPDX license list.
 
+import "strings"
+
+// IsException does a case-insensitive lookup for the exception id in the exceptions map.
+// It returns true and the case-sensitive ID if found, otherwise false and the original id.
+func IsException(id string) (bool, string) {
+	foundID, ok := exceptionsMap[strings.ToUpper(id)]
+	if ok {
+		return true, foundID
+	}
+	return false, id
+}
+
+// GetExceptionsMap returns a map of exception license IDs keyed by uppercase ID.
+func GetExceptionsMap() map[string]string {
+   	copied := make(map[string]string, len(exceptionsMap))  
+    for k, v := range exceptionsMap {  
+        copied[k] = v  
+    }  
+    return copied
+}
+
 // GetExceptions returns a slice of exception license IDs.
 func GetExceptions() []string {
 	return []string{
@@ -64,8 +87,23 @@ func GetExceptions() []string {
 	}
 	getExceptionsContents = append(getExceptionsContents, `	}
 }
+
+`...)
+
+	getExceptionsContents = append(getExceptionsContents, `var exceptionsMap = map[string]string{
+`...)
+	for _, id := range exceptionLicenseIDs {
+		getExceptionsContents = append(getExceptionsContents, `		"`+strings.ToUpper(id)+`": "`+id+`",
+`...)
+	}
+	getExceptionsContents = append(getExceptionsContents, `}
 `...)
 
+	getExceptionsContents, err = format.Source(getExceptionsContents)
+	if err != nil {
+		return fmt.Errorf("format generated get_exceptions.go: %w", err)
+	}
+
 	err = os.WriteFile("../spdxexp/spdxlicenses/get_exceptions.go", getExceptionsContents, 0600)
 	if err != nil {
 		return err

cmd/license.go

@@ -3,7 +3,9 @@ package main
 import (
 	"encoding/json"
 	"fmt"
+	"go/format"
 	"os"
+	"strings"
 )
 
 type LicenseData struct {
@@ -59,6 +61,27 @@ func extractLicenseIDs() error {
 // Code generated by go-spdx cmd/license.go. DO NOT EDIT.
 // Source: https://github.com/spdx/license-list-data specifies official SPDX license list.
 
+import "strings"
+
+// IsActiveLicense does a case-insensitive lookup for the license id in the active licenses map.
+// It returns true and the case-sensitive ID if found, otherwise false and the original id.
+func IsActiveLicense(id string) (bool, string) {
+	foundID, ok := licensesMap[strings.ToUpper(id)]
+	if ok {
+		return true, foundID
+	}
+	return false, id
+}
+
+// GetLicensesMap returns a map of active license IDs keyed by uppercase ID.
+func GetLicensesMap() map[string]string {
+   	copied := make(map[string]string, len(licensesMap))  
+    for k, v := range licensesMap {  
+        copied[k] = v  
+    }  
+    return copied
+}
+
 // GetLicenses returns a slice of active license IDs.
 func GetLicenses() []string {
 	return []string{
@@ -69,8 +92,23 @@ func GetLicenses() []string {
 	}
 	getLicensesContents = append(getLicensesContents, `	}
 }
+
+`...)
+
+	getLicensesContents = append(getLicensesContents, `var licensesMap = map[string]string{
+`...)
+	for _, id := range activeLicenseIDs {
+		getLicensesContents = append(getLicensesContents, `		"`+strings.ToUpper(id)+`": "`+id+`",
+`...)
+	}
+	getLicensesContents = append(getLicensesContents, `}
 `...)
 
+	getLicensesContents, err = format.Source(getLicensesContents)
+	if err != nil {
+		return fmt.Errorf("format generated get_licenses.go: %w", err)
+	}
+
 	err = os.WriteFile("../spdxexp/spdxlicenses/get_licenses.go", getLicensesContents, 0600)
 	if err != nil {
 		return err
@@ -83,6 +121,27 @@ func GetLicenses() []string {
 // Code generated by go-spdx cmd/license.go. DO NOT EDIT.
 // Source: https://github.com/spdx/license-list-data specifies official SPDX license list.
 
+import "strings"
+
+// IsDeprecatedLicense does a case-insensitive lookup for the license id in the deprecated licenses map.
+// It returns true and the case-sensitive ID if found, otherwise false and the original id.
+func IsDeprecatedLicense(id string) (bool, string) {
+	foundID, ok := deprecatedMap[strings.ToUpper(id)]
+	if ok {
+		return true, foundID
+	}
+	return false, id
+}
+
+// GetDeprecatedMap returns a map of deprecated license IDs keyed by uppercase ID.
+func GetDeprecatedMap() map[string]string {
+   	copied := make(map[string]string, len(deprecatedMap))  
+    for k, v := range deprecatedMap {  
+        copied[k] = v  
+    }  
+    return copied
+}
+
 // GetDeprecated returns a slice of deprecated license IDs.
 func GetDeprecated() []string {
 	return []string{
@@ -93,8 +152,23 @@ func GetDeprecated() []string {
 	}
 	getDeprecatedContents = append(getDeprecatedContents, `	}
 }
+
+`...)
+
+	getDeprecatedContents = append(getDeprecatedContents, `var deprecatedMap = map[string]string{
+`...)
+	for _, id := range deprecatedLicenseIDs {
+		getDeprecatedContents = append(getDeprecatedContents, `		"`+strings.ToUpper(id)+`": "`+id+`",
+`...)
+	}
+	getDeprecatedContents = append(getDeprecatedContents, `}
 `...)
 
+	getDeprecatedContents, err = format.Source(getDeprecatedContents)
+	if err != nil {
+		return fmt.Errorf("format generated get_deprecated.go: %w", err)
+	}
+
 	err = os.WriteFile("../spdxexp/spdxlicenses/get_deprecated.go", getDeprecatedContents, 0600)
 	if err != nil {
 		return err

spdxexp/benchmark_satisfies_test.go

@@ -12,14 +12,18 @@ type satisfiesBenchmarkScenario struct {
 
 var satisfiesBenchmarkScenarios = []satisfiesBenchmarkScenario{
 	// Scenario order is used as-is in the summary table.
-	{"MIT", "MIT"},
-	{"mit", "mit"},
-	{"Apache-2.0", "Apache-2.0"},
-	{"Zed", "Zed"},
-	{"MIT AND Apache-2.0", "MIT AND Apache-2.0"},
-	{"MIT AND Apache-2.0 OR Zed", "MIT AND Apache-2.0 OR Zed"},
-	{"GPL-2.0-or-later", "GPL-2.0-or-later"},
-	{"GPL-2.0+", "GPL-2.0+"},
+	{"MIT--exact", "MIT"},
+	{"mit--caseinsensitive", "mit"},
+	{"mit--extra-space", " MIT  "},
+	{"Apache-2.0--active-early", "Apache-2.0"},
+	{"Zed--active-end", "Zed"},
+	{"MIT AND Apache-2.0--complex", "MIT AND Apache-2.0"},
+	{"MIT AND Apache-2.0 OR Zed--complex", "MIT AND Apache-2.0 OR Zed"},
+	{"BSD-2-Clause-FreeBSD--deprecated", "BSD-2-Clause-FreeBSD"},
+	{"GPL-2.0-or-later--range", "GPL-2.0-or-later"},
+	{"Apache-1.0+--plus-range", "Apache-1.0+"},
+	{"LicenseRef-scancode-adobe-postscript", "LicenseRef-scancode-adobe-postscript"},
+	{"DocumentRef-spdx-tool-1.2:LicenseRef-MIT-Style-2", "DocumentRef-spdx-tool-1.2:LicenseRef-MIT-Style-2"},
 }
 
 func BenchmarkSatisfies(b *testing.B) {

spdxexp/benchmark_validate_licenses_test.go

@@ -12,14 +12,18 @@ type validateLicensesBenchmarkScenario struct {
 
 var validateLicensesBenchmarkScenarios = []validateLicensesBenchmarkScenario{
 	// Scenario order is used as-is in the summary table.
-	{"MIT", []string{"MIT"}},
-	{"mit", []string{"mit"}},
-	{"Apache-2.0", []string{"Apache-2.0"}},
-	{"Zed", []string{"Zed"}},
-	{"MIT AND Apache-2.0", []string{"MIT", "Apache-2.0"}},
-	{"MIT AND Apache-2.0 OR Zed", []string{"MIT", "Apache-2.0", "Zed"}},
-	{"GPL-2.0-or-later", []string{"GPL-2.0-or-later"}},
-	{"GPL-2.0+", []string{"GPL-2.0+"}},
+	{"MIT--exact", []string{"MIT"}},
+	{"mit--caseinsensitive", []string{"mit"}},
+	{"mit--extra-space", []string{" MIT  "}},
+	{"Apache-2.0--active-early", []string{"Apache-2.0"}},
+	{"Zed--active-end", []string{"Zed"}},
+	{"MIT AND Apache-2.0--complex", []string{"MIT", "Apache-2.0"}},
+	{"MIT AND Apache-2.0 OR Zed--complex", []string{"MIT", "Apache-2.0", "Zed"}},
+	{"BSD-2-Clause-FreeBSD--deprecated", []string{"BSD-2-Clause-FreeBSD"}},
+	{"GPL-2.0-or-later--range", []string{"GPL-2.0-or-later"}},
+	{"Apache-1.0+--plus-range", []string{"Apache-1.0+"}},
+	{"LicenseRef-scancode-adobe-postscript", []string{"LicenseRef-scancode-adobe-postscript"}},
+	{"DocumentRef-spdx-tool-1.2:LicenseRef-MIT-Style-2", []string{"DocumentRef-spdx-tool-1.2:LicenseRef-MIT-Style-2"}},
 }
 
 func BenchmarkValidateLicenses(b *testing.B) {

spdxexp/license.go

@@ -8,7 +8,7 @@ import (
 
 // activeLicense returns true if the id is an active license.
 func activeLicense(id string) (bool, string) {
-	return inLicenseList(spdxlicenses.GetLicenses(), id)
+	return spdxlicenses.IsActiveLicense(id)
 }
 
 // ActiveLicense returns true if the id is an active license.
@@ -18,22 +18,12 @@ func ActiveLicense(id string) (bool, string) {
 
 // deprecatedLicense returns true if the id is a deprecated license.
 func deprecatedLicense(id string) (bool, string) {
-	return inLicenseList(spdxlicenses.GetDeprecated(), id)
+	return spdxlicenses.IsDeprecatedLicense(id)
 }
 
 // exceptionLicense returns true if the id is an exception license.
 func exceptionLicense(id string) (bool, string) {
-	return inLicenseList(spdxlicenses.GetExceptions(), id)
-}
-
-// inLicenseList looks for id in the list of licenses.  The check is case-insensitive (e.g. "mit" will match "MIT").
-func inLicenseList(licenses []string, id string) (bool, string) {
-	for _, license := range licenses {
-		if strings.EqualFold(license, id) {
-			return true, license
-		}
-	}
-	return false, id
+	return spdxlicenses.IsException(id)
 }
 
 const (