internal/gps: add isPreservedNonGoFile

Signed-off-by: Ibrahim AshShohail <[email protected]>

Author: ibrasho

internal/gps/prune.go

@@ -32,9 +32,24 @@ const (
 )
 
 var (
-	preservedNonGoFiles = []string{
-		"LICENSE",
-		"COPYING",
+	// licenseFilePrefixes is a list of name prefixes for licesnse files.
+	licenseFilePrefixes = []string{
+		"license",
+		"licence",
+		"copying",
+		"unlicense",
+		"copyright",
+		"copyleft",
+	}
+	// legalFileSubstrings contains substrings that are likey part of a legal
+	// declaration file.
+	legalFileSubstrings = []string{
+		"legal",
+		"notice",
+		"disclaimer",
+		"patent",
+		"third-party",
+		"thirdparty",
 	}
 )
 
@@ -215,22 +230,36 @@ func calculateNonGoFiles(baseDir string) ([]string, error) {
 			return nil
 		}
 
-		// Ignore preserved non-Go files. We check for prefix incase the file
-		// has an extension. For example: LICENSE.md.
-		for _, prefix := range preservedNonGoFiles {
-			if strings.HasPrefix(info.Name(), prefix) {
-				return nil
-			}
+		if !isPreservedNonGoFile(info.Name()) {
+			files = append(files, path)
 		}
 
-		files = append(files, path)
-
 		return nil
 	})
 
 	return files, err
 }
 
+// isPreservedNonGoFile checks if the file name idicates that the file should be
+// preserved. It assumes the file is not a Go file (doesn't have a .go suffix).
+func isPreservedNonGoFile(name string) bool {
+	name = strings.ToLower(name)
+
+	for _, prefix := range licenseFilePrefixes {
+		if strings.HasPrefix(name, prefix) {
+			return true
+		}
+	}
+
+	for _, substring := range legalFileSubstrings {
+		if strings.Contains(name, substring) {
+			return true
+		}
+	}
+
+	return false
+}
+
 // pruneGoTestFiles deletes all Go test files (*_test.go) within baseDirr.
 func pruneGoTestFiles(baseDir string, logger *log.Logger) error {
 	files, err := calculateGoTestFiles(baseDir)