ScanCode: Gracefully handle underscores in license keys

ScanCode may return SPDX license keys containing underscores characters
which is not allwed, see [1]. This results in ORT's scanner crashing
due to an SpdxException when it tries to parse the SPDX license key.

This issue has first occured in 2020 and been fixed by [2]. It got
re-introduced recently by [3].

Fix the issue based on the idea of the orignal fix [2]. The touched
function becomes less efficient. Delierately don't refactor for
efficiency because this fix can be reverted as soon as [3] is fixed.

[1] aboutcode-org/scancode-toolkit#2813
[2] fb0370f
[3] #4523

Signed-off-by: Frank Viernau <frank.viernau@here.com>

scanner/src/main/kotlin/scanners/scancode/ScanCodeResultParser.kt

@@ -209,7 +209,7 @@ private fun getLicenseFindings(result: JsonNode, parseExpressions: Boolean): Lis
  */
 private fun getSpdxLicenseId(license: JsonNode): String {
     // There is a bug in ScanCode 3.0.2 that returns an empty string instead of null for licenses unknown to SPDX.
-    val id = license["spdx_license_key"].textValueOrEmpty()
+    val id = license["spdx_license_key"].textValueOrEmpty().replace('_', '-')
 
     // For regular SPDX IDs, return early here.
     if (id.isNotEmpty() && !id.startsWith(LICENSE_REF_PREFIX)) return id

utils/spdx/src/main/kotlin/Utils.kt

@@ -160,17 +160,23 @@ private fun getLicenseTextResource(id: String): URL? =
 
 private val LICENSE_REF_FILENAME_REGEX by lazy { Regex("^LicenseRef-\\w+-") }
 
-private fun getLicenseTextFile(id: String, dir: File): File? =
-    id.replace(LICENSE_REF_FILENAME_REGEX, "").let { idWithoutLicenseRefNamespace ->
-        sequenceOf(
-            id,
-            id.removePrefix("LicenseRef-"),
-            idWithoutLicenseRefNamespace,
-            "$idWithoutLicenseRefNamespace.LICENSE"
-        ).mapNotNull { filename ->
-            dir.resolve(filename).takeIf { it.isFile }
-        }.firstOrNull()
+private fun getLicenseTextFile(id: String, dir: File): File? {
+    val idWithoutLicenseRefNamespace = id.replace(LICENSE_REF_FILENAME_REGEX, "")
+    val candidateFilenames = setOf(
+        id,
+        id.removePrefix("LicenseRef-"),
+        idWithoutLicenseRefNamespace,
+        "$idWithoutLicenseRefNamespace.LICENSE"
+    )
+
+    dir.listFiles().filter { it.isFile }.forEach { file ->
+        // Normalize the filenames to work around https://github.com/nexB/scancode-toolkit/issues/2813.
+        val normalizedFilename = "${file.nameWithoutExtension.replace('_', '-')}.${file.extension}"
+        if (normalizedFilename in candidateFilenames) return file
     }
 
+    return null
+}
+
 private fun addScanCodeLicenseTextsDir(licenseTextDirectories: List<File>): List<File> =
     (listOfNotNull(scanCodeLicenseTextDir) + licenseTextDirectories).distinct()