feat(spdx): Expose the validation of license text files' existence

ORT has a complicate logic to search for a file containing the full license
text of a given license: several search paths are tried. Unfortunately,
the latter are not exposed, making it hard for an alternative
`LicenseTextProvider` to have the same logic. Additionally, the current
code assumes that the license files are on a local storage and the I/Os are
therefore cheap. To allow the current utility function to be used with a
`LicenseTextProvider` that works with license text files stored on a
network storage, this commit exposes the validation function.

Signed-off-by: Nicolas Nobelis <[email protected]>

Author: nnobelis

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

@@ -116,15 +116,42 @@ fun getLicenseText(
     licenseTextDirectories: List<File> = emptyList()
 ): String? = getLicenseTextReader(id, handleExceptions, addScanCodeLicenseTextsDir(licenseTextDirectories))?.invoke()
 
+/**
+ * Get a lambda that allows to read the full text of the license with the provided SPDX [id]. Several locations are
+ * searched for the license text: [licenseTextDirectories], the ScanCode license texts directory, the ORT's list of SPDX
+ * licenses and the resources of this module. If [handleExceptions] is enabled, the [id] may also refer to a license
+ * with an SPDX exception.
+ */
 fun getLicenseTextReader(
     id: String,
     handleExceptions: Boolean = false,
     licenseTextDirectories: List<File> = emptyList()
+): (() -> String)? {
+    return getLicenseTextReader(id, handleExceptions, licenseTextDirectories) { dir, filename ->
+        dir.resolve(filename).takeIf { it.isFile }
+    }
+}
+
+/**
+ * Get a lambda that allows to read the full text of the license with the provided SPDX [id]. Several locations are
+ * searched for the license text: [licenseTextDirectories], the ScanCode license texts directory, the ORT's list of SPDX
+ * licenses and the resources of this module. If [handleExceptions] is enabled, the [id] may also refer to a license
+ * with an SPDX exception.
+ * [validateFile] is a lambda that is called with the directory and filename to look if the license file exists. It
+ * should return the file if it exists, null otherwise.
+ */
+fun getLicenseTextReader(
+    id: String,
+    handleExceptions: Boolean = false,
+    licenseTextDirectories: List<File> = emptyList(),
+    validateFile: (File, String) -> File?
 ): (() -> String)? {
     return if (id.startsWith(LICENSE_REF_PREFIX)) {
         getLicenseTextResource(id)?.let { { it.readText() } }
             ?: addScanCodeLicenseTextsDir(licenseTextDirectories).firstNotNullOfOrNull { dir ->
-                getLicenseTextFile(id, dir)?.let { file ->
+                getLicenseTextFile(id) { filename ->
+                    validateFile(dir, filename)
+                }?.let { file ->
                     {
                         file.readText().removeYamlFrontMatter()
                     }
@@ -140,7 +167,7 @@ private fun getLicenseTextResource(id: String): URL? = object {}.javaClass.getRe
 
 private val LICENSE_REF_FILENAME_REGEX by lazy { Regex("^$LICENSE_REF_PREFIX\\w+-") }
 
-private fun getLicenseTextFile(id: String, dir: File): File? =
+fun getLicenseTextFile(id: String, validateFile: (String) -> File?): File? =
     id.replace(LICENSE_REF_FILENAME_REGEX, "").let { idWithoutLicenseRefNamespace ->
         listOfNotNull(
             id,
@@ -153,7 +180,7 @@ private fun getLicenseTextFile(id: String, dir: File): File? =
                 id == "LicenseRef-scancode-x11-xconsortium-veillard"
             }
         ).firstNotNullOfOrNull { filename ->
-            dir.resolve(filename).takeIf { it.isFile }
+            validateFile(filename)
         }
     }
 

utils/spdx/src/test/kotlin/UtilsTest.kt

@@ -198,6 +198,22 @@ class UtilsTest : WordSpec() {
             }
         }
 
+        "getLicenseTextReader provided a custom dir" should {
+            "call the custom validation lambda if provided" {
+                val id = "LicenseRef-ort-abc"
+                val text = "a\nb\nc"
+                val candidateFilenames = mutableListOf<String>()
+
+                setupTempFile(id, text)
+
+                getLicenseTextReader(id, true, listOf(tempDir)) { directory, filename ->
+                    candidateFilenames += filename
+                    directory.resolve(filename).takeIf { it.isFile }
+                }?.invoke() shouldBe text
+                candidateFilenames shouldBe listOf("LicenseRef-ort-abc")
+            }
+        }
+
         "removeYamlFrontMatter" should {
             "remove a YAML front matter" {
                 val text = """