fix(scancode): Filter out non-originary findings that are just refere…

…nces License findings that are references to license findings in other files are now ignored, because they already appear as findings for those other files. Fixes #8190. Signed-off-by: Sebastian Schuberth <sebastian@doubleopen.org>
oss-review-toolkit · Apr 2, 2024 · 8a19a58 · 8a19a58
1 parent 7df19c3
commit 8a19a58
Show file tree

Hide file tree

Showing 3 changed files with 11 additions and 7 deletions.
diff --git a/plugins/scanners/scancode/src/main/kotlin/ScanCodeResultModel.kt b/plugins/scanners/scancode/src/main/kotlin/ScanCodeResultModel.kt
@@ -157,7 +157,8 @@ sealed interface LicenseEntry {
         override val score: Float,
         override val startLine: Int,
         override val endLine: Int,
-        override val licenseExpression: String
+        override val licenseExpression: String,
+        val fromFile: String? = null // This might be missing in JSON.
     ) : LicenseEntry
 }
 

diff --git a/plugins/scanners/scancode/src/main/kotlin/ScanCodeResultModelMapper.kt b/plugins/scanners/scancode/src/main/kotlin/ScanCodeResultModelMapper.kt
@@ -64,6 +64,7 @@ fun ScanCodeResult.toScanSummary(preferFileLicense: Boolean = false): ScanSummar
     val issues = mutableListOf<Issue>()
 
     val header = headers.single()
+    val inputName = header.options.input.first().substringAfterLast('/')
 
     val outputFormatVersion = header.outputFormatVersion?.let { Semver(it) }
     if (outputFormatVersion != null && outputFormatVersion.major > MAX_SUPPORTED_OUTPUT_FORMAT_MAJOR_VERSION) {
@@ -82,9 +83,16 @@ fun ScanCodeResult.toScanSummary(preferFileLicense: Boolean = false): ScanSummar
         ?: files.flatMap { it.scanCodeKeyToSpdxIdMappings }.toMap()
 
     filesOfTypeFile.forEach { file ->
+        val licensesWithoutReferences = file.licenses.filter {
+            // Note that "fromFile" contains the name of the input directory, see
+            // https://github.com/nexB/scancode-toolkit/issues/3712.
+            it !is LicenseEntry.Version3 || it.fromFile == null || it.fromFile == "$inputName/${file.path}"
+                || it.fromFile == inputName // Input is a single file.
+        }
+
         // ScanCode creates separate license entries for each license in an expression. Deduplicate these by grouping by
         // the same expression.
-        val licenses = file.licenses.groupBy {
+        val licenses = licensesWithoutReferences.groupBy {
             LicenseMatch(it.licenseExpression, it.startLine, it.endLine, it.score)
         }.map {
             // Arbitrarily take the first of the duplicate license entries.

diff --git a/plugins/scanners/scancode/src/test/kotlin/ScanCodeResultParserTest.kt b/plugins/scanners/scancode/src/test/kotlin/ScanCodeResultParserTest.kt
@@ -175,11 +175,6 @@ class ScanCodeResultParserTest : FreeSpec({
                         location = TextLocation("COPYING", 59),
                         score = 100.0f
                     ),
-                    LicenseFinding(
-                        license = "LGPL-2.1-only",
-                        location = TextLocation("COPYING", 1, 502),
-                        score = 100.0f
-                    ),
                     LicenseFinding(
                         license = "LGPL-2.1-only",
                         location = TextLocation("COPYING.LGPLv2.1", 1, 502),