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

…nces Fixes #8190. Signed-off-by: Sebastian Schuberth <sebastian@doubleopen.org>
oss-review-toolkit · Mar 28, 2024 · a1396c1 · a1396c1
1 parent f8ecff6
commit a1396c1
Show file tree

Hide file tree

Showing 3 changed files with 10 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 inputDirName = 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 +83,15 @@ 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 == "$inputDirName/${file.path}"
+        }
+
         // 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),