feat(fossid): Improve PURL mapping

When a snippet has no artifact name and if it contains a GitHub URL, map
instead the URL to a PURL with "GitHub" type. Also, if a snippet has no
artifact name, fallback to a PURL with "Generic" type.

Signed-off-by: Nicolas Nobelis <nicolas.nobelis@bosch.com>

Author: nnobelis

plugins/scanners/fossid/src/main/kotlin/Utils.kt

@@ -28,6 +28,7 @@ import org.apache.logging.log4j.kotlin.loggerOf
 import org.ossreviewtoolkit.clients.fossid.model.result.Snippet
 import org.ossreviewtoolkit.clients.fossid.model.rules.IgnoreRule
 import org.ossreviewtoolkit.clients.fossid.model.rules.RuleType
+import org.ossreviewtoolkit.downloader.VcsHost
 import org.ossreviewtoolkit.model.Issue
 import org.ossreviewtoolkit.model.PackageProvider
 import org.ossreviewtoolkit.model.Severity
@@ -150,17 +151,37 @@ private fun urlToPackageType(url: String): String =
         }
     }
 
-internal fun Snippet.toPurl(url: String): String =
-    when {
-        purl != null -> checkNotNull(purl)
-
-        artifact.isNullOrEmpty() -> ""
+internal fun Snippet.toPurl(url: String): String {
+    purl?.also { return it }
+
+    if (artifact.isNullOrEmpty()) {
+        val vcsHost = VcsHost.fromUrl(url)
+        if (vcsHost != null) {
+            val orga = vcsHost.getUserOrOrganization(url)
+            val repo = vcsHost.getProject(url)
+
+            if (orga != null && repo != null) {
+                return PackageURLBuilder.aPackageURL().withType("github")
+                    .withNamespace(orga)
+                    .withName(repo)
+                    .withVersion(version)
+                    .build()
+                    .canonicalize()
+            }
+        }
 
-        else -> PackageURLBuilder.aPackageURL()
-            .withType(urlToPackageType(url))
+        return PackageURLBuilder.aPackageURL().withType("generic")
             .withNamespace(author)
-            .withName(artifact)
+            .withName(url)
             .withVersion(version)
             .build()
             .canonicalize()
     }
+
+    return PackageURLBuilder.aPackageURL().withType(urlToPackageType(url))
+        .withNamespace(author)
+        .withName(artifact)
+        .withVersion(version)
+        .build()
+        .canonicalize()
+}

plugins/scanners/fossid/src/test/kotlin/FossIdSnippetMappingTest.kt

@@ -170,7 +170,54 @@ class FossIdSnippetMappingTest : WordSpec({
                 with(finding.snippets) {
                     this shouldHaveSize 2
                     first().purl shouldBe "pkg:maven/com/vdurmont/semver4j@3.1.0"
-                    last().purl shouldBe ""
+                    last().purl shouldBe "pkg:generic/com/vdurmont/https%3A%2F%2Frepo1.maven.org%2Fmaven2%2Fcom%2F" +
+                        "vdurmont%2Fsemver4j%2F3.1.0%2F@3.1.0"
+                }
+            }
+        }
+
+        "map github url to purl" {
+            val issues = mutableListOf<Issue>()
+            val listSnippets = flowOf(
+                "src/main/java/Tokenizer.java" to
+                    setOf(
+                        createSnippet(
+                            2,
+                            MatchType.FULL,
+                            "pkg:maven/com.vdurmont/semver4j@3.1.0",
+                            "MIT",
+                            "com/vdurmont/semver4j/Tokenizer.java"
+                        ).copy(
+                            author = "com/vdurmont",
+                            artifact = null,
+                            version = "3.1.0",
+                            url = "https://github.com/MartinPulec/UltraGrid/archive/refs/heads/devel.tar.gz",
+                            purl = null
+                        )
+                    )
+            )
+            val rawResults = RawResults(
+                emptyList(),
+                emptyList(),
+                emptyList(),
+                emptyList(),
+                listSnippets,
+                emptyMap()
+            )
+
+            val mappedSnippets = mapSnippetFindings(
+                rawResults,
+                500,
+                issues,
+                emptyMap(),
+                emptyList(),
+                mutableSetOf()
+            )
+
+            issues should beEmpty()
+            mappedSnippets.shouldBeSingleton { finding ->
+                finding.snippets.shouldBeSingleton { snippet ->
+                    snippet.purl shouldBe "pkg:github/martinpulec/ultragrid@3.1.0"
                 }
             }
         }