fix(model): Rework purl conversion according to the specs

Implement the pseudo-algorithm described at [1]. Most importantly, '/' in namespaces are now not escaped anymore (also see the lengthy discussion at [2]), key names are lower-cased, and qualifiers are sorted for comparability. [1]: https://github.com/package-url/purl-spec/blob/master/PURL-SPECIFICATION.rst#how-to-build-purl-string-from-its-components [2]: package-url/purl-spec#176 Signed-off-by: Sebastian Schuberth <[email protected]>
oss-review-toolkit · Oct 23, 2024 · 6978919 · 6978919
1 parent 036bcb4
commit 6978919
Show file tree

Hide file tree

Showing 4 changed files with 37 additions and 21 deletions.
diff --git a/model/src/main/kotlin/utils/PurlExtensions.kt b/model/src/main/kotlin/utils/PurlExtensions.kt
@@ -114,9 +114,8 @@ fun String.toProvenance(): Provenance {
         "download_url=" in extras -> {
             val encodedUrl = getQualifierValue("download_url")
 
-            val percentEncodedColon = "%3A"
             val checksum = getQualifierValue("checksum")
-            val (algorithm, value) = checksum.split(percentEncodedColon, limit = 2)
+            val (algorithm, value) = checksum.split(':', limit = 2)
 
             ArtifactProvenance(
                 sourceArtifact = RemoteArtifact(

diff --git a/model/src/main/kotlin/utils/PurlUtils.kt b/model/src/main/kotlin/utils/PurlUtils.kt
@@ -21,6 +21,7 @@
 
 package org.ossreviewtoolkit.model.utils
 
+import org.ossreviewtoolkit.model.HashAlgorithm
 import org.ossreviewtoolkit.utils.common.percentEncode
 
 /**
@@ -92,28 +93,44 @@ internal fun createPurl(
 ): String =
     buildString {
         append("pkg:")
-        append(type)
+        append(type.lowercase())
+        append('/')
 
         if (namespace.isNotEmpty()) {
+            append(namespace.trim('/').split('/').joinToString("/") { it.percentEncode() })
             append('/')
-            append(namespace.percentEncode())
+            append(name.trim('/').percentEncode())
+        } else {
+            append(name.percentEncode())
         }
 
-        append('/')
-        append(name.percentEncode())
+        if (version.isNotEmpty()) {
+            append('@')
 
-        append('@')
-        append(version.percentEncode())
+            // See https://github.com/package-url/purl-spec/blob/master/PURL-SPECIFICATION.rst#character-encoding which
+            // says "the '#', '?', '@' and ':' characters must NOT be encoded when used as separators".
+            val isChecksum = HashAlgorithm.VERIFIABLE.any { version.startsWith("${it.name.lowercase()}:") }
+            if (isChecksum) append(version) else append(version.percentEncode())
+        }
 
-        qualifiers.onEachIndexed { index, entry ->
+        qualifiers.filterValues { it.isNotEmpty() }.toSortedMap().onEachIndexed { index, entry ->
             if (index == 0) append("?") else append("&")
-            append(entry.key.percentEncode())
+
+            val key = entry.key.lowercase()
+            append(key)
+
             append("=")
-            append(entry.value.percentEncode())
+
+            // See https://github.com/package-url/purl-spec/blob/master/PURL-SPECIFICATION.rst#known-qualifiers-keyvalue-pairs.
+            if (key in KNOWN_KEYS) append(entry.value) else append(entry.value.percentEncode())
         }
 
         if (subpath.isNotEmpty()) {
-            val value = subpath.split('/').joinToString("/", prefix = "#") { it.percentEncode() }
+            val value = subpath.trim('/').split('/')
+                .filter { it.isNotEmpty() && it != "." && it != ".." }
+                .joinToString("/", prefix = "#") { it.percentEncode() }
             append(value)
         }
     }
+
+private val KNOWN_KEYS = setOf("repository_url", "download_url", "vcs_url", "file_name", "checksum")
diff --git a/model/src/test/kotlin/utils/PurlExtensionsTest.kt b/model/src/test/kotlin/utils/PurlExtensionsTest.kt
@@ -68,10 +68,10 @@ class PurlExtensionsTest : WordSpec({
             purl shouldBe "pkg:generic/name@version"
         }
 
-        "percent-encode namespaces with segments" {
-            val purl = Identifier("generic", "name/space", "name", "version").toPurl()
+        "percent-encode namespace segments" {
+            val purl = Identifier("generic", "name space/with spaces", "name", "version").toPurl()
 
-            purl shouldBe "pkg:generic/name%2Fspace/name@version"
+            purl shouldBe "pkg:generic/name%20space/with%20spaces/name@version"
         }
 
         "percent-encode the name" {
@@ -104,8 +104,8 @@ class PurlExtensionsTest : WordSpec({
             val purl = id.toPurl(extras.qualifiers, extras.subpath)
 
             purl shouldBe "pkg:maven/com.example/[email protected]?" +
-                "download_url=https%3A%2F%2Fexample.com%2Fsources.zip&" +
-                "checksum=md5%3Addce269a1e3d054cae349621c198dd52"
+                "checksum=md5:ddce269a1e3d054cae349621c198dd52&" +
+                "download_url=https://example.com/sources.zip"
             purl.toProvenance() shouldBe provenance
         }
 
@@ -125,10 +125,10 @@ class PurlExtensionsTest : WordSpec({
             val purl = id.toPurl(extras.qualifiers, extras.subpath)
 
             purl shouldBe "pkg:maven/com.example/[email protected]?" +
-                "vcs_type=Git&" +
-                "vcs_url=https%3A%2F%2Fgithub.com%2Fapache%2Fcommons-text.git&" +
+                "resolved_revision=7643b12421100d29fd2b78053e77bcb04a251b2e&" +
                 "vcs_revision=7643b12421100d29fd2b78053e77bcb04a251b2e&" +
-                "resolved_revision=7643b12421100d29fd2b78053e77bcb04a251b2e" +
+                "vcs_type=Git&" +
+                "vcs_url=https://github.com/apache/commons-text.git" +
                 "#subpath"
             purl.toProvenance() shouldBe provenance
         }

diff --git a/...nagers/swiftpm/src/funTest/assets/projects/synthetic/expected-output-only-lockfile-v1.yml b/...nagers/swiftpm/src/funTest/assets/projects/synthetic/expected-output-only-lockfile-v1.yml
@@ -51,7 +51,7 @@ packages:
     revision: "eb51f949cdd0c9d88abba9ce79d37eb7ea1231d0"
     path: ""
 - id: "Swift::github.com/apple/swift-crypto:"
-  purl: "pkg:swift/github.com%2Fapple%2Fswift-crypto@"
+  purl: "pkg:swift/github.com%2Fapple%2Fswift-crypto"
   declared_licenses: []
   declared_licenses_processed: {}
   description: ""