fix: escape % in ArchivePath segments

`:` was escaped as `%3A`, but `%` itself wasn't, so an entry named `a:b.jar` and one named `a%3Ab.jar` got the same `ArchivePath`. Their contents then collide in `FileMetadata` / `FileDigests` (last writer wins) and share artifact IDs.

`%` is now escaped as `%25` before `:`, which makes the encoding injective. The encoding is documented on `Coordinates.ArchivePath` and in the schema.

Signed-off-by: Alex Goodman <wagoodman@users.noreply.github.com>

internal/archive/traversal.go

@@ -21,16 +21,21 @@ type Traversal struct {
 }
 
 // VirtualPath returns the colon-delimited chain of archives from the scan root to loc, ending in loc
-// itself: "app.war:WEB-INF/lib/dep.jar". Outside any archive it is the location's path unchanged. A
+// itself: "app.war:WEB-INF/lib/dep.jar". Outside any archive it is the location's path unchanged. In an
-// colon in an entry name is escaped so splitting on ':' recovers the archive boundaries; the leading
+// entry name '%' is escaped as "%25" and then ':' as "%3A", so splitting on ':' recovers the archive
-// slash is dropped to match the java cataloger, which joins slash-less zip entry names onto the
+// boundaries and no two entry names share an escaped form. The leading slash is dropped to match the
-// containing archive's path.
+// java cataloger, which joins slash-less zip entry names onto the containing archive's path.
 func VirtualPath(loc file.Location) string {
 	if loc.ArchivePath == "" {
 		return loc.Path()
 	}
-	entryPath := strings.ReplaceAll(loc.Path(), ":", "%3A")
+	return loc.ArchivePath + ":" + strings.TrimPrefix(escapeEntryPath(loc.Path()), "/")
-	return loc.ArchivePath + ":" + strings.TrimPrefix(entryPath, "/")
+}
+
+var entryPathEscaper = strings.NewReplacer("%", "%25", ":", "%3A")
+
+func escapeEntryPath(p string) string {
+	return entryPathEscaper.Replace(p)
 }
 
 type traversalCtxKey struct{}

internal/archive/traversal_test.go

@@ -85,14 +85,14 @@ func TestVirtualPath(t *testing.T) {
 	}
 }
 
-func TestVirtualPath_theEscapeCharacterIsNotItselfEscaped(t *testing.T) {
+func TestVirtualPath_theEscapeCharacterIsItselfEscaped(t *testing.T) {
-	// an accepted limit: an entry path holding the literal text %3A is indistinguishable from one holding
+	// an entry path holding the literal text %3A must not collide with one holding a colon, or their
-	// a colon. Escaping % too would move many more real identities than it protects.
+	// contents share an ArchivePath and overwrite each other's file metadata
 	fromColon := VirtualPath(locationIn("outer.zip", "a:b.jar"))
 	fromLiteral := VirtualPath(locationIn("outer.zip", "a%3Ab.jar"))
 
 	assert.Equal(t, "outer.zip:a%3Ab.jar", fromColon)
-	assert.Equal(t, fromColon, fromLiteral)
+	assert.Equal(t, "outer.zip:a%253Ab.jar", fromLiteral)
 }
 
 func TestVirtualPath_escapingComposesAcrossNestingLevels(t *testing.T) {

schema/json/schema-16.1.11.json

@@ -724,7 +724,7 @@
         },
         "archivePath": {
           "type": "string",
-          "description": "ArchivePath is the colon-delimited chain of archive paths, from the scan root, traversed to reach this file (e.g. \"app.war:WEB-INF/lib/dep.jar\"). Blank for files not found within an archive. It disambiguates identically-named files in different archives on the same filesystem."
+          "description": "ArchivePath is the colon-delimited chain of archive paths, from the scan root, traversed to reach this file (e.g. \"app.war:WEB-INF/lib/dep.jar\"). Blank for files not found within an archive. It disambiguates identically-named files in different archives on the same filesystem. Within each segment after the first, '%' is escaped as \"%25\" and ':' as \"%3A\", so splitting on ':' recovers the archive boundaries."
         }
       },
       "type": "object",
@@ -2490,7 +2490,7 @@
         },
         "archivePath": {
           "type": "string",
-          "description": "ArchivePath is the colon-delimited chain of archive paths, from the scan root, traversed to reach this file (e.g. \"app.war:WEB-INF/lib/dep.jar\"). Blank for files not found within an archive. It disambiguates identically-named files in different archives on the same filesystem."
+          "description": "ArchivePath is the colon-delimited chain of archive paths, from the scan root, traversed to reach this file (e.g. \"app.war:WEB-INF/lib/dep.jar\"). Blank for files not found within an archive. It disambiguates identically-named files in different archives on the same filesystem. Within each segment after the first, '%' is escaped as \"%25\" and ':' as \"%3A\", so splitting on ':' recovers the archive boundaries."
         },
         "accessPath": {
           "type": "string",

schema/json/schema-latest.json

@@ -724,7 +724,7 @@
         },
         "archivePath": {
           "type": "string",
-          "description": "ArchivePath is the colon-delimited chain of archive paths, from the scan root, traversed to reach this file (e.g. \"app.war:WEB-INF/lib/dep.jar\"). Blank for files not found within an archive. It disambiguates identically-named files in different archives on the same filesystem."
+          "description": "ArchivePath is the colon-delimited chain of archive paths, from the scan root, traversed to reach this file (e.g. \"app.war:WEB-INF/lib/dep.jar\"). Blank for files not found within an archive. It disambiguates identically-named files in different archives on the same filesystem. Within each segment after the first, '%' is escaped as \"%25\" and ':' as \"%3A\", so splitting on ':' recovers the archive boundaries."
         }
       },
       "type": "object",
@@ -2490,7 +2490,7 @@
         },
         "archivePath": {
           "type": "string",
-          "description": "ArchivePath is the colon-delimited chain of archive paths, from the scan root, traversed to reach this file (e.g. \"app.war:WEB-INF/lib/dep.jar\"). Blank for files not found within an archive. It disambiguates identically-named files in different archives on the same filesystem."
+          "description": "ArchivePath is the colon-delimited chain of archive paths, from the scan root, traversed to reach this file (e.g. \"app.war:WEB-INF/lib/dep.jar\"). Blank for files not found within an archive. It disambiguates identically-named files in different archives on the same filesystem. Within each segment after the first, '%' is escaped as \"%25\" and ':' as \"%3A\", so splitting on ':' recovers the archive boundaries."
         },
         "accessPath": {
           "type": "string",

syft/file/coordinates.go

@@ -15,7 +15,7 @@ type Coordinates struct {
 	// FileSystemID is an ID representing an entire filesystem. For container images, this is a layer digest. For directories or a root filesystem, this is blank. A file inside an archive carries the FileSystemID of the filesystem the archive was found in.
 	FileSystemID string `json:"layerID,omitempty" cyclonedx:"layerID"`
 
-	// ArchivePath is the colon-delimited chain of archive paths, from the scan root, traversed to reach this file (e.g. "app.war:WEB-INF/lib/dep.jar"). Blank for files not found within an archive. It disambiguates identically-named files in different archives on the same filesystem.
+	// ArchivePath is the colon-delimited chain of archive paths, from the scan root, traversed to reach this file (e.g. "app.war:WEB-INF/lib/dep.jar"). Blank for files not found within an archive. It disambiguates identically-named files in different archives on the same filesystem. Within each segment after the first, '%' is escaped as "%25" and ':' as "%3A", so splitting on ':' recovers the archive boundaries.
 	ArchivePath string `json:"archivePath,omitempty" cyclonedx:"archivePath"`
 }