downloader: CacheKey: include decompression flag

The caches are now created separately for compressed and decompressed contents.
When a decompressed content is cached and a compressed content is requested,
the downloader now correctly returns the compressed content.

- URLSHA/data:                 unmodified original data
- URLSHA/sha256.digest:        digest of the original data
- URLSHA+decomp/data:          decompressed data
- URLSHA+decomp/sha256.digest: digest of the *original* (i.e., compressed) data

Caching a decompressed content does not automatically cache the original compressed content.

Signed-off-by: Akihiro Suda <akihiro.suda.cz@hco.ntt.co.jp>

Author: AkihiroSuda

cmd/limactl/prune.go

@@ -114,20 +114,24 @@ func knownLocations(ctx context.Context) (map[string]limatype.File, error) {
 
 func locationsFromLimaYAML(y *limatype.LimaYAML) map[string]limatype.File {
 	locations := make(map[string]limatype.File)
-	for _, f := range y.Images {
-		locations[downloader.CacheKey(f.Location)] = f.File
-		if f.Kernel != nil {
-			locations[downloader.CacheKey(f.Kernel.Location)] = f.Kernel.File
+	// decompress=false and decompress=true are cached separately
+	// because the same URL may be used for both compressed and uncompressed files.
+	for _, decompres := range []bool{false, true} {
+		for _, f := range y.Images {
+			locations[downloader.CacheKey(f.Location, decompres)] = f.File
+			if f.Kernel != nil {
+				locations[downloader.CacheKey(f.Kernel.Location, decompres)] = f.Kernel.File
+			}
+			if f.Initrd != nil {
+				locations[downloader.CacheKey(f.Initrd.Location, decompres)] = *f.Initrd
+			}
 		}
-		if f.Initrd != nil {
-			locations[downloader.CacheKey(f.Initrd.Location)] = *f.Initrd
+		for _, f := range y.Containerd.Archives {
+			locations[downloader.CacheKey(f.Location, decompres)] = f
+		}
+		for _, f := range y.Firmware.Images {
+			locations[downloader.CacheKey(f.Location, decompres)] = f.File
 		}
-	}
-	for _, f := range y.Containerd.Archives {
-		locations[downloader.CacheKey(f.Location)] = f
-	}
-	for _, f := range y.Firmware.Images {
-		locations[downloader.CacheKey(f.Location)] = f.File
 	}
 	return locations
 }

pkg/downloader/downloader.go

@@ -230,7 +230,7 @@ func Download(ctx context.Context, local, remote string, opts ...Opt) (*Result,
 		return res, nil
 	}
 
-	shad := cacheDirectoryPath(o.cacheDir, remote)
+	shad := cacheDirectoryPath(o.cacheDir, remote, o.decompress)
 	if err := os.MkdirAll(shad, 0o700); err != nil {
 		return nil, err
 	}
@@ -255,7 +255,7 @@ func Download(ctx context.Context, local, remote string, opts ...Opt) (*Result,
 // nil if the file was copied, nil, nil if the file is not in the cache or the
 // cache needs update, or nil, error on fatal error.
 func getCached(ctx context.Context, localPath, remote string, o options) (*Result, error) {
-	shad := cacheDirectoryPath(o.cacheDir, remote)
+	shad := cacheDirectoryPath(o.cacheDir, remote, o.decompress)
 	shadData := filepath.Join(shad, "data")
 	shadTime := filepath.Join(shad, "time")
 	shadType := filepath.Join(shad, "type")
@@ -301,7 +301,7 @@ func getCached(ctx context.Context, localPath, remote string, o options) (*Resul
 
 // fetch downloads remote to the cache and copy the cached file to local path.
 func fetch(ctx context.Context, localPath, remote string, o options) (*Result, error) {
-	shad := cacheDirectoryPath(o.cacheDir, remote)
+	shad := cacheDirectoryPath(o.cacheDir, remote, o.decompress)
 	shadData := filepath.Join(shad, "data")
 	shadTime := filepath.Join(shad, "time")
 	shadType := filepath.Join(shad, "type")
@@ -354,7 +354,7 @@ func Cached(remote string, opts ...Opt) (*Result, error) {
 		return nil, errors.New("local files are not cached")
 	}
 
-	shad := cacheDirectoryPath(o.cacheDir, remote)
+	shad := cacheDirectoryPath(o.cacheDir, remote, o.decompress)
 	shadData := filepath.Join(shad, "data")
 	shadTime := filepath.Join(shad, "time")
 	shadType := filepath.Join(shad, "type")
@@ -404,8 +404,8 @@ func Cached(remote string, opts ...Opt) (*Result, error) {
 //   - "data" file contains the data
 //   - "time" file contains the time (Last-Modified header)
 //   - "type" file contains the type (Content-Type header)
-func cacheDirectoryPath(cacheDir, remote string) string {
-	return filepath.Join(cacheDir, "download", "by-url-sha256", CacheKey(remote))
+func cacheDirectoryPath(cacheDir, remote string, decompress bool) string {
+	return filepath.Join(cacheDir, "download", "by-url-sha256", CacheKey(remote, decompress))
 }
 
 // cacheDigestPath returns the cache digest file path.
@@ -776,8 +776,12 @@ func CacheEntries(opts ...Opt) (map[string]string, error) {
 }
 
 // CacheKey returns the key for a cache entry of the remote URL.
-func CacheKey(remote string) string {
-	return fmt.Sprintf("%x", sha256.Sum256([]byte(remote)))
+func CacheKey(remote string, decompress bool) string {
+	k := fmt.Sprintf("%x", sha256.Sum256([]byte(remote)))
+	if decompress {
+		k += "+decomp"
+	}
+	return k
 }
 
 // RemoveAllCacheDir removes the cache directory.