Files
git-pages/src/garbage.go
2026-04-14 05:01:37 +00:00

84 lines
2.2 KiB
Go

package git_pages
import (
"context"
"fmt"
"github.com/c2h5oh/datasize"
"github.com/dghubble/trie"
)
func trieReduce(data trie.Trier) (items, total int64) {
data.Walk(func(key string, value any) error {
items += 1
total += *value.(*int64)
return nil
})
return
}
func TraceGarbage(ctx context.Context) error {
allBlobs := trie.NewRuneTrie()
liveBlobs := trie.NewRuneTrie()
traceManifest := func(manifestName string, manifest *Manifest) error {
for _, entry := range manifest.GetContents() {
if entry.GetType() == Type_ExternalFile {
blobName := string(entry.Data)
if size := allBlobs.Get(blobName); size == nil {
return fmt.Errorf("%s: dangling reference %s", manifestName, blobName)
} else {
liveBlobs.Put(blobName, size)
}
}
}
return nil
}
// Enumerate all blobs.
for metadata, err := range backend.EnumerateBlobs(ctx) {
if err != nil {
return fmt.Errorf("trace blobs err: %w", err)
}
allBlobs.Put(metadata.Name, &metadata.Size)
}
// Enumerate blobs live via site manifests.
for item, err := range backend.GetAllManifests(ctx) {
metadata, manifest := item.Splat()
if err != nil {
return fmt.Errorf("trace sites err: %w", err)
}
err = traceManifest(metadata.Name, manifest)
if err != nil {
return fmt.Errorf("trace sites err: %w", err)
}
}
// Enumerate blobs live via audit records.
auditIDs := backend.SearchAuditLog(ctx, SearchAuditLogOptions{})
for record, err := range backend.GetAuditLogRecords(ctx, auditIDs) {
if err != nil {
logc.Fatalln(ctx, err)
}
if record.Manifest != nil {
err = traceManifest(record.GetAuditID().String(), record.Manifest)
if err != nil {
return fmt.Errorf("trace audit err: %w", err)
}
}
}
allBlobsCount, allBlobsSize := trieReduce(allBlobs)
liveBlobsCount, liveBlobsSize := trieReduce(liveBlobs)
logc.Printf(ctx, "trace all: %d blobs, %s",
allBlobsCount, datasize.ByteSize(allBlobsSize).HR())
logc.Printf(ctx, "trace live: %d blobs, %s",
liveBlobsCount, datasize.ByteSize(liveBlobsSize).HR())
logc.Printf(ctx, "trace dead: %d blobs, %s",
allBlobsCount-liveBlobsCount, datasize.ByteSize(allBlobsSize-liveBlobsSize).HR())
return nil
}