From 515d2ed7aea870cdbb083f2bbdd51831d134f198 Mon Sep 17 00:00:00 2001
From: Gilles Chehade <gilles@poolp.org>
Date: Mon, 11 Nov 2024 23:10:52 +0100
Subject: [PATCH] add vfs cardinality

---
 snapshot/backup.go    | 53 ++++++++++++++++++++++++++++++++++++++++++-
 snapshot/vfs/types.go |  2 ++
 2 files changed, 54 insertions(+), 1 deletion(-)

diff --git a/snapshot/backup.go b/snapshot/backup.go
index 6b6ca416..3d9d6d41 100644
--- a/snapshot/backup.go
+++ b/snapshot/backup.go
@@ -1,6 +1,8 @@
 package snapshot
 
 import (
+	"bytes"
+	"encoding/binary"
 	"fmt"
 	"io"
 	"math"
@@ -115,6 +117,18 @@ func (cache *scanCache) RecordChecksum(pathname string, checksum [32]byte) error
 	return cache.db.Put([]byte(fmt.Sprintf("__checksum__:%s", pathname)), checksum[:], nil)
 }
 
+func (cache *scanCache) RecordCardinality(pathname string, files uint64, dirs uint64) error {
+	pathname = strings.TrimSuffix(pathname, "/")
+	if pathname == "" {
+		pathname = "/"
+	}
+
+	buffer := bytes.NewBuffer(make([]byte, 0, 16))
+	binary.Write(buffer, binary.LittleEndian, files)
+	binary.Write(buffer, binary.LittleEndian, dirs)
+	return cache.db.Put([]byte(fmt.Sprintf("__cardinality__:%s", pathname)), buffer.Bytes(), nil)
+}
+
 func (cache *scanCache) GetChecksum(pathname string) ([32]byte, error) {
 	data, err := cache.db.Get([]byte(fmt.Sprintf("__checksum__:%s", pathname)), nil)
 	if err != nil {
@@ -130,6 +144,24 @@ func (cache *scanCache) GetChecksum(pathname string) ([32]byte, error) {
 	return ret, nil
 }
 
+func (cache *scanCache) GetCardinality(pathname string) (uint64, uint64, error) {
+	data, err := cache.db.Get([]byte(fmt.Sprintf("__cardinality__:%s", pathname)), nil)
+	if err != nil {
+		return 0, 0, err
+	}
+
+	if len(data) != 16 {
+		return 0, 0, fmt.Errorf("invalid cardinality length: %d", len(data))
+	}
+
+	buffer := bytes.NewReader(data)
+	var files uint64
+	var dirs uint64
+	binary.Read(buffer, binary.LittleEndian, &files)
+	binary.Read(buffer, binary.LittleEndian, &dirs)
+	return files, dirs, nil
+}
+
 func (cache *scanCache) EnumerateKeysWithPrefixReverse(prefix string, isDirectory bool) (<-chan importer.ScanRecord, error) {
 	// Create a channel to return the keys
 	keyChan := make(chan importer.ScanRecord)
@@ -534,10 +566,24 @@ func (snap *Snapshot) Backup(scanDir string, options *PushOptions) error {
 		dirEntry := vfs.NewDirectoryEntry(filepath.Dir(record.Pathname), &record)
 
 		for _, child := range record.Children {
-			value, err := sc.GetChecksum(filepath.Join(record.Pathname, child.Name()))
+			childpath := filepath.Join(record.Pathname, child.Name())
+			value, err := sc.GetChecksum(childpath)
 			if err != nil {
 				continue
 			}
+
+			if child.IsDir() {
+				dirEntry.DirCardinality++
+
+				files, dirs, err := sc.GetCardinality(childpath)
+				if err != nil {
+					continue
+				}
+				dirEntry.FileCardinality += files
+				dirEntry.DirCardinality += dirs
+			} else {
+				dirEntry.FileCardinality++
+			}
 			dirEntry.AddChild(value, child)
 		}
 
@@ -558,6 +604,11 @@ func (snap *Snapshot) Backup(scanDir string, options *PushOptions) error {
 		if err != nil {
 			return err
 		}
+		err = sc.RecordCardinality(record.Pathname, dirEntry.FileCardinality, dirEntry.DirCardinality)
+		if err != nil {
+			return err
+		}
+
 		atomic.AddUint64(&snap.statistics.VFSDirectoriesCount, 1)
 		atomic.AddUint64(&snap.statistics.VFSDirectoriesSize, dirEntrySize)
 		snap.Event(events.DirectoryOKEvent(snap.Header.SnapshotID, record.Pathname))
diff --git a/snapshot/vfs/types.go b/snapshot/vfs/types.go
index f5026936..4aa75c2d 100644
--- a/snapshot/vfs/types.go
+++ b/snapshot/vfs/types.go
@@ -195,6 +195,8 @@ type DirEntry struct {
 	CustomMetadata     []CustomMetadata    `msgpack:"customMetadata,omitempty"`     // Custom key-value metadata defined by the user (optional)
 	Tags               []string            `msgpack:"tags,omitempty"`               // List of tags associated with the directory (optional)
 	ParentPath         string              `msgpack:"parentPath,omitempty"`         // Path to the parent directory (optional)
+	FileCardinality    uint64              `msgpack:"fileCardinality,omitempty"`    // Number of files in the directory
+	DirCardinality     uint64              `msgpack:"dirCardinality,omitempty"`     // Number of subdirectories in the directory
 }
 
 func (*DirEntry) fsEntry() {}