shell: add minCacheAge flag to remote.uncache command (#8225)

* add minCacheAge flag to remote.uncache command #8221

* address code review feedback: add nil check and improve test isolation

* address code review feedback: use consistent timestamp in FileFilter

weed/shell/command_remote_uncache.go

@@ -36,7 +36,8 @@ func (c *commandRemoteUncache) Help() string {
 	remote.uncache -dir=/xxx/some/sub/dir -include=*.pdf
 	remote.uncache -dir=/xxx/some/sub/dir -exclude=*.txt
 	remote.uncache -minSize=1024000    # uncache files larger than 100K
-	remote.uncache -minAge=3600        # uncache files older than 1 hour
+	remote.uncache -minAge=3600        # uncache files older than 1 hour (created time)
+	remote.uncache -minCacheAge=3600   # uncache files older than 1 hour (cached time)
 
 `
 }
@@ -128,12 +129,14 @@ func (c *commandRemoteUncache) uncacheContentData(commandEnv *CommandEnv, writer
 }
 
 type FileFilter struct {
-	include *string
-	exclude *string
-	minSize *int64
-	maxSize *int64
-	minAge  *int64
-	maxAge  *int64
+	include     *string
+	exclude     *string
+	minSize     *int64
+	maxSize     *int64
+	minAge      *int64
+	maxAge      *int64
+	minCacheAge *int64
+	now         int64
 }
 
 func newFileFilter(remoteMountCommand *flag.FlagSet) (ff *FileFilter) {
@@ -142,12 +145,17 @@ func newFileFilter(remoteMountCommand *flag.FlagSet) (ff *FileFilter) {
 	ff.exclude = remoteMountCommand.String("exclude", "", "pattens of file names, e.g., *.pdf, *.html, ab?d.txt")
 	ff.minSize = remoteMountCommand.Int64("minSize", -1, "minimum file size in bytes")
 	ff.maxSize = remoteMountCommand.Int64("maxSize", -1, "maximum file size in bytes")
-	ff.minAge = remoteMountCommand.Int64("minAge", -1, "minimum file age in seconds")
-	ff.maxAge = remoteMountCommand.Int64("maxAge", -1, "maximum file age in seconds")
+	ff.minAge = remoteMountCommand.Int64("minAge", -1, "minimum file age in seconds (created time)")
+	ff.maxAge = remoteMountCommand.Int64("maxAge", -1, "maximum file age in seconds (created time)")
+	ff.minCacheAge = remoteMountCommand.Int64("minCacheAge", -1, "minimum file cache age in seconds (last cached time)")
+	ff.now = time.Now().Unix()
 	return
 }
 
 func (ff *FileFilter) matches(entry *filer_pb.Entry) bool {
+	if entry.Attributes == nil {
+		return false
+	}
 	if *ff.include != "" {
 		if ok, _ := filepath.Match(*ff.include, entry.Name); !ok {
 			return false
@@ -169,12 +177,21 @@ func (ff *FileFilter) matches(entry *filer_pb.Entry) bool {
 		}
 	}
 	if *ff.minAge != -1 {
-		if entry.Attributes.Crtime+*ff.minAge > time.Now().Unix() {
+		if entry.Attributes.Crtime+*ff.minAge > ff.now {
 			return false
 		}
 	}
 	if *ff.maxAge != -1 {
-		if entry.Attributes.Crtime+*ff.maxAge < time.Now().Unix() {
+		if entry.Attributes.Crtime+*ff.maxAge < ff.now {
+			return false
+		}
+	}
+	if *ff.minCacheAge != -1 {
+		lastCachedTime := entry.Attributes.Crtime
+		if entry.RemoteEntry != nil && entry.RemoteEntry.LastLocalSyncTsNs > 0 {
+			lastCachedTime = entry.RemoteEntry.LastLocalSyncTsNs / 1e9
+		}
+		if lastCachedTime+*ff.minCacheAge > ff.now {
 			return false
 		}
 	}

weed/shell/command_remote_uncache_test.go

@@ -0,0 +1,104 @@
+package shell
+
+import (
+	"testing"
+	"time"
+
+	"github.com/seaweedfs/seaweedfs/weed/pb/filer_pb"
+)
+
+func TestFileFilter_matches_minCacheAge(t *testing.T) {
+	now := time.Now().Unix()
+
+	tests := []struct {
+		name        string
+		minCacheAge int64
+		entry       *filer_pb.Entry
+		want        bool
+	}{
+		{
+			name:        "no minCacheAge",
+			minCacheAge: -1,
+			entry: &filer_pb.Entry{
+				Attributes: &filer_pb.FuseAttributes{Crtime: now - 100},
+			},
+			want: true,
+		},
+		{
+			name:        "recent cache, should not match",
+			minCacheAge: 3600,
+			entry: &filer_pb.Entry{
+				Attributes: &filer_pb.FuseAttributes{Crtime: now - 7200},
+				RemoteEntry: &filer_pb.RemoteEntry{
+					LastLocalSyncTsNs: now * 1e9,
+				},
+			},
+			want: false,
+		},
+		{
+			name:        "old cache, should match",
+			minCacheAge: 3600,
+			entry: &filer_pb.Entry{
+				Attributes: &filer_pb.FuseAttributes{Crtime: now - 7200},
+				RemoteEntry: &filer_pb.RemoteEntry{
+					LastLocalSyncTsNs: (now - 4000) * 1e9,
+				},
+			},
+			want: true,
+		},
+		{
+			name:        "no remote entry, uses crtime - recent, should not match",
+			minCacheAge: 3600,
+			entry: &filer_pb.Entry{
+				Attributes: &filer_pb.FuseAttributes{Crtime: now - 100},
+			},
+			want: false,
+		},
+		{
+			name:        "no remote entry, uses crtime - old, should match",
+			minCacheAge: 3600,
+			entry: &filer_pb.Entry{
+				Attributes: &filer_pb.FuseAttributes{Crtime: now - 4000},
+			},
+			want: true,
+		},
+		{
+			name:        "remote entry with 0 sync ts, uses crtime - recent, should not match",
+			minCacheAge: 3600,
+			entry: &filer_pb.Entry{
+				Attributes:  &filer_pb.FuseAttributes{Crtime: now - 100},
+				RemoteEntry: &filer_pb.RemoteEntry{LastLocalSyncTsNs: 0},
+			},
+			want: false,
+		},
+		{
+			name:        "nil attributes, should not match",
+			minCacheAge: 3600,
+			entry: &filer_pb.Entry{
+				Attributes: nil,
+			},
+			want: false,
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			defaultString := ""
+			defaultInt64 := int64(-1)
+			ff := &FileFilter{
+				include:     &defaultString,
+				exclude:     &defaultString,
+				minSize:     &defaultInt64,
+				maxSize:     &defaultInt64,
+				minAge:      &defaultInt64,
+				maxAge:      &defaultInt64,
+				minCacheAge: &tt.minCacheAge,
+				now:         now,
+			}
+
+			if got := ff.matches(tt.entry); got != tt.want {
+				t.Errorf("FileFilter.matches() = %v, want %v", got, tt.want)
+			}
+		})
+	}
+}