fix(efficiency): report reclaimable bytes, not the sum of every copy

cmd/dive/cli/internal/command/ci/evaluator.go

@@ -98,7 +98,7 @@ func (e Evaluator) Evaluate(ctx context.Context, analysis *image.Analysis) Evalu
 
 		e.InefficientFiles = append(e.InefficientFiles, ReferenceFile{
 			References: len(fileData.Nodes),
-			SizeBytes:  uint64(fileData.CumulativeSize),
+			SizeBytes:  uint64(fileData.WastedSize()),
 			Path:       fileData.Path,
 		})
 	}

cmd/dive/cli/internal/command/export/testdata/snapshots/export_test.snap

@@ -20,7 +20,7 @@
     "sizeBytes": 6405
    }
   ],
-  "inefficientBytes": 32025,
+  "inefficientBytes": 19215,
   "sizeBytes": 1220598
  },
  "layer": [

cmd/dive/cli/internal/ui/v1/view/image_details.go

@@ -86,7 +86,7 @@ func (v *ImageDetails) Render() error {
 	var wastedSpace int64
 	for idx := 0; idx < len(v.inefficiencies); idx++ {
 		data := v.inefficiencies[len(v.inefficiencies)-1-idx]
-		wastedSpace += data.CumulativeSize
+		wastedSpace += data.WastedSize()
 
 		inefficiencyReport += fmt.Sprintf(analysisTemplate, strconv.Itoa(len(data.Nodes)), humanize.Bytes(uint64(data.CumulativeSize)), data.Path)
 	}

cmd/dive/cli/testdata/snapshots/cli_build_test.snap

@@ -116,19 +116,19 @@ View build details: docker-desktop://<redacted>
 [Test_Build_CI_gate_fail - 1]
 Analysis:
   efficiency:        100.00 %
-  wastedBytes:       131 bytes (131 B)
-  userWastedPercent: 71.98 %
+  wastedBytes:       97 bytes (97 B)
+  userWastedPercent: 53.30 %
 
 Inefficient Files:
   Count  Wasted Space  File Path
-  3      80 B          /root/saved.txt
-  2      34 B          /root/example/somefile1.txt
+  3      63 B          /root/saved.txt
+  2      17 B          /root/example/somefile1.txt
   2      17 B          /root/example/somefile3.txt
   2      0 B           /root
   10     0 B           /etc
 
 Evaluation:
-  FAIL  highestUserWastedPercent (too many bytes wasted, relative to the user bytes added (%-user-wasted-bytes=0.72 > threshold=0.1))
+  FAIL  highestUserWastedPercent (too many bytes wasted, relative to the user bytes added (%-user-wasted-bytes=0.53 > threshold=0.1))
   SKIP  highestWastedBytes (disabled)
   PASS  lowestEfficiency (0.9)
 

cmd/dive/cli/testdata/snapshots/cli_ci_test.snap

@@ -2,19 +2,19 @@
 [Test_CI_Fail - 1]
 Analysis:
   efficiency:        100.00 %
-  wastedBytes:       131 bytes (131 B)
-  userWastedPercent: 71.98 %
+  wastedBytes:       97 bytes (97 B)
+  userWastedPercent: 53.30 %
 
 Inefficient Files:
   Count  Wasted Space  File Path
-  3      80 B          /root/saved.txt
-  2      34 B          /root/example/somefile1.txt
+  3      63 B          /root/saved.txt
+  2      17 B          /root/example/somefile1.txt
   2      17 B          /root/example/somefile3.txt
   2      0 B           /root
   10     0 B           /etc
 
 Evaluation:
-  FAIL  highestUserWastedPercent (too many bytes wasted, relative to the user bytes added (%-user-wasted-bytes=0.72 > threshold=0.1))
+  FAIL  highestUserWastedPercent (too many bytes wasted, relative to the user bytes added (%-user-wasted-bytes=0.53 > threshold=0.1))
   SKIP  highestWastedBytes (disabled)
   PASS  lowestEfficiency (0.9)
 

dive/filetree/efficiency.go

@@ -14,6 +14,17 @@ type EfficiencyData struct {
 	minDiscoveredSize int64
 }
 
+// WastedSize returns the number of bytes that could be reclaimed for this
+// path by keeping a single copy. CumulativeSize sums every copy across
+// layers, including the smallest one we'd keep, so the recoverable amount
+// is the total minus the smallest copy.
+func (e *EfficiencyData) WastedSize() int64 {
+	if e.minDiscoveredSize < 0 {
+		return 0
+	}
+	return e.CumulativeSize - e.minDiscoveredSize
+}
+
 // EfficiencySlice represents an ordered set of EfficiencyData data structures.
 type EfficiencySlice []*EfficiencyData
 

dive/image/analysis.go

@@ -30,7 +30,7 @@ func Analyze(ctx context.Context, img *Image) (*Analysis, error) {
 
 	var wastedBytes uint64
 	for _, file := range inefficiencies {
-		wastedBytes += uint64(file.CumulativeSize)
+		wastedBytes += uint64(file.WastedSize())
 	}
 
 	return &Analysis{

dive/image/docker/image_archive_analysis_test.go

@@ -14,7 +14,7 @@ func Test_Analysis(t *testing.T) {
 		wastedPercent float64
 		path          string
 	}{
-		"docker-image": {0.9844212134184309, 1220598, 66237, 32025, 0.4834911001404049, "../../../.data/test-docker-image.tar"},
+		"docker-image": {0.9844212134184309, 1220598, 66237, 19215, 0.29009466008424295, "../../../.data/test-docker-image.tar"},
 	}
 
 	for name, test := range table {