Add statistics to JSON and YAML reports (#730)

* Add statistics to JSON and YAML reports Signed-off-by: egibs <[email protected]> * Add JSON stats test case Signed-off-by: egibs <[email protected]> * Avoid rendering empty stats object when running diffs Signed-off-by: egibs <[email protected]> * Use final report field instead Signed-off-by: egibs <[email protected]> --------- Signed-off-by: egibs <[email protected]>
chainguard-dev · Dec 31, 2024 · 3b49925 · 3b49925
1 parent bbd5349
commit 3b49925
Show file tree

Hide file tree

Showing 18 changed files with 263 additions and 27 deletions.
diff --git a/cmd/mal/mal.go b/cmd/mal/mal.go
@@ -424,7 +424,7 @@ func main() {
 						return err
 					}
 
-					err = renderer.Full(ctx, res)
+					err = renderer.Full(ctx, &mc, res)
 					if err != nil {
 						returnCode = ExitRenderFailed
 						return err
@@ -467,7 +467,7 @@ func main() {
 						return err
 					}
 
-					err = renderer.Full(ctx, res)
+					err = renderer.Full(ctx, &mc, res)
 					if err != nil {
 						returnCode = ExitRenderFailed
 						return err
@@ -550,7 +550,7 @@ func main() {
 						return length
 					}(&res.Files)
 
-					err = renderer.Full(ctx, res)
+					err = renderer.Full(ctx, &mc, res)
 					if err != nil {
 						returnCode = ExitRenderFailed
 						return err

diff --git a/pkg/action/archive_test.go b/pkg/action/archive_test.go
@@ -243,7 +243,7 @@ func TestScanArchive(t *testing.T) {
 	if err != nil {
 		t.Fatal(err)
 	}
-	if err := r.Full(ctx, res); err != nil {
+	if err := r.Full(ctx, nil, res); err != nil {
 		t.Fatalf("full: %v", err)
 	}
 

diff --git a/pkg/action/oci_test.go b/pkg/action/oci_test.go
@@ -46,7 +46,7 @@ func TestOCI(t *testing.T) {
 	if err != nil {
 		t.Fatal(err)
 	}
-	if err := r.Full(ctx, res); err != nil {
+	if err := r.Full(ctx, nil, res); err != nil {
 		t.Fatalf("full: %v", err)
 	}
 

diff --git a/pkg/action/scan.go b/pkg/action/scan.go
@@ -572,7 +572,7 @@ func Scan(ctx context.Context, c malcontent.Config) (*malcontent.Report, error)
 		}
 		return true
 	})
-	if c.Stats {
+	if c.Stats && c.Renderer.Name() != "JSON" && c.Renderer.Name() != "YAML" {
 		err = render.Statistics(&c, r)
 		if err != nil {
 			return r, fmt.Errorf("stats: %w", err)

diff --git a/pkg/malcontent/malcontent.go b/pkg/malcontent/malcontent.go
@@ -17,7 +17,7 @@ import (
 type Renderer interface {
 	Scanning(context.Context, string)
 	File(context.Context, *FileReport) error
-	Full(context.Context, *Report) error
+	Full(context.Context, *Config, *Report) error
 	Name() string
 }
 

diff --git a/pkg/refresh/refresh.go b/pkg/refresh/refresh.go
@@ -195,7 +195,7 @@ func executeRefresh(ctx context.Context, testData []TestData) error {
 					return fmt.Errorf("refresh sample data for %s: %w", data.OutputPath, err)
 				}
 
-				if err := data.Config.Renderer.Full(ctx, res); err != nil {
+				if err := data.Config.Renderer.Full(ctx, nil, res); err != nil {
 					return fmt.Errorf("render results for %s: %w", data.OutputPath, err)
 				}
 

diff --git a/pkg/render/json.go b/pkg/render/json.go
@@ -28,7 +28,7 @@ func (r JSON) File(_ context.Context, _ *malcontent.FileReport) error {
 	return nil
 }
 
-func (r JSON) Full(_ context.Context, rep *malcontent.Report) error {
+func (r JSON) Full(_ context.Context, c *malcontent.Config, rep *malcontent.Report) error {
 	jr := Report{
 		Diff:   rep.Diff,
 		Files:  make(map[string]*malcontent.FileReport),
@@ -52,6 +52,10 @@ func (r JSON) Full(_ context.Context, rep *malcontent.Report) error {
 		return true
 	})
 
+	if c != nil && c.Stats && jr.Diff == nil {
+		jr.Stats = serializedStats(c, rep)
+	}
+
 	j, err := json.MarshalIndent(jr, "", "    ")
 	if err != nil {
 		return err

diff --git a/pkg/render/markdown.go b/pkg/render/markdown.go
@@ -59,7 +59,7 @@ func (r Markdown) File(ctx context.Context, fr *malcontent.FileReport) error {
 	return nil
 }
 
-func (r Markdown) Full(ctx context.Context, rep *malcontent.Report) error {
+func (r Markdown) Full(ctx context.Context, _ *malcontent.Config, rep *malcontent.Report) error {
 	if rep.Diff == nil {
 		return nil
 	}

diff --git a/pkg/render/render.go b/pkg/render/render.go
@@ -6,6 +6,7 @@ package render
 import (
 	"fmt"
 	"io"
+	"sort"
 
 	"github.com/chainguard-dev/malcontent/pkg/malcontent"
 )
@@ -15,6 +16,17 @@ type Report struct {
 	Diff   *malcontent.DiffReport            `json:",omitempty" yaml:",omitempty"`
 	Files  map[string]*malcontent.FileReport `json:",omitempty" yaml:",omitempty"`
 	Filter string                            `json:",omitempty" yaml:",omitempty"`
+	Stats  *Stats                            `json:",omitempty" yaml:",omitempty"`
+}
+
+// Stats stores a JSON- or YAML-friendly Statistics report.
+type Stats struct {
+	PkgStats       []malcontent.StrMetric `json:",omitempty" yaml:",omitempty"`
+	ProcessedFiles int                    `json:",omitempty" yaml:",omitempty"`
+	RiskStats      []malcontent.IntMetric `json:",omitempty" yaml:",omitempty"`
+	SkippedFiles   int                    `json:",omitempty" yaml:",omitempty"`
+	TotalBehaviors int                    `json:",omitempty" yaml:",omitempty"`
+	TotalRisks     int                    `json:",omitempty" yaml:",omitempty"`
 }
 
 // New returns a new Renderer.
@@ -56,3 +68,25 @@ func riskEmoji(score int) string {
 
 	return symbol
 }
+
+func serializedStats(c *malcontent.Config, r *malcontent.Report) *Stats {
+	pkgStats, _, totalBehaviors := PkgStatistics(c, &r.Files)
+	riskStats, totalRisks, processedFiles, skippedFiles := RiskStatistics(c, &r.Files)
+
+	sort.Slice(pkgStats, func(i, j int) bool {
+		return pkgStats[i].Key < pkgStats[j].Key
+	})
+
+	sort.Slice(riskStats, func(i, j int) bool {
+		return riskStats[i].Key < riskStats[j].Key
+	})
+
+	return &Stats{
+		PkgStats:       pkgStats,
+		ProcessedFiles: processedFiles,
+		RiskStats:      riskStats,
+		SkippedFiles:   skippedFiles,
+		TotalBehaviors: totalBehaviors,
+		TotalRisks:     totalRisks,
+	}
+}
diff --git a/pkg/render/simple.go b/pkg/render/simple.go
@@ -48,7 +48,7 @@ func (r Simple) File(_ context.Context, fr *malcontent.FileReport) error {
 	return nil
 }
 
-func (r Simple) Full(_ context.Context, rep *malcontent.Report) error {
+func (r Simple) Full(_ context.Context, _ *malcontent.Config, rep *malcontent.Report) error {
 	if rep.Diff == nil {
 		return nil
 	}

diff --git a/pkg/render/stats.go b/pkg/render/stats.go
@@ -19,7 +19,7 @@ func smLength(m *sync.Map) int {
 	return length
 }
 
-func riskStatistics(c *malcontent.Config, files *sync.Map) ([]malcontent.IntMetric, int, int, int) {
+func RiskStatistics(c *malcontent.Config, files *sync.Map) ([]malcontent.IntMetric, int, int, int) {
 	length := smLength(files)
 
 	riskMap := make(map[int][]string, length)
@@ -73,7 +73,7 @@ func riskStatistics(c *malcontent.Config, files *sync.Map) ([]malcontent.IntMetr
 	return stats, total(), processedFiles, skippedFiles
 }
 
-func pkgStatistics(_ *malcontent.Config, files *sync.Map) ([]malcontent.StrMetric, int, int) {
+func PkgStatistics(_ *malcontent.Config, files *sync.Map) ([]malcontent.StrMetric, int, int) {
 	length := smLength(files)
 	numBehaviors := 0
 	pkgMap := make(map[string]int, length)
@@ -117,8 +117,8 @@ func pkgStatistics(_ *malcontent.Config, files *sync.Map) ([]malcontent.StrMetri
 }
 
 func Statistics(c *malcontent.Config, r *malcontent.Report) error {
-	riskStats, totalRisks, processedFiles, skippedFiles := riskStatistics(c, &r.Files)
-	pkgStats, width, totalBehaviors := pkgStatistics(c, &r.Files)
+	riskStats, totalRisks, processedFiles, skippedFiles := RiskStatistics(c, &r.Files)
+	pkgStats, width, totalBehaviors := PkgStatistics(c, &r.Files)
 
 	statsSymbol := "📊"
 	riskSymbol := "⚠️ "

diff --git a/pkg/render/strings.go b/pkg/render/strings.go
@@ -109,7 +109,7 @@ func (r StringMatches) File(_ context.Context, fr *malcontent.FileReport) error
 	return nil
 }
 
-func (r StringMatches) Full(_ context.Context, rep *malcontent.Report) error {
+func (r StringMatches) Full(_ context.Context, _ *malcontent.Config, rep *malcontent.Report) error {
 	// Non-diff files are handled on the fly by File()
 	if rep.Diff == nil {
 		return nil

diff --git a/pkg/render/tea.go b/pkg/render/tea.go
@@ -385,7 +385,7 @@ func (r *Interactive) File(ctx context.Context, fr *malcontent.FileReport) error
 	return nil
 }
 
-func (r *Interactive) Full(ctx context.Context, rep *malcontent.Report) error {
+func (r *Interactive) Full(ctx context.Context, _ *malcontent.Config, rep *malcontent.Report) error {
 	defer func() {
 		r.program.Send(scanCompleteMsg{})
 		r.wg.Wait()

diff --git a/pkg/render/terminal.go b/pkg/render/terminal.go
@@ -92,7 +92,7 @@ func (r Terminal) File(ctx context.Context, fr *malcontent.FileReport) error {
 	return nil
 }
 
-func (r Terminal) Full(ctx context.Context, rep *malcontent.Report) error {
+func (r Terminal) Full(ctx context.Context, _ *malcontent.Config, rep *malcontent.Report) error {
 	// Non-diff files are handled on the fly by File()
 	if rep.Diff == nil {
 		return nil

diff --git a/pkg/render/terminal_brief.go b/pkg/render/terminal_brief.go
@@ -69,7 +69,7 @@ func (r TerminalBrief) File(_ context.Context, fr *malcontent.FileReport) error
 	return nil
 }
 
-func (r TerminalBrief) Full(_ context.Context, rep *malcontent.Report) error {
+func (r TerminalBrief) Full(_ context.Context, _ *malcontent.Config, rep *malcontent.Report) error {
 	// Non-diff files are handled on the fly by File()
 	if rep.Diff == nil {
 		return nil

diff --git a/pkg/render/yaml.go b/pkg/render/yaml.go
@@ -28,7 +28,7 @@ func (r YAML) File(_ context.Context, _ *malcontent.FileReport) error {
 	return nil
 }
 
-func (r YAML) Full(_ context.Context, rep *malcontent.Report) error {
+func (r YAML) Full(_ context.Context, c *malcontent.Config, rep *malcontent.Report) error {
 	// Make the sync.Map YAML-friendly
 	yr := Report{
 		Diff:   rep.Diff,
@@ -52,6 +52,10 @@ func (r YAML) Full(_ context.Context, rep *malcontent.Report) error {
 		return true
 	})
 
+	if c != nil && c.Stats && yr.Diff == nil {
+		yr.Stats = serializedStats(c, rep)
+	}
+
 	yaml, err := yaml.Marshal(yr)
 	if err != nil {
 		return err

diff --git a/tests/macOS/clean/ls.stats.json b/tests/macOS/clean/ls.stats.json
@@ -0,0 +1,113 @@
+{
+    "Files": {
+        "macOS/clean/ls": {
+            "Path": "macOS/clean/ls",
+            "SHA256": "461b7ef5288c9c4b0d10aefc9ca42f7ddab9954a3f3d032e3a783e3da0c970b6",
+            "Size": 154352,
+            "Syscalls": [
+                "getdents",
+                "openat",
+                "readlink"
+            ],
+            "Pledge": [
+                "rpath"
+            ],
+            "Behaviors": [
+                {
+                    "Description": "binary contains hardcoded URL",
+                    "MatchStrings": [
+                        "http://crl.apple.com/codesigning.crl0",
+                        "http://www.apple.com/DTDs/PropertyList",
+                        "http://www.apple.com/appleca/root.crl0",
+                        "https://www.apple.com/appleca/0"
+                    ],
+                    "RiskScore": 1,
+                    "RiskLevel": "LOW",
+                    "RuleURL": "https://github.com/chainguard-dev/malcontent/blob/main/rules/c2/addr/url.yara#binary_with_url",
+                    "ID": "c2/addr/url",
+                    "RuleName": "binary_with_url"
+                },
+                {
+                    "Description": "Look up or override terminal settings",
+                    "MatchStrings": [
+                        "TERM"
+                    ],
+                    "RiskScore": 1,
+                    "RiskLevel": "LOW",
+                    "RuleURL": "https://github.com/chainguard-dev/malcontent/blob/main/rules/exec/shell/TERM.yara#TERM",
+                    "ReferenceURL": "https://www.gnu.org/software/gettext/manual/html_node/The-TERM-variable.html",
+                    "ID": "exec/shell/TERM",
+                    "RuleName": "TERM"
+                },
+                {
+                    "Description": "traverse filesystem hierarchy",
+                    "MatchStrings": [
+                        "_fts_children",
+                        "_fts_close",
+                        "_fts_open",
+                        "_fts_read",
+                        "_fts_set"
+                    ],
+                    "RiskScore": 1,
+                    "RiskLevel": "LOW",
+                    "RuleURL": "https://github.com/chainguard-dev/malcontent/blob/main/rules/fs/directory/directory-traverse.yara#fts",
+                    "ID": "fs/directory/traverse",
+                    "RuleName": "fts"
+                },
+                {
+                    "Description": "read value of a symbolic link",
+                    "MatchStrings": [
+                        "readlink"
+                    ],
+                    "RiskScore": 1,
+                    "RiskLevel": "LOW",
+                    "RuleURL": "https://github.com/chainguard-dev/malcontent/blob/main/rules/fs/link-read.yara#readlink",
+                    "ReferenceURL": "https://man7.org/linux/man-pages/man2/readlink.2.html",
+                    "ID": "fs/link_read",
+                    "RuleName": "readlink"
+                }
+            ],
+            "RiskScore": 1,
+            "RiskLevel": "LOW"
+        }
+    },
+    "Stats": {
+        "PkgStats": [
+            {
+                "Count": 1,
+                "Key": "c2/addr/url",
+                "Total": 4,
+                "Value": 25
+            },
+            {
+                "Count": 1,
+                "Key": "exec/shell/TERM",
+                "Total": 4,
+                "Value": 25
+            },
+            {
+                "Count": 1,
+                "Key": "fs/directory/traverse",
+                "Total": 4,
+                "Value": 25
+            },
+            {
+                "Count": 1,
+                "Key": "fs/link_read",
+                "Total": 4,
+                "Value": 25
+            }
+        ],
+        "ProcessedFiles": 1,
+        "RiskStats": [
+            {
+                "Count": 1,
+                "Key": 1,
+                "Total": 1,
+                "Value": 100
+            }
+        ],
+        "TotalBehaviors": 4,
+        "TotalRisks": 1
+    }
+}