From ac58687d760e52627181c335443ed6cb3bbf3b8e Mon Sep 17 00:00:00 2001
From: desmax74 <mdessi@redhat.com>
Date: Wed, 11 Oct 2023 20:21:45 +0200
Subject: [PATCH] Add Guacone collect files json.bz2 capability

Signed-off-by: desmax74 <mdessi@redhat.com>
---
 cmd/guacone/cmd/gcs_test.go                   |  21 +++++++++++++++++-
 .../exampledata/busybox-cyclonedx.json.bz2    | Bin 0 -> 543 bytes
 pkg/handler/processor/process/process.go      |  14 +++++++++---
 pkg/handler/processor/processor.go            |   5 +++++
 4 files changed, 36 insertions(+), 4 deletions(-)
 create mode 100644 internal/testing/testdata/exampledata/busybox-cyclonedx.json.bz2

diff --git a/cmd/guacone/cmd/gcs_test.go b/cmd/guacone/cmd/gcs_test.go
index 687daa7fe1..eb106bd994 100644
--- a/cmd/guacone/cmd/gcs_test.go
+++ b/cmd/guacone/cmd/gcs_test.go
@@ -15,7 +15,11 @@
 
 package cmd
 
-import "testing"
+import (
+	"testing"
+
+	"github.com/spf13/cobra"
+)
 
 func TestValidateGCSFlags(t *testing.T) {
 	testCases := []struct {
@@ -78,3 +82,18 @@ func TestValidateGCSFlags(t *testing.T) {
 	}
 
 }
+
+func TestJsonBz2Ingestion(t *testing.T) {
+	rootCmd := &cobra.Command{
+		Use:   "guacone",
+		Short: "guacone",
+	}
+	rootCmd.AddCommand(collectCmd)
+	rootCmd.AddCommand(filesCmd)
+	bz2Path := "./../../../internal/testing/testdata/exampledata/busybox-cyclonedx.json.bz2"
+	rootCmd.SetArgs([]string{"collect", "files", bz2Path})
+	err := rootCmd.Execute()
+	if err != nil {
+		t.Fatal(err)
+	}
+}
diff --git a/internal/testing/testdata/exampledata/busybox-cyclonedx.json.bz2 b/internal/testing/testdata/exampledata/busybox-cyclonedx.json.bz2
new file mode 100644
index 0000000000000000000000000000000000000000..f39d16880a9683ac5d9b5d9cc5d7738e90be14f0
GIT binary patch
literal 543
zcmV+)0^t2ZT4*^jL0KkKS*Of5i~s^I-+)yRPzV3;9VG=yKkwh_Kmq*b<Twoy6V*S|
z(wTr0De0OMBm+T{$`MU8^)`)D^)hJyGynhq0Mk>{rh`BLZA}0GXa;~KG)+@YJyX<d
znt`AI&;SD@vkY0SZ0YK($jK8lAfSj%0RZ5W6luu70ajpc7Gcu~2_%ro0@gb|7$Smm
z*UOXJjPB04P=@!$&EPnRy(JxF)d&wP6Q^mSikK1*(h%Yd3C03NiC~@(PXX+L!2mH3
z0V4qCQ8sJ&<CC^^&cX}?^*eXX-R)XJ<Rhk*AIkpH0u-2iQj}3~pyn7p=&YAJ2_JFt
z19S~jFeIH%l*&`dJ62@(c%d0a2!fn2SP5E~B{F0wiNTXOt#U?`MKn^W005*!I9dfI
ziSY<sU$=FvU-c7ow6H#K(N}AniRsLDXdR^hJ4RGarpx2X)!>7OE($NB7(XDrmPd0C
zy&$~u#X>#N0B_AFhM*{W8JP&8jaD4GvppZDf0rebQFOdonK?z$21ipWqOu$u-zsTd
z)v9fc57qD#WRZa+3u>=$2~;e)j23geuu6a<Bs7lc<SnKpq=A0<8+EqbTQ!v%PXVKF
zgL=8|lx0uNd|ym(36L?2V5y5TeD?%e0%L1dZ7vU%X@Qz1=b%nYs9Y89j7$d6;!t7h
hpbf_=hS->KSa~l$ko|X|5e-lAcO+AV2?~6}c)<Nk=urRw

literal 0
HcmV?d00001

diff --git a/pkg/handler/processor/process/process.go b/pkg/handler/processor/process/process.go
index 8af4d6f433..1e3c0a314e 100644
--- a/pkg/handler/processor/process/process.go
+++ b/pkg/handler/processor/process/process.go
@@ -22,6 +22,8 @@ import (
 	"encoding/xml"
 	"fmt"
 	"io"
+	"path/filepath"
+	"strings"
 
 	uuid "github.com/gofrs/uuid"
 	"github.com/guacsec/guac/pkg/emitter"
@@ -216,6 +218,13 @@ func decodeDocument(ctx context.Context, i *processor.Document) error {
 	logger := logging.FromContext(ctx)
 	var reader io.Reader
 	var err error
+	if i.Encoding == "" {
+		ext := filepath.Ext(i.SourceInformation.Source)
+		encoding, ok := processor.EncodingExts[strings.ToLower(ext)]
+		if ok {
+			i.Encoding = encoding
+		}
+	}
 	logger.Infof("Decoding document with encoding:  %v", i.Encoding)
 	switch i.Encoding {
 	case processor.EncodingBzip2:
@@ -225,17 +234,16 @@ func decodeDocument(ctx context.Context, i *processor.Document) error {
 		if err != nil {
 			return fmt.Errorf("unable to create zstd reader: %w", err)
 		}
-	case processor.EncodingUnknown:
 	}
 	if reader != nil {
-		if err := decompressDocument(ctx, i, reader); err != nil {
+		if err := decompressDocument(i, reader); err != nil {
 			return fmt.Errorf("unable to decode document: %w", err)
 		}
 	}
 	return nil
 }
 
-func decompressDocument(ctx context.Context, i *processor.Document, reader io.Reader) error {
+func decompressDocument(i *processor.Document, reader io.Reader) error {
 	uncompressed, err := io.ReadAll(reader)
 	if err != nil {
 		return fmt.Errorf("unable to decompress document: %w", err)
diff --git a/pkg/handler/processor/processor.go b/pkg/handler/processor/processor.go
index 6a4eb0aba4..adefc680f4 100644
--- a/pkg/handler/processor/processor.go
+++ b/pkg/handler/processor/processor.go
@@ -87,6 +87,11 @@ const (
 	EncodingUnknown EncodingType = "UNKNOWN"
 )
 
+var EncodingExts = map[string]EncodingType{
+	".bz2": EncodingBzip2,
+	".zst": EncodingZstd,
+}
+
 // SourceInformation provides additional information about where the document comes from
 type SourceInformation struct {
 	// Collector describes the name of the collector providing this information