add --max-size & --refresh

Signed-off-by: Hadi <[email protected]>

gh-recon/deep.go

@@ -82,9 +82,10 @@ type DeepResult struct {
 	Repository string
 	Owner      string
 	Name       string
+	Size       int
 }
 
-func (r Recon) Deep(username, excludeRepos string) (response []DeepResult) {
+func (r Recon) Deep(username, excludeRepos string, refresh bool) (response []DeepResult) {
 	excludeReposList := strings.Split(excludeRepos, ",")
 	repos, resp, err := r.client.Repositories.ListByUser(
 		r.ctx,
@@ -107,6 +108,7 @@ func (r Recon) Deep(username, excludeRepos string) (response []DeepResult) {
 				Repository: repo.GetCloneURL(),
 				Owner:      repo.GetOwner().GetLogin(),
 				Name:       repo.GetName(),
+				Size:       repo.GetSize(),
 			})
 		}
 	}
@@ -119,15 +121,43 @@ func (r Recon) Deep(username, excludeRepos string) (response []DeepResult) {
 	}
 
 	tmp_folder := "/tmp/ghrecon-" + username
+
+	if folderExists(tmp_folder) {
+		if refresh {
+			r.PrintInfo("INFO", "Deleting existing folder "+tmp_folder)
+			err := os.RemoveAll(tmp_folder)
+			if err != nil {
+				r.PrintInfo("ERROR", "Failed to delete existing folder "+tmp_folder)
+			}
+		}
+	}
+
 	for _, repo := range response {
 		if slices.Contains(excludeReposList, repo.Name) ||
 			slices.Contains(excludeReposList, repo.Owner+"/"+repo.Name) {
 			r.PrintInfo("INFO", "Skipping repository", repo.Owner+"/"+repo.Name)
 			continue
 		}
+
+		maxRepoSize := r.maxRepoSize * 1024
+
+		if repo.Size > maxRepoSize {
+			r.PrintInfo(
+				"INFO",
+				"Skipping repository "+repo.Owner+"/"+repo.Name+" due to size", fmt.Sprintf(
+					"%d",
+					repo.Size/1024,
+				)+"MB > "+fmt.Sprintf(
+					"%d",
+					maxRepoSize/1024,
+				)+"MB",
+			)
+			continue
+		}
 		r.PrintInfo(
 			"Downloading",
 			repo.Owner+"/"+repo.Name,
+			fmt.Sprintf("%d", repo.Size/1024)+"MB",
 		)
 
 		destination := tmp_folder + "/" + repo.Owner + "/" + repo.Name

gh-recon/main.go

@@ -8,11 +8,12 @@ import (
 )
 
 type Recon struct {
-	client   *github.Client
-	logger   *log.Logger
-	ctx      context.Context
-	silent   bool
-	jsonFile string
+	client      *github.Client
+	logger      *log.Logger
+	ctx         context.Context
+	silent      bool
+	jsonFile    string
+	maxRepoSize int
 }
 
 func NewRecon(
@@ -21,12 +22,14 @@ func NewRecon(
 	ctx context.Context,
 	silent bool,
 	jsonFile string,
+	maxRepoSize int,
 ) *Recon {
 	return &Recon{
-		client:   client,
-		logger:   logger,
-		ctx:      ctx,
-		silent:   silent,
-		jsonFile: jsonFile,
+		client:      client,
+		logger:      logger,
+		ctx:         ctx,
+		silent:      silent,
+		jsonFile:    jsonFile,
+		maxRepoSize: maxRepoSize,
 	}
 }

main.go

@@ -19,9 +19,17 @@ func main() {
 	var silent bool
 	var jsonFile string
 	var excludeRepos string
+	var maxRepoSize int
+	var refresh bool
 	flag.StringVarP(&username, "username", "u", "", "GitHub username to analyze")
 	flag.StringVarP(&token, "token", "t", "", "GitHub personal access token (e.g. ghp_...)")
 	flag.StringVarP(&fromEmail, "email", "e", "", "Search accounts by email address")
+	flag.IntVar(
+		&maxRepoSize,
+		"max-size",
+		150,
+		"Limit the size of repositories to scan (in MB) (Only for deep scan)",
+	)
 	flag.BoolVarP(
 		&onlyCommitsLeak,
 		"only-commits",
@@ -29,6 +37,13 @@ func main() {
 		false,
 		"Display only commits with author info",
 	)
+	flag.BoolVarP(
+		&refresh,
+		"refresh",
+		"r",
+		false,
+		"Refresh the cache (deep scan only)",
+	)
 	flag.BoolVarP(
 		&deep,
 		"deep",
@@ -85,6 +100,7 @@ func main() {
 		ctx,
 		silent,
 		jsonFile,
+		maxRepoSize,
 	)
 
 	r.Header()
@@ -130,7 +146,7 @@ func main() {
 	}
 
 	if deep {
-		results["Deep"] = r.Deep(username, excludeRepos)
+		results["Deep"] = r.Deep(username, excludeRepos, refresh)
 	}
 
 	r.WriteJson(results)