add url parser and test for git remote url parser

Author: wahyudibo

internal/command/scan/handler.go

@@ -67,7 +67,7 @@ func (h *Handler) Execute(ctx context.Context) (constant.ExitCode, error) {
 	h.stopLoadingMessage()
 
 	if !h.Args.Quiet {
-		fmt.Fprintf(w, "Project name: %s\nGit provider: %s\n", repoMetadata.Name, repoMetadata.Provider)
+		fmt.Fprintf(w, "Project name: %s\nGit provider: %s\n", repoMetadata.RepoName, repoMetadata.Provider)
 		if h.Args.Format == "" || h.Args.Format == FormatPretty {
 			fmt.Fprintf(w, "Format: %s (default)\n", FormatPretty)
 		} else {
@@ -88,15 +88,15 @@ func (h *Handler) Execute(ctx context.Context) (constant.ExitCode, error) {
 	}
 
 	// pass the list of the tracked files and compress it into zip file.
-	h.displayCompressingMessage(repoMetadata.Name)
-	projectZipBuf, err := h.Archiver.OutputZipToIOReader(repoMetadata.Path, filepaths)
+	h.displayCompressingMessage(repoMetadata.RepoName)
+	projectZipBuf, err := h.Archiver.OutputZipToIOReader(repoMetadata.DirPath, filepaths)
 	if err != nil {
 		return constant.ErrorExitCode, err
 	}
 	h.stopLoadingMessage()
 
 	// create presigned url for uploading the compressed file
-	projectZipName := fmt.Sprintf("%s_%s.tar.gz", repoMetadata.Name, repoMetadata.CommitHash)
+	projectZipName := fmt.Sprintf("%s_%s.tar.gz", repoMetadata.RepoName, repoMetadata.CommitHash)
 	createUploadURLReq := &grclient.CreateUploadURLReq{
 		File: projectZipName,
 	}
@@ -119,7 +119,7 @@ func (h *Handler) Execute(ctx context.Context) (constant.ExitCode, error) {
 
 	// call GuardRails trigger scan API
 	triggerScanReq := &grclient.TriggerScanReq{
-		Repository: repoMetadata.Name,
+		Repository: repoMetadata.RepoName,
 		SHA:        repoMetadata.CommitHash,
 		Branch:     repoMetadata.Branch,
 		FileName:   projectZipName,
@@ -129,7 +129,7 @@ func (h *Handler) Execute(ctx context.Context) (constant.ExitCode, error) {
 		return constant.ErrorExitCode, err
 	}
 
-	h.displayRetrievingScanResultMessage(repoMetadata.Name)
+	h.displayRetrievingScanResultMessage(repoMetadata.RepoName)
 	bo := backoff.NewConstantBackOff(h.Config.HttpClient.PollingInterval)
 	backoffCtx, backoffCtxCancel := context.WithTimeout(ctx, h.Config.HttpClient.RetryTimeout)
 	defer backoffCtxCancel()

internal/command/scan/handler_test.go

@@ -51,10 +51,10 @@ func TestScanCommandExecuteSuccess(t *testing.T) {
 	commitHash := fmt.Sprintf("%x", encrypted)
 
 	repoMetadata := &repository.Metadata{
-		Path:       args.Path,
+		DirPath:    args.Path,
 		Protocol:   "https",
 		Provider:   "github",
-		Name:       gofakeit.AppName(),
+		RepoName:   gofakeit.AppName(),
 		Branch:     gofakeit.Word(),
 		CommitHash: commitHash,
 	}
@@ -92,7 +92,7 @@ func TestScanCommandExecuteSuccess(t *testing.T) {
 		"internal/tools/spinner/spinner.go",
 		"main.go",
 	}
-	fileZipName := fmt.Sprintf("%s_%s.tar.gz", repoMetadata.Name, repoMetadata.CommitHash)
+	fileZipName := fmt.Sprintf("%s_%s.tar.gz", repoMetadata.RepoName, repoMetadata.CommitHash)
 	fileZipByte := bytes.NewReader([]byte{})
 	createUploadURLReq := &grclient.CreateUploadURLReq{
 		File: fileZipName,
@@ -105,7 +105,7 @@ func TestScanCommandExecuteSuccess(t *testing.T) {
 		File:      fileZipByte,
 	}
 	triggerScanReq := &grclient.TriggerScanReq{
-		Repository: repoMetadata.Name,
+		Repository: repoMetadata.RepoName,
 		SHA:        repoMetadata.CommitHash,
 		Branch:     repoMetadata.Branch,
 		FileName:   fileZipName,
@@ -129,7 +129,7 @@ func TestScanCommandExecuteSuccess(t *testing.T) {
 	gomock.InOrder(
 		mockRepo.EXPECT().GetMetadataFromRemoteURL().Return(repoMetadata, nil),
 		mockRepo.EXPECT().ListFiles().Return(listOfFiles, nil),
-		mockArc.EXPECT().OutputZipToIOReader(repoMetadata.Path, listOfFiles).Return(fileZipByte, nil),
+		mockArc.EXPECT().OutputZipToIOReader(repoMetadata.DirPath, listOfFiles).Return(fileZipByte, nil),
 		mockGrClient.EXPECT().CreateUploadURL(ctx, createUploadURLReq).Return(createUploadURLResp, nil),
 		mockGrClient.EXPECT().UploadProject(ctx, uploadProjectReq).Return(nil),
 		mockGrClient.EXPECT().TriggerScan(ctx, triggerScanReq).Return(triggerScanResp, nil),

internal/repository/repository.go

@@ -5,6 +5,7 @@ import (
 	"fmt"
 	"io"
 	"regexp"
+	"strings"
 
 	"github.com/go-git/go-git/v5"
 	"github.com/go-git/go-git/v5/plumbing/object"
@@ -40,12 +41,14 @@ type repository struct {
 
 // Metadata contains repository metadata.
 type Metadata struct {
-	Path       string
-	Protocol   string
-	Provider   string
-	Name       string
-	Branch     string
-	CommitHash string
+	DirPath     string
+	Protocol    string
+	Provider    string
+	UserAccount string
+	RemoteURL   string
+	RepoName    string
+	Branch      string
+	CommitHash  string
 }
 
 // New instantiates new repository.
@@ -59,7 +62,7 @@ func New(projectPath string) (Repository, error) {
 		return nil, err
 	}
 
-	return &repository{client: client, Metadata: &Metadata{Path: projectPath}}, nil
+	return &repository{client: client, Metadata: &Metadata{DirPath: projectPath}}, nil
 }
 
 // GetMetadataFromRemoteURL implements repository.Repository interface.
@@ -75,19 +78,8 @@ func (r *repository) GetMetadataFromRemoteURL() (*Metadata, error) {
 		return nil, ErrGitRemoteURLNotFound
 	}
 
-	remoteURL := remoteURLs[0]
-
-	re := regexp.MustCompile(`(?P<Protocol>git@|http(s)?:\/\/)(.+@)*(?P<Provider>[\w\d\.-]+)(:[\d]+){0,1}(\/scm)?\/*(?P<Name>.*)`)
-	matches := re.FindStringSubmatch(remoteURL)
-
-	protocolRe := regexp.MustCompile(`[^\w]`)
-	protocol := protocolRe.ReplaceAllString(matches[re.SubexpIndex("Protocol")], "")
-
-	providerRe := regexp.MustCompile(`^([\w\d-]+)([\.\w\d]*)`)
-	provider := providerRe.FindStringSubmatch(matches[re.SubexpIndex("Provider")])
-
-	nameRe := regexp.MustCompile(`\/(.*)\.git$`)
-	name := nameRe.FindStringSubmatch(matches[re.SubexpIndex("Name")])
+	// retrieves all possible information from git remote url.
+	getMetadataFromRemoteURL(r.Metadata, remoteURLs[0])
 
 	ref, err := r.client.Head()
 	if err != nil {
@@ -97,15 +89,41 @@ func (r *repository) GetMetadataFromRemoteURL() (*Metadata, error) {
 	branchRe := regexp.MustCompile(`([^\/]+$)`)
 	branch := branchRe.FindString(ref.Name().String())
 
-	r.Metadata.Protocol = protocol
-	r.Metadata.Provider = provider[1]
-	r.Metadata.Name = name[1]
+	r.Metadata.RemoteURL = remoteURLs[0]
 	r.Metadata.Branch = branch
 	r.Metadata.CommitHash = ref.Hash().String()
 
 	return r.Metadata, nil
 }
 
+func getMetadataFromRemoteURL(metadata *Metadata, gitRemoteURL string) {
+	re := regexp.MustCompile(`(?P<Protocol>git@|http(s)?:\/\/)(.+@)*(?P<Provider>[\w\d\.-]+)(:[\d]+){0,1}(\/scm)?\/*(?P<Name>.*)`)
+	matches := re.FindStringSubmatch(gitRemoteURL)
+
+	protocolRe := regexp.MustCompile(`[^\w]`)
+	protocol := protocolRe.ReplaceAllString(matches[re.SubexpIndex("Protocol")], "")
+
+	providerRe := regexp.MustCompile(`^([\w\d-]+)([\.\w\d]*)`)
+	provider := providerRe.FindStringSubmatch(matches[re.SubexpIndex("Provider")])
+
+	// use url parser to extract username and repository name by extracting the last 2 parts of the url
+	urlParts := strings.Split(gitRemoteURL, "/")
+
+	userAccount := urlParts[len(urlParts)-2]
+	// remove the "[email protected]:" from user account part if exists
+	userAccountParts := strings.Split(userAccount, ":")
+	if len(userAccountParts) > 1 {
+		userAccount = userAccountParts[1]
+	}
+
+	repoName := strings.TrimSuffix(urlParts[len(urlParts)-1], ".git")
+
+	metadata.Protocol = protocol
+	metadata.Provider = provider[1]
+	metadata.UserAccount = userAccount
+	metadata.RepoName = repoName
+}
+
 // ListFiles implements repository.Repository interface.
 func (r *repository) ListFiles() ([]string, error) {
 	filepaths := make([]string, 0)

internal/repository/repository_test.go

@@ -0,0 +1,34 @@
+package repository
+
+import (
+	"fmt"
+	"testing"
+
+	"github.com/brianvoe/gofakeit/v6"
+	"github.com/stretchr/testify/assert"
+)
+
+func TestParseGitRemoteURLToMetadataSuccess(t *testing.T) {
+	userAccount := gofakeit.Username()
+	repoName := fmt.Sprintf("%s-%s", gofakeit.LoremIpsumWord(), gofakeit.LoremIpsumWord())
+
+	urls := []string{
+		fmt.Sprintf("[email protected]:%s/%s.git", userAccount, repoName),
+		fmt.Sprintf("http://github.com/%s/%s.git", userAccount, repoName),
+		fmt.Sprintf("https://github.com/%s/%s.git", userAccount, repoName),
+		fmt.Sprintf("http://username:password@bitbucket-server:7990/scm/%s/%s.git", userAccount, repoName),
+		fmt.Sprintf("https://gitlab-on-premise.com/%s/%s.git", userAccount, repoName),
+		fmt.Sprintf("http://username:password@gitlab-ee/scm/%s/%s.git", userAccount, repoName),
+		fmt.Sprintf("https://username:[email protected]/prefix1/prefix2/prefix3/%s/%s.git", userAccount, repoName),
+	}
+
+	metadata := new(Metadata)
+
+	for _, url := range urls {
+		getMetadataFromRemoteURL(metadata, url)
+		assert.NotEmpty(t, metadata.Protocol)
+		assert.NotEmpty(t, metadata.Provider)
+		assert.Equal(t, metadata.UserAccount, userAccount)
+		assert.Equal(t, metadata.RepoName, repoName)
+	}
+}