Don't scan subgroups past gitlab-rails limit

3 files changed, 64 insertions, 11 deletions

diff --git a/internal/domain/domain.go b/internal/domain/domain.go
index 3638c372..8345f69c 100644
--- a/internal/domain/domain.go
+++ b/internal/domain/domain.go
@@ -22,10 +22,11 @@ import (
 )
 
 const (
-	// MaxProjectDepth is set to the maximum nested project depth in gitlab (21) plus 3.
+	subgroupScanLimit int = 21
+	// maxProjectDepth is set to the maximum nested project depth in gitlab (21) plus 3.
 	// One for the project, one for the first empty element of the split (URL.Path starts with /),
 	// and one for the real file path
-	MaxProjectDepth int = 24
+	maxProjectDepth int = subgroupScanLimit + 3
 )
 
 type locationDirectoryError struct {
@@ -120,7 +121,7 @@ func getHost(r *http.Request) string {
 func (d *D) getProjectWithSubpath(r *http.Request) (*project, string, string) {
 	// Check for a project specified in the URL: http://group.gitlab.io/projectA
 	// If present, these projects shadow the group domain.
-	split := strings.SplitN(r.URL.Path, "/", MaxProjectDepth)
+	split := strings.SplitN(r.URL.Path, "/", maxProjectDepth)
 	if len(split) >= 2 {
 		project, projectPath, urlPath := d.digProjectWithSubpath("", split[1:])
 		if project != nil {
diff --git a/internal/domain/map.go b/internal/domain/map.go
index 26657a1c..2891a272 100644
--- a/internal/domain/map.go
+++ b/internal/domain/map.go
@@ -60,7 +60,7 @@ func (dm Map) updateGroupDomain(rootDomain, groupName, projectPath string, https
 		}
 	}
 
-	split := strings.SplitN(strings.ToLower(projectPath), "/", MaxProjectDepth)
+	split := strings.SplitN(strings.ToLower(projectPath), "/", maxProjectDepth)
 	projectName := split[len(split)-1]
 	g := &groupDomain.group
 
@@ -108,7 +108,7 @@ func (dm Map) readProjectConfig(rootDomain string, group, projectName string, co
 	}
 }
 
-func readProject(group, parent, projectName string, fanIn chan<- jobResult) {
+func readProject(group, parent, projectName string, level int, fanIn chan<- jobResult) {
 	if strings.HasPrefix(projectName, ".") {
 		return
 	}
@@ -121,8 +121,10 @@ func readProject(group, parent, projectName string, fanIn chan<- jobResult) {
 	projectPath := filepath.Join(parent, projectName)
 	if _, err := os.Lstat(filepath.Join(group, projectPath, "public")); err != nil {
 		// maybe it's a subgroup
-		buf := make([]byte, 2*os.Getpagesize())
-		readProjects(group, projectPath, buf, fanIn)
+		if level <= subgroupScanLimit {
+			buf := make([]byte, 2*os.Getpagesize())
+			readProjects(group, projectPath, level+1, buf, fanIn)
+		}
 
 		return
 	}
@@ -137,7 +139,7 @@ func readProject(group, parent, projectName string, fanIn chan<- jobResult) {
 	fanIn <- jobResult{group: group, project: projectPath, config: config}
 }
 
-func readProjects(group, parent string, buf []byte, fanIn chan<- jobResult) {
+func readProjects(group, parent string, level int, buf []byte, fanIn chan<- jobResult) {
 	subgroup := filepath.Join(group, parent)
 	fis, err := godirwalk.ReadDirents(subgroup, buf)
 	if err != nil {
@@ -154,7 +156,7 @@ func readProjects(group, parent string, buf []byte, fanIn chan<- jobResult) {
 			continue
 		}
 
-		readProject(group, parent, project.Name(), fanIn)
+		readProject(group, parent, project.Name(), level, fanIn)
 	}
 }
 
@@ -178,7 +180,7 @@ func (dm Map) ReadGroups(rootDomain string, fis godirwalk.Dirents) {
 			for group := range fanOutGroups {
 				started := time.Now()
 
-				readProjects(group, "", buf, fanIn)
+				readProjects(group, "", 0, buf, fanIn)
 
 				log.WithFields(log.Fields{
 					"group":    group,
diff --git a/internal/domain/map_test.go b/internal/domain/map_test.go
index f7b065e4..dc5e8648 100644
--- a/internal/domain/map_test.go
+++ b/internal/domain/map_test.go
@@ -5,6 +5,7 @@ import (
 	"fmt"
 	"io/ioutil"
 	"os"
+	"strings"
 	"testing"
 	"time"
 
@@ -78,7 +79,56 @@ func TestReadProjects(t *testing.T) {
 	subgroup, ok := domain.subgroups["subgroup"]
 	require.True(t, ok, "missing group.test.io subgroup")
 	_, ok = subgroup.projects["project"]
-	require.True(t, ok, "missing project for subgrup in group.test.io domain")
+	require.True(t, ok, "missing project for subgroup in group.test.io domain")
+}
+
+func TestReadProjectsMaxDepth(t *testing.T) {
+	nGroups := 3
+	levels := subgroupScanLimit + 5
+	cleanup := buildFakeDomainsDirectory(t, nGroups, levels)
+	defer cleanup()
+
+	defaultDomain := "test.io"
+	dm := make(Map)
+	dm.ReadGroups(defaultDomain, getEntries(t))
+
+	var domains []string
+	for d := range dm {
+		domains = append(domains, d)
+	}
+
+	var expectedDomains []string
+	for i := 0; i < nGroups; i++ {
+		expectedDomains = append(expectedDomains, fmt.Sprintf("group-%d.%s", i, defaultDomain))
+	}
+
+	for _, expected := range domains {
+		assert.Contains(t, domains, expected)
+	}
+
+	for _, actual := range domains {
+		// we are not checking config.json domains here
+		if !strings.HasSuffix(actual, defaultDomain) {
+			continue
+		}
+		assert.Contains(t, expectedDomains, actual)
+	}
+
+	// check subgroups
+	domain, ok := dm["group-0.test.io"]
+	require.True(t, ok, "missing group-0.test.io domain")
+	subgroup := &domain.group
+	for i := 0; i < levels; i++ {
+		subgroup, ok = subgroup.subgroups["sub"]
+		if i <= subgroupScanLimit {
+			require.True(t, ok, "missing group-0.test.io subgroup at level %d", i)
+			_, ok = subgroup.projects["project-0"]
+			require.True(t, ok, "missing project for subgroup in group-0.test.io domain at level %d", i)
+		} else {
+			require.False(t, ok, "subgroup level %d. Maximum allowed nesting level is %d", i, subgroupScanLimit)
+			break
+		}
+	}
 }
 
 // This write must be atomic, otherwise we cannot predict the state of the