Harden health checks and refresh GitHub health cache

cmd/awesome-docker/main.go

@@ -200,6 +200,12 @@ func healthCmd() *cobra.Command {
 			for _, e := range errs {
 				fmt.Printf("  error: %v\n", e)
 			}
+			if len(infos) == 0 {
+				if len(errs) > 0 {
+					return fmt.Errorf("failed to fetch GitHub metadata for all repositories (%d errors); check network/DNS and GITHUB_TOKEN", len(errs))
+				}
+				return fmt.Errorf("no GitHub repositories found in README")
+			}
 
 			scored := scorer.ScoreAll(infos)
 			cacheEntries := scorer.ToCacheEntries(scored)

internal/checker/github.go

@@ -27,21 +27,35 @@ type RepoInfo struct {
 
 // ExtractGitHubRepo extracts owner/name from a GitHub URL.
 // Returns false for non-repo URLs (issues, wiki, apps, etc.).
-func ExtractGitHubRepo(url string) (owner, name string, ok bool) {
+func ExtractGitHubRepo(rawURL string) (owner, name string, ok bool) {
-	if !strings.HasPrefix(url, "https://github.com/") {
+	u, err := url.Parse(rawURL)
+	if err != nil {
 		return "", "", false
 	}
-	path := strings.TrimPrefix(url, "https://github.com/")
+
-	path = strings.TrimRight(path, "/")
+	host := strings.ToLower(u.Hostname())
+	if host != "github.com" && host != "www.github.com" {
+		return "", "", false
+	}
+
+	path := strings.Trim(u.Path, "/")
 	parts := strings.Split(path, "/")
 	if len(parts) != 2 || parts[0] == "" || parts[1] == "" {
 		return "", "", false
 	}
-	// Skip non-repo paths
+
-	if parts[0] == "apps" || parts[0] == "features" || parts[0] == "topics" {
+	// Skip known non-repository top-level routes.
+	switch parts[0] {
+	case "apps", "features", "topics":
 		return "", "", false
 	}
-	return parts[0], parts[1], true
+
+	name = strings.TrimSuffix(parts[1], ".git")
+	if name == "" {
+		return "", "", false
+	}
+
+	return parts[0], name, true
 }
 
 func isHTTPURL(raw string) bool {
@@ -52,6 +66,16 @@ func isHTTPURL(raw string) bool {
 	return u.Scheme == "http" || u.Scheme == "https"
 }
 
+func isGitHubAuthError(err error) bool {
+	if err == nil {
+		return false
+	}
+	s := strings.ToLower(err.Error())
+	return strings.Contains(s, "401 unauthorized") ||
+		strings.Contains(s, "bad credentials") ||
+		strings.Contains(s, "resource not accessible by integration")
+}
+
 // PartitionLinks separates URLs into GitHub repos and external HTTP(S) links.
 func PartitionLinks(urls []string) (github, external []string) {
 	for _, url := range urls {
@@ -134,6 +158,9 @@ func (gc *GitHubChecker) CheckRepos(ctx context.Context, urls []string, batchSiz
 		info, err := gc.CheckRepo(ctx, owner, name)
 		if err != nil {
 			errs = append(errs, err)
+			if isGitHubAuthError(err) {
+				break
+			}
 			continue
 		}
 		results = append(results, info)

internal/checker/github_test.go

@@ -1,6 +1,7 @@
 package checker
 
 import (
+	"errors"
 	"testing"
 )
 
@@ -14,6 +15,10 @@ func TestExtractGitHubRepo(t *testing.T) {
 		{"https://github.com/docker/compose", "docker", "compose", true},
 		{"https://github.com/moby/moby", "moby", "moby", true},
 		{"https://github.com/user/repo/", "user", "repo", true},
+		{"https://github.com/user/repo?tab=readme-ov-file", "user", "repo", true},
+		{"https://github.com/user/repo#readme", "user", "repo", true},
+		{"https://github.com/user/repo.git", "user", "repo", true},
+		{"https://www.github.com/user/repo", "user", "repo", true},
 		{"https://github.com/user/repo/issues", "", "", false},
 		{"https://github.com/user/repo/wiki", "", "", false},
 		{"https://github.com/apps/dependabot", "", "", false},
@@ -52,3 +57,22 @@ func TestPartitionLinks(t *testing.T) {
 		t.Errorf("external links = %d, want 2", len(ext))
 	}
 }
+
+func TestIsGitHubAuthError(t *testing.T) {
+	tests := []struct {
+		err  error
+		want bool
+	}{
+		{errors.New("non-200 OK status code: 401 Unauthorized body: \"Bad credentials\""), true},
+		{errors.New("Resource not accessible by integration"), true},
+		{errors.New("dial tcp: lookup api.github.com: no such host"), false},
+		{errors.New("context deadline exceeded"), false},
+	}
+
+	for _, tt := range tests {
+		got := isGitHubAuthError(tt.err)
+		if got != tt.want {
+			t.Errorf("isGitHubAuthError(%q) = %v, want %v", tt.err, got, tt.want)
+		}
+	}
+}