sbolint: github URL checking initial (incomplete) implementation.

1 files changed, 62 insertions, 1 deletions

diff --git a/sbolint b/sbolint
index bb7355b..e08703f 100755
--- a/sbolint
+++ b/sbolint
@@ -1146,7 +1146,7 @@ sub check_dl_and_md5 {
 			next;
 		}
 
-		#check_github_url($file, $u);
+		check_github_url($file, $u);
 
 		if($url_head) {
 			curl_head_request($file, $u) || do {
@@ -1235,6 +1235,67 @@ sub curl_head_request {
 ## 	# and sometimes it's a different letter (r, or g, or capital V, etc).
 ## }
 
+sub check_github_url {
+	my $file = shift;
+	my $url = shift;
+	my $old_url = $url;
+	my $new_url;
+	my $ext;
+
+	# do not police releases/ or raw/ URLs, only archive/
+	return unless $url =~ m{github\.com/.*archive/};
+
+	if($url =~ s,refs/tags/,,) {
+		log_error "$file: github URLs should not have refs/tags/";
+	}
+
+	    #https:  //    site/  .../   .../  archive/
+	(my $proto, undef, undef, $user, $proj, undef, $tag, $filename, $extra) = split /\//, $url;
+
+	log_error "$file: github URLs must be https://" unless $proto eq 'https:';
+
+	if(!defined $filename) {
+		log_error "$file: github URL is non-canonical (not enough components)";
+		$tag =~ s,(\.tar\.gz|\.zip)$,,;
+		$ext = $1;
+	}
+
+	# TODO: this is not an error, because github projects are allowed to have /
+	# in their tag names. So we get this *valid* URL:
+	# https://github.com/zfsonlinux/zfs-auto-snapshot/archive/upstream/1.2.4/zfs-auto-snapshot-upstream-1.2.4.tar.gz
+	if(defined $extra) {
+		#log_error "$file: github URL is non-canonical (too many components)";
+		log_note "$file: github URL has extra components, don't (yet) know how to check it.";
+		return;
+	}
+
+	# TODO: commit hashes have to be complete in the filename, and may be
+	# truncated in the dir name after arvhive/
+
+	my $ver = $tag;
+	$ver =~ s,^v(\d),$1,;
+
+	if(defined $filename) {
+		$filename =~ /(\.tar\.gz|\.zip)$/;
+		$ext = $1;
+	}
+
+	if(!defined $ext) {
+		log_warning "$file: github URL not .tar.gz or .zip";
+		$ext = "";
+	}
+
+	if(defined $filename && $filename ne "$proj-$ver$ext") {
+		log_error "$file: github URL has wrong filename $filename";
+	}
+
+	$new_url = "https://github.com/$user/$proj/archive/$tag/$proj-$ver$ext";
+
+	if($old_url ne $new_url) {
+		log_warning("$file: canonical github URL should be: $new_url");
+	}
+}
+
 sub script_exists {
 	my $file = $buildname . ".SlackBuild";
 	unless(-e $file) {