sbolint github URL checking, WIP.

1 files changed, 55 insertions, 64 deletions

diff --git a/sbolint b/sbolint
index 6a021e0..28b5ac9 100755
--- a/sbolint
+++ b/sbolint
@@ -1199,41 +1199,21 @@ sub curl_head_request {
 	return close($pipe);
 }
 
-# WIP, maybe no longer needed
-## sub check_github_url {
-## 	my $file = shift;
-## 	my $url = shift;
-## 	return unless $url =~ m{(https?:)//github\.com};
-## 
-## 	if($1 eq "http:") {
-## 		log_warning("$file: github URL $url should be https");
-## 	}
-## 
-## 	(my $expect_filename = $url) =~ s,.*/,,;
-## 	my(undef, undef, undef, $user, $prog, $archive, $ver, $filename) = split /\//, $url;
-## 	warn "user $user, prog $prog, archive $archive, ver $ver, filename $filename, expect_filename $expect_filename\n";
-## 
-## 	# assume these are correct, for now
-## 	return if $user eq 'downloads';
-## 	return if $archive eq 'releases';
-## 
-## 	# TODO: work out what to do about /raw/
-## 	return if $archive eq 'raw';
-## 
-## 	if($archive ne 'archive') {
-## 		log_warning("$file: unknown github URL type: $url");
-## 		return;
-## 	}
-## 
-## 	# OK, good URLs look like this:
-## 	# https://github.com/jeetsukumaran/DendroPy/archive/v4.4.0/DendroPy-4.4.0.tar.gz
-## 	# ...and bad ones look like this:
-## 	# https://github.com/haiwen/seafile-client/archive/v4.4.2.tar.gz
-## 	# Corrected version of the bad one would be:
-## 	# https://github.com/haiwen/seafile-client/archive/v4.4.2/seafile-client-4.4.2.tar.gz
-## 	# Notice the "v" isn't part of the version number. It's not always there,
-## 	# and sometimes it's a different letter (r, or g, or capital V, etc).
-## }
+# github is a mess...
+# OK, good URLs look like this:
+# https://github.com/jeetsukumaran/DendroPy/archive/v4.4.0/DendroPy-4.4.0.tar.gz
+# ...and bad ones look like this:
+# https://github.com/haiwen/seafile-client/archive/v4.4.2.tar.gz
+# Corrected version of the bad one would be:
+# https://github.com/haiwen/seafile-client/archive/v4.4.2/seafile-client-4.4.2.tar.gz
+# Notice the "v" isn't part of the version number. It's not always there.
+# If there's a "v" and it's immediately followed by a number, it's not part of
+# the version number. If it's followed by something other than a number, e.g.
+# ver-1.0 or v.1.0, it *is* part of the version number.
+# Since git allows / characters in tag names, we sometimes get a URL like:
+# https://github.com/jeremysalwen/lv2file/archive/upstream/0.95/lv2file-upstream-0.95.tar.gz
+# ...which is perfectly valid. However, this is invalid:
+# https://github.com/jeremysalwen/lv2file/archive/upstream/0.95.tar.gz
 
 sub check_github_url {
 	my $file = shift;
@@ -1241,55 +1221,66 @@ sub check_github_url {
 	my $old_url = $url;
 	my $new_url;
 	my $ext;
+	my $tag;
+	my $ver;
+	my $dir;
 
 	# do not police releases/ or raw/ URLs, only archive/
-	return unless $url =~ m{github\.com/.*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;
+	    #https:  //    site/  .../   .../  archive/  ...everything else.
+	(my $proto, undef, undef, $user, $proj, undef,   @parts) = 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;
-	}
+	my $filename = pop @parts;
+	$filename =~ m,(\.tar\.gz|\.zip)$,;
+	$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;
+	if(!defined $ext) {
+		log_warning("$file: github URL should end in .tar.gz or .zip");
+		$ext = "";
 	}
 
-	# 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 = $1;
+	if(@parts == 0) {
+		# filename loox like: tag.tar.gz
+		($tag = $filename) =~ s,(\.tar\.gz|\.zip)$,,;
+		log_error "$file: github URL is non-canonical (not enough components)";
+		$dir = $tag;
+	} elsif(@parts == 1) {
+		$tag = $dir = $parts[0];
+		if($tag =~ /^[0-9a-f]{6,}$/ && $tag !~ /^20\d{6,}/) {
+			# commit hash. the /^20\d{6,}/ is to exclude ISO dates like 20240402
+			if(length($tag) < 40) {
+				# shortened, maybe the full hash is in the filename?
+				if($filename =~ /([0-9a-f]{40})\./) {
+					$tag = $1; # leave $dir alone! it's allowed to be shortened.
+				} else {
+					# if not, we give up.
+					log_error("$file: github commit URL needs full 40-digit commit hash in filename");
+					return;
+				}
+			}
+		}
+	} else {
+		$tag = join("-", @parts);
+		$dir = join("/", @parts);
 	}
 
-	if(!defined $ext) {
-		log_warning "$file: github URL not .tar.gz or .zip";
-		$ext = "";
-	}
+	($ver = $tag) =~ s,^v(\d),$1,i;
+
+	# TODO: commit hashes have to be complete in the filename, and may be
+	# truncated in the dir name after archive/
 
 	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";
+	$new_url = "https://github.com/$user/$proj/archive/$dir/$proj-$ver$ext";
 
 	if($old_url ne $new_url) {
 		log_warning("$file: canonical github URL should be: $new_url");