builder: defer submodule detection to git

src/tito/builder/submodule_aware_builder.py

@@ -27,7 +27,6 @@
     find_spec_like_file,
     get_commit_timestamp,
 )
-from tito.exception import TitoException
 from tito.tar import TarFixer
 
 
@@ -109,54 +108,31 @@ def run_git_archive(self, relative_git_dir, prefix, commit, dest_tar, subdir=Non
                 "%s > /dev/null" % git_archive_cmd,
             )
 
-    # pylint: disable=too-many-locals, too-many-arguments, consider-using-f-string
-    def _submodule_archives(self, relative_git_dir, prefix, commit, initial_tar,
-                            submodule_tree="."):
-        with chdir(submodule_tree):
-            submodules_cmd = "git config --file .gitmodules --get-regexp path"
-            submodules_output = run_command(submodules_cmd)
-
-        # split submodules output on newline
-        # then on tab, and the directory is the last entry
-        submodules_list = [
-            line.split(" ")[-1] for line in submodules_output.split("\n")
-        ]
-
-        # We ignore the hash in the sub modules list as we'll have to get the correct one
-        # from the commit id in commit
-        for submodule in submodules_list:
-            with chdir(submodule_tree):
-                submodule_git_dir = os.path.join(submodule, ".git")
-                if not os.path.exists(submodule_git_dir):
-                    raise TitoException("%r path does not contain '.git'. "
-                                        "Have you cloned the repository recursively?\n"
-                                        "Try `git submodule update --init --recursive`!" %
-                                        os.path.abspath(submodule))
-                # to find the submodule shars:
-                # git rev-parse <commit>:./<submodule>
-                rev_parse_cmd = "git rev-parse %s:./%s" % (commit, submodule)
-                submodule_commit = run_command(rev_parse_cmd)
-                submodule_tar_file = "%s.%s" % (initial_tar, submodule.replace("/", "_"))
-                # prefix should be <prefix>/<submodule>
-                submodule_prefix = "%s/%s" % (prefix, submodule)
-
-                self.run_git_archive(
-                    relative_git_dir,
-                    submodule_prefix,
-                    submodule_commit,
-                    submodule_tar_file,
-                    submodule,
-                )
-            yield (submodule_tar_file)
-
-            submodule_dir = os.path.join(submodule_tree, submodule)
-            gitmodules_path = os.path.join(submodule_dir, ".gitmodules")
-            if os.path.exists(gitmodules_path):
-                for archive in self._submodule_archives(
-                        relative_git_dir, submodule_prefix, submodule_commit,
-                        submodule_tar_file, submodule_dir
-                ):
-                    yield archive
+    def _submodule_archives(self, relative_git_dir, prefix, initial_tar, source_tree="."):
+        with chdir(source_tree):
+            # Let git handle edge cases for .gitmodules (eg: empty files etc)
+            submodules_status_cmd = "git submodule status --recursive"
+            submodules_status_output = run_command(submodules_status_cmd)
+
+        for line in submodules_status_output.strip().split("\n"):
+            row = line.split()
+            submodule_commit, submodule_relative_dir = row[0], row[1]
+
+            submodule_tar_file_suffix = submodule_relative_dir.replace("/", "_")
+            submodule_tar_file = f"{initial_tar}.{submodule_tar_file_suffix}"
+
+            # prefix should be <prefix>/<submodule>
+            submodule_prefix = f"{prefix}/{submodule_relative_dir}"
+
+            self.run_git_archive(
+                relative_git_dir,
+                submodule_prefix,
+                submodule_commit,
+                submodule_tar_file,
+                submodule_relative_dir,
+            )
+
+            yield submodule_tar_file
 
     def create_tgz(self, git_root, prefix, commit, relative_dir, dest_tgz):
         """
@@ -176,12 +152,12 @@ def create_tgz(self, git_root, prefix, commit, relative_dir, dest_tgz):
         timestamp = get_commit_timestamp(commit)
 
         # Accommodate standalone projects with specfile in root of git repo:
-        relative_git_dir = "%s" % relative_dir
+        relative_git_dir = str(relative_dir)
         if relative_git_dir in ["/", "./"]:
             relative_git_dir = ""
 
         basename = os.path.splitext(dest_tgz)[0]
-        initial_tar = "%s.initial" % basename
+        initial_tar = f"{basename}.initial"
 
         # We need to tar up the following:
         # 1. the current repo
@@ -190,11 +166,11 @@ def create_tgz(self, git_root, prefix, commit, relative_dir, dest_tgz):
         # 2. all of the submodules
         # then combine those into a single archive.
         for submodule_tar_file in self._submodule_archives(
-            relative_git_dir, prefix, commit, initial_tar
+            relative_git_dir, prefix, initial_tar
         ):
-            run_command("tar -Af %s %s" % (initial_tar, submodule_tar_file))
+            run_command(f"tar -Af {initial_tar} {submodule_tar_file}")
 
-        fixed_tar = "%s.tar" % basename
+        fixed_tar = f"{basename}.tar"
         fixed_tar_fh = open(fixed_tar, "wb")
         try:
             tarfixer = TarFixer(
@@ -205,4 +181,4 @@ def create_tgz(self, git_root, prefix, commit, relative_dir, dest_tgz):
             fixed_tar_fh.close()
 
         # It's a pity we can't use Python's gzip, but it doesn't offer an equivalent of -n
-        return run_command("gzip -n -c < %s > %s" % (fixed_tar, dest_tgz))
+        return run_command(f"gzip -n -c < {fixed_tar} > {dest_tgz}")