From 8e82992446aec238e65c6c43a949474406203b9f Mon Sep 17 00:00:00 2001
From: "John T. Wodder II" <git@varonathe.org>
Date: Fri, 7 Jun 2024 10:00:01 -0400
Subject: [PATCH] Set `github-access-status` in superdataset's `.gitmodules`

---
 src/backups2datalad/adataset.py   | 69 ++++++++++++++++++++++++-------
 src/backups2datalad/datasetter.py | 19 +++++++++
 2 files changed, 72 insertions(+), 16 deletions(-)

diff --git a/src/backups2datalad/adataset.py b/src/backups2datalad/adataset.py
index 73ab403..8f96ee0 100644
--- a/src/backups2datalad/adataset.py
+++ b/src/backups2datalad/adataset.py
@@ -183,6 +183,25 @@ async def is_dirty(self) -> bool:
             != ""
         )
 
+    async def has_changes(
+        self, paths: Sequence[str | Path] = (), cached: bool = False
+    ) -> bool:
+        args: list[str | Path] = ["diff", "--quiet"]
+        if cached:
+            args.append("--cached")
+        if paths:
+            args.append("--")
+            args.extend(paths)
+        try:
+            await self.call_git(*args, quiet_rcs=[1])
+        except subprocess.CalledProcessError as e:
+            if e.returncode == 1:
+                return True
+            else:
+                raise
+        else:
+            return False
+
     async def get_repo_config(self, key: str, file: str | None = None) -> str | None:
         args = ["--file", file] if file is not None else []
         try:
@@ -193,6 +212,18 @@ async def get_repo_config(self, key: str, file: str | None = None) -> str | None
             else:
                 raise
 
+    async def set_repo_config(
+        self, key: str, value: str, file: str | None = None
+    ) -> None:
+        args = ["--file", file] if file is not None else ["--local"]
+        await self.call_git(
+            "config",
+            *args,
+            "--replace-all",
+            key,
+            value,
+        )
+
     async def get_datalad_id(self) -> str:
         r = await self.get_repo_config("datalad.dataset.id", file=".datalad/config")
         assert r is not None
@@ -206,13 +237,8 @@ async def get_embargo_status(self) -> EmbargoStatus:
             return EmbargoStatus(value)
 
     async def set_embargo_status(self, status: EmbargoStatus) -> None:
-        await self.call_git(
-            "config",
-            "--file",
-            ".datalad/config",
-            "--replace-all",
-            EMBARGO_STATUS_KEY,
-            status.value,
+        await self.set_repo_config(
+            EMBARGO_STATUS_KEY, status.value, file=".datalad/config"
         )
 
     async def call_git(self, *args: str | Path, **kwargs: Any) -> None:
@@ -289,6 +315,22 @@ async def commit(
                 "  Please check if all changes were staged."
             )
 
+    async def commit_if_changed(
+        self,
+        message: str,
+        commit_date: datetime | None = None,
+        paths: Sequence[str | Path] = (),
+        check_dirty: bool = True,
+    ) -> None:
+        await self.call_git("add", "-A", *paths)
+        if await self.has_changes(paths=paths, cached=True):
+            await self.commit(
+                message,
+                commit_date=commit_date,
+                paths=paths,
+                check_dirty=check_dirty,
+            )
+
     async def push(self, to: str, jobs: int, data: str | None = None) -> None:
         waits = exp_wait(attempts=6, base=2.1)
         while True:
@@ -512,7 +554,7 @@ async def create_github_sibling(
                 (f"branch.{DEFAULT_BRANCH}.remote", "github"),
                 (f"branch.{DEFAULT_BRANCH}.merge", f"refs/heads/{DEFAULT_BRANCH}"),
             ]:
-                await self.call_git("config", "--local", "--replace-all", key, value)
+                await self.set_repo_config(key, value)
             return True
         else:
             log.debug("GitHub remote already exists for %s", name)
@@ -658,13 +700,10 @@ async def uninstall_subdatasets(self) -> None:
 
     async def add_submodule(self, path: str, url: str, datalad_id: str) -> None:
         await self.call_git("submodule", "add", "--", url, path)
-        await self.call_git(
-            "config",
-            "--file",
-            ".gitmodules",
-            "--replace-all",
+        await self.set_repo_config(
             f"submodule.{path}.datalad-id",
             datalad_id,
+            file=".gitmodules",
         )
         await self.add(".gitmodules")
 
@@ -684,9 +723,7 @@ async def populate_up_to_date(self) -> bool:
 
     async def update_populate_status(self) -> None:
         head = await self.get_commit_hash()
-        await self.call_git(
-            "config", "--local", "--replace-all", "dandi.populated", head
-        )
+        await self.set_repo_config("dandi.populated", head)
 
 
 class ObjectType(Enum):
diff --git a/src/backups2datalad/datasetter.py b/src/backups2datalad/datasetter.py
index e2386f1..52dfdc3 100644
--- a/src/backups2datalad/datasetter.py
+++ b/src/backups2datalad/datasetter.py
@@ -80,14 +80,33 @@ async def update_from_backup(
             workers=self.config.workers,
         )
         to_save: list[str] = []
+        access_status: dict[str, str] = {}
         for d, changed in report.results:
             if changed:
                 to_save.append(d.identifier)
+                if self.config.gh_org is not None:
+                    access_status[d.identifier] = (
+                        "public"
+                        if d.embargo_status is EmbargoStatus.OPEN
+                        else "private"
+                    )
         if to_save:
             log.debug("Committing superdataset")
             superds.assert_no_duplicates_in_gitmodules()
             msg = await self.get_superds_commit_message(superds, to_save)
             await superds.save(message=msg, path=to_save)
+            if access_status:
+                for did, access in access_status.items():
+                    await superds.set_repo_config(
+                        f"submodule.{did}.github-access-status",
+                        access,
+                        file=".gitmodules",
+                    )
+                await superds.commit_if_changed(
+                    "[backups2datalad] Update github-access-status keys in .gitmodules",
+                    paths=[".gitmodules"],
+                    check_dirty=False,
+                )
             superds.assert_no_duplicates_in_gitmodules()
             log.debug("Superdataset committed")
         if report.failed: