Make repo update script fill in the latest TS commit

I want to publish a new version of the Zed plugin, and it points to the
Tree-sitter grammar by commit. This is what got the whole "rcl patch"
thing started!

So now, upon export of the Zed extension, we'll pin it to the HEAD of
the adjacent tree-sitter-rcl repository.

grammar/zed/extension.rcl

@@ -12,11 +12,9 @@ let root = import "//Cargo.rcl";
   grammars = {
     rcl = {
       repository = "https://github.com/rcl-lang/tree-sitter-rcl",
-      // TODO: We should load this from a file, so that `update_repos.py` can
+      // When tools/update_repos.py exports this to the zed-rcl repository, it
-      // write the new commit after it exports the new Tree-sitter grammar.
+      // replaces this commit with the HEAD of the tree-sitter-rcl repository.
-      // But even nicer would be if we had an `rcl patch` command that could
-      // update the value in this file in-place.
       commit = "e371920de7d1921a4ac9c6d78bd74ea3e94dfade",
-    }
+    },
-  }
+  },
 }

tools/update_repos.py

@@ -39,15 +39,24 @@ repositories as a Git submodule here?
     an external repository, we can have both.
 """
 
+import json
 import shutil
 import sys
 import textwrap
+import tomllib
 
 from os import path
-from subprocess import check_output
+from subprocess import check_output, check_call
 from typing import List
 
 
+def get_rcl_version() -> str:
+    # TODO: Would be nice to load Cargo.rcl instead, with the RCL module.
+    with open("Cargo.toml", "rb") as f:
+        version: str = tomllib.load(f)["package"]["version"]
+        return version
+
+
 def ignores_tree_sitter_rcl(dirname: str, _entries: List[str]) -> List[str]:
     if dirname == "grammar/tree-sitter-rcl":
         return [".gitignore", "Cargo.rcl"]
@@ -62,76 +71,98 @@ def ignores_zed(dirname: str, _entries: List[str]) -> List[str]:
         return []
 
 
+def git_head(repo_dir: str) -> str:
+    return check_output(
+        ["git", "-C", repo_dir, "rev-parse", "HEAD"], encoding="ascii"
+    ).strip()
+
+
 def git_commit(repo_dir: str, message: str) -> None:
     print(f"Creating commit in {repo_dir} ...")
     git = ["git", "-C", repo_dir]
     check_output([*git, "add", "."])
-    out = check_output([*git, "commit", "--message", message.strip()])
+    check_call([*git, "commit", "--message", message.strip()])
-    print(out.decode("utf-8"))
+
+
+def get_message(src_dir: str) -> str:
+    rcl_version = get_rcl_version()
+    git_describe = check_output(["git", "describe", "--dirty"]).decode("utf-8").strip()
+    commit_hash = check_output(["git", "rev-parse", "HEAD"]).decode("ascii").strip()
+    return textwrap.dedent(
+        f"""
+        Update to RCL {rcl_version}
+
+        This is a generated export based on files in the {src_dir}
+        directory of the main RCL repository. It is generated by
+        tools/update_repos.py.
+
+        Upstream-Describe: {git_describe}
+        Upstream-Commit: {commit_hash}
+        """
+    )
+
+
+def update_zed_rcl() -> None:
+    assert path.isdir("../zed-rcl/.git")
+
+    src_dir = "grammar/zed"
+    dst_dir = "../zed-rcl"
+
+    # Before we export, update `extension.toml` to point to the correct Git commit.
+    ts_commit = git_head("../tree-sitter-rcl")
+    check_call(
+        [
+            "rcl",
+            "patch",
+            "--in-place",
+            "grammar/zed/extension.rcl",
+            "grammars.rcl.commit",
+            json.dumps(ts_commit),
+        ]
+    )
+    check_call(["rcl", "build"])
+
+    # TODO: It would be nice to skip the index and just generate the right
+    # Git tree. This is possible with `git fast-import`, which is also what
+    # MkDocs uses under the hood. But let's not go there right now.
+    shutil.copytree(
+        src=src_dir,
+        dst=dst_dir,
+        ignore=ignores_zed,
+        dirs_exist_ok=True,
+    )
+    git_commit(dst_dir, get_message(src_dir))
+
+
+def update_tree_sitter_rcl() -> None:
+    assert path.isdir("../tree-sitter-rcl/.git")
+
+    src_dir = "grammar/tree-sitter-rcl"
+    dst_dir = "../tree-sitter-rcl"
+
+    # Rebuild the Tree-sitter grammar to ensure that we don't copy anything
+    # stale.
+    check_output(["tree-sitter", "generate", "--build"], cwd=src_dir)
+
+    shutil.copytree(
+        src=src_dir,
+        dst=dst_dir,
+        ignore=ignores_tree_sitter_rcl,
+        dirs_exist_ok=True,
+    )
+    git_commit(dst_dir, get_message(src_dir))
 
 
 def main() -> None:
-    assert path.isdir("../tree-sitter-rcl/.git")
-    assert path.isdir("../zed-rcl/.git")
-
-    current_desc = check_output(["git", "describe", "--dirty"]).decode("utf-8")
-    current_commit = check_output(["git", "rev-parse", "HEAD"]).decode("ascii")
-
     repos = sys.argv[1:]
     if repos == []:
         repos = ["tree-sitter-rcl", "zed-rcl"]
 
-    if "zed-rcl" in repos:
-        src_dir = "grammar/zed"
-        dst_dir = "../zed-rcl"
-        # TODO: It would be nice to skip the index and just generate the right
-        # Git tree. This is possible with `git fast-import`, which is also what
-        # MkDocs uses under the hood. But let's not go there right now.
-        shutil.copytree(
-            src=src_dir,
-            dst=dst_dir,
-            ignore=ignores_zed,
-            dirs_exist_ok=True,
-        )
-        message = textwrap.dedent(
-            f"""
-            Update to RCL {current_desc}
-
-            This is a generated export based on files in the {src_dir}
-            directory of the main RCL repository. It is generated by
-            tools/update_repos.py.
-
-            Upstream-Commit: {current_commit}
-            """
-        )
-        git_commit(dst_dir, message)
-
     if "tree-sitter-rcl" in repos:
-        src_dir = "grammar/tree-sitter-rcl"
+        update_tree_sitter_rcl()
-        dst_dir = "../tree-sitter-rcl"
 
-        # Rebuild the Tree-sitter grammar to ensure that we don't copy anything
+    if "zed-rcl" in repos:
-        # stale.
+        update_zed_rcl()
-        check_output(["tree-sitter", "generate", "--build"], cwd=src_dir)
-
-        shutil.copytree(
-            src=src_dir,
-            dst=dst_dir,
-            ignore=ignores_tree_sitter_rcl,
-            dirs_exist_ok=True,
-        )
-        message = textwrap.dedent(
-            f"""
-            Update to RCL {current_desc}
-
-            This is a generated export based on files in the {src_dir}
-            directory of the main RCL repository. It is generated by
-            tools/update_repos.py.
-
-            Upstream-Commit: {current_commit}
-            """
-        )
-        git_commit(dst_dir, message)
 
 
 if __name__ == "__main__":