nvim-treesitter/update.py: source from nvim-treesitter

Nurr hasn't always updated at the time nvim-treesitter updates, so it's
better to get the data straight from the source to ensure it is correct.

This uses lua itself to evaluate nvim-treesitter's lockfile and convert
it to a json.

pkgs/applications/editors/vim/plugins/utils/nvim-treesitter/tojson.lua

@@ -0,0 +1,3 @@
+local data = loadfile()()
+local json = require"json".encode(data)
+io.write(json)

pkgs/applications/editors/vim/plugins/utils/nvim-treesitter/update-shell.nix

@@ -10,11 +10,17 @@ let
       requests
     ]
   );
+  luaWithPackages = luajit.withPackages (
+    ps: with ps; [
+      json
+    ]
+  );
 in
 
 mkShell {
   packages = [
     nurl
     pythonWithPackages
+    luaWithPackages
   ];
 }

pkgs/applications/editors/vim/plugins/utils/nvim-treesitter/update.py

@@ -12,8 +12,8 @@ import requests
 
 log = logging.getLogger("vim-updater")
 
-NURR_JSON_URL = (
-    "https://raw.githubusercontent.com/lumen-oss/nurr/main/tree-sitter-parsers.json"
+NVIM_TREESITTER_PARSERS_URL = (
+    "https://raw.githubusercontent.com/nvim-treesitter/nvim-treesitter/refs/heads/main/lua/nvim-treesitter/parsers.lua"
 )
 NVIM_TREESITTER_QUERIES_URL = "https://api.github.com/repos/nvim-treesitter/nvim-treesitter/contents/runtime/queries"
 
@@ -95,9 +95,9 @@ def generate_query(lang: str, parser_info: dict | None, queries_set: set[str]):
     return generated
 
 
-def fetch_nurr_parsers():
-    """Fetch the parser information from nurr repository"""
-    log.info("Fetching parser data from %s", NURR_JSON_URL)
+def fetch_pinned_parsers():
+    """Fetch the parser information from the nvim-treesitter repository"""
+    log.info("Fetching parser data from %s", NVIM_TREESITTER_PARSERS_URL)
 
     headers = {}
     github_token = os.environ.get("GITHUB_TOKEN")
@@ -107,18 +107,18 @@ def fetch_nurr_parsers():
     else:
         log.warning("No GITHUB_TOKEN found. GitHub API requests may be rate-limited.")
 
-    response = requests.get(NURR_JSON_URL, headers=headers, timeout=30)
+    response = requests.get(NVIM_TREESITTER_PARSERS_URL, headers=headers, timeout=30)
     response.raise_for_status()
-    data = response.json()
+    lua_path = Path(__file__).parent / "tojson.lua"
+    lua_output = subprocess.check_output(
+        ["luajit", lua_path],
+        text=True,
+        input=response.text,
+    )
+    data = json.loads(lua_output)
 
-    try:
-        parsers = data["parsers"]
-    except KeyError:
-        raise ValueError(
-            "Unexpected response from NURR:\n" + json.dumps(data, indent=2)
-        )
-    log.info(f"Successfully fetched {len(parsers)} parsers")
-    return parsers
+    log.info(f"Successfully fetched {len(data)} parsers")
+    return data
 
 
 def fetch_available_queries():
@@ -140,21 +140,18 @@ def fetch_available_queries():
     return languages
 
 
-def process_parser_info(parser_info, parsers_map):
+def process_parser_info(parser, parsers):
     """Process a single parser info entry and generate grammar for it"""
-    return generate_grammar(parser_info["lang"], parser_info, parsers_map)
+    return generate_grammar(parser, parsers[parser], parsers)
 
 
 def update_grammars():
-    """Update grammar definitions using nurr's parser information"""
-    parsers_info = fetch_nurr_parsers()
+    """Update grammar definitions using nvim-treesitter's pinned parsers"""
+    parsers = fetch_pinned_parsers()
     queries_list = fetch_available_queries()
 
-    # Create a mapping of lang -> parser_info for quick lookup
-    parsers_map = {p["lang"]: p for p in parsers_info}
-
     generated_file = """# generated by pkgs/applications/editors/vim/plugins/utils/nvim-treesitter/update.py
-# Using parser data from https://github.com/nvim-neorocks/nurr/blob/main/tree-sitter-parsers.json
+# Using parser data from https://github.com/nvim-treesitter/nvim-treesitter/blob/main/lua/nvim-treesitter/parsers.lua
 
 {
   buildGrammar,
@@ -174,7 +171,7 @@ def update_grammars():
     # Process parsers in parallel for better performance
     with ThreadPoolExecutor(max_workers=5) as executor:
         for generated in executor.map(
-            lambda p: process_parser_info(p, parsers_map), parsers_info
+            lambda p: process_parser_info(p, parsers), sorted(parsers.keys())
         ):
             generated_file += generated
 
@@ -188,7 +185,7 @@ def update_grammars():
 
     # Process queries - include parser info if available for requires field
     for lang in queries_list:
-        parser_info = parsers_map.get(lang)
+        parser_info = parsers.get(lang)
         generated_file += generate_query(lang, parser_info, queries_set)
 
     generated_file += "  };\n}\n"