From a953387d22f56bc62c67b14cbdfca846f1eef69f Mon Sep 17 00:00:00 2001
From: Profpatsch <mail@profpatsch.de>
Date: Mon, 12 Sep 2022 16:25:42 +0200
Subject: [PATCH] tree-sitter/update: move checkTreeSitterRepos into python
 impl

Direct translation of the jq set logic.
---
 .../tools/parsing/tree-sitter/default.nix     |  4 +---
 .../tools/parsing/tree-sitter/update.nix      | 23 ++++--------------
 .../tools/parsing/tree-sitter/update_impl.py  | 24 ++++++++++++++++---
 3 files changed, 27 insertions(+), 24 deletions(-)

diff --git a/pkgs/development/tools/parsing/tree-sitter/default.nix b/pkgs/development/tools/parsing/tree-sitter/default.nix
index fcc81236f4f4..dcc8be0640f4 100644
--- a/pkgs/development/tools/parsing/tree-sitter/default.nix
+++ b/pkgs/development/tools/parsing/tree-sitter/default.nix
@@ -42,9 +42,7 @@ let
     fetchSubmodules = true;
   };
 
-  update-all-grammars = callPackage ./update.nix {
-    inherit src;
-  };
+  update-all-grammars = callPackage ./update.nix {};
 
   fetchGrammar = (v: fetchgit { inherit (v) url rev sha256 fetchSubmodules; });
 
diff --git a/pkgs/development/tools/parsing/tree-sitter/update.nix b/pkgs/development/tools/parsing/tree-sitter/update.nix
index a51208a5bfde..17b13bad8b0a 100644
--- a/pkgs/development/tools/parsing/tree-sitter/update.nix
+++ b/pkgs/development/tools/parsing/tree-sitter/update.nix
@@ -6,9 +6,7 @@
 , lib
 , coreutils
 , curl
-, jq
 , xe
-, src
 }:
 
 # Grammar list:
@@ -389,21 +387,6 @@ let
 
   jsonFile = name: val: (formats.json { }).generate name val;
 
-  # check the tree-sitter orga repos
-  checkTreeSitterRepos = writeShellScript "get-grammars.sh" ''
-    set -euo pipefail
-    res=$(${jq}/bin/jq \
-      --slurpfile known "${knownTreeSitterOrgGrammarReposJson}" \
-      --slurpfile ignore "${ignoredTreeSitterOrgReposJson}" \
-      '. - ($known[0] + $ignore[0])' \
-      )
-    if [ ! "$res" == "[]" ]; then
-      echo "These repositories are neither known nor ignored:" 1>&2
-      echo "$res" 1>&2
-      exit 1
-    fi
-  '';
-
   # implementation of the fetching of repo information from github
   fetchImpl = passArgs "fetchImpl-with-args" {
       binaries = {
@@ -411,6 +394,10 @@ let
         nix-prefetch-git = "${nix-prefetch-git}/bin/nix-prefetch-git";
         inherit atomically-write;
       };
+      inherit
+        knownTreeSitterOrgGrammarRepos
+        ignoredTreeSitterOrgRepos
+        ;
     }
     (writers.writePython3 "fetchImpl" {
         flakeIgnore = ["E501"];
@@ -443,7 +430,7 @@ let
     echo "fetching list of grammars" 1>&2
     treeSitterRepos=$(${fetchImpl} fetch-orga-latest-repos '{"orga": "tree-sitter"}')
     echo "checking the tree-sitter repo list against the grammars we know" 1>&2
-    printf '%s' "$treeSitterRepos" | ${checkTreeSitterRepos}
+    printf '%s' "$treeSitterRepos" | ${fetchImpl} check-tree-sitter-repos '{}'
     echo "writing files to ${outputDir}" 1>&2
     mkdir -p "${outputDir}"
     ${forEachParallel
diff --git a/pkgs/development/tools/parsing/tree-sitter/update_impl.py b/pkgs/development/tools/parsing/tree-sitter/update_impl.py
index 1bff7aac1abf..a65b7bef664a 100644
--- a/pkgs/development/tools/parsing/tree-sitter/update_impl.py
+++ b/pkgs/development/tools/parsing/tree-sitter/update_impl.py
@@ -16,6 +16,10 @@ jsonArg: dict = json.loads(sys.argv[2])
 Args = Iterator[str]
 
 
+def log(msg: str) -> None:
+    print(msg, file=sys.stderr)
+
+
 def curl_github_args(token: str | None, url: str) -> Args:
     """Query the github API via curl"""
     yield bins["curl"]
@@ -58,7 +62,7 @@ def nix_prefetch_git_args(url: str, version_rev: str) -> Args:
 def run_cmd(args: Args) -> bytes:
     all = list(args)
     if debug:
-        print(all, file=sys.stderr)
+        log(str(all))
     return sub.check_output(all)
 
 
@@ -91,14 +95,14 @@ def fetchRepo() -> None:
             match curl_result(out):
                 case "not found":
                     # github sometimes returns an empty list even tough there are releases
-                    print(f"uh-oh, latest for {orga}/{repo} is not there, using HEAD", file=sys.stderr)
+                    log(f"uh-oh, latest for {orga}/{repo} is not there, using HEAD")
                     release = "HEAD"
                 case {"tag_name": tag_name}:
                     release = tag_name
                 case _:
                     sys.exit(f"git result for {orga}/{repo} did not have a `tag_name` field")
 
-            print(f"Fetching latest release ({release}) of {orga}/{repo} …", file=sys.stderr)
+            log(f"Fetching latest release ({release}) of {orga}/{repo} …")
             res = run_cmd(
                 atomically_write_args(
                     os.path.join(
@@ -144,10 +148,24 @@ def fetchOrgaLatestRepos() -> None:
             sys.exit("input json must have `orga` key")
 
 
+def checkTreeSitterRepos() -> None:
+    """Make sure we know about all tree sitter repos on the tree sitter orga."""
+    github_tree_sitter_repos: set[str] = set(json.load(sys.stdin))
+    known: set[str] = set(args["knownTreeSitterOrgGrammarRepos"])
+    ignored: set[str] = set(args["ignoredTreeSitterOrgRepos"])
+
+    unknown = github_tree_sitter_repos - (known | ignored)
+
+    if unknown:
+        sys.exit(f"These repositories are neither known nor ignored:\n{unknown}")
+
+
 match mode:
     case "fetch-repo":
         fetchRepo()
     case "fetch-orga-latest-repos":
         fetchOrgaLatestRepos()
+    case "check-tree-sitter-repos":
+        checkTreeSitterRepos()
     case _:
         sys.exit(f"mode {mode} unknown")