[primer] Separate Comparator data loading from comparison logic

Constructor now takes pre-loaded PackageMessages dicts, making it
directly testable with in-memory data.  I/O and batch merging move to
a from_json() factory method; _load_json() becomes module-level.

Author: Pierre-Sassoulas

pylint/testutils/_primer/comparator.py

@@ -16,32 +16,15 @@
 
 
 class Comparator:
-    def __init__(
-        self, base_file: str, new_file: str, batches: int | None = None
-    ) -> None:
-        self._base_file = base_file
-        self._new_file = new_file
-        self._batches = batches
+    """Cross-reference two primer JSON outputs and iterate over differences."""
+
+    def __init__(self, main_data: PackageMessages, pr_data: PackageMessages) -> None:
+        self._main_data = main_data
+        self._pr_data = pr_data
 
     def __iter__(self) -> Generator[tuple[str, PackageData, PackageData]]:
-        main_data: PackageMessages
-        if self._batches is None:
-            main_data = self._load_json(self._base_file)
-            pr_data = self._load_json(self._new_file)
-        else:
-            main_data = {}
-            pr_data = {}
-            for idx in range(self._batches):
-                main_data.update(
-                    self._load_json(
-                        self._base_file.replace("BATCHIDX", "batch" + str(idx))
-                    )
-                )
-                pr_data.update(
-                    self._load_json(
-                        self._new_file.replace("BATCHIDX", "batch" + str(idx))
-                    )
-                )
+        main_data = self._main_data
+        pr_data = self._pr_data
 
         missing_messages: PackageMessages = {}
         for package, data in main_data.items():
@@ -63,7 +46,26 @@ def __iter__(self) -> Generator[tuple[str, PackageData, PackageData]]:
             yield package, pkg_missing, new_messages
 
     @staticmethod
-    def _load_json(file_path: Path | str) -> PackageMessages:
-        with open(file_path, encoding="utf-8") as f:
-            result: PackageMessages = json.load(f)
-        return result
+    def from_json(
+        base_file: Path | str, new_file: Path | str, batches: int | None = None
+    ) -> Comparator:
+        """Build a Comparator from JSON file paths, handling batched runs."""
+        main_data: PackageMessages
+        pr_data: PackageMessages
+        if batches is None:
+            main_data = _load_json(base_file)
+            pr_data = _load_json(new_file)
+        else:
+            main_data = {}
+            pr_data = {}
+            for idx in range(batches):
+                suffix = f"batch{idx}"
+                main_data.update(_load_json(str(base_file).replace("BATCHIDX", suffix)))
+                pr_data.update(_load_json(str(new_file).replace("BATCHIDX", suffix)))
+        return Comparator(main_data, pr_data)
+
+
+def _load_json(file_path: Path | str) -> PackageMessages:
+    with open(file_path, encoding="utf-8") as f:
+        result: PackageMessages = json.load(f)
+    return result

pylint/testutils/_primer/primer_compare_command.py

@@ -13,7 +13,7 @@
 
 class CompareCommand(PrimerCommand):
     def run(self) -> None:
-        comparator = Comparator(
+        comparator = Comparator.from_json(
             self.config.base_file, self.config.new_file, self.config.batches
         )
         comment = self._create_comment(comparator)

tests/testutils/_primer/test_comparator.py

@@ -18,7 +18,7 @@
 )
 def test_comparator(directory: Path) -> None:
     """Test Comparator with each fixture directory."""
-    comparator = Comparator(str(directory / "main.json"), str(directory / "pr.json"))
+    comparator = Comparator.from_json(directory / "main.json", directory / "pr.json")
     expected = json.loads((directory / "expected_comparator.json").read_text("utf-8"))
     results = list(comparator)
     assert len(results) == len(expected)
@@ -30,9 +30,9 @@ def test_comparator(directory: Path) -> None:
 
 def test_comparator_batched() -> None:
     fixture = Path(__file__).parent / "batched_cases"
-    comparator = Comparator(
-        str(fixture / "main_BATCHIDX.json"),
-        str(fixture / "pr_BATCHIDX.json"),
+    comparator = Comparator.from_json(
+        fixture / "main_BATCHIDX.json",
+        fixture / "pr_BATCHIDX.json",
         batches=2,
     )
     expected = json.loads((fixture / "expected_comparator.json").read_text("utf-8"))