Move reading and diffing from __init__ to __iter__

Avoid side effects in __init__ by deferring JSON loading and
message diffing to iteration time.

Author: Pierre-Sassoulas

pylint/testutils/_primer/comparator.py

@@ -19,42 +19,48 @@ 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
+
+    def __iter__(self) -> Generator[tuple[str, PackageData, PackageData]]:
         main_data: PackageMessages
-        if batches is None:
-            main_data = self._load_json(base_file)
-            self.pr_data = self._load_json(new_file)
+        if self._batches is None:
+            main_data = self._load_json(self._base_file)
+            pr_data = self._load_json(self._new_file)
         else:
             main_data = {}
-            self.pr_data = {}
-            for idx in range(batches):
+            pr_data = {}
+            for idx in range(self._batches):
                 main_data.update(
-                    self._load_json(base_file.replace("BATCHIDX", "batch" + str(idx)))
+                    self._load_json(
+                        self._base_file.replace("BATCHIDX", "batch" + str(idx))
+                    )
                 )
-                self.pr_data.update(
-                    self._load_json(new_file.replace("BATCHIDX", "batch" + str(idx)))
+                pr_data.update(
+                    self._load_json(
+                        self._new_file.replace("BATCHIDX", "batch" + str(idx))
+                    )
                 )
 
-        self.missing_messages: PackageMessages = {}
+        missing_messages: PackageMessages = {}
         for package, data in main_data.items():
             package_missing_messages: list[OldJsonExport] = []
             for message in data["messages"]:
                 try:
-                    self.pr_data[package]["messages"].remove(message)
+                    pr_data[package]["messages"].remove(message)
                 except ValueError:
                     package_missing_messages.append(message)
-            self.missing_messages[package] = PackageData(
-                commit=self.pr_data[package]["commit"],
+            missing_messages[package] = PackageData(
+                commit=pr_data[package]["commit"],
                 messages=package_missing_messages,
             )
 
-    def __iter__(
-        self,
-    ) -> Generator[tuple[str, PackageData, PackageData]]:
-        for package, missing_messages in self.missing_messages.items():
-            new_messages = self.pr_data[package]
-            if not missing_messages["messages"] and not new_messages["messages"]:
+        for package, pkg_missing in missing_messages.items():
+            new_messages = pr_data[package]
+            if not pkg_missing["messages"] and not new_messages["messages"]:
                 continue
-            yield package, missing_messages, new_messages
+            yield package, pkg_missing, new_messages
 
     @staticmethod
     def _load_json(file_path: Path | str) -> PackageMessages: