baseline cache

Author: KotlinIsland

.mypy/baseline.json

@@ -83,7 +83,7 @@
         "code": "no-any-explicit",
         "column": 0,
         "message": "Explicit \"Any\" is not allowed",
-        "offset": 14,
+        "offset": 15,
         "src": "def cache_meta_from_dict(meta: dict[str, Any], data_json: str) -> CacheMeta:",
         "target": "mypy.build.cache_meta_from_dict"
       },
@@ -463,6 +463,22 @@
         "src": "meta.get(\"plugin_data\", None),",
         "target": "mypy.build.cache_meta_from_dict"
       },
+      {
+        "code": "no-any-expr",
+        "column": 8,
+        "message": "Expression type contains \"Any\" (has type \"dict[str, Any]\")",
+        "offset": 1,
+        "src": "meta.get(\"baseline_hash\", None),",
+        "target": "mypy.build.cache_meta_from_dict"
+      },
+      {
+        "code": "no-any-expr",
+        "column": 8,
+        "message": "Expression has type \"Any\"",
+        "offset": 0,
+        "src": "meta.get(\"baseline_hash\", None),",
+        "target": "mypy.build.cache_meta_from_dict"
+      },
       {
         "code": "no-any-expr",
         "column": 26,
@@ -1343,11 +1359,19 @@
         "src": "if m.plugin_data != plugin_data:",
         "target": "mypy.build.find_cache_meta"
       },
+      {
+        "code": "no-any-expr",
+        "column": 24,
+        "message": "Expression type contains \"Any\" (has type \"CacheMeta\")",
+        "offset": 17,
+        "src": "if baseline_hash != m.baseline_hash:",
+        "target": "mypy.build.find_cache_meta"
+      },
       {
         "code": "no-any-expr",
         "column": 11,
         "message": "Expression type contains \"Any\" (has type \"CacheMeta\")",
-        "offset": 5,
+        "offset": 7,
         "src": "return m",
         "target": "mypy.build.find_cache_meta"
       },
@@ -1595,15 +1619,15 @@
         "code": "no-any-expr",
         "column": 18,
         "message": "Expression has type \"Any\"",
-        "offset": 55,
+        "offset": 56,
         "src": "plugin_data = manager.plugin.report_config_data(ReportConfigContext(id, path, is_check=False))",
         "target": "mypy.build.write_cache"
       },
       {
         "code": "no-any-expr",
         "column": 11,
-        "message": "Expression type contains \"Any\" (has type \"dict[str, str | int | list[str] | Mapping[str, object] | list[int] | Any]\")",
-        "offset": 51,
+        "message": "Expression type contains \"Any\" (has type \"dict[str, str | int | list[str] | Mapping[str, object] | list[int] | Any | None]\")",
+        "offset": 64,
         "src": "meta = {",
         "target": "mypy.build.write_cache"
       },
@@ -1626,8 +1650,8 @@
       {
         "code": "no-any-expr",
         "column": 26,
-        "message": "Expression type contains \"Any\" (has type \"dict[str, str | int | list[str] | Mapping[str, object] | list[int] | Any]\")",
-        "offset": 4,
+        "message": "Expression type contains \"Any\" (has type \"dict[str, str | int | list[str] | Mapping[str, object] | list[int] | Any | None]\")",
+        "offset": 5,
         "src": "meta_str = json_dumps(meta, manager.options.debug_cache)",
         "target": "mypy.build.write_cache"
       },
@@ -1650,7 +1674,7 @@
       {
         "code": "no-any-expr",
         "column": 48,
-        "message": "Expression type contains \"Any\" (has type \"dict[str, str | int | list[str] | Mapping[str, object] | list[int] | Any]\")",
+        "message": "Expression type contains \"Any\" (has type \"dict[str, str | int | list[str] | Mapping[str, object] | list[int] | Any | None]\")",
         "offset": 0,
         "src": "return interface_hash, cache_meta_from_dict(meta, data_json)",
         "target": "mypy.build.write_cache"
@@ -1875,7 +1899,7 @@
         "code": "no-any-expr",
         "column": 15,
         "message": "Expression type contains \"Any\" (has type \"CacheMeta | None\")",
-        "offset": 79,
+        "offset": 80,
         "src": "if self.meta:",
         "target": "mypy.build.State.generate_unused_ignore_notes"
       },
@@ -2887,7 +2911,7 @@
         "code": "explicit-override",
         "column": 4,
         "message": "Method \"visit_type_alias_expr\" is not using @override but is overriding a method in class \"mypy.visitor.ExpressionVisitor\"",
-        "offset": 46,
+        "offset": 47,
         "src": "def visit_type_alias_expr(self, alias: TypeAliasExpr) -> Type:",
         "target": "mypy.checkexpr.ExpressionChecker.visit_type_alias_expr"
       },
@@ -6811,7 +6835,7 @@
         "code": "truthy-bool",
         "column": 11,
         "message": "Member \"options\" has type \"Options\" which does not implement __bool__ or __len__ so it could always be true in boolean context",
-        "offset": 518,
+        "offset": 516,
         "src": "if self.options:",
         "target": "mypy.errors.Errors.is_error_code_enabled"
       },
@@ -6835,7 +6859,7 @@
         "code": "redundant-expr",
         "column": 19,
         "message": "Condition is always false",
-        "offset": 371,
+        "offset": 375,
         "src": "if e.type is None:",
         "target": "mypy.errors.Errors.render_messages"
       },
@@ -10735,7 +10759,7 @@
         "code": "redundant-expr",
         "column": 18,
         "message": "Condition is always true",
-        "offset": 105,
+        "offset": 104,
         "src": "while True:",
         "target": "mypy.ipc.IPCBase.read"
       }
@@ -11341,7 +11365,7 @@
         "code": "helpful-string",
         "column": 12,
         "message": "The string for \"None\" isn't helpful in a user-facing or semantic string",
-        "offset": 235,
+        "offset": 239,
         "src": "\"Warning: unused section(s) in {}: {}\".format(",
         "target": "mypy.main.run_build"
       },
@@ -36453,7 +36477,7 @@
         "code": "no-any-expr",
         "column": 18,
         "message": "Expression type contains \"Any\" (has type \"list[(str, Any)]\")",
-        "offset": 365,
+        "offset": 364,
         "src": "members = inspect.getmembers(",
         "target": "mypy.util.get_class_descriptors"
       },

CHANGELOG.md

@@ -3,6 +3,8 @@
 ## [Unreleased]
 ### Added
 - `collections.User*` should have `__repr__`
+### Fixed
+- cache modules that only have baseline errors
 
 ## [2.8.1]
 ### Fixes

mypy/build.py

@@ -344,6 +344,7 @@ class CacheMeta(NamedTuple):
     version_id: str  # mypy version for cache invalidation
     ignore_all: bool  # if errors were ignored
     plugin_data: Any  # config data from plugins
+    baseline_hash: str | None  # hash representing the baselined errors of this module
 
 
 # NOTE: dependencies + suppressed == all reachable imports;
@@ -382,6 +383,7 @@ def cache_meta_from_dict(meta: dict[str, Any], data_json: str) -> CacheMeta:
         meta.get("version_id", sentinel),
         meta.get("ignore_all", True),
         meta.get("plugin_data", None),
+        meta.get("baseline_hash", None),
     )
 
 
@@ -1105,7 +1107,7 @@ def save_baseline(manager: BuildManager):
         manager.stdout.write(f"Baseline successfully updated at {file}\n")
 
 
-def load_baseline(options: Options, errors: Errors, stdout: TextIO) -> None:
+def load_baseline(options: Options, errors: Errors, stdout: TextIO):
     if not options.baseline_file:
         return
 
@@ -1485,6 +1487,25 @@ def find_cache_meta(id: str, path: str, manager: BuildManager) -> CacheMeta | No
         manager.log(f"Metadata abandoned for {id}: plugin configuration differs")
         return None
 
+    baseline_hash = None
+    baseline_errors = manager.errors.baseline.get(path)
+    if baseline_errors:
+        errors = [
+            {
+                "code": error["code"],
+                "column": error["column"],
+                "line": error["line"],
+                "message": error["message"],
+            }
+            for error in baseline_errors
+        ]
+        baseline_hash = hash_digest(json.dumps(errors).encode())
+    if baseline_hash != m.baseline_hash:
+        manager.log(f"Metadata abandoned for {id}: baseline is different")
+        return None
+    if baseline_errors:
+        manager.errors.all_errors[path] = "fresh"
+
     manager.add_stats(fresh_metas=1)
     return m
 
@@ -1651,6 +1672,7 @@ def write_cache(
     source_hash: str,
     ignore_all: bool,
     manager: BuildManager,
+    baseline_errors: list[ErrorInfo] | None,
 ) -> tuple[str, CacheMeta | None]:
     """Write cache files for a module.
 
@@ -1736,6 +1758,19 @@ def write_cache(
         manager.log(f"Error in os.stat({data_json!r}), skipping cache write")
         return interface_hash, None
 
+    baseline_hash = None
+    if baseline_errors:
+        errors = [
+            {
+                "code": error.code and error.code.code,
+                "column": error.column,
+                "line": error.line,
+                "message": error.message,
+            }
+            for error in baseline_errors
+        ]
+        baseline_hash = hash_digest(json.dumps(errors).encode())
+
     mtime = 0 if bazel else int(st.st_mtime)
     size = st.st_size
     # Note that the options we store in the cache are the options as
@@ -1761,6 +1796,7 @@ def write_cache(
         "version_id": manager.version_id,
         "ignore_all": ignore_all,
         "plugin_data": plugin_data,
+        "baseline_hash": baseline_hash,
     }
 
     # Write meta cache file
@@ -2609,7 +2645,7 @@ def valid_references(self) -> set[str]:
 
         return valid_refs
 
-    def write_cache(self) -> None:
+    def write_cache(self, *, baseline_errors: list[ErrorInfo] | None) -> None:
         assert self.tree is not None, "Internal error: method must be called on parsed file only"
         # We don't support writing cache files in fine-grained incremental mode.
         if (
@@ -2648,6 +2684,7 @@ def write_cache(self) -> None:
             self.source_hash,
             self.ignore_all,
             self.manager,
+            baseline_errors,
         )
         if new_interface_hash == self.interface_hash:
             self.manager.log(f"Cached module {self.id} has same interface")
@@ -3599,12 +3636,17 @@ def process_stale_scc(graph: Graph, scc: list[str], manager: BuildManager) -> No
         for id in stale:
             graph[id].transitive_error = True
     for id in stale:
-        if graph[id].xpath not in manager.errors.ignored_files:
-            errors = manager.errors.file_messages(
-                graph[id].xpath, formatter=manager.error_formatter
-            )
-            manager.flush_errors(manager.errors.simplify_path(graph[id].xpath), errors, False)
-        graph[id].write_cache()
+        file = graph[id].xpath
+        if file not in manager.errors.ignored_files:
+            errors = manager.errors.file_messages(file, formatter=manager.error_formatter)
+            manager.flush_errors(manager.errors.simplify_path(file), errors, False)
+        all_errors = manager.errors.all_errors.get(file)
+        if all_errors == "fresh":
+            raise ValueError("huh?")
+        if file in manager.errors.ignored_files:
+            all_errors = None
+        # here, the `all_errors` will only be cached if not `transitive_error`, so `all_errors` is baselined errors
+        graph[id].write_cache(baseline_errors=all_errors)
         graph[id].mark_as_rechecked()