Add basic support for is_generated

Summary: Right now ig manually filters this out by manually filtering on file names, they have requested that generated code not be included in roll ups and other tools in the dashboard. This is the first step in getting that data into the pipelines and then into the tool itself. We should also be able to index on it, which should make filtering this data faster and simpler from both a code and user perspective.

Reviewed By: inseokhwang

Differential Revision: D52545219

fbshipit-source-id: cecd68dd48cd79107db81648bf6256dcd0322f2d

client/coverage_data.py

@@ -91,6 +91,7 @@ class ModuleMode(str, Enum):
 class ModuleModeInfo(json_mixins.SnakeCaseAndExcludeJsonMixin):
     mode: ModuleMode
     explicit_comment_line: Optional[LineNumber]
+    is_generated: bool
 
 
 class FunctionAnnotationStatus(str, Enum):
@@ -461,6 +462,7 @@ class ModuleModeCollector(VisitorWithPositionData):
     ignore_all_by_code_regex: Pattern[str] = compile(
         r" ?#+ *pyre-ignore-all-errors\[[0-9]+[0-9, ]*\]"
     )
+    is_generated_regex: Pattern[str] = compile(r"@" + "generated")
 
     def __init__(self, strict_by_default: bool) -> None:
         self.strict_by_default: bool = strict_by_default
@@ -471,6 +473,7 @@ def __init__(self, strict_by_default: bool) -> None:
             ModuleMode.STRICT if strict_by_default else ModuleMode.UNSAFE
         )
         self.explicit_comment_line: Optional[int] = None
+        self.is_generated: bool = False
 
     def is_strict_module(self) -> bool:
         return self.mode == ModuleMode.STRICT
@@ -488,6 +491,9 @@ def visit_Comment(self, node: libcst.Comment) -> None:
             self.mode = ModuleMode.IGNORE_ALL
             self.explicit_comment_line = self.location(node).start_line
 
+        if self.is_generated_regex.search(node.value):
+            self.is_generated = True
+
 
 def collect_mode(
     module: libcst.MetadataWrapper,
@@ -500,6 +506,7 @@ def collect_mode(
     return ModuleModeInfo(
         mode=mode,
         explicit_comment_line=visitor.explicit_comment_line,
+        is_generated=visitor.is_generated,
     )
 
 

client/tests/coverage_data_tests.py

@@ -1051,13 +1051,16 @@ def assert_counts(
         default_strict: bool,
         mode: ModuleMode,
         explicit_comment_line: Optional[int],
+        is_generated: bool = False,
     ) -> None:
         source_module = parse_code(source)
         result = coverage_data.collect_mode(source_module, default_strict)
         self.assertEqual(mode, result.mode)
         self.assertEqual(explicit_comment_line, result.explicit_comment_line)
+        self.assertEqual(is_generated, result.is_generated)
 
     def test_strict_files(self) -> None:
+        generated_string = "generated"
         self.assert_counts(
             """
             # pyre-unsafe
@@ -1136,6 +1139,18 @@ def foo(x: str) -> int:
             mode=ModuleMode.STRICT,
             explicit_comment_line=None,
         )
+        self.assert_counts(
+            f"""
+            # @{generated_string}
+
+            def foo(x: str) -> int:
+                return x
+            """,
+            default_strict=True,
+            mode=ModuleMode.STRICT,
+            explicit_comment_line=None,
+            is_generated=True,
+        )
 
 
 class ModuleFindingHelpersTest(testslide.TestCase):