[mypyc] Don't simplify module prefixes if using separate compilation (#19206)

Mypyc shortens module prefixes to generate nicer, shorter C names.
However, this assumes that we know all possible module prefixes. When
doing separate compilation, we only have access to a subset of possible
module prefixes, so there's no good way to shorten module prefixes while
keeping names unique.

Author: JukkaL

mypyc/codegen/emitmodule.py

@@ -306,7 +306,10 @@ def compile_ir_to_c(
         for source in sources
     }
 
-    names = NameGenerator([[source.module for source in sources] for sources, _ in groups])
+    names = NameGenerator(
+        [[source.module for source in sources] for sources, _ in groups],
+        separate=compiler_options.separate,
+    )
 
     # Generate C code for each compilation group. Each group will be
     # compiled into a separate extension module.

mypyc/namegen.py

@@ -34,20 +34,30 @@ class NameGenerator:
 
     The generated should be internal to a build and thus the mapping is
     arbitrary. Just generating names '1', '2', ... would be correct,
-    though not very usable.
+    though not very usable. The generated names may be visible in CPU
+    profiles and when debugging using native debuggers.
     """
 
-    def __init__(self, groups: Iterable[list[str]]) -> None:
+    def __init__(self, groups: Iterable[list[str]], *, separate: bool = False) -> None:
         """Initialize with a list of modules in each compilation group.
 
         The names of modules are used to shorten names referring to
         modules, for convenience. Arbitrary module
         names are supported for generated names, but uncompiled modules
         will use long names.
+
+        If separate is True, assume separate compilation. This implies
+        that we don't have knowledge of all sources that will be linked
+        together. In this case we won't trim module prefixes, since we
+        don't have enough information to determine common module prefixes.
         """
         self.module_map: dict[str, str] = {}
         for names in groups:
-            self.module_map.update(make_module_translation_map(names))
+            if not separate:
+                self.module_map.update(make_module_translation_map(names))
+            else:
+                for name in names:
+                    self.module_map[name] = name + "."
         self.translations: dict[tuple[str, str], str] = {}
         self.used_names: set[str] = set()
 

mypyc/test/test_namegen.py

@@ -52,3 +52,17 @@ def test_name_generator(self) -> None:
         assert g.private_name("foo", "C_x_y") == "foo___C_x_y"
         assert g.private_name("foo", "C_x_y") == "foo___C_x_y"
         assert g.private_name("foo", "___") == "foo______3_"
+
+        g = NameGenerator([["foo.zar"]])
+        assert g.private_name("foo.zar", "f") == "f"
+
+    def test_name_generator_with_separate(self) -> None:
+        g = NameGenerator([["foo", "foo.zar"]], separate=True)
+        assert g.private_name("foo", "f") == "foo___f"
+        assert g.private_name("foo", "C.x.y") == "foo___C___x___y"
+        assert g.private_name("foo.zar", "C.x.y") == "foo___zar___C___x___y"
+        assert g.private_name("foo", "C.x_y") == "foo___C___x_y"
+        assert g.private_name("foo", "___") == "foo______3_"
+
+        g = NameGenerator([["foo.zar"]], separate=True)
+        assert g.private_name("foo.zar", "f") == "foo___zar___f"