fix[tool]: compile multiple files (#4053)

fix compilation for multiple files where they initialize the same
module. the analysis was getting cached between compilations, leading to
a compiler panic on allocating the storage layout (because the module
was previously touched by the allocator).

this was not caught in previous testing because the pattern in the test
suite is to run a single compilation per test, with a fresh input
bundle.

tests/unit/cli/storage_layout/test_storage_layout.py

@@ -1,4 +1,5 @@
 from vyper.compiler import compile_code
+from vyper.evm.opcodes import version_check
 
 from .utils import adjust_storage_layout_for_cancun
 
@@ -342,3 +343,40 @@ def foo() -> uint256:
 
     out = compile_code(code, input_bundle=input_bundle, output_formats=["layout"])
     assert out["layout"] == expected_layout
+
+
+def test_multiple_compile_codes(make_input_bundle):
+    # test calling compile_code multiple times with the same library allocated
+    # in different locations
+    lib = """
+x: uint256
+    """
+    input_bundle = make_input_bundle({"lib.vy": lib})
+
+    main1 = """
+import lib
+
+initializes: lib
+t: uint256
+    """
+    main2 = """
+import lib
+
+t: uint256
+initializes: lib
+    """
+    out1 = compile_code(main1, input_bundle=input_bundle, output_formats=["layout"])["layout"]
+    out2 = compile_code(main2, input_bundle=input_bundle, output_formats=["layout"])["layout"]
+
+    layout1 = out1["storage_layout"]["lib"]
+    layout2 = out2["storage_layout"]["lib"]
+
+    assert layout1 != layout2  # for clarity
+
+    if version_check(begin="cancun"):
+        start_slot = 0
+    else:
+        start_slot = 1
+
+    assert layout1 == {"x": {"slot": start_slot, "type": "uint256", "n_slots": 1}}
+    assert layout2 == {"x": {"slot": start_slot + 1, "type": "uint256", "n_slots": 1}}

vyper/compiler/__init__.py

@@ -103,6 +103,7 @@ def compile_from_file_input(
         output_formats = ("bytecode",)
 
     # make IR output the same between runs
+    # TODO: move this to CompilerData.__init__()
     codegen.reset_names()
 
     compiler_data = CompilerData(

vyper/compiler/phases.py

@@ -94,6 +94,9 @@ def __init__(
         self.input_bundle = input_bundle or FilesystemInputBundle([Path(".")])
         self.expected_integrity_sum = integrity_sum
 
+        # ast cache, hitchhike onto the input_bundle object
+        self.input_bundle._cache._ast_of: dict[int, vy_ast.Module] = {}  # type: ignore
+
     @cached_property
     def source_code(self):
         return self.file_input.source_code

vyper/semantics/analysis/module.py

@@ -199,10 +199,6 @@ def __init__(
 
         self.module_t: Optional[ModuleT] = None
 
-        # ast cache, hitchhike onto the input_bundle object
-        if not hasattr(self.input_bundle._cache, "_ast_of"):
-            self.input_bundle._cache._ast_of: dict[int, vy_ast.Module] = {}  # type: ignore
-
     def analyze_module_body(self):
         # generate a `ModuleT` from the top-level node
         # note: also validates unique method ids