[inferpython] generate qualified Textual procdesc

Summary: instead of generating Textual toplevel procedures, we use the filepath+module name to qualify them. Now we can analyze multiple files in a package, even if they have same names.

Reviewed By: skcho

Differential Revision:
D65819620

Privacy Context Container: L1208441

fbshipit-source-id: 962d86e922a456be078944ecf89a62c920983232

infer/src/pulse/PulseModelsPython.ml

@@ -135,11 +135,13 @@ let import_name name _fromlist _level : model =
   start_model
   @@ fun () ->
   let* opt_str = as_constant_string name in
-  let function_name =
+  let module_name =
     Option.value_or_thunk opt_str ~default:(fun () ->
         L.die InternalError "frontend should always give a string here" )
   in
-  let proc_name = Procname.make_python ~class_name:None ~function_name in
+  let class_name = PythonClassName.make module_name in
+  let function_name = "__module_body__" in
+  let proc_name = Procname.make_python ~class_name:(Some class_name) ~function_name in
   let* globals = Dict.make [] [] in
   (* TODO: call it only once! *)
   let* _ = python_call proc_name [("globals", globals)] in

infer/src/python/PyIR.mli

@@ -63,7 +63,7 @@ module ScopedIdent : sig
 end
 
 module QualName : sig
-  type t
+  type t = {module_name: Ident.t; function_name: Ident.t}
 
   val pp : Format.formatter -> t -> unit
 

infer/src/python/PyIR2Textual.ml

@@ -31,22 +31,23 @@ module Typ = struct
   let value = Textual.(Typ.Ptr (Typ.Struct (TypeName.of_string "PyObject")))
 end
 
-let str_module_body name = F.asprintf "%a.__module_body__" Ident.pp name
+let str_module_body = "__module_body__"
 
 type proc_kind = ModuleBody of Ident.t | RegularFunction of QualName.t
 
 let is_module_body = function ModuleBody _ -> true | _ -> false
 
-let mk_qualified_proc_name ?loc kind =
-  let qual_name_str =
-    match kind with
-    | ModuleBody name ->
-        str_module_body name
-    | RegularFunction qual_name ->
-        F.asprintf "%a" QualName.pp qual_name
-  in
-  { Textual.QualifiedProcName.enclosing_class= TopLevel
-  ; name= Textual.ProcName.of_string ?loc qual_name_str }
+let typename_of_ident ?loc ident = Textual.TypeName.of_string ?loc (F.asprintf "%a" Ident.pp ident)
+
+let mk_qualified_proc_name ?loc kind : Textual.QualifiedProcName.t =
+  let procname_of_ident ident = Textual.ProcName.of_string ?loc (F.asprintf "%a" Ident.pp ident) in
+  match kind with
+  | ModuleBody name ->
+      { enclosing_class= Enclosing (typename_of_ident name)
+      ; name= Textual.ProcName.of_string ?loc str_module_body }
+  | RegularFunction {module_name; function_name} ->
+      { enclosing_class= Enclosing (typename_of_ident ?loc module_name)
+      ; name= procname_of_ident function_name }
 
 
 let mk_procdecl_attributes {CodeInfo.co_argcount; co_varnames} =
@@ -128,9 +129,7 @@ let rec of_exp exp : Textual.Exp.t =
   | LoadClassDeref {name; slot= _} ->
       call_builtin "py_load_class_deref" [exp_of_ident_str name] (* TODO: more arg needed *)
   | ImportName {name; fromlist; level} ->
-      let str =
-        str_module_body name |> Textual.ProcName.of_string |> F.asprintf "%a" Textual.ProcName.pp
-      in
+      let str = typename_of_ident name |> F.asprintf "%a" Textual.TypeName.pp in
       call_builtin "py_import_name" [Textual.Exp.Const (Str str); of_exp fromlist; of_exp level]
   | ImportFrom {name; exp} ->
       call_builtin "py_import_from" [exp_of_ident_str name; of_exp exp]

infer/src/python/unit/PyIR2TextualTest.ml

@@ -59,20 +59,20 @@ def g():
     TRANSFORMATION PyIR -> Textual
     .source_language = "python"
 
-    define dummy::__module_body__(globals: *PyGlobals) : *PyObject {
+    define dummy.__module_body__(globals: *PyGlobals) : *PyObject {
       local locals: *PyLocals
       #b0:
           store &locals <- globals
           _ = $builtins.py_store_name("x", locals, globals, $builtins.py_make_int(0))
-          n0 = $builtins.py_make_function(fun (globals, locals) -> dummy::f(globals, locals), $builtins.py_make_none(), $builtins.py_make_none(), $builtins.py_make_none(), $builtins.py_make_none())
+          n0 = $builtins.py_make_function(fun (globals, locals) -> dummy.f(globals, locals), $builtins.py_make_none(), $builtins.py_make_none(), $builtins.py_make_none(), $builtins.py_make_none())
           _ = $builtins.py_store_name("f", locals, globals, n0)
-          n1 = $builtins.py_make_function(fun (globals, locals) -> dummy::g(globals, locals), $builtins.py_make_none(), $builtins.py_make_none(), $builtins.py_make_none(), $builtins.py_make_none())
+          n1 = $builtins.py_make_function(fun (globals, locals) -> dummy.g(globals, locals), $builtins.py_make_none(), $builtins.py_make_none(), $builtins.py_make_none(), $builtins.py_make_none())
           _ = $builtins.py_store_name("g", locals, globals, n1)
           ret $builtins.py_make_none()
 
     }
 
-    define .args = "y,l" dummy::f(globals: *PyGlobals, locals: *PyLocals) : *PyObject {
+    define .args = "y,l" dummy.f(globals: *PyGlobals, locals: *PyLocals) : *PyObject {
       #b0:
           n0 = $builtins.py_load_fast("y", locals)
           if n0 then jmp b1 else jmp b2
@@ -107,7 +107,7 @@ def g():
 
     }
 
-    define dummy::g(globals: *PyGlobals, locals: *PyLocals) : *PyObject {
+    define dummy.g(globals: *PyGlobals, locals: *PyLocals) : *PyObject {
       #b0:
           n0 = $builtins.py_load_global("print", globals)
           n1 = $builtins.py_load_global("x", globals)
@@ -119,20 +119,20 @@ def g():
     TYPE INFERENCE
     .source_language = "python"
 
-    define dummy::__module_body__(globals: *PyGlobals) : *PyObject {
+    define dummy.__module_body__(globals: *PyGlobals) : *PyObject {
       local locals: *PyLocals
       #b0:
           store &locals <- [&globals:*PyGlobals]:*PyGlobals
           _ = $builtins.py_store_name("x", [&locals:*PyLocals], [&globals:*PyGlobals], $builtins.py_make_int(0))
-          n0 = $builtins.py_make_function(fun (globals, locals) -> dummy::f(globals, locals), $builtins.py_make_none(), $builtins.py_make_none(), $builtins.py_make_none(), $builtins.py_make_none())
+          n0 = $builtins.py_make_function(fun (globals, locals) -> dummy.f(globals, locals), $builtins.py_make_none(), $builtins.py_make_none(), $builtins.py_make_none(), $builtins.py_make_none())
           _ = $builtins.py_store_name("f", [&locals:*PyLocals], [&globals:*PyGlobals], n0)
-          n1 = $builtins.py_make_function(fun (globals, locals) -> dummy::g(globals, locals), $builtins.py_make_none(), $builtins.py_make_none(), $builtins.py_make_none(), $builtins.py_make_none())
+          n1 = $builtins.py_make_function(fun (globals, locals) -> dummy.g(globals, locals), $builtins.py_make_none(), $builtins.py_make_none(), $builtins.py_make_none(), $builtins.py_make_none())
           _ = $builtins.py_store_name("g", [&locals:*PyLocals], [&globals:*PyGlobals], n1)
           ret $builtins.py_make_none()
 
     }
 
-    define .args = "y,l" dummy::f(globals: *PyGlobals, locals: *PyLocals) : *PyObject {
+    define .args = "y,l" dummy.f(globals: *PyGlobals, locals: *PyLocals) : *PyObject {
       #b0:
           n0 = $builtins.py_load_fast("y", [&locals:*PyLocals])
           if n0 then jmp b1 else jmp b2
@@ -167,7 +167,7 @@ def g():
 
     }
 
-    define dummy::g(globals: *PyGlobals, locals: *PyLocals) : *PyObject {
+    define dummy.g(globals: *PyGlobals, locals: *PyLocals) : *PyObject {
       #b0:
           n0 = $builtins.py_load_global("print", [&globals:*PyGlobals])
           n1 = $builtins.py_load_global("x", [&globals:*PyGlobals])
@@ -185,7 +185,7 @@ def g():
       #entry:
           n0:*PyGlobals = load &globals
           n1:*PyLocals = load &locals
-          n2 = dummy::g(n0, n1)
+          n2 = dummy.g(n0, n1)
           ret n2
 
     }
@@ -196,12 +196,12 @@ def g():
       #entry:
           n0:*PyGlobals = load &globals
           n1:*PyLocals = load &locals
-          n2 = dummy::f(n0, n1)
+          n2 = dummy.f(n0, n1)
           ret n2
 
     }
 
-    define dummy::__module_body__(globals: *PyGlobals) : *PyObject {
+    define dummy.__module_body__(globals: *PyGlobals) : *PyObject {
       local locals: *PyLocals
       #b0:
           n2:*PyGlobals = load &globals
@@ -233,7 +233,7 @@ def g():
 
     }
 
-    define .args = "y,l" dummy::f(globals: *PyGlobals, locals: *PyLocals) : *PyObject {
+    define .args = "y,l" dummy.f(globals: *PyGlobals, locals: *PyLocals) : *PyObject {
       #b0:
           n13:*PyLocals = load &locals
           n0 = $builtins.py_load_fast("y", n13)
@@ -292,7 +292,7 @@ def g():
 
     }
 
-    define dummy::g(globals: *PyGlobals, locals: *PyLocals) : *PyObject {
+    define dummy.g(globals: *PyGlobals, locals: *PyLocals) : *PyObject {
       #b0:
           n3:*PyGlobals = load &globals
           n0 = $builtins.py_load_global("print", n3)

infer/tests/codetoanalyze/python/pulse/.inferconfig

@@ -4,10 +4,10 @@
   "pulse-specialization-partial": true,
   "pulse-taint-short-traces": true,
   "pulse-taint-sources": [
-    { "procedure": "taint::source" }
+    { "procedure": "taint.source" }
   ],
   "pulse-taint-sinks": [
-    { "procedure": "taint::sink",
+    { "procedure": "taint.sink",
       "taint_target": ["ArgumentPositions", [1]]
     }
   ]

infer/tests/codetoanalyze/python/pulse/Makefile

@@ -9,7 +9,7 @@ INFERPRINT_OPTIONS = --issues-tests
 
 include $(TESTS_DIR)/infer.make
 
-SOURCES = $(sort $(wildcard *.py))
+SOURCES = $(sort $(wildcard *.py) $(wildcard */*.py) $(wildcard */*/*.py))
 
 default: test
 

infer/tests/codetoanalyze/python/pulse/dir1/dir3/testmod.py

@@ -0,0 +1,10 @@
+# Copyright (c) Facebook, Inc. and its affiliates.
+#
+# This source code is licensed under the MIT license found in the
+# LICENSE file in the root directory of this source tree.
+
+import random
+import taint
+
+tainted_global3 = taint.source()
+untainted_global3 = random.random()

infer/tests/codetoanalyze/python/pulse/dir1/dir4/testmod.py

@@ -0,0 +1,10 @@
+# Copyright (c) Facebook, Inc. and its affiliates.
+#
+# This source code is licensed under the MIT license found in the
+# LICENSE file in the root directory of this source tree.
+
+import random
+import taint
+
+tainted_global4 = taint.source()
+untainted_global4 = random.random()

infer/tests/codetoanalyze/python/pulse/dir1/testmod.py

@@ -0,0 +1,10 @@
+# Copyright (c) Facebook, Inc. and its affiliates.
+#
+# This source code is licensed under the MIT license found in the
+# LICENSE file in the root directory of this source tree.
+
+import random
+import taint
+
+tainted_global1 = taint.source()
+untainted_global1 = random.random()

infer/tests/codetoanalyze/python/pulse/dir2/dir5/testmod.py

@@ -0,0 +1,10 @@
+# Copyright (c) Facebook, Inc. and its affiliates.
+#
+# This source code is licensed under the MIT license found in the
+# LICENSE file in the root directory of this source tree.
+
+import random
+import taint
+
+tainted_global5 = taint.source()
+untainted_global5 = random.random()

infer/tests/codetoanalyze/python/pulse/dir2/dir6/testmod.py

@@ -0,0 +1,10 @@
+# Copyright (c) Facebook, Inc. and its affiliates.
+#
+# This source code is licensed under the MIT license found in the
+# LICENSE file in the root directory of this source tree.
+
+import random
+import taint
+
+tainted_global6 = taint.source()
+untainted_global6 = random.random()

infer/tests/codetoanalyze/python/pulse/dir2/testmod.py

@@ -0,0 +1,10 @@
+# Copyright (c) Facebook, Inc. and its affiliates.
+#
+# This source code is licensed under the MIT license found in the
+# LICENSE file in the root directory of this source tree.
+
+import random
+import taint
+
+tainted_global2 = taint.source()
+untainted_global2 = random.random()