[infer/py] adding support for BUILD_TUPLE and BUILD_SET

Summary: BUILD_TUPLE and BUILD_SET are virtually clones of the already implemented BUILD_LIST

Reviewed By: ngorogiannis

Differential Revision: D49369159

fbshipit-source-id: feb0ba1f24fe266cfbc79680e90c52f962caeee7

infer/src/python/PyBuiltin.ml

@@ -103,6 +103,8 @@ module Builtin = struct
     | PythonIter
     | PythonIterNext
     | PythonBuildList
+    | PythonBuildSet
+    | PythonBuildTuple
     | PythonSubscriptGet
     | PythonSubscriptSet
     | CompareOp of Compare.t
@@ -158,6 +160,10 @@ let to_proc_name = function
             "python_iter_next"
         | PythonBuildList ->
             "python_build_list"
+        | PythonBuildSet ->
+            "python_build_set"
+        | PythonBuildTuple ->
+            "python_build_tuple"
         | PythonSubscriptGet ->
             "python_subscript_get"
         | PythonSubscriptSet ->
@@ -298,6 +304,10 @@ module Set = struct
           ; used_struct_types= [PyCommon.pyIterItemStruct] } )
       ; ( Builtin.PythonBuildList
         , {formals_types= None; result_type= annot PyCommon.pyList; used_struct_types= []} )
+      ; ( Builtin.PythonBuildSet
+        , {formals_types= None; result_type= annot PyCommon.pySet; used_struct_types= []} )
+      ; ( Builtin.PythonBuildTuple
+        , {formals_types= None; result_type= annot PyCommon.pyTuple; used_struct_types= []} )
       ; ( Builtin.PythonSubscriptGet
         , { formals_types= Some [annotatedObject; annotatedObject]
           ; result_type= annotatedObject

infer/src/python/PyBuiltin.mli

@@ -40,6 +40,8 @@ type textual =
   | PythonIter
   | PythonIterNext
   | PythonBuildList
+  | PythonBuildSet
+  | PythonBuildTuple
   | PythonSubscriptGet
   | PythonSubscriptSet
   | CompareOp of Compare.t

infer/src/python/PyCommon.ml

@@ -46,6 +46,10 @@ let pyClass = mk_type "PyClass"
 
 let pyList = mk_type "PyList"
 
+let pySet = mk_type "PySet"
+
+let pyTuple = mk_type "PyTuple"
+
 let py_iter_item = type_name "PyIterItem"
 
 let pyIterItem = T.Typ.Ptr (Struct py_iter_item)

infer/src/python/PyCommon.mli

@@ -86,6 +86,12 @@ val pyNone : Textual.Typ.t
 val pyList : Textual.Typ.t
 (** Python's builtin [list] type *)
 
+val pySet : Textual.Typ.t
+(** Python's builtin [set] type *)
+
+val pyTuple : Textual.Typ.t
+(** Python's builtin [tuple] type *)
+
 val mk_int : int64 -> Textual.Exp.t
 (** Helper function to define typed Textual expression for literal integers. *)
 

infer/src/python/PyTrans.ml

@@ -108,7 +108,7 @@ let rec py_to_exp env c =
             Ok (env, e) )
       in
       let exp = T.Exp.call_non_virtual PyCommon.python_tuple args in
-      Ok (env, exp, PyCommon.pyObject)
+      Ok (env, exp, PyCommon.pyTuple)
 
 
 let annotated_type_of_annotation typ =
@@ -1201,7 +1201,7 @@ module BUILD = struct
   module LIST = struct
     (** {v BUILD_LIST(count) v}
 
-        Creates a list consuming count items from the stack, and pushes the resulting tuple onto the
+        Creates a list consuming count items from the stack, and pushes the resulting list onto the
         stack. *)
     let run env code {FFI.Instruction.opname; arg= count} =
       let open IResult.Let_syntax in
@@ -1212,6 +1212,36 @@ module BUILD = struct
       let env = Env.push env (DataStack.Temp id) in
       Ok (env, None)
   end
+
+  module SET = struct
+    (** {v BUILD_SET(count) v}
+
+        Creates a set consuming count items from the stack, and pushes the resulting set onto the
+        stack. *)
+    let run env code {FFI.Instruction.opname; arg= count} =
+      let open IResult.Let_syntax in
+      Debug.p "[%s] count = %d\n" opname count ;
+      let* env, items = pop_n_datastack opname env count in
+      let* env, items = cells_to_textual env code items in
+      let env, id, _typ = Env.mk_builtin_call env Builtin.PythonBuildSet items in
+      let env = Env.push env (DataStack.Temp id) in
+      Ok (env, None)
+  end
+
+  module TUPLE = struct
+    (** {v BUILD_TUPLE(count) v}
+
+        Creates a tuple consuming count items from the stack, and pushes the resulting tuple onto
+        the stack. *)
+    let run env code {FFI.Instruction.opname; arg= count} =
+      let open IResult.Let_syntax in
+      Debug.p "[%s] count = %d\n" opname count ;
+      let* env, items = pop_n_datastack opname env count in
+      let* env, items = cells_to_textual env code items in
+      let env, id, _typ = Env.mk_builtin_call env Builtin.PythonBuildTuple items in
+      let env = Env.push env (DataStack.Temp id) in
+      Ok (env, None)
+  end
 end
 
 (** Return the offset of the next opcode, if any *)
@@ -1768,6 +1798,10 @@ let run_instruction env code ({FFI.Instruction.opname; starts_line} as instr) ne
       JUMP.IF_OR_POP.run ~jump_if:false env code instr next_offset_opt
   | "BUILD_LIST" ->
       BUILD.LIST.run env code instr
+  | "BUILD_SET" ->
+      BUILD.SET.run env code instr
+  | "BUILD_TUPLE" ->
+      BUILD.TUPLE.run env code instr
   | "STORE_SUBSCR" ->
       STORE.SUBSCR.run env code instr
   | "BINARY_SUBSCR" ->

infer/src/python/unit/PyTransTest.ml

@@ -2640,6 +2640,98 @@ l[x] = 10
           declare $builtins.python_float(float) : *PyFloat
 
           declare $builtins.python_int(int) : *PyInt |}]
+
+
+    let%expect_test _ =
+      let source =
+        {|
+t = (1, 2, 3) # will be a constant, not a BUILD_TUPLE
+def f(x, y, z):
+        return (x, y, z) # should be BUILD_TUPLE
+|}
+      in
+      test source ;
+      [%expect
+        {|
+        .source_language = "python"
+
+        define dummy.$toplevel() : *PyNone {
+          #b0:
+              store &dummy::t <- $builtins.python_tuple($builtins.python_int(1), $builtins.python_int(2), $builtins.python_int(3)):*PyTuple
+              n0 = $builtins.python_code("dummy.f")
+              ret null
+
+        }
+
+        define dummy.f(x: *PyObject, y: *PyObject, z: *PyObject) : *PyObject {
+          #b0:
+              n0:*PyObject = load &x
+              n1:*PyObject = load &y
+              n2:*PyObject = load &z
+              n3 = $builtins.python_build_tuple(n0, n1, n2)
+              ret n3
+
+        }
+
+        global dummy::t: *PyObject
+
+        global $python_implicit_names::__name__: *PyString
+
+        global $python_implicit_names::__file__: *PyString
+
+        declare $builtins.python_build_tuple(...) : *PyTuple
+
+        declare $builtins.python_code(*String) : *PyCode
+
+        declare $builtins.python_tuple(...) : *PyObject
+
+        declare $builtins.python_bytes(*Bytes) : *PyBytes
+
+        declare $builtins.python_string(*String) : *PyString
+
+        declare $builtins.python_bool(int) : *PyBool
+
+        declare $builtins.python_float(float) : *PyFloat
+
+        declare $builtins.python_int(int) : *PyInt |}]
+
+
+    let%expect_test _ =
+      let source = {|
+s = {1, 2, 3}
+|} in
+      test source ;
+      [%expect
+        {|
+          .source_language = "python"
+
+          define dummy.$toplevel() : *PyNone {
+            #b0:
+                n0 = $builtins.python_build_set($builtins.python_int(1), $builtins.python_int(2), $builtins.python_int(3))
+                store &dummy::s <- n0:*PySet
+                ret null
+
+          }
+
+          global dummy::s: *PyObject
+
+          global $python_implicit_names::__name__: *PyString
+
+          global $python_implicit_names::__file__: *PyString
+
+          declare $builtins.python_build_set(...) : *PySet
+
+          declare $builtins.python_tuple(...) : *PyObject
+
+          declare $builtins.python_bytes(*Bytes) : *PyBytes
+
+          declare $builtins.python_string(*String) : *PyString
+
+          declare $builtins.python_bool(int) : *PyBool
+
+          declare $builtins.python_float(float) : *PyFloat
+
+          declare $builtins.python_int(int) : *PyInt |}]
   end )