From 5601e359593439a9e1cc1f38768020bbceb7f576 Mon Sep 17 00:00:00 2001
From: Jingchen Ye <97littleleaf11@gmail.com>
Date: Tue, 22 Jun 2021 04:14:37 +0800
Subject: [PATCH 01/14] Add branch for str

---
 mypyc/irbuild/ll_builder.py |  9 ++++++++-
 mypyc/primitives/str_ops.py | 10 +++++++++-
 2 files changed, 17 insertions(+), 2 deletions(-)

diff --git a/mypyc/irbuild/ll_builder.py b/mypyc/irbuild/ll_builder.py
index ef2e482e45c2..11a3563a8c7e 100644
--- a/mypyc/irbuild/ll_builder.py
+++ b/mypyc/irbuild/ll_builder.py
@@ -63,7 +63,7 @@
 )
 from mypyc.primitives.int_ops import int_comparison_op_mapping
 from mypyc.primitives.exc_ops import err_occurred_op, keep_propagating_op
-from mypyc.primitives.str_ops import unicode_compare, str_check_if_true
+from mypyc.primitives.str_ops import unicode_compare, str_check_if_true, str_size_op
 from mypyc.primitives.set_ops import new_set_op
 from mypyc.rt_subtype import is_runtime_subtype
 from mypyc.subtype import is_subtype
@@ -1132,6 +1132,13 @@ def builtin_len(self, val: Value, line: int, use_pyssize_t: bool = False) -> Val
             offset = Integer(1, c_pyssize_t_rprimitive, line)
             return self.int_op(short_int_rprimitive, size_value, offset,
                                IntOp.LEFT_SHIFT, line)
+        elif is_str_rprimitive(typ):
+            size_value = self.call_c(str_size_op, [val], line)
+            if use_pyssize_t:
+                return size_value
+            offset = Integer(1, c_pyssize_t_rprimitive, line)
+            return self.int_op(short_int_rprimitive, size_value, offset,
+                               IntOp.LEFT_SHIFT, line)
         elif is_set_rprimitive(typ):
             elem_address = self.add(GetElementPtr(val, PySetObject, 'used'))
             size_value = self.add(LoadMem(c_pyssize_t_rprimitive, elem_address))
diff --git a/mypyc/primitives/str_ops.py b/mypyc/primitives/str_ops.py
index 4e56d885528b..9e9706b2d7e6 100644
--- a/mypyc/primitives/str_ops.py
+++ b/mypyc/primitives/str_ops.py
@@ -5,7 +5,8 @@
 from mypyc.ir.ops import ERR_MAGIC, ERR_NEVER
 from mypyc.ir.rtypes import (
     RType, object_rprimitive, str_rprimitive, int_rprimitive, list_rprimitive,
-    c_int_rprimitive, pointer_rprimitive, bool_rprimitive, bit_rprimitive
+    c_int_rprimitive, pointer_rprimitive, bool_rprimitive, bit_rprimitive,
+    c_pyssize_t_rprimitive
 )
 from mypyc.primitives.registry import (
     method_op, binary_op, function_op,
@@ -134,3 +135,10 @@
     c_function_name="CPyStr_IsTrue",
     error_kind=ERR_NEVER,
 )
+
+str_size_op = custom_op(
+    arg_types=[str_rprimitive],
+    return_type=c_pyssize_t_rprimitive,
+    c_function_name='PyUnicode_Size',
+    error_kind=ERR_NEVER)
+

From 96213011be8ef3b0ee14e8dc8d78083124c1d573 Mon Sep 17 00:00:00 2001
From: Jingchen Ye <97littleleaf11@gmail.com>
Date: Tue, 22 Jun 2021 04:20:54 +0800
Subject: [PATCH 02/14] Fix method name

---
 mypyc/primitives/str_ops.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/mypyc/primitives/str_ops.py b/mypyc/primitives/str_ops.py
index 9e9706b2d7e6..c12827513927 100644
--- a/mypyc/primitives/str_ops.py
+++ b/mypyc/primitives/str_ops.py
@@ -139,6 +139,6 @@
 str_size_op = custom_op(
     arg_types=[str_rprimitive],
     return_type=c_pyssize_t_rprimitive,
-    c_function_name='PyUnicode_Size',
+    c_function_name='PyUnicode_GET_LENGTH',
     error_kind=ERR_NEVER)
 

From cc98072e00cef78c7c2fc8ec230daaeb87e8bb1b Mon Sep 17 00:00:00 2001
From: Jingchen Ye <97littleleaf11@gmail.com>
Date: Tue, 22 Jun 2021 14:06:39 +0800
Subject: [PATCH 03/14] Support str in builtin_len

---
 mypyc/irbuild/ll_builder.py      | 17 ++++++++++-------
 mypyc/primitives/dict_ops.py     |  2 +-
 mypyc/primitives/str_ops.py      |  2 +-
 mypyc/test-data/irbuild-str.test | 11 +++++++----
 4 files changed, 19 insertions(+), 13 deletions(-)

diff --git a/mypyc/irbuild/ll_builder.py b/mypyc/irbuild/ll_builder.py
index 11a3563a8c7e..43e15fc7bee9 100644
--- a/mypyc/irbuild/ll_builder.py
+++ b/mypyc/irbuild/ll_builder.py
@@ -51,7 +51,7 @@
     list_tuple_op, new_tuple_op, new_tuple_with_length_op
 )
 from mypyc.primitives.dict_ops import (
-    dict_update_in_display_op, dict_new_op, dict_build_op, dict_size_op
+    dict_update_in_display_op, dict_new_op, dict_build_op, dict_ssize_t_size_op
 )
 from mypyc.primitives.generic_ops import (
     py_getattr_op, py_call_op, py_call_with_kwargs_op, py_method_call_op,
@@ -63,7 +63,9 @@
 )
 from mypyc.primitives.int_ops import int_comparison_op_mapping
 from mypyc.primitives.exc_ops import err_occurred_op, keep_propagating_op
-from mypyc.primitives.str_ops import unicode_compare, str_check_if_true, str_size_op
+from mypyc.primitives.str_ops import (
+    unicode_compare, str_check_if_true, str_ssize_t_size_op
+)
 from mypyc.primitives.set_ops import new_set_op
 from mypyc.rt_subtype import is_runtime_subtype
 from mypyc.subtype import is_subtype
@@ -982,10 +984,11 @@ def add_bool_branch(self, value: Value, true: BasicBlock, false: BasicBlock) ->
             zero = Integer(0, short_int_rprimitive)
             self.compare_tagged_condition(value, zero, '!=', true, false, value.line)
             return
-        elif is_same_type(value.type, str_rprimitive):
-            value = self.call_c(str_check_if_true, [value], value.line)
+        # elif is_same_type(value.type, str_rprimitive):
+        #     value = self.call_c(str_check_if_true, [value], value.line)
         elif (is_same_type(value.type, list_rprimitive)
-                or is_same_type(value.type, dict_rprimitive)):
+                or is_same_type(value.type, dict_rprimitive)
+                or is_same_type(value.type, str_rprimitive)):
             length = self.builtin_len(value, value.line)
             zero = Integer(0)
             value = self.binary_op(length, zero, '!=', value.line)
@@ -1126,14 +1129,14 @@ def builtin_len(self, val: Value, line: int, use_pyssize_t: bool = False) -> Val
             return self.int_op(short_int_rprimitive, size_value, offset,
                                IntOp.LEFT_SHIFT, line)
         elif is_dict_rprimitive(typ):
-            size_value = self.call_c(dict_size_op, [val], line)
+            size_value = self.call_c(dict_ssize_t_size_op, [val], line)
             if use_pyssize_t:
                 return size_value
             offset = Integer(1, c_pyssize_t_rprimitive, line)
             return self.int_op(short_int_rprimitive, size_value, offset,
                                IntOp.LEFT_SHIFT, line)
         elif is_str_rprimitive(typ):
-            size_value = self.call_c(str_size_op, [val], line)
+            size_value = self.call_c(str_ssize_t_size_op, [val], line)
             if use_pyssize_t:
                 return size_value
             offset = Integer(1, c_pyssize_t_rprimitive, line)
diff --git a/mypyc/primitives/dict_ops.py b/mypyc/primitives/dict_ops.py
index e176ffcac316..5ae96a174904 100644
--- a/mypyc/primitives/dict_ops.py
+++ b/mypyc/primitives/dict_ops.py
@@ -240,7 +240,7 @@
     c_function_name='CPyDict_CheckSize',
     error_kind=ERR_FALSE)
 
-dict_size_op = custom_op(
+dict_ssize_t_size_op = custom_op(
     arg_types=[dict_rprimitive],
     return_type=c_pyssize_t_rprimitive,
     c_function_name='PyDict_Size',
diff --git a/mypyc/primitives/str_ops.py b/mypyc/primitives/str_ops.py
index c12827513927..a94cd480ec8b 100644
--- a/mypyc/primitives/str_ops.py
+++ b/mypyc/primitives/str_ops.py
@@ -136,7 +136,7 @@
     error_kind=ERR_NEVER,
 )
 
-str_size_op = custom_op(
+str_ssize_t_size_op = custom_op(
     arg_types=[str_rprimitive],
     return_type=c_pyssize_t_rprimitive,
     c_function_name='PyUnicode_GET_LENGTH',
diff --git a/mypyc/test-data/irbuild-str.test b/mypyc/test-data/irbuild-str.test
index 3de91be40ea2..ce98c746c474 100644
--- a/mypyc/test-data/irbuild-str.test
+++ b/mypyc/test-data/irbuild-str.test
@@ -148,14 +148,17 @@ def is_true(x: str) -> bool:
 [out]
 def is_true(x):
     x :: str
-    r0 :: bit
+    r0 :: native_int
+    r1 :: short_int
+    r2 :: bit
 L0:
-    r0 = CPyStr_IsTrue(x)
-    if r0 goto L1 else goto L2 :: bool
+    r0 = PyUnicode_GET_LENGTH(x)
+    r1 = r0 << 1
+    r2 = r1 != 0
+    if r2 goto L1 else goto L2 :: bool
 L1:
     return 1
 L2:
     return 0
 L3:
     unreachable
-

From 52bd1ef1e74d743a3ab1a8d18d1382e93f8310db Mon Sep 17 00:00:00 2001
From: Jingchen Ye <97littleleaf11@gmail.com>
Date: Tue, 22 Jun 2021 14:37:06 +0800
Subject: [PATCH 04/14] Revert and simplify

---
 mypyc/irbuild/ll_builder.py      | 46 ++++++++++++--------------------
 mypyc/test-data/irbuild-str.test | 10 +++----
 2 files changed, 20 insertions(+), 36 deletions(-)

diff --git a/mypyc/irbuild/ll_builder.py b/mypyc/irbuild/ll_builder.py
index 43e15fc7bee9..3ad1eed91303 100644
--- a/mypyc/irbuild/ll_builder.py
+++ b/mypyc/irbuild/ll_builder.py
@@ -984,11 +984,10 @@ def add_bool_branch(self, value: Value, true: BasicBlock, false: BasicBlock) ->
             zero = Integer(0, short_int_rprimitive)
             self.compare_tagged_condition(value, zero, '!=', true, false, value.line)
             return
-        # elif is_same_type(value.type, str_rprimitive):
-        #     value = self.call_c(str_check_if_true, [value], value.line)
+        elif is_same_type(value.type, str_rprimitive):
+            value = self.call_c(str_check_if_true, [value], value.line)
         elif (is_same_type(value.type, list_rprimitive)
-                or is_same_type(value.type, dict_rprimitive)
-                or is_same_type(value.type, str_rprimitive)):
+                or is_same_type(value.type, dict_rprimitive)):
             length = self.builtin_len(value, value.line)
             zero = Integer(0)
             value = self.binary_op(length, zero, '!=', value.line)
@@ -1119,39 +1118,28 @@ def builtin_len(self, val: Value, line: int, use_pyssize_t: bool = False) -> Val
         Return c_pyssize_t if use_pyssize_t is true (unshifted).
         """
         typ = val.type
+        size_value = None
         if is_list_rprimitive(typ) or is_tuple_rprimitive(typ):
             elem_address = self.add(GetElementPtr(val, PyVarObject, 'ob_size'))
             size_value = self.add(LoadMem(c_pyssize_t_rprimitive, elem_address))
             self.add(KeepAlive([val]))
-            if use_pyssize_t:
-                return size_value
-            offset = Integer(1, c_pyssize_t_rprimitive, line)
-            return self.int_op(short_int_rprimitive, size_value, offset,
-                               IntOp.LEFT_SHIFT, line)
-        elif is_dict_rprimitive(typ):
-            size_value = self.call_c(dict_ssize_t_size_op, [val], line)
-            if use_pyssize_t:
-                return size_value
-            offset = Integer(1, c_pyssize_t_rprimitive, line)
-            return self.int_op(short_int_rprimitive, size_value, offset,
-                               IntOp.LEFT_SHIFT, line)
-        elif is_str_rprimitive(typ):
-            size_value = self.call_c(str_ssize_t_size_op, [val], line)
-            if use_pyssize_t:
-                return size_value
-            offset = Integer(1, c_pyssize_t_rprimitive, line)
-            return self.int_op(short_int_rprimitive, size_value, offset,
-                               IntOp.LEFT_SHIFT, line)
         elif is_set_rprimitive(typ):
             elem_address = self.add(GetElementPtr(val, PySetObject, 'used'))
             size_value = self.add(LoadMem(c_pyssize_t_rprimitive, elem_address))
             self.add(KeepAlive([val]))
+        elif is_dict_rprimitive(typ):
+            size_value = self.call_c(dict_ssize_t_size_op, [val], line)
+        elif is_str_rprimitive(typ):
+            size_value = self.call_c(str_ssize_t_size_op, [val], line)
+
+        if size_value is not None:
             if use_pyssize_t:
                 return size_value
             offset = Integer(1, c_pyssize_t_rprimitive, line)
             return self.int_op(short_int_rprimitive, size_value, offset,
                                IntOp.LEFT_SHIFT, line)
-        elif isinstance(typ, RInstance):
+
+        if isinstance(typ, RInstance):
             # TODO: Support use_pyssize_t
             assert not use_pyssize_t
             length = self.gen_method_call(val, '__len__', [], int_rprimitive, line)
@@ -1165,12 +1153,12 @@ def builtin_len(self, val: Value, line: int, use_pyssize_t: bool = False) -> Val
             self.add(Unreachable())
             self.activate_block(ok)
             return length
+
+        # generic case
+        if use_pyssize_t:
+            return self.call_c(generic_ssize_t_len_op, [val], line)
         else:
-            # generic case
-            if use_pyssize_t:
-                return self.call_c(generic_ssize_t_len_op, [val], line)
-            else:
-                return self.call_c(generic_len_op, [val], line)
+            return self.call_c(generic_len_op, [val], line)
 
     def new_tuple(self, items: List[Value], line: int) -> Value:
         size = Integer(len(items), c_pyssize_t_rprimitive)  # type: Value
diff --git a/mypyc/test-data/irbuild-str.test b/mypyc/test-data/irbuild-str.test
index ce98c746c474..936cd4b0b4f0 100644
--- a/mypyc/test-data/irbuild-str.test
+++ b/mypyc/test-data/irbuild-str.test
@@ -148,14 +148,10 @@ def is_true(x: str) -> bool:
 [out]
 def is_true(x):
     x :: str
-    r0 :: native_int
-    r1 :: short_int
-    r2 :: bit
+    r0 :: bit
 L0:
-    r0 = PyUnicode_GET_LENGTH(x)
-    r1 = r0 << 1
-    r2 = r1 != 0
-    if r2 goto L1 else goto L2 :: bool
+    r0 = CPyStr_IsTrue(x)
+    if r0 goto L1 else goto L2 :: bool
 L1:
     return 1
 L2:

From 2a880c75cb63278c9fe507b40c56448814511b29 Mon Sep 17 00:00:00 2001
From: Jingchen Ye <97littleleaf11@gmail.com>
Date: Tue, 22 Jun 2021 15:03:16 +0800
Subject: [PATCH 05/14] Add test about list comprehension for str

---
 mypyc/irbuild/for_helpers.py   | 5 +++--
 mypyc/test-data/run-lists.test | 4 ++++
 2 files changed, 7 insertions(+), 2 deletions(-)

diff --git a/mypyc/irbuild/for_helpers.py b/mypyc/irbuild/for_helpers.py
index 2e44ec3afaf5..98878cea64c6 100644
--- a/mypyc/irbuild/for_helpers.py
+++ b/mypyc/irbuild/for_helpers.py
@@ -16,7 +16,7 @@
 )
 from mypyc.ir.rtypes import (
     RType, is_short_int_rprimitive, is_list_rprimitive, is_sequence_rprimitive,
-    is_tuple_rprimitive, is_dict_rprimitive,
+    is_tuple_rprimitive, is_dict_rprimitive, is_str_rprimitive,
     RTuple, short_int_rprimitive, int_rprimitive
 )
 from mypyc.primitives.registry import CFunctionDescription
@@ -164,7 +164,8 @@ def sequence_from_generator_preallocate_helper(
     """
     if len(gen.sequences) == 1 and len(gen.indices) == 1 and len(gen.condlists[0]) == 0:
         rtype = builder.node_type(gen.sequences[0])
-        if is_list_rprimitive(rtype) or is_tuple_rprimitive(rtype):
+        if (is_list_rprimitive(rtype) or is_tuple_rprimitive(rtype)
+                or is_str_rprimitive(rtype)):
             sequence = builder.accept(gen.sequences[0])
             length = builder.builder.builtin_len(sequence, gen.line, use_pyssize_t=True)
             target_op = empty_op_llbuilder(length, gen.line)
diff --git a/mypyc/test-data/run-lists.test b/mypyc/test-data/run-lists.test
index a78b2d65de4b..9ef337c4e075 100644
--- a/mypyc/test-data/run-lists.test
+++ b/mypyc/test-data/run-lists.test
@@ -357,3 +357,7 @@ def test() -> None:
     source_e = [0, 1, 2]
     e = list((x ** 2) for x in (y + 2 for y in source_e))
     assert e == [4, 9, 16]
+    source_str = "abcd"
+    f = list("str:" + x for x in source_str)
+    assert f == ["str:a", "str:b", "str:c", "str:d"]
+

From 3b6b9342b9c6d212bda09c0c9aeddee6b47c7b49 Mon Sep 17 00:00:00 2001
From: Jingchen Ye <97littleleaf11@gmail.com>
Date: Fri, 25 Jun 2021 16:32:28 +0800
Subject: [PATCH 06/14] Modify irbuild test and add run test for tuple creation

---
 mypyc/irbuild/specialize.py        |  4 +-
 mypyc/test-data/irbuild-tuple.test | 87 ++++++++++--------------------
 mypyc/test-data/run-tuples.test    |  4 ++
 3 files changed, 34 insertions(+), 61 deletions(-)

diff --git a/mypyc/irbuild/specialize.py b/mypyc/irbuild/specialize.py
index ef2986cda0b1..367a2fc33268 100644
--- a/mypyc/irbuild/specialize.py
+++ b/mypyc/irbuild/specialize.py
@@ -121,7 +121,7 @@ def dict_methods_fast_path(
 def translate_list_from_generator_call(
         builder: IRBuilder, expr: CallExpr, callee: RefExpr) -> Optional[Value]:
     # Special case for simplest list comprehension, for example
-    #     list(f(x) for x in other_list/other_tuple)
+    #     list(f(x) for x in some_list/some_tuple/some_str)
     # translate_list_comprehension() would take care of other cases if this fails.
     if (len(expr.args) == 1
             and expr.arg_kinds[0] == ARG_POS
@@ -137,7 +137,7 @@ def translate_list_from_generator_call(
 def translate_tuple_from_generator_call(
         builder: IRBuilder, expr: CallExpr, callee: RefExpr) -> Optional[Value]:
     # Special case for simplest tuple creation from a generator, for example
-    #     tuple(f(x) for x in other_list/other_tuple)
+    #     tuple(f(x) for x in some_list/some_tuple/some_str)
     # translate_safe_generator_call() would take care of other cases if this fails.
     if (len(expr.args) == 1
             and expr.arg_kinds[0] == ARG_POS
diff --git a/mypyc/test-data/irbuild-tuple.test b/mypyc/test-data/irbuild-tuple.test
index 20590645d5d3..00fb66aa6a02 100644
--- a/mypyc/test-data/irbuild-tuple.test
+++ b/mypyc/test-data/irbuild-tuple.test
@@ -350,14 +350,13 @@ L4:
     return 1
 
 
-[case testTupleBuiltFromList2]
+[case testTupleBuiltFromStr]
 def f2(val: str) -> str:
     return val + "f2"
 
 def test() -> None:
-    source = ["a", "b", "c"]
+    source = "abc"
     a = tuple(f2(x) for x in source)
-    print(a)
 [out]
 def f2(val):
     val, r0, r1 :: str
@@ -366,71 +365,41 @@ L0:
     r1 = PyUnicode_Concat(val, r0)
     return r1
 def test():
-    r0, r1, r2 :: str
-    r3 :: list
-    r4, r5, r6, r7 :: ptr
-    source :: list
-    r8 :: ptr
-    r9 :: native_int
-    r10 :: tuple
-    r11 :: short_int
-    r12 :: ptr
-    r13 :: native_int
-    r14 :: short_int
-    r15 :: bit
-    r16 :: object
-    r17, x, r18 :: str
-    r19 :: bit
-    r20 :: short_int
+    r0, source :: str
+    r1 :: native_int
+    r2 :: tuple
+    r3 :: short_int
+    r4 :: native_int
+    r5 :: short_int
+    r6 :: bit
+    r7, x, r8 :: str
+    r9 :: bit
+    r10 :: short_int
     a :: tuple
-    r21 :: object
-    r22 :: str
-    r23, r24 :: object
 L0:
-    r0 = 'a'
-    r1 = 'b'
-    r2 = 'c'
-    r3 = PyList_New(3)
-    r4 = get_element_ptr r3 ob_item :: PyListObject
-    r5 = load_mem r4 :: ptr*
-    set_mem r5, r0 :: builtins.object*
-    r6 = r5 + WORD_SIZE*1
-    set_mem r6, r1 :: builtins.object*
-    r7 = r5 + WORD_SIZE*2
-    set_mem r7, r2 :: builtins.object*
-    keep_alive r3
-    source = r3
-    r8 = get_element_ptr source ob_size :: PyVarObject
-    r9 = load_mem r8 :: native_int*
-    keep_alive source
-    r10 = PyTuple_New(r9)
-    r11 = 0
+    r0 = 'abc'
+    source = r0
+    r1 = PyUnicode_GET_LENGTH(source)
+    r2 = PyTuple_New(r1)
+    r3 = 0
 L1:
-    r12 = get_element_ptr source ob_size :: PyVarObject
-    r13 = load_mem r12 :: native_int*
-    keep_alive source
-    r14 = r13 << 1
-    r15 = r11 < r14 :: signed
-    if r15 goto L2 else goto L4 :: bool
+    r4 = PyUnicode_GET_LENGTH(source)
+    r5 = r4 << 1
+    r6 = r3 < r5 :: signed
+    if r6 goto L2 else goto L4 :: bool
 L2:
-    r16 = CPyList_GetItemUnsafe(source, r11)
-    r17 = cast(str, r16)
-    x = r17
-    r18 = f2(x)
-    r19 = CPySequenceTuple_SetItemUnsafe(r10, r11, r18)
+    r7 = CPyStr_GetItem(source, r3)
+    x = r7
+    r8 = f2(x)
+    r9 = CPySequenceTuple_SetItemUnsafe(r2, r3, r8)
 L3:
-    r20 = r11 + 2
-    r11 = r20
+    r10 = r3 + 2
+    r3 = r10
     goto L1
 L4:
-    a = r10
-    r21 = builtins :: module
-    r22 = 'print'
-    r23 = CPyObject_GetAttr(r21, r22)
-    r24 = PyObject_CallFunctionObjArgs(r23, a, 0)
+    a = r2
     return 1
 
-
 [case testTupleBuiltFromVariableLengthTuple]
 from typing import Tuple
 
diff --git a/mypyc/test-data/run-tuples.test b/mypyc/test-data/run-tuples.test
index c4abfb1d3a36..5ef243b10aa0 100644
--- a/mypyc/test-data/run-tuples.test
+++ b/mypyc/test-data/run-tuples.test
@@ -196,6 +196,10 @@ def test_sequence_generator() -> None:
     a = tuple(f8(x) for x in source_fixed_length_tuple)
     assert a == (False, True, False, True)
 
+    source_str = "abbc"
+    a = tuple("s:" + x for x in source_str)
+    assert a == ("s:a", "s:b", "s:b", "s:c")
+
 TUPLE: Final[Tuple[str, ...]] = ('x', 'y')
 
 def test_final_boxed_tuple() -> None:

From 253a99b7ff15d7b653e68fd1b20678c0d1cd86a6 Mon Sep 17 00:00:00 2001
From: Jingchen Ye <97littleleaf11@gmail.com>
Date: Mon, 28 Jun 2021 03:01:07 +0800
Subject: [PATCH 07/14] Add a wrapper function

---
 mypyc/lib-rt/str_ops.c      | 8 ++++++++
 mypyc/primitives/str_ops.py | 7 +++----
 2 files changed, 11 insertions(+), 4 deletions(-)

diff --git a/mypyc/lib-rt/str_ops.c b/mypyc/lib-rt/str_ops.c
index 7208a9b7e206..a933b16b0b82 100644
--- a/mypyc/lib-rt/str_ops.c
+++ b/mypyc/lib-rt/str_ops.c
@@ -102,6 +102,14 @@ PyObject *CPyStr_GetSlice(PyObject *obj, CPyTagged start, CPyTagged end) {
     }
     return CPyObject_GetSlice(obj, start, end);
 }
+
+Py_ssize_t CPyStr_Size_size_t(PyObject *str) {
+    if (PyUnicode_READY(str) != -1) {
+        return PyUnicode_GET_LENGTH(str);
+    }
+    return -1;
+}
+
 /* Check if the given string is true (i.e. it's length isn't zero) */
 bool CPyStr_IsTrue(PyObject *obj) {
     Py_ssize_t length = PyUnicode_GET_LENGTH(obj);
diff --git a/mypyc/primitives/str_ops.py b/mypyc/primitives/str_ops.py
index a94cd480ec8b..4ffb0e7dc8c7 100644
--- a/mypyc/primitives/str_ops.py
+++ b/mypyc/primitives/str_ops.py
@@ -10,7 +10,7 @@
 )
 from mypyc.primitives.registry import (
     method_op, binary_op, function_op,
-    load_address_op, custom_op
+    load_address_op, custom_op, ERR_NEG_INT
 )
 
 
@@ -139,6 +139,5 @@
 str_ssize_t_size_op = custom_op(
     arg_types=[str_rprimitive],
     return_type=c_pyssize_t_rprimitive,
-    c_function_name='PyUnicode_GET_LENGTH',
-    error_kind=ERR_NEVER)
-
+    c_function_name='CPyStr_Size_size_t',
+    error_kind=ERR_NEG_INT)

From 4c717faa4ea4f4cf24aed0583d4e8da5ad664ccf Mon Sep 17 00:00:00 2001
From: Jingchen Ye <97littleleaf11@gmail.com>
Date: Tue, 29 Jun 2021 00:31:26 +0800
Subject: [PATCH 08/14] Fix test

---
 mypyc/lib-rt/str_ops.c          |  7 -------
 mypyc/primitives/str_ops.py     | 10 +++++-----
 mypyc/test-data/run-tuples.test |  6 +++---
 3 files changed, 8 insertions(+), 15 deletions(-)

diff --git a/mypyc/lib-rt/str_ops.c b/mypyc/lib-rt/str_ops.c
index a933b16b0b82..328020d9fb17 100644
--- a/mypyc/lib-rt/str_ops.c
+++ b/mypyc/lib-rt/str_ops.c
@@ -103,13 +103,6 @@ PyObject *CPyStr_GetSlice(PyObject *obj, CPyTagged start, CPyTagged end) {
     return CPyObject_GetSlice(obj, start, end);
 }
 
-Py_ssize_t CPyStr_Size_size_t(PyObject *str) {
-    if (PyUnicode_READY(str) != -1) {
-        return PyUnicode_GET_LENGTH(str);
-    }
-    return -1;
-}
-
 /* Check if the given string is true (i.e. it's length isn't zero) */
 bool CPyStr_IsTrue(PyObject *obj) {
     Py_ssize_t length = PyUnicode_GET_LENGTH(obj);
diff --git a/mypyc/primitives/str_ops.py b/mypyc/primitives/str_ops.py
index 4ffb0e7dc8c7..1a3efcaead48 100644
--- a/mypyc/primitives/str_ops.py
+++ b/mypyc/primitives/str_ops.py
@@ -89,7 +89,7 @@
 
 # str1 += str2
 #
-# PyUnicodeAppend makes an effort to reuse the LHS when the refcount
+# PyUnicode_Append makes an effort to reuse the LHS when the refcount
 # is 1. This is super dodgy but oh well, the interpreter does it.
 binary_op(name='+=',
           arg_types=[str_rprimitive, str_rprimitive],
@@ -116,7 +116,7 @@
     name='replace',
     arg_types=[str_rprimitive, str_rprimitive, str_rprimitive],
     return_type=str_rprimitive,
-    c_function_name="PyUnicode_Replace",
+    c_function_name='PyUnicode_Replace',
     error_kind=ERR_MAGIC,
     extra_int_constants=[(-1, c_int_rprimitive)])
 
@@ -125,19 +125,19 @@
     name='replace',
     arg_types=[str_rprimitive, str_rprimitive, str_rprimitive, int_rprimitive],
     return_type=str_rprimitive,
-    c_function_name="CPyStr_Replace",
+    c_function_name='CPyStr_Replace',
     error_kind=ERR_MAGIC)
 
 # check if a string is true (isn't an empty string)
 str_check_if_true = custom_op(
     arg_types=[str_rprimitive],
     return_type=bit_rprimitive,
-    c_function_name="CPyStr_IsTrue",
+    c_function_name='CPyStr_IsTrue',
     error_kind=ERR_NEVER,
 )
 
 str_ssize_t_size_op = custom_op(
     arg_types=[str_rprimitive],
     return_type=c_pyssize_t_rprimitive,
-    c_function_name='CPyStr_Size_size_t',
+    c_function_name='PyUnicode_GET_LENGTH',
     error_kind=ERR_NEG_INT)
diff --git a/mypyc/test-data/run-tuples.test b/mypyc/test-data/run-tuples.test
index 5ef243b10aa0..759177342fa9 100644
--- a/mypyc/test-data/run-tuples.test
+++ b/mypyc/test-data/run-tuples.test
@@ -196,9 +196,9 @@ def test_sequence_generator() -> None:
     a = tuple(f8(x) for x in source_fixed_length_tuple)
     assert a == (False, True, False, True)
 
-    source_str = "abbc"
-    a = tuple("s:" + x for x in source_str)
-    assert a == ("s:a", "s:b", "s:b", "s:c")
+    source_str = 'abbc'
+    b = tuple('s:' + x for x in source_str)
+    assert b == ('s:a', 's:b', 's:b', 's:c')
 
 TUPLE: Final[Tuple[str, ...]] = ('x', 'y')
 

From 872fd8258610f5860c5187be6103a0d5d401c96c Mon Sep 17 00:00:00 2001
From: Jingchen Ye <97littleleaf11@gmail.com>
Date: Tue, 29 Jun 2021 01:02:09 +0800
Subject: [PATCH 09/14] Fix irbuild test

---
 mypyc/test-data/irbuild-tuple.test | 44 ++++++++++++++++--------------
 1 file changed, 24 insertions(+), 20 deletions(-)

diff --git a/mypyc/test-data/irbuild-tuple.test b/mypyc/test-data/irbuild-tuple.test
index 00fb66aa6a02..81996109644f 100644
--- a/mypyc/test-data/irbuild-tuple.test
+++ b/mypyc/test-data/irbuild-tuple.test
@@ -367,37 +367,41 @@ L0:
 def test():
     r0, source :: str
     r1 :: native_int
-    r2 :: tuple
-    r3 :: short_int
-    r4 :: native_int
-    r5 :: short_int
+    r2 :: bit
+    r3 :: tuple
+    r4 :: short_int
+    r5 :: native_int
     r6 :: bit
-    r7, x, r8 :: str
-    r9 :: bit
-    r10 :: short_int
+    r7 :: short_int
+    r8 :: bit
+    r9, x, r10 :: str
+    r11 :: bit
+    r12 :: short_int
     a :: tuple
 L0:
     r0 = 'abc'
     source = r0
     r1 = PyUnicode_GET_LENGTH(source)
-    r2 = PyTuple_New(r1)
-    r3 = 0
+    r2 = r1 >= 0 :: signed
+    r3 = PyTuple_New(r1)
+    r4 = 0
 L1:
-    r4 = PyUnicode_GET_LENGTH(source)
-    r5 = r4 << 1
-    r6 = r3 < r5 :: signed
-    if r6 goto L2 else goto L4 :: bool
+    r5 = PyUnicode_GET_LENGTH(source)
+    r6 = r5 >= 0 :: signed
+    r7 = r5 << 1
+    r8 = r4 < r7 :: signed
+    if r8 goto L2 else goto L4 :: bool
 L2:
-    r7 = CPyStr_GetItem(source, r3)
-    x = r7
-    r8 = f2(x)
-    r9 = CPySequenceTuple_SetItemUnsafe(r2, r3, r8)
+    r9 = CPyStr_GetItem(source, r4)
+    x = r9
+    r10 = f2(x)
+    r11 = CPySequenceTuple_SetItemUnsafe(r3, r4, r10)
 L3:
-    r10 = r3 + 2
-    r3 = r10
+    r12 = r4 + 2
+    r4 = r12
     goto L1
 L4:
-    a = r2
+    a = r3
     return 1
 
 [case testTupleBuiltFromVariableLengthTuple]

From 0689db6d9b87fc49f7bfeb87cb0036add49c51f0 Mon Sep 17 00:00:00 2001
From: Jingchen Ye <97littleleaf11@gmail.com>
Date: Tue, 29 Jun 2021 22:39:33 +0800
Subject: [PATCH 10/14] Add a wrapper function

---
 mypyc/lib-rt/CPy.h          | 1 +
 mypyc/lib-rt/str_ops.c      | 4 ++++
 mypyc/primitives/str_ops.py | 2 +-
 3 files changed, 6 insertions(+), 1 deletion(-)

diff --git a/mypyc/lib-rt/CPy.h b/mypyc/lib-rt/CPy.h
index f1a342527b11..cd57380160ce 100644
--- a/mypyc/lib-rt/CPy.h
+++ b/mypyc/lib-rt/CPy.h
@@ -391,6 +391,7 @@ PyObject *CPyStr_GetSlice(PyObject *obj, CPyTagged start, CPyTagged end);
 bool CPyStr_Startswith(PyObject *self, PyObject *subobj);
 bool CPyStr_Endswith(PyObject *self, PyObject *subobj);
 bool CPyStr_IsTrue(PyObject *obj);
+Py_ssize_t CPyStr_Size_size_t(PyObject *str);
 
 
 // Set operations
diff --git a/mypyc/lib-rt/str_ops.c b/mypyc/lib-rt/str_ops.c
index 328020d9fb17..60b462b76fa2 100644
--- a/mypyc/lib-rt/str_ops.c
+++ b/mypyc/lib-rt/str_ops.c
@@ -108,3 +108,7 @@ bool CPyStr_IsTrue(PyObject *obj) {
     Py_ssize_t length = PyUnicode_GET_LENGTH(obj);
     return length != 0;
 }
+
+Py_ssize_t CPyStr_Size_size_t(PyObject *str) {
+     return PyUnicode_GET_LENGTH(str);
+}
\ No newline at end of file
diff --git a/mypyc/primitives/str_ops.py b/mypyc/primitives/str_ops.py
index 1a3efcaead48..b8480a2831c2 100644
--- a/mypyc/primitives/str_ops.py
+++ b/mypyc/primitives/str_ops.py
@@ -139,5 +139,5 @@
 str_ssize_t_size_op = custom_op(
     arg_types=[str_rprimitive],
     return_type=c_pyssize_t_rprimitive,
-    c_function_name='PyUnicode_GET_LENGTH',
+    c_function_name='CPyStr_Size_size_t',
     error_kind=ERR_NEG_INT)

From f46d88306fd71c96850b5f3183a23210369653b5 Mon Sep 17 00:00:00 2001
From: Jingchen Ye <97littleleaf11@gmail.com>
Date: Tue, 29 Jun 2021 22:49:10 +0800
Subject: [PATCH 11/14] Fix

---
 mypyc/primitives/str_ops.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/mypyc/primitives/str_ops.py b/mypyc/primitives/str_ops.py
index b8480a2831c2..23013c40eb26 100644
--- a/mypyc/primitives/str_ops.py
+++ b/mypyc/primitives/str_ops.py
@@ -10,7 +10,7 @@
 )
 from mypyc.primitives.registry import (
     method_op, binary_op, function_op,
-    load_address_op, custom_op, ERR_NEG_INT
+    load_address_op, custom_op
 )
 
 
@@ -140,4 +140,4 @@
     arg_types=[str_rprimitive],
     return_type=c_pyssize_t_rprimitive,
     c_function_name='CPyStr_Size_size_t',
-    error_kind=ERR_NEG_INT)
+    error_kind=ERR_NEVER)

From 6148994b30c3844a4c585a34e307239018cacdf2 Mon Sep 17 00:00:00 2001
From: Jingchen Ye <97littleleaf11@gmail.com>
Date: Wed, 30 Jun 2021 00:23:11 +0800
Subject: [PATCH 12/14] Fix irbuild test

---
 mypyc/test-data/irbuild-tuple.test | 46 ++++++++++++++----------------
 1 file changed, 21 insertions(+), 25 deletions(-)

diff --git a/mypyc/test-data/irbuild-tuple.test b/mypyc/test-data/irbuild-tuple.test
index 81996109644f..28ceea1e026d 100644
--- a/mypyc/test-data/irbuild-tuple.test
+++ b/mypyc/test-data/irbuild-tuple.test
@@ -367,41 +367,37 @@ L0:
 def test():
     r0, source :: str
     r1 :: native_int
-    r2 :: bit
-    r3 :: tuple
-    r4 :: short_int
-    r5 :: native_int
+    r2 :: tuple
+    r3 :: short_int
+    r4 :: native_int
+    r5 :: short_int
     r6 :: bit
-    r7 :: short_int
-    r8 :: bit
-    r9, x, r10 :: str
-    r11 :: bit
-    r12 :: short_int
+    r7, x, r8 :: str
+    r9 :: bit
+    r10 :: short_int
     a :: tuple
 L0:
     r0 = 'abc'
     source = r0
-    r1 = PyUnicode_GET_LENGTH(source)
-    r2 = r1 >= 0 :: signed
-    r3 = PyTuple_New(r1)
-    r4 = 0
+    r1 = CPyStr_Size_size_t(source)
+    r2 = PyTuple_New(r1)
+    r3 = 0
 L1:
-    r5 = PyUnicode_GET_LENGTH(source)
-    r6 = r5 >= 0 :: signed
-    r7 = r5 << 1
-    r8 = r4 < r7 :: signed
-    if r8 goto L2 else goto L4 :: bool
+    r4 = CPyStr_Size_size_t(source)
+    r5 = r4 << 1
+    r6 = r3 < r5 :: signed
+    if r6 goto L2 else goto L4 :: bool
 L2:
-    r9 = CPyStr_GetItem(source, r4)
-    x = r9
-    r10 = f2(x)
-    r11 = CPySequenceTuple_SetItemUnsafe(r3, r4, r10)
+    r7 = CPyStr_GetItem(source, r3)
+    x = r7
+    r8 = f2(x)
+    r9 = CPySequenceTuple_SetItemUnsafe(r2, r3, r8)
 L3:
-    r12 = r4 + 2
-    r4 = r12
+    r10 = r3 + 2
+    r3 = r10
     goto L1
 L4:
-    a = r3
+    a = r2
     return 1
 
 [case testTupleBuiltFromVariableLengthTuple]

From 2f2a959a23e5f497b8229d5ed8071802261f4167 Mon Sep 17 00:00:00 2001
From: Jingchen Ye <97littleleaf11@gmail.com>
Date: Wed, 30 Jun 2021 23:59:38 +0800
Subject: [PATCH 13/14] Revert previous revert

---
 mypyc/lib-rt/str_ops.c      | 5 ++++-
 mypyc/primitives/str_ops.py | 4 ++--
 2 files changed, 6 insertions(+), 3 deletions(-)

diff --git a/mypyc/lib-rt/str_ops.c b/mypyc/lib-rt/str_ops.c
index 60b462b76fa2..1b1454bed714 100644
--- a/mypyc/lib-rt/str_ops.c
+++ b/mypyc/lib-rt/str_ops.c
@@ -110,5 +110,8 @@ bool CPyStr_IsTrue(PyObject *obj) {
 }
 
 Py_ssize_t CPyStr_Size_size_t(PyObject *str) {
-     return PyUnicode_GET_LENGTH(str);
+    if (PyUnicode_READY(str) != -1) {
+        return PyUnicode_GET_LENGTH(str);
+    }
+    return -1;
 }
\ No newline at end of file
diff --git a/mypyc/primitives/str_ops.py b/mypyc/primitives/str_ops.py
index 23013c40eb26..b8480a2831c2 100644
--- a/mypyc/primitives/str_ops.py
+++ b/mypyc/primitives/str_ops.py
@@ -10,7 +10,7 @@
 )
 from mypyc.primitives.registry import (
     method_op, binary_op, function_op,
-    load_address_op, custom_op
+    load_address_op, custom_op, ERR_NEG_INT
 )
 
 
@@ -140,4 +140,4 @@
     arg_types=[str_rprimitive],
     return_type=c_pyssize_t_rprimitive,
     c_function_name='CPyStr_Size_size_t',
-    error_kind=ERR_NEVER)
+    error_kind=ERR_NEG_INT)

From 70f6e25614d4bdc78050166d3667f3e9fb40733b Mon Sep 17 00:00:00 2001
From: Jingchen Ye <97littleleaf11@gmail.com>
Date: Thu, 1 Jul 2021 02:39:11 +0800
Subject: [PATCH 14/14] Fix irbuild test

---
 mypyc/test-data/irbuild-tuple.test | 44 ++++++++++++++++--------------
 1 file changed, 24 insertions(+), 20 deletions(-)

diff --git a/mypyc/test-data/irbuild-tuple.test b/mypyc/test-data/irbuild-tuple.test
index 28ceea1e026d..564a4bf74d50 100644
--- a/mypyc/test-data/irbuild-tuple.test
+++ b/mypyc/test-data/irbuild-tuple.test
@@ -367,37 +367,41 @@ L0:
 def test():
     r0, source :: str
     r1 :: native_int
-    r2 :: tuple
-    r3 :: short_int
-    r4 :: native_int
-    r5 :: short_int
+    r2 :: bit
+    r3 :: tuple
+    r4 :: short_int
+    r5 :: native_int
     r6 :: bit
-    r7, x, r8 :: str
-    r9 :: bit
-    r10 :: short_int
+    r7 :: short_int
+    r8 :: bit
+    r9, x, r10 :: str
+    r11 :: bit
+    r12 :: short_int
     a :: tuple
 L0:
     r0 = 'abc'
     source = r0
     r1 = CPyStr_Size_size_t(source)
-    r2 = PyTuple_New(r1)
-    r3 = 0
+    r2 = r1 >= 0 :: signed
+    r3 = PyTuple_New(r1)
+    r4 = 0
 L1:
-    r4 = CPyStr_Size_size_t(source)
-    r5 = r4 << 1
-    r6 = r3 < r5 :: signed
-    if r6 goto L2 else goto L4 :: bool
+    r5 = CPyStr_Size_size_t(source)
+    r6 = r5 >= 0 :: signed
+    r7 = r5 << 1
+    r8 = r4 < r7 :: signed
+    if r8 goto L2 else goto L4 :: bool
 L2:
-    r7 = CPyStr_GetItem(source, r3)
-    x = r7
-    r8 = f2(x)
-    r9 = CPySequenceTuple_SetItemUnsafe(r2, r3, r8)
+    r9 = CPyStr_GetItem(source, r4)
+    x = r9
+    r10 = f2(x)
+    r11 = CPySequenceTuple_SetItemUnsafe(r3, r4, r10)
 L3:
-    r10 = r3 + 2
-    r3 = r10
+    r12 = r4 + 2
+    r4 = r12
     goto L1
 L4:
-    a = r2
+    a = r3
     return 1
 
 [case testTupleBuiltFromVariableLengthTuple]