Addressing review comments.

* Fixing code comments.
* Adding tests with more unicode chars.
* Adding commit ID for code copied from cpython.git.

Author: advait-dixit

mypyc/lib-rt/str_ops.c

@@ -5,7 +5,7 @@
 #include <Python.h>
 #include "CPy.h"
 
-// Copied from cpython.git:Objects/unicodeobject.c.
+// Copied from cpython.git:Objects/unicodeobject.c@0ef4ffeefd1737c18dc9326133c7894d58108c2e.
 #define BLOOM_MASK unsigned long
 #define BLOOM(mask, ch)     ((mask &  (1UL << ((ch) & (BLOOM_WIDTH - 1)))))
 #if LONG_BIT >= 128
@@ -18,7 +18,8 @@
 #error "LONG_BIT is smaller than 32"
 #endif
 
-// Copied from cpython.git:Objects/unicodeobject.c. This is needed for str.strip("...").
+// Copied from cpython.git:Objects/unicodeobject.c@0ef4ffeefd1737c18dc9326133c7894d58108c2e.
+// This is needed for str.strip("...").
 static inline BLOOM_MASK
 make_bloom_mask(int kind, const void* ptr, Py_ssize_t len)
 {
@@ -226,14 +227,18 @@ PyObject *CPyStr_RSplit(PyObject *str, PyObject *sep, CPyTagged max_split) {
     return PyUnicode_RSplit(str, sep, temp_max_split);
 }
 
-// This function has been copied from _PyUnicode_XStrip in cpython.git:Objects/unicodeobject.c.
+// This function has been copied from _PyUnicode_XStrip in cpython.git:Objects/unicodeobject.c@0ef4ffeefd1737c18dc9326133c7894d58108c2e.
 static PyObject *_PyStr_XStrip(PyObject *self, int striptype, PyObject *sepobj) {
     const void *data;
     int kind;
     Py_ssize_t i, j, len;
     BLOOM_MASK sepmask;
     Py_ssize_t seplen;
 
+    // This check is needed from Python 3.9 and earlier.
+    if (PyUnicode_READY(self) == -1 || PyUnicode_READY(sepobj) == -1)
+        return NULL;
+
     kind = PyUnicode_KIND(self);
     data = PyUnicode_DATA(self);
     len = PyUnicode_GET_LENGTH(self);
@@ -272,11 +277,15 @@ static PyObject *_PyStr_XStrip(PyObject *self, int striptype, PyObject *sepobj)
     return PyUnicode_Substring(self, i, j);
 }
 
-// Copied from do_strip function in cpython.git/Objects/unicodeobject.c.
+// Copied from do_strip function in cpython.git/Objects/unicodeobject.c@0ef4ffeefd1737c18dc9326133c7894d58108c2e.
 PyObject *_CPyStr_Strip(PyObject *self, int strip_type, PyObject *sep) {
     if (sep == NULL || sep == Py_None) {
         Py_ssize_t len, i, j;
 
+        // This check is needed from Python 3.9 and earlier.
+        if (PyUnicode_READY(self) == -1)
+            return NULL;
+
         len = PyUnicode_GET_LENGTH(self);
 
         if (PyUnicode_IS_ASCII(self)) {

mypyc/primitives/str_ops.py

@@ -136,7 +136,6 @@
 )
 
 # str.strip, str.lstrip, str.rstrip
-# Order of iteration matters. It should correspond with LEFTSTRIP, RIGHTSTRIP and BOTHSTRIP macros defined in CPy.h.
 for strip_prefix in ["l", "r", ""]:
     method_op(
         name=f"{strip_prefix}strip",

mypyc/test-data/run-strings.test

@@ -776,7 +776,6 @@ def test_surrogate() -> None:
     assert repr("foobar\x00\xab\ud912\U00012345") == r"'foobar\x00«\ud912𒍅'"
 
 [case testStrip]
-# This is a negative test. strip variants without args does not use efficient primitives.
 def test_all_strips_default() -> None:
     s = " a1\t"
     assert s.lstrip() == "a1\t"
@@ -787,3 +786,10 @@ def test_all_strips() -> None:
     assert s.lstrip("xy") == "b2yy"
     assert s.strip("xy") == "b2"
     assert s.rstrip("xy") == "xxb2"
+def test_unicode_whitespace() -> None:
+    assert "\u200A\u000D\u2009\u2020\u000Dtt\u0085\u000A".strip() == "\u2020\u000Dtt"
+def test_unicode_range() -> None:
+    assert "\u2029 \U00107581 ".lstrip() == "\U00107581 "
+    assert "\u2029 \U0010AAAA\U00104444B\u205F ".strip() == "\U0010AAAA\U00104444B"
+    assert " \u3000\u205F ".strip() == ""
+    assert "\u2029 \U00102865\u205F ".rstrip() == "\u2029 \U00102865"