From 173204639e5c555dadbf5465e8d8b9b0d6b736ae Mon Sep 17 00:00:00 2001
From: "Ralf W. Grosse-Kunstleve" <rwgk@google.com>
Date: Mon, 10 Aug 2020 17:49:14 -0700
Subject: [PATCH] Adding tests specifically to exercise pybind11::str::raw_str.
 (#2366)

These tests will also alert us to any behavior changes across Python and PyPy versions.

Hardening tests in preparation for changing `pybind11::str` to only hold `PyUnicodeObject` (NOT also `bytes`). Note that this test exposes that `pybind11::str` can also hold `bytes`.
---
 tests/test_pytypes.cpp |  2 ++
 tests/test_pytypes.py  | 32 ++++++++++++++++++++++++++++++++
 2 files changed, 34 insertions(+)

diff --git a/tests/test_pytypes.cpp b/tests/test_pytypes.cpp
index 9f7bc37dc..9dae6e7d6 100644
--- a/tests/test_pytypes.cpp
+++ b/tests/test_pytypes.cpp
@@ -237,6 +237,8 @@ TEST_SUBMODULE(pytypes, m) {
         );
     });
 
+    m.def("convert_to_pybind11_str", [](py::object o) { return py::str(o); });
+
     m.def("get_implicit_casting", []() {
         py::dict d;
         d["char*_i1"] = "abc";
diff --git a/tests/test_pytypes.py b/tests/test_pytypes.py
index 4cfc707a3..1c7b1dd1f 100644
--- a/tests/test_pytypes.py
+++ b/tests/test_pytypes.py
@@ -220,6 +220,38 @@ def test_constructors():
         assert noconv2[k] is expected[k]
 
 
+def test_pybind11_str_raw_str():
+    # specifically to exercise pybind11::str::raw_str
+    cvt = m.convert_to_pybind11_str
+    assert cvt(u"Str") == u"Str"
+    assert cvt(b'Bytes') == u"Bytes" if str is bytes else "b'Bytes'"
+    assert cvt(None) == u"None"
+    assert cvt(False) == u"False"
+    assert cvt(True) == u"True"
+    assert cvt(42) == u"42"
+    assert cvt(2**65) == u"36893488147419103232"
+    assert cvt(-1.50) == u"-1.5"
+    assert cvt(()) == u"()"
+    assert cvt((18,)) == u"(18,)"
+    assert cvt([]) == u"[]"
+    assert cvt([28]) == u"[28]"
+    assert cvt({}) == u"{}"
+    assert cvt({3: 4}) == u"{3: 4}"
+    assert cvt(set()) == u"set([])" if str is bytes else "set()"
+    assert cvt({3, 3}) == u"set([3])" if str is bytes else "{3}"
+
+    valid_orig = u"DZ"
+    valid_utf8 = valid_orig.encode("utf-8")
+    valid_cvt = cvt(valid_utf8)
+    assert type(valid_cvt) == bytes  # Probably surprising.
+    assert valid_cvt == b'\xc7\xb1'
+
+    malformed_utf8 = b'\x80'
+    malformed_cvt = cvt(malformed_utf8)
+    assert type(malformed_cvt) == bytes  # Probably surprising.
+    assert malformed_cvt == b'\x80'
+
+
 def test_implicit_casting():
     """Tests implicit casting when assigning or appending to dicts and lists."""
     z = m.get_implicit_casting()