Limited API: Add a fast path to _PepUnicode_AsString()

Add a helper that returns the char * UTF-8 data from a
PyUnicode object without triggering conversions.

Change _PepUnicode_AsString() to return const char *
to match Python 3 on this occasion.

[ChangeLog][shiboken6] A fast path for converting
Python string to UTF-8 has been added to the limited API.

Change-Id: I619933633e226525fc248b27fd42cd1ce5178be0
Reviewed-by: Christian Tismer <tismer@stackless.com>

2 files changed, 24 insertions, 3 deletions

diff --git a/sources/shiboken6/libshiboken/pep384impl.cpp b/sources/shiboken6/libshiboken/pep384impl.cpp
index d2a2dd68a..d9b7bd4fb 100644
--- a/sources/shiboken6/libshiboken/pep384impl.cpp
+++ b/sources/shiboken6/libshiboken/pep384impl.cpp
@@ -359,8 +359,26 @@ void *_PepUnicode_DATA(PyObject *str)
            ? _PepUnicode_COMPACT_DATA(str) : _PepUnicode_NONCOMPACT_DATA(str);
 }
 
-char *
-_PepUnicode_AsString(PyObject *str)
+// Fast path accessing UTF8 data without doing a conversion similar
+// to _PyUnicode_AsUTF8String
+static const char *utf8FastPath(PyObject *str)
+{
+    if (PyUnicode_GetLength(str) == 0)
+        return "";
+    auto *asciiObj = reinterpret_cast<PepASCIIObject *>(str);
+    if (asciiObj->state.kind != PepUnicode_1BYTE_KIND)
+        return nullptr; // Empirical: PyCompactUnicodeObject.utf8 is only valid for 1 byte
+    if (asciiObj->state.ascii) {
+        auto *data = asciiObj + 1;
+        return reinterpret_cast<const char *>(data);
+    }
+    auto *compactObj = reinterpret_cast<PepCompactUnicodeObject *>(str);
+    if (compactObj->utf8_length)
+        return compactObj->utf8;
+    return nullptr;
+}
+
+const char *_PepUnicode_AsString(PyObject *str)
 {
     /*
      * We need to keep the string alive but cannot borrow the Python object.
@@ -373,6 +391,9 @@ _PepUnicode_AsString(PyObject *str)
 #define TOSTRING(x) STRINGIFY(x)
 #define AT __FILE__ ":" TOSTRING(__LINE__)
 
+    if (const auto *utf8 = utf8FastPath(str))
+        return utf8;
+
     static PyObject *cstring_dict = nullptr;
     if (cstring_dict == nullptr) {
         cstring_dict = PyDict_New();
diff --git a/sources/shiboken6/libshiboken/pep384impl.h b/sources/shiboken6/libshiboken/pep384impl.h
index 2c9551ea6..551038f84 100644
--- a/sources/shiboken6/libshiboken/pep384impl.h
+++ b/sources/shiboken6/libshiboken/pep384impl.h
@@ -240,7 +240,7 @@ LIBSHIBOKEN_API int Pep_GetVerboseFlag(void);
 
 #ifdef Py_LIMITED_API
 
-LIBSHIBOKEN_API char *_PepUnicode_AsString(PyObject *);
+LIBSHIBOKEN_API const char *_PepUnicode_AsString(PyObject *);
 
 enum PepUnicode_Kind {
     PepUnicode_WCHAR_KIND = 0,