Return Py_ssize_t instead of size_t for len(char*) and len(Py_UNICODE*).

Closes GH-2992.

Cython/Compiler/Optimize.py

@@ -2557,12 +2557,20 @@ class OptimizeBuiltinCalls(Visitor.NodeRefCleanupMixin,
     Pyx_strlen_func_type = PyrexTypes.CFuncType(
         PyrexTypes.c_size_t_type, [
             PyrexTypes.CFuncTypeArg("bytes", PyrexTypes.c_const_char_ptr_type, None)
-        ])
+        ],
+        nogil=True)
+
+    Pyx_ssize_strlen_func_type = PyrexTypes.CFuncType(
+        PyrexTypes.c_py_ssize_t_type, [
+            PyrexTypes.CFuncTypeArg("bytes", PyrexTypes.c_const_char_ptr_type, None)
+        ],
+        exception_value="-1")
 
     Pyx_Py_UNICODE_strlen_func_type = PyrexTypes.CFuncType(
-        PyrexTypes.c_size_t_type, [
+        PyrexTypes.c_py_ssize_t_type, [
             PyrexTypes.CFuncTypeArg("unicode", PyrexTypes.c_const_py_unicode_ptr_type, None)
-        ])
+        ],
+        exception_value="-1")
 
     PyObject_Size_func_type = PyrexTypes.CFuncType(
         PyrexTypes.c_py_ssize_t_type, [
@@ -2596,15 +2604,16 @@ class OptimizeBuiltinCalls(Visitor.NodeRefCleanupMixin,
             arg = arg.arg
         if arg.type.is_string:
             new_node = ExprNodes.PythonCapiCallNode(
-                node.pos, "strlen", self.Pyx_strlen_func_type,
+                node.pos, "__Pyx_ssize_strlen", self.Pyx_ssize_strlen_func_type,
                 args = [arg],
                 is_temp = node.is_temp,
-                utility_code = UtilityCode.load_cached("IncludeStringH", "StringTools.c"))
+                utility_code = UtilityCode.load_cached("ssize_strlen", "StringTools.c"))
         elif arg.type.is_pyunicode_ptr:
             new_node = ExprNodes.PythonCapiCallNode(
-                node.pos, "__Pyx_Py_UNICODE_strlen", self.Pyx_Py_UNICODE_strlen_func_type,
+                node.pos, "__Pyx_Py_UNICODE_ssize_strlen", self.Pyx_Py_UNICODE_strlen_func_type,
                 args = [arg],
-                is_temp = node.is_temp)
+                is_temp = node.is_temp,
+                utility_code = UtilityCode.load_cached("ssize_pyunicode_strlen", "StringTools.c"))
         elif arg.type.is_memoryviewslice:
             func_type = PyrexTypes.CFuncType(
                 PyrexTypes.c_py_ssize_t_type, [

Cython/Utility/StringTools.c

@@ -7,6 +7,40 @@
 
 #include <string>
 
+
+//////////////////// ssize_strlen.proto ////////////////////
+
+static CYTHON_INLINE Py_ssize_t __Pyx_ssize_strlen(const char *s);/*proto*/
+
+//////////////////// ssize_strlen ////////////////////
+//@requires: IncludeStringH
+
+static CYTHON_INLINE Py_ssize_t __Pyx_ssize_strlen(const char *s) {
+    size_t len = strlen(s);
+    if (unlikely(len > PY_SSIZE_T_MAX)) {
+        PyErr_SetString(PyExc_OverflowError, "byte string is too long");
+        return -1;
+    }
+    return (Py_ssize_t) len;
+}
+
+
+//////////////////// ssize_pyunicode_strlen.proto ////////////////////
+
+static CYTHON_INLINE Py_ssize_t __Pyx_Py_UNICODE_ssize_strlen(const Py_UNICODE *u);/*proto*/
+
+//////////////////// ssize_pyunicode_strlen ////////////////////
+
+static CYTHON_INLINE Py_ssize_t __Pyx_Py_UNICODE_ssize_strlen(const Py_UNICODE *u) {
+    size_t len = __Pyx_Py_UNICODE_strlen(u);
+    if (unlikely(len > PY_SSIZE_T_MAX)) {
+        PyErr_SetString(PyExc_OverflowError, "Py_UNICODE string is too long");
+        return -1;
+    }
+    return (Py_ssize_t) len;
+}
+
+
 //////////////////// InitStrings.proto ////////////////////
 
 static int __Pyx_InitStrings(__Pyx_StringTabEntry *t); /*proto*/