Switch to CPython's str.find and str.rfind

from_cpython/Objects/stringobject.c

@@ -49,3 +49,56 @@ PyObject* string_rsplit(PyStringObject* self, PyObject* args) {
     return stringlib_rsplit((PyObject*)self, s, len, sub, n, maxsplit);
 }
 
+Py_LOCAL_INLINE(Py_ssize_t)
+string_find_internal(PyStringObject *self, PyObject *args, int dir)
+{
+    PyObject *subobj;
+    const char *sub;
+    Py_ssize_t sub_len;
+    Py_ssize_t start=0, end=PY_SSIZE_T_MAX;
+
+    if (!stringlib_parse_args_finds("find/rfind/index/rindex",
+                                    args, &subobj, &start, &end))
+        return -2;
+
+    if (PyString_Check(subobj)) {
+        sub = PyString_AS_STRING(subobj);
+        sub_len = PyString_GET_SIZE(subobj);
+    }
+#ifdef Py_USING_UNICODE
+    else if (PyUnicode_Check(subobj))
+        return PyUnicode_Find(
+            (PyObject *)self, subobj, start, end, dir);
+#endif
+    else if (PyObject_AsCharBuffer(subobj, &sub, &sub_len))
+        /* XXX - the "expected a character buffer object" is pretty
+           confusing for a non-expert.  remap to something else ? */
+        return -2;
+
+    if (dir > 0)
+        return stringlib_find_slice(
+            PyString_AS_STRING(self), PyString_GET_SIZE(self),
+            sub, sub_len, start, end);
+    else
+        return stringlib_rfind_slice(
+            PyString_AS_STRING(self), PyString_GET_SIZE(self),
+            sub, sub_len, start, end);
+}
+
+PyObject *
+string_rfind(PyStringObject *self, PyObject *args)
+{
+    Py_ssize_t result = string_find_internal(self, args, -1);
+    if (result == -2)
+        return NULL;
+    return PyInt_FromSsize_t(result);
+}
+
+PyObject *
+string_find(PyStringObject *self, PyObject *args)
+{
+    Py_ssize_t result = string_find_internal(self, args, +1);
+    if (result == -2)
+        return NULL;
+    return PyInt_FromSsize_t(result);
+}

src/runtime/str.cpp

@@ -35,6 +35,8 @@
 #include "runtime/util.h"
 
 extern "C" PyObject* string_rsplit(PyStringObject* self, PyObject* args) noexcept;
+extern "C" PyObject* string_find(PyStringObject* self, PyObject* args) noexcept;
+extern "C" PyObject* string_rfind(PyStringObject* self, PyObject* args) noexcept;
 
 namespace pyston {
 
@@ -1917,49 +1919,6 @@ Box* strEncode(BoxedString* self, Box* encoding, Box* error) {
     return result;
 }
 
-Box* strFind(BoxedString* self, Box* elt, Box* _start) {
-    if (self->cls != str_cls)
-        raiseExcHelper(TypeError, "descriptor 'find' requires a 'str' object but received a '%s'", getTypeName(self));
-
-    if (elt->cls != str_cls)
-        raiseExcHelper(TypeError, "expected a character buffer object");
-
-    if (_start->cls != int_cls) {
-        raiseExcHelper(TypeError, "'start' must be an int for now");
-        // Real error message:
-        // raiseExcHelper(TypeError, "slice indices must be integers or None or have an __index__ method");
-    }
-
-    int64_t start = static_cast<BoxedInt*>(_start)->n;
-    if (start < 0) {
-        start += self->s.size();
-        start = std::max(0L, start);
-    }
-
-    BoxedString* sub = static_cast<BoxedString*>(elt);
-
-    size_t r = self->s.find(sub->s, start);
-    if (r == std::string::npos)
-        return boxInt(-1);
-    return boxInt(r);
-}
-
-Box* strRfind(BoxedString* self, Box* elt) {
-    if (self->cls != str_cls)
-        raiseExcHelper(TypeError, "descriptor 'rfind' requires a 'str' object but received a '%s'", getTypeName(self));
-
-    if (elt->cls != str_cls)
-        raiseExcHelper(TypeError, "expected a character buffer object");
-
-    BoxedString* sub = static_cast<BoxedString*>(elt);
-
-    size_t r = self->s.rfind(sub->s);
-    if (r == std::string::npos)
-        return boxInt(-1);
-    return boxInt(r);
-}
-
-
 extern "C" Box* strGetitem(BoxedString* self, Box* slice) {
     assert(self->cls == str_cls);
 
@@ -2233,6 +2192,8 @@ void strDestructor(Box* b) {
 
 static PyMethodDef string_methods[] = {
     { "rsplit", (PyCFunction)string_rsplit, METH_VARARGS, NULL },
+    { "find", (PyCFunction)string_find, METH_VARARGS, NULL },
+    { "rfind", (PyCFunction)string_rfind, METH_VARARGS, NULL },
 };
 
 void setupStr() {
@@ -2288,10 +2249,6 @@ void setupStr() {
     str_cls->giveAttr("endswith",
                       new BoxedFunction(boxRTFunction((void*)strEndswith, BOXED_BOOL, 4, 2, 0, 0), { NULL, NULL }));
 
-    str_cls->giveAttr("find",
-                      new BoxedFunction(boxRTFunction((void*)strFind, BOXED_INT, 3, 1, false, false), { boxInt(0) }));
-    str_cls->giveAttr("rfind", new BoxedFunction(boxRTFunction((void*)strRfind, BOXED_INT, 2)));
-
     str_cls->giveAttr("partition", new BoxedFunction(boxRTFunction((void*)strPartition, UNKNOWN, 2)));
 
     str_cls->giveAttr("format", new BoxedFunction(boxRTFunction((void*)strFormat, UNKNOWN, 1, 0, true, true)));

test/tests/str_functions.py

@@ -80,7 +80,8 @@ print "hello world".translate(translation_map, "llo")
 print "hello world".translate(None, "llo")
 
 for i in xrange(-10, 10):
-    print i, "aaaaa".find("a", i)
+    print i, "aaaaa".find("a", i), "aaaa".find("a", 2, i)
+    print i, "aaaaa".rfind("a", i), "aaaa".rfind("a", 2, i)
 
 print "hello world".partition("hi")
 print "hello world".partition("hello")