Implement bytes.startswith in mypyc

Author: esarp

mypyc/lib-rt/CPy.h

@@ -784,7 +784,7 @@ PyObject *CPyBytes_Join(PyObject *sep, PyObject *iter);
 CPyTagged CPyBytes_Ord(PyObject *obj);
 PyObject *CPyBytes_Multiply(PyObject *bytes, CPyTagged count);
 PyObject *CPyBytes_Translate(PyObject *bytes, PyObject *table);
-
+int CPyBytes_Startswith(PyObject *self, PyObject *subobj);
 
 int CPyBytes_Compare(PyObject *left, PyObject *right);
 

mypyc/lib-rt/bytes_ops.c

@@ -220,3 +220,42 @@ PyObject *CPyBytes_Translate(PyObject *bytes, PyObject *table) {
     }
     return PyObject_CallMethodOneArg(bytes, name, table);
 }
+
+int CPyBytes_Startswith(PyObject *self, PyObject *subobj) {
+    if (PyBytes_CheckExact(self) && PyBytes_CheckExact(subobj)) {
+        if (self == subobj) {
+            return 1;
+        }
+
+        Py_ssize_t self_len = PyBytes_GET_SIZE(self);
+        Py_ssize_t subobj_len = PyBytes_GET_SIZE(subobj);
+
+        if (subobj_len > self_len) {
+            return 0;
+        }
+
+        const char *self_buf = PyBytes_AS_STRING(self);
+        const char *subobj_buf = PyBytes_AS_STRING(subobj);
+
+        if (subobj_len == 0) {
+            return 1;
+        }
+
+        return memcmp(self_buf, subobj_buf, (size_t)subobj_len) == 0 ? 1 : 0;
+    }
+    _Py_IDENTIFIER(startswith);
+    PyObject *name = _PyUnicode_FromId(&PyId_startswith);
+    if (name == NULL) {
+        return 2;
+    }
+    PyObject *result = PyObject_CallMethodOneArg(self, name, subobj);
+    if (result == NULL) {
+        return 2;
+    }
+    int ret = PyObject_IsTrue(result);
+    Py_DECREF(result);
+    if (ret < 0) {
+        return 2;
+    }
+    return ret;
+}

mypyc/primitives/bytes_ops.py

@@ -6,6 +6,7 @@
 from mypyc.ir.rtypes import (
     RUnion,
     bit_rprimitive,
+    bool_rprimitive,
     bytes_rprimitive,
     c_int_rprimitive,
     c_pyssize_t_rprimitive,
@@ -137,6 +138,16 @@
     error_kind=ERR_MAGIC,
 )
 
+# bytes.startswith(bytes)
+method_op(
+    name="startswith",
+    arg_types=[bytes_rprimitive, bytes_rprimitive],
+    return_type=c_int_rprimitive,
+    c_function_name="CPyBytes_Startswith",
+    truncated_type=bool_rprimitive,
+    error_kind=ERR_MAGIC,
+)
+
 # Join bytes objects and return a new bytes.
 # The first argument is the total number of the following bytes.
 bytes_build_op = custom_op(

mypyc/test-data/fixtures/ir.py

@@ -179,6 +179,7 @@ def __getitem__(self, i: slice) -> bytes: ...
     def join(self, x: Iterable[object]) -> bytes: ...
     def decode(self, encoding: str=..., errors: str=...) -> str: ...
     def translate(self, t: bytes) -> bytes: ...
+    def startswith(self, t: bytes) -> bool: ...
     def __iter__(self) -> Iterator[int]: ...
 
 class bytearray:

mypyc/test-data/irbuild-bytes.test

@@ -248,3 +248,17 @@ def f(b, table):
 L0:
     r0 = CPyBytes_Translate(b, table)
     return r0
+
+[case testBytesStartsWith]
+def f(a: bytes, b: bytes) -> bool:
+    return a.startswith(b)
+[out]
+def f(a, b):
+    a, b :: bytes
+    r0 :: i32
+    r1 :: bool
+L0:
+    r0 = CPyBytes_Startswith(a, b)
+    r1 = truncate r0: i32 to builtins.bool
+    return r1
+

mypyc/test-data/run-bytes.test

@@ -200,6 +200,19 @@ def test_translate() -> None:
     with assertRaises(ValueError, "translation table must be 256 characters long"):
         b'test'.translate(bytes(100))
 
+def test_startswith() -> None:
+    # Test default behavior
+    test = b'some string'
+    assert test.startswith(b'some')
+    assert test.startswith(b'some string')
+    assert not test.startswith(b'other')
+    assert not test.startswith(b'some string but longer')
+
+    # Test empty cases
+    assert test.startswith(b'')
+    assert b''.startswith(b'')
+    assert not b''.startswith(test)
+
 [case testBytesSlicing]
 def test_bytes_slicing() -> None:
     b = b'abcdefg'