Objects/bytesobject.c (part 2)

Source:

cpython 3.14 @ ab2d84fe1023/Objects/bytesobject.c ↗

This annotation covers the bytes type's formatting, concatenation, interning, and protocol methods. See also objects_bytesobject_detail for construction and the buffer protocol.

Map

Lines	Symbol	Role
1-200	PyBytes_FromStringAndSize, PyBytes_FromFormat	Construction
201-600	bytes_concat, PyBytes_Concat	+ operator
601-900	bytes_repeat	* operator
901-1200	bytes_% / _PyBytes_FormatLong	% formatting
1201-1600	Short-string interning	characters[256] cache
1601-1900	bytes_decode	bytes.decode(encoding)
1901-2200	bytes_contains	in operator (substring search)
2201-3500	tp_methods: find, count, replace, split, join, upper, lower, etc.	String-like methods

Reading

Single-byte interning

// CPython: Objects/bytesobject.c:1623 single_byte_interning
static PyBytesObject *characters[256];

PyObject *
PyBytes_FromStringAndSize(const char *str, Py_ssize_t size)
{
    if (size == 0 && (op = nullstring) != NULL) {
        return Py_NewRef(op);
    }
    if (size == 1 && str != NULL) {
        op = characters[(unsigned char)str[0]];
        if (op != NULL) return Py_NewRef(op);
    }
    ...
}

A 256-element static array caches all single-byte bytes objects. b'\x00' through b'\xff' are singletons allocated at startup.

PyBytes_Concat

// CPython: Objects/bytesobject.c:1050 PyBytes_Concat
void
PyBytes_Concat(PyObject **pv, PyObject *w)
{
    PyObject *v = *pv;
    if (Py_SIZE(v) > PY_SSIZE_T_MAX - Py_SIZE(w)) {
        PyErr_NoMemory();
        goto error;
    }
    PyObject *sv = _PyBytes_Resize(&v, Py_SIZE(v) + Py_SIZE(w));
    ...
    memcpy((void *)(((PyBytesObject *)sv)->ob_val + Py_SIZE(v)),
           ((PyBytesObject *)w)->ob_val, Py_SIZE(w));
    *pv = sv;
    return;
error:
    Py_CLEAR(*pv);
}

_PyBytes_Resize tries to resize in-place (if refcount == 1); otherwise allocates a new object and copies.

bytes_decode

Calls PyUnicode_Decode(buf, len, encoding, errors), which dispatches to the codec registry for the given encoding name.

bytes as a sequence

bytes implements sq_length, sq_item, sq_ass_item, sq_contains. b'sub' in b'string' uses _PyBytes_Find (Boyer-Moore-Horspool search for long strings, naive for short).

gopy notes

The gopy equivalent is objects/str.go (which handles both str and bytes). The interning of single bytes maps to a [256]*Bytes array in the package. bytes.decode() calls the codec registry. in operator maps to Go's bytes.Contains.