Modules/binascii.c (part 2)

Source:

cpython 3.14 @ ab2d84fe1023/Modules/binascii.c ↗

This annotation covers hex encoding and quoted-printable. See modules_binascii_detail for b2a_base64, a2b_base64, crc32, crc_hqx, and the Error exception.

Map

Lines	Symbol	Role
1-80	b2a_hex / hexlify	Convert bytes to hex ASCII
81-180	a2b_hex / unhexlify	Decode hex ASCII back to bytes
181-320	b2a_uu	UU-encode a line (legacy Unix-to-Unix encoding)
321-460	a2b_uu	Decode a UU-encoded line
461-620	b2a_qp	Encode bytes as quoted-printable (email)
621-800	a2b_qp	Decode quoted-printable back to bytes

Reading

b2a_hex

// CPython: Modules/binascii.c:680 binascii_b2a_hex_impl
static PyObject *
binascii_b2a_hex_impl(PyObject *module, Py_buffer *data, PyObject *sep,
                      int bytes_per_sep)
{
    /* Convert each byte to two hex characters. */
    const unsigned char *pin = data->buf;
    PyObject *rv = PyBytes_FromStringAndSize(NULL, data->len * 2);
    unsigned char *pout = (unsigned char *)PyBytes_AS_STRING(rv);
    for (Py_ssize_t i = 0; i < data->len; i++) {
        unsigned char c = pin[i];
        pout[i*2]   = Py_hexdigits[c >> 4];
        pout[i*2+1] = Py_hexdigits[c & 0x0f];
    }
    return rv;
}

binascii.hexlify(b'\xde\xad') returns b'dead'. The optional sep parameter (Python 3.8+) inserts a separator every bytes_per_sep bytes: hexlify(b'\xde\xad', sep='-') returns b'de-ad'.

a2b_hex

// CPython: Modules/binascii.c:740 binascii_a2b_hex_impl
static PyObject *
binascii_a2b_hex_impl(PyObject *module, Py_buffer *hexstr)
{
    const unsigned char *argbuf = hexstr->buf;
    Py_ssize_t arglen = hexstr->len;
    if (arglen % 2 != 0) {
        PyErr_SetString(Error, "Odd-length string");
        return NULL;
    }
    PyObject *rv = PyBytes_FromStringAndSize(NULL, arglen / 2);
    unsigned char *p = (unsigned char *)PyBytes_AS_STRING(rv);
    for (Py_ssize_t i = 0; i < arglen; i += 2) {
        int top = _PyLong_DigitValue[argbuf[i]];
        int bot = _PyLong_DigitValue[argbuf[i+1]];
        if (top >= 16 || bot >= 16) {
            PyErr_SetString(Error, "Non-hexadecimal digit found");
            ...
        }
        *p++ = (top << 4) + bot;
    }
    return rv;
}

binascii.unhexlify('dead') returns b'\xde\xad'. _PyLong_DigitValue is a 256-entry lookup table mapping ASCII characters to their digit values (255 for non-hex characters).

b2a_qp

// CPython: Modules/binascii.c:1020 binascii_b2a_qp_impl
/* Quoted-Printable encoding (RFC 2045):
   - Non-printable bytes and '=' are encoded as =XX (uppercase hex)
   - Lines are wrapped at 76 characters with '=\r\n' soft line breaks
   - Tabs and spaces at end of line are encoded
   - istext=True: treat '\r\n' and '\n' as line endings (not encoded)
*/

Quoted-Printable is used for email bodies. It preserves ASCII text readably while encoding binary data as =XX sequences. quotetabs=True also encodes horizontal whitespace.

a2b_qp

// CPython: Modules/binascii.c:1120 binascii_a2b_qp_impl
/* Decode quoted-printable:
   - '=XX' -> single byte with value 0xXX
   - '=\n' or '=\r\n' -> soft line break (deleted)
   - Everything else passes through unchanged
*/

gopy notes

binascii.hexlify/unhexlify are in module/binascii/module.go. b2a_hex uses fmt.Sprintf or a lookup table. a2b_hex uses hex.DecodeString. b2a_qp/a2b_qp are Go ports of the RFC 2045 state machines.