Modules/binascii.c (part 7)

Source:

cpython 3.14 @ ab2d84fe1023/Modules/binascii.c ↗

This annotation covers binary-to-ASCII encoding. See modules_binascii6_detail for b2a_base64, a2b_base64, b2a_hex, a2b_hex, and the encode table.

Map

Lines	Symbol	Role
1-80	b2a_base64	Encode bytes to base64 with optional newline
81-160	a2b_base64	Decode base64 to bytes; skip whitespace
161-240	b2a_hex / a2b_hex	Hexadecimal encoding/decoding
241-340	b2a_uu / a2b_uu	Unix-to-Unix encoding (uuencode)
341-600	crc_hqx / crc32	CRC checksums

Reading

b2a_base64

// CPython: Modules/binascii.c:640 binascii_b2a_base64_impl
static PyObject *
binascii_b2a_base64_impl(PyObject *module, Py_buffer *data, int newline)
{
    Py_ssize_t datalength = data->len;
    /* Output length: ceil(n/3)*4 + newline */
    Py_ssize_t ngroups = (datalength + 2) / 3;
    Py_ssize_t out_len = ngroups * 4 + newline;
    PyObject *rv = PyBytes_FromStringAndSize(NULL, out_len);
    unsigned char *p = (unsigned char *)PyBytes_AS_STRING(rv);
    unsigned char *bin_data = data->buf;
    for (Py_ssize_t i = 0; i < ngroups; i++) {
        /* Encode 3 bytes -> 4 base64 characters */
        unsigned int t = bin_data[0] << 16 | bin_data[1] << 8 | bin_data[2];
        *p++ = table_b2a_base64[(t >> 18) & 0x3f];
        *p++ = table_b2a_base64[(t >> 12) & 0x3f];
        *p++ = table_b2a_base64[(t >>  6) & 0x3f];
        *p++ = table_b2a_base64[(t      ) & 0x3f];
        bin_data += 3;
    }
    if (newline) *p++ = '\n';
    return rv;
}

Base64 encodes 3 input bytes as 4 ASCII characters. The lookup table table_b2a_base64 maps 6-bit values to A-Z, a-z, 0-9, +, /. Padding = is added when len(data) % 3 != 0.

a2b_hex

// CPython: Modules/binascii.c:340 binascii_a2b_hex_impl
static PyObject *
binascii_a2b_hex_impl(PyObject *module, Py_buffer *hexstr)
{
    if (hexstr->len % 2 != 0) {
        PyErr_SetString(PyExc_ValueError, "Odd-length string");
        return NULL;
    }
    PyObject *rv = PyBytes_FromStringAndSize(NULL, hexstr->len / 2);
    unsigned char *p = (unsigned char *)PyBytes_AS_STRING(rv);
    const unsigned char *s = hexstr->buf;
    for (Py_ssize_t i = 0; i < hexstr->len; i += 2) {
        int top = table_hex[s[i]], bot = table_hex[s[i+1]];
        if (top == -1 || bot == -1) {
            PyErr_SetString(PyExc_ValueError, "Non-hexadecimal digit found");
            return NULL;
        }
        *p++ = (top << 4) | bot;
    }
    return rv;
}

binascii.unhexlify('deadbeef') converts hex pairs to bytes. table_hex maps ASCII hex digits to 4-bit values; invalid characters map to -1.

gopy notes

b2a_base64 is module/binascii.B2ABase64 in module/binascii/module.go. It uses Go's encoding/base64. a2b_hex uses encoding/hex.DecodeString. crc_hqx is a polynomial CRC table lookup; crc32 uses hash/crc32.ChecksumIEEE.