Modules/_json.c (part 2)

Source:

cpython 3.14 @ ab2d84fe1023/Modules/_json.c ↗

This annotation covers the C encoder. See modules_json_detail for the scanner (decoder), JSONDecoder, and loads.

Map

Lines	Symbol	Role
1-80	py_encode_basestring	Escape a Python str for JSON encoding
81-200	py_encode_basestring_ascii	ASCII-safe variant (non-ASCII → \uXXXX)
201-350	encoder_encode_float	Handle nan / inf and format floats
351-500	encoder_encode_string	Top-level string encoding with cache
501-700	encoder_listencode_obj	Recursively encode a Python object
701-850	encoder_listencode_list	Encode a list/tuple to [...]
851-1000	encoder_listencode_dict	Encode a dict to {...}, check key types

Reading

py_encode_basestring_ascii

// CPython: Modules/_json.c:120 py_encode_basestring_ascii
static PyObject *
py_encode_basestring_ascii(PyObject *pystr)
{
    /* Escape control characters, backslash, double-quote.
       Encode non-ASCII as \uXXXX (or \uXXXX\uXXXX surrogates for BMP). */
    Py_UCS4 c;
    while (input_chars--) {
        c = *input++;
        if (c == '"')  { *output++ = '\\'; *output++ = '"';  continue; }
        if (c == '\\') { *output++ = '\\'; *output++ = '\\'; continue; }
        if (c < 0x20)  { /* \b \f \n \r \t or \uXXXX */ ... continue; }
        if (c > 0x7F)  { /* \uXXXX or surrogate pair */ ... continue; }
        *output++ = (char)c;
    }
}

encoder_encode_float

// CPython: Modules/_json.c:260 encoder_encode_float
static PyObject *
encoder_encode_float(PyEncoderObject *s, PyObject *obj)
{
    double d = PyFloat_AS_DOUBLE(obj);
    if (Py_IS_NAN(d)) {
        if (!s->allow_nan) {
            PyErr_SetString(PyExc_ValueError,
                "Out of range float values are not JSON compliant");
            return NULL;
        }
        return PyUnicode_FromString("NaN");
    }
    if (Py_IS_INFINITY(d))
        return PyUnicode_FromString(d > 0 ? "Infinity" : "-Infinity");
    return PyFloat_Type.tp_repr(obj);
}

JSON does not have NaN or Infinity literals. allow_nan=True (the default in Python's json) encodes them as bare NaN/Infinity which is technically non-compliant.

encoder_listencode_obj

// CPython: Modules/_json.c:520 encoder_listencode_obj
static int
encoder_listencode_obj(PyEncoderObject *s, _PyUnicodeWriter *writer,
                       PyObject *obj, Py_ssize_t indent_level)
{
    /* Dispatch by type */
    if (obj == Py_None)           return encoder_encode_null(s, writer);
    if (obj == Py_True)           return encoder_encode_true(s, writer);
    if (obj == Py_False)          return encoder_encode_false(s, writer);
    if (PyUnicode_Check(obj))     return encoder_encode_string(s, writer, obj);
    if (PyLong_Check(obj))        return encoder_encode_long(s, writer, obj);
    if (PyFloat_Check(obj))       return encoder_encode_float(s, writer, obj);
    if (PyList_Check(obj) || PyTuple_Check(obj))
                                  return encoder_listencode_list(s, writer, obj, indent_level);
    if (PyDict_Check(obj))        return encoder_listencode_dict(s, writer, obj, indent_level);
    /* default: call s->default(obj) */
    newobj = PyObject_CallOneArg(s->defaultfn, obj);
    ...
}

Circular reference detection

// CPython: Modules/_json.c:580 encoder_listencode_dict
/* Before encoding a container, add its id to check_circular set.
   If id is already in the set, raise ValueError("Circular reference"). */
if (s->check_circular) {
    ident = PyLong_FromVoidPtr(dct);
    if (PySet_Contains(markers, ident)) {
        PyErr_SetString(PyExc_ValueError, "Circular reference detected");
        goto bail;
    }
    PySet_Add(markers, ident);
}
/* ... encode dict ... */
PySet_Discard(markers, ident);

gopy notes

_json C accelerator is in module/json/encoder.go as a pure Go implementation. py_encode_basestring_ascii is json.EncodeASCII. Circular reference detection uses a map[uintptr]bool (Go id set). The default hook calls a Python callable via vm.CallOne.