Function scanstring_unicode

simplejson/_speedups.c:1483–1669

#if PY_VERSION_HEX >= 0x030E0000
static PyObject *
scanstring_unicode(_speedups_state *state, PyObject *pystr, Py_ssize_t end,
                   int strict, Py_ssize_t *next_end_ptr)
{
    /* Python 3.14+: use PyUnicodeWriter instead of a chunks list.
     * The writer is lazily created on the first escape sequence so that
     * the common no-escape path returns a cheap PyUnicode_Substring. */
    PyObject *rval;
    Py_ssize_t begin = end - 1;
    Py_ssize_t next = begin;
    int kind = PyUnicode_KIND(pystr);
    Py_ssize_t len = PyUnicode_GET_LENGTH(pystr);
    void *buf = PyUnicode_DATA(pystr);
    PyUnicodeWriter *writer = NULL;
    Py_ssize_t literal_start;

    if (len == end) {
        raise_errmsg(state, ERR_STRING_UNTERMINATED, pystr, begin);
        goto bail;
    }
    else if (end < 0 || len < end) {
        /* Out-of-range end: match py_scanstring, which raises
         * JSONDecodeError("Unterminated string starting at") so that
         * user code using `except JSONDecodeError` catches the C path
         * the same way it catches the pure-Python path. */
        raise_errmsg(state, ERR_STRING_UNTERMINATED, pystr, begin);
        goto bail;
    }

    literal_start = end;
    while (1) {
        /* Find the end of the string or the next escape */
        JSON_UNICHR c = 0;
        for (next = end; next < len; next++) {
            c = PyUnicode_READ(kind, buf, next);
            if (c == '"' || c == '\\') {
                break;
            }
            else if (strict && c <= 0x1f) {
                raise_errmsg(state, ERR_STRING_CONTROL, pystr, next);
                goto bail;
            }
        }
        if (!(c == '"' || c == '\\')) {
            raise_errmsg(state, ERR_STRING_UNTERMINATED, pystr, begin);
            goto bail;
        }
        next++;
        if (c == '"') {
            end = next;
            break;
        }
        /* Backslash escape — ensure writer exists and flush the
         * literal span [literal_start, next-1). */
        if (writer == NULL) {
            writer = PyUnicodeWriter_Create(len - begin);
            if (writer == NULL)
                goto bail;