Function export_escaped_unquoted_to_utf8_buffer

python/stringzilla.c:6581–6634

* @brief Exports a string to a UTF-8 buffer, escaping single quotes. * @param[in] cstr The input string to export. * @param[in] cstr_length The length of the input string. * @param[out] buffer The output buffer to write to. * @param[in] buffer_length The size of the output buffer. * @param[out] did_fit Populated with 1 if the string is fully exported, 0 if it didn't fit, -1 if invalid U

 *  @return Pointer to the end of the written data in the buffer, or buffer position where error occurred.
 */
sz_cptr_t export_escaped_unquoted_to_utf8_buffer(sz_cptr_t cstr, sz_size_t cstr_length,    //
                                                 sz_ptr_t buffer, sz_size_t buffer_length, //
                                                 int *did_fit) {
    sz_cptr_t const cstr_end = cstr + cstr_length;
    sz_ptr_t buffer_ptr = buffer;
    *did_fit = 1;

    // Validate UTF-8 first
    if (!sz_utf8_valid(cstr, cstr_length)) {
        *did_fit = -1; // Signal UTF-8 error
        return buffer_ptr;
    }

    // First pass: calculate required buffer size (input already validated)
    sz_size_t required_bytes = 2; // Opening and closing quotes
    sz_cptr_t scan_ptr = cstr;
    while (scan_ptr < cstr_end) {
        sz_rune_t rune;
        sz_rune_length_t rune_length;
        sz_rune_parse(scan_ptr, &rune, &rune_length);

        if (rune_length == 1 && *scan_ptr == '\'') { required_bytes += 2; } // Escaped quote: \'
        else { required_bytes += rune_length; }                             // Normal rune
        scan_ptr += rune_length;
    }

    // Check if we have enough buffer space
    if (required_bytes > buffer_length) {
        *did_fit = 0;
        return buffer_ptr;
    }

    // Second pass: actually write to buffer
    *(buffer_ptr++) = '\''; // Opening quote

    while (cstr < cstr_end) {
        sz_rune_t rune;
        sz_rune_length_t rune_length;
        sz_rune_parse(cstr, &rune, &rune_length);

        if (rune_length == 1 && *cstr == '\'') {
            *(buffer_ptr++) = '\\';
            *(buffer_ptr++) = '\'';
        }
        else {
            sz_copy(buffer_ptr, cstr, rune_length);
            buffer_ptr += rune_length;
        }
        cstr += rune_length;
    }

    *(buffer_ptr++) = '\''; // Closing quote
    return buffer_ptr;
}

/**
 *  @brief Exports a binary string to a buffer in Python bytes representation (b'\\x..').