TON DNS 解析器

介绍

TON DNS 是一个强大的工具。它不仅允许将 TON 网站/存储包分配给域名，还可以设置子域名解析。

相关链接

1. TON 智能合约地址系统
2. TEP-0081 - TON DNS 标准
3. .ton DNS 集合的源代码
4. .t.me DNS 集合的源代码
5. 域名合约搜索器
6. 简单子域名管理器代码

域名合约搜索器

子域名具有实际用途。例如，区块链浏览器目前没有提供通过名称查找域名合约的方法。让我们探索如何创建一个合约，提供查找这类域名的机会。

信息

This contract is deployed at EQDkAbAZNb4uk-6pzTPDO2s0tXZweN-2R08T2Wy6Z3qzH_Zp and linked to resolve-contract.ton. To test it, you may write <your-domain.ton>.resolve-contract.ton in the address bar of your favourite TON explorer and get to the page of TON DNS domain contract. Subdomains and .t.me domains are supported as well.

您可以尝试通过访问 resolve-contract.ton.resolve-contract.ton 来查看解析器代码。不幸的是，这将不会显示子解析器（那是不同的智能合约），您将看到域名合约本身的页面。

dnsresolve() 代码

部分重复部分已省略。

(int, cell) dnsresolve(slice subdomain, int category) method_id {
  int subdomain_bits = slice_bits(subdomain);
  throw_unless(70, (subdomain_bits % 8) == 0);
  
  int starts_with_zero_byte = subdomain.preload_int(8) == 0;  ;; assuming that 'subdomain' is not empty
  if (starts_with_zero_byte) {
    subdomain~load_uint(8);
    if (subdomain.slice_bits() == 0) {   ;; current contract has no DNS records by itself
      return (8, null());
    }
  }
  
  ;; we are loading some subdomain
  ;; supported subdomains are "ton\0", "me\0t\0" and "address\0"
  
  slice subdomain_sfx = null();
  builder domain_nft_address = null();
  
  if (subdomain.starts_with("746F6E00"s)) {
    ;; we're resolving
    ;; "ton" \0 <subdomain> \0 [subdomain_sfx]
    subdomain~skip_bits(32);
    
    ;; reading domain name
    subdomain_sfx = subdomain;
    while (subdomain_sfx~load_uint(8)) { }
    
    subdomain~skip_last_bits(8 + slice_bits(subdomain_sfx));
    
    domain_nft_address = get_ton_dns_nft_address_by_index(slice_hash(subdomain));
  } elseif (subdomain.starts_with("6164647265737300"s)) {
    subdomain~skip_bits(64);
    
    domain_nft_address = subdomain~decode_base64_address_to(begin_cell());
    
    subdomain_sfx = subdomain;
    if (~ subdomain_sfx.slice_empty?()) {
      throw_unless(71, subdomain_sfx~load_uint(8) == 0);
    }
  } else {
    return (0, null());
  }
  
  if (slice_empty?(subdomain_sfx)) {
    ;; example of domain being resolved:
    ;; [initial, not accessible in this contract] "ton\0resolve-contract\0ton\0ratelance\0"
    ;; [what is accessible by this contract]      "ton\0ratelance\0"
    ;; subdomain          "ratelance"
    ;; subdomain_sfx      ""
    
    ;; we want the resolve result to point at contract of 'ratelance.ton', not its owner
    ;; so we must answer that resolution is complete + "wallet"H is address of 'ratelance.ton' contract
    
    ;; dns_smc_address#9fd3 smc_addr:MsgAddressInt flags:(## 8) { flags <= 1 } cap_list:flags . 0?SmcCapList = DNSRecord;
    ;; _ (HashmapE 256 ^DNSRecord) = DNS_RecordSet;
    
    cell wallet_record = begin_cell().store_uint(0x9fd3, 16).store_builder(domain_nft_address).store_uint(0, 8).end_cell();
    
    if (category == 0) {
      cell dns_dict = new_dict();
      dns_dict~udict_set_ref(256, "wallet"H, wallet_record);
      return (subdomain_bits, dns_dict);
    } elseif (category == "wallet"H) {
      return (subdomain_bits, wallet_record);
    } else {
      return (subdomain_bits, null());
    }
  } else {
    ;; subdomain          "resolve-contract"
    ;; subdomain_sfx      "ton\0ratelance\0"
    ;; we want to pass \0 further, so that next resolver has opportunity to process only one byte
    
    ;; next resolver is contract of 'resolve-contract<.ton>'
    ;; dns_next_resolver#ba93 resolver:MsgAddressInt = DNSRecord;
    cell resolver_record = begin_cell().store_uint(0xba93, 16).store_builder(domain_nft_address).end_cell();
    return (subdomain_bits - slice_bits(subdomain_sfx) - 8, resolver_record);
  }
}

dnsresolve() 解释

• 用户请求 "stabletimer.ton.resolve-contract.ton"。
• 应用程序将其转换为 "\0ton\0resolve-contract\0ton\0stabletimer\0"（第一个零字节是可选的）。
• 根 DNS 解析器将请求定向到 TON DNS 集合，剩余部分为 "\0resolve-contract\0ton\0stabletimer\0"。
• TON DNS 集合将请求委托给特定域名，留下 "\0ton\0stabletimer\0"。
• .TON DNS 域名合约将解析传递给编辑器指定的子解析器，子域名为 "ton\0stabletimer\0"。

这是 dnsresolve() 被调用的点。 分步解释其工作方式：

1. 它将子域名和类别作为输入。
2. 如果开头有零字节，则跳过。
3. 检查子域名是否以 "ton\0" 开头。如果是，
   1. 跳过前32位（子域名 = "resolve-contract\0"）
   2. 设置 subdomain_sfx 的值为 subdomain，并读取直到零字节的字节
   3. （子域名 = "resolve-contract\0"，subdomain_sfx = ""）
   4. 从子域名切片的末尾裁剪零字节和 subdomain_sfx（子域名 = "resolve-contract"）
   5. 使用 slice_hash 和 get_ton_dns_nft_address_by_index 函数将域名转换为合约地址。您可以在 [[Subresolvers#Appendix 1. resolve-contract.ton 的代码|附录 1]] 中看到它们。
4. 否则，dnsresolve() 检查子域名是否以 "address\0" 开头。如果是，它跳过该前缀并读取 base64 地址。
5. 如果提供的用于解析的子域名与这些前缀都不匹配，函数通过返回 (0, null())（零字节前缀解析无 DNS 条目）表示失败。
6. 然后检查子域名后缀是否为空。空后缀表示请求已完全满足。如果后缀为空：
   1. dnsresolve() 为域名的 "wallet" 子部分创建一个 DNS 记录，使用它检索到的 TON 域名合约地址。
   2. 如果请求类别 0（所有 DNS 条目），则将记录包装在字典中并返回。
   3. 如果请求类别为 "wallet"H，则按原样返回记录。
   4. 否则，指定类别没有 DNS 条目，因此函数表示解析成功但未找到任何结果。
7. 如果后缀不为空：
   1. 之前获得的合约地址用作下一个解析器。函数构建指向它的下一个解析器记录。
   2. "\0ton\0stabletimer\0" 被传递给该合约：处理的位是子域名的位。

总结来说，dnsresolve() 要么：

• 将子域名完全解析为 DNS 记录
• 部分解析为解析器记录，以将解析传递给另一个合约
• 为未知子域名返回“未找到域名”的结果

危险

实际上，base64 地址解析不起作用：如果您尝试输入 <some-address>.address.resolve-contract.ton，您将收到一个错误，表明域名配置错误或不存在。原因是域名不区分大小写（从真实 DNS 继承的功能），因此会转换为小写，将您带到不存在的工作链的某个地址。

绑定解析器

现在子解析器合约已部署，我们需要将域名指向它，即更改域名的 dns_next_resolver 记录。我们可以通过将以下 TL-B 结构的消息发送到域名合约来实现。

1. change_dns_record#4eb1f0f9 query_id:uint64 record_key#19f02441ee588fdb26ee24b2568dd035c3c9206e11ab979be62e55558a1d17ff record:^[dns_next_resolver#ba93 resolver:MsgAddressInt]

创建自己的子域名管理器

子域名对普通用户来说可能有用 - 例如，将几个项目链接到单个域名，或链接到朋友的钱包。

合约数据

我们需要在合约数据中存储所有者的地址和 域名->记录哈希->记录值 字典。

global slice owner;
global cell domains;

() load_data() impure {
  slice ds = get_data().begin_parse();
  owner = ds~load_msg_addr();
  domains = ds~load_dict();
}
() save_data() impure {
  set_data(begin_cell().store_slice(owner).store_dict(domains).end_cell());
}

处理记录更新

const int op::update_record = 0x537a3491;
;; op::update_record#537a3491 domain_name:^Cell record_key:uint256
;;     value:(Maybe ^Cell) = InMsgBody;

() recv_internal(cell in_msg, slice in_msg_body) {
  if (in_msg_body.slice_empty?()) { return (); }   ;; simple money transfer

  slice in_msg_full = in_msg.begin_parse();
  if (in_msg_full~load_uint(4) & 1) { return (); } ;; bounced message

  slice sender = in_msg_full~load_msg_addr();
  load_data();
  throw_unless(501, equal_slices(sender, owner));
  
  int op = in_msg_body~load_uint(32);
  if (op == op::update_record) {
    slice domain = in_msg_body~load_ref().begin_parse();
    (cell records, _) = domains.udict_get_ref?(256, string_hash(domain));

    int key = in_msg_body~load_uint(256);
    throw_if(502, key == 0);  ;; cannot update "all records" record

    if (in_msg_body~load_uint(1) == 1) {
      cell value = in_msg_body~load_ref();
      records~udict_set_ref(256, key, value);
    } else {
      records~udict_delete?(256, key);
    }

    domains~udict_set_ref(256, string_hash(domain), records);
    save_data();
  }
}

我们检查传入消息是否包含某些请求，不是弹回的，来自所有者，且请求为 op::update_record。

然后，我们从消息中加载域名。我们不能将域名按原样存储在字典中：它们可能有不同的长度，但 TVM 非前缀字典只能包含等长的键。因此，我们计算 string_hash(domain) - 域名的 SHA-256；域名保证有整数个八位字节，因此这是有效的。

之后，我们为指定域名更新记录，并将新数据保存到合约存储中。

解析域名

(slice, slice) ~parse_sd(slice subdomain) {
  ;; "test\0qwerty\0" -> "test" "qwerty\0"
  slice subdomain_sfx = subdomain;
  while (subdomain_sfx~load_uint(8)) { }  ;; searching zero byte
  subdomain~skip_last_bits(slice_bits(subdomain_sfx));
  return (subdomain, subdomain_sfx);
}

(int, cell) dnsresolve(slice subdomain, int category) method_id {
  int subdomain_bits = slice_bits(subdomain);
  throw_unless(70, subdomain_bits % 8 == 0);
  if (subdomain.preload_uint(8) == 0) { subdomain~skip_bits(8); }
  
  slice subdomain_suffix = subdomain~parse_sd();  ;; "test\0" -> "test" ""
  int subdomain_suffix_bits = slice_bits(subdomain_suffix);

  load_data();
  (cell records, _) = domains.udict_get_ref?(256, string_hash(subdomain));

  if (subdomain_suffix_bits > 0) { ;; more than "<SUBDOMAIN>\0" requested
    category = "dns_next_resolver"H;
  }

  int resolved = subdomain_bits - subdomain_suffix_bits;

  if (category == 0) { ;; all categories are requested
    return (resolved, records);
  }

  (cell value, int found) = records.udict_get_ref?(256, category);
  return (resolved, value);
}

dnsresolve 函数检查请求的子域名是否包含整数个八位字节，跳过子域名切片开头的可选零字节，然后将其分割为最高级别的域和其他部分（test\0qwerty\0 被分割为 test 和 qwerty\0）。加载与请求的域名对应的记录字典。

如果存在非空子域名后缀，函数返回已解析的字节数和在 "dns_next_resolver"H 键下找到的下一个解析器记录。否则，函数返回已解析的字节数（即整个切片长度）和请求的记录。

可以通过更优雅地处理错误来改进此函数，但这不是绝对必需的。

附录 1. resolve-contract.ton 的代码

Details

(builder, ()) ~store_slice(builder to, slice s) asm "STSLICER";
int starts_with(slice a, slice b) asm "SDPFXREV";

const slice ton_dns_minter = "EQC3dNlesgVD8YbAazcauIrXBPfiVhMMr5YYk2in0Mtsz0Bz"a;
cell ton_dns_domain_code() asm """
  B{<TON DNS NFT code in HEX format>}
  B>boc
  PUSHREF
""";

const slice tme_minter = "EQCA14o1-VWhS2efqoh_9M1b_A9DtKTuoqfmkn83AbJzwnPi"a;
cell tme_domain_code() asm """
  B{<T.ME NFT code in HEX format>}
  B>boc
  PUSHREF
""";

cell calculate_ton_dns_nft_item_state_init(int item_index) inline {
  cell data = begin_cell().store_uint(item_index, 256).store_slice(ton_dns_minter).end_cell();
  return begin_cell().store_uint(0, 2).store_dict(ton_dns_domain_code()).store_dict(data).store_uint(0, 1).end_cell();
}

cell calculate_tme_nft_item_state_init(int item_index) inline {
  cell config = begin_cell().store_uint(item_index, 256).store_slice(tme_minter).end_cell();
  cell data = begin_cell().store_ref(config).store_maybe_ref(null()).end_cell();
  return begin_cell().store_uint(0, 2).store_dict(tme_domain_code()).store_dict(data).store_uint(0, 1).end_cell();
}

builder calculate_nft_item_address(int wc, cell state_init) inline {
  return begin_cell()
      .store_uint(4, 3)
      .store_int(wc, 8)
      .store_uint(cell_hash(state_init), 256);
}

builder get_ton_dns_nft_address_by_index(int index) inline {
  cell state_init = calculate_ton_dns_nft_item_state_init(index);
  return calculate_nft_item_address(0, state_init);
}

builder get_tme_nft_address_by_index(int index) inline {
  cell state_init = calculate_tme_nft_item_state_init(index);
  return calculate_nft_item_address(0, state_init);
}

(slice, builder) decode_base64_address_to(slice readable, builder target) inline {
  builder addr_with_flags = begin_cell();
  repeat(48) {
    int char = readable~load_uint(8);
    if (char >= "a"u) {
      addr_with_flags~store_uint(char - "a"u + 26, 6);
    } elseif ((char == "_"u) | (char == "/"u)) {
      addr_with_flags~store_uint(63, 6);
    } elseif (char >= "A"u) {
      addr_with_flags~store_uint(char - "A"u, 6);
    } elseif (char >= "0"u) {
      addr_with_flags~store_uint(char - "0"u + 52, 6);
    } else {
      addr_with_flags~store_uint(62, 6);
    }
  }
  
  slice addr_with_flags = addr_with_flags.end_cell().begin_parse();
  addr_with_flags~skip_bits(8);
  addr_with_flags~skip_last_bits(16);
  
  target~store_uint(4, 3);
  target~store_slice(addr_with_flags);
  return (readable, target);
}

slice decode_base64_address(slice readable) method_id {
  (slice _remaining, builder addr) = decode_base64_address_to(readable, begin_cell());
  return addr.end_cell().begin_parse();
}

(int, cell) dnsresolve(slice subdomain, int category) method_id {
  int subdomain_bits = slice_bits(subdomain);

  throw_unless(70, (subdomain_bits % 8) == 0);
  
  int starts_with_zero_byte = subdomain.preload_int(8) == 0;  ;; assuming that 'subdomain' is not empty
  if (starts_with_zero_byte) {
    subdomain~load_uint(8);
    if (subdomain.slice_bits() == 0) {   ;; current contract has no DNS records by itself
      return (8, null());
    }
  }
  
  ;; we are loading some subdomain
  ;; supported subdomains are "ton\0", "me\0t\0" and "address\0"
  
  slice subdomain_sfx = null();
  builder domain_nft_address = null();
  
  if (subdomain.starts_with("746F6E00"s)) {
    ;; we're resolving
    ;; "ton" \0 <subdomain> \0 [subdomain_sfx]
    subdomain~skip_bits(32);
    
    ;; reading domain name
    subdomain_sfx = subdomain;
    while (subdomain_sfx~load_uint(8)) { }
    
    subdomain~skip_last_bits(8 + slice_bits(subdomain_sfx));
    
    domain_nft_address = get_ton_dns_nft_address_by_index(slice_hash(subdomain));
  } elseif (subdomain.starts_with("6D65007400"s)) {
    ;; "t" \0 "me" \0 <subdomain> \0 [subdomain_sfx]
    subdomain~skip_bits(40);
    
    ;; reading domain name
    subdomain_sfx = subdomain;
    while (subdomain_sfx~load_uint(8)) { }
    
    subdomain~skip_last_bits(8 + slice_bits(subdomain_sfx));
    
    domain_nft_address = get_tme_nft_address_by_index(string_hash(subdomain));
  } elseif (subdomain.starts_with("6164647265737300"s)) {
    subdomain~skip_bits(64);
    
    domain_nft_address = subdomain~decode_base64_address_to(begin_cell());
    
    subdomain_sfx = subdomain;
    if (~ subdomain_sfx.slice_empty?()) {
      throw_unless(71, subdomain_sfx~load_uint(8) == 0);
    }
  } else {
    return (0, null());
  }
  
  if (slice_empty?(subdomain_sfx)) {
    ;; example of domain being resolved:
    ;; [initial, not accessible in this contract] "ton\0resolve-contract\0ton\0ratelance\0"
    ;; [what is accessible by this contract]      "ton\0ratelance\0"
    ;; subdomain          "ratelance"
    ;; subdomain_sfx      ""
    
    ;; we want the resolve result to point at contract of 'ratelance.ton', not its owner
    ;; so we must answer that resolution is complete + "wallet"H is address of 'ratelance.ton' contract
    
    ;; dns_smc_address#9fd3 smc_addr:MsgAddressInt flags:(## 8) { flags <= 1 } cap_list:flags . 0?SmcCapList = DNSRecord;
    ;; _ (HashmapE 256 ^DNSRecord) = DNS_RecordSet;
    
    cell wallet_record = begin_cell().store_uint(0x9fd3, 16).store_builder(domain_nft_address).store_uint(0, 8).end_cell();
    
    if (category == 0) {
      cell dns_dict = new_dict();
      dns_dict~udict_set_ref(256, "wallet"H, wallet_record);
      return (subdomain_bits, dns_dict);
    } elseif (category == "wallet"H) {
      return (subdomain_bits, wallet_record);
    } else {
      return (subdomain_bits, null());
    }
  } else {
    ;; example of domain being resolved:
    ;; [initial, not accessible in this contract] "ton\0resolve-contract\0ton\0resolve-contract\0ton\0ratelance\0"
    ;; [what is accessible by this contract]      "ton\0resolve-contract\0ton\0ratelance\0"
    ;; subdomain          "resolve-contract"
    ;; subdomain_sfx      "ton\0ratelance\0"
    ;; and we want to pass \0 further, so that next resolver has opportunity to process only one byte
    
    ;; next resolver is contract of 'resolve-contract<.ton>'
    ;; dns_next_resolver#ba93 resolver:MsgAddressInt = DNSRecord;
    cell resolver_record = begin_cell().store_uint(0xba93, 16).store_builder(domain_nft_address).end_cell();
    return (subdomain_bits - slice_bits(subdomain_sfx) - 8, resolver_record);
  }
}

() recv_internal() {
  return ();
}