Domain.cpp

   1 #include "Domain.hpp"
   2
   3 #include "IP.hpp"
   4 #include "IP4.hpp"
   5 #include "IP6.hpp"
   6
   7 #include <algorithm>
   8
   9 #include <idn2.h>
  10 #include <uninorm.h>
  11
  12 #include <glog/logging.h>
  13
  14 #include <stdexcept>
  15
  16 namespace {
  17 size_t constexpr max_length = 255;
  18 }
  19
  20 // Normalization Form KC (NFKC) Compatibility Decomposition, followed
  21 // by Canonical Composition, see <http://unicode.org/reports/tr15/>
  22
  23 std::string nfkc(std::string_view str)
  24 {
  25   size_t length = max_length;
  26   char   bfr[max_length];
  27   CHECK_LE(str.length(), max_length);
  28   auto udata = reinterpret_cast<uint8_t const*>(str.data());
  29   auto ubfr  = reinterpret_cast<uint8_t*>(bfr);
  30   CHECK_NOTNULL(u8_normalize(UNINORM_NFKC, udata, str.size(), ubfr, &length));
  31   return std::string{bfr, length};
  32 }
  33
  34 bool Domain::validate(std::string_view dom)
  35 {
  36   if (dom.length() > max_length) {
  37     return false;
  38   }
  39
  40   // Handle "bare" IP addresses, without the brackets.
  41   if (IP::is_address(dom)) {
  42     return true;
  43   }
  44
  45   if (IP::is_address_literal(dom)) {
  46     return true;
  47   }
  48
  49   dom = remove_trailing_dot(dom);
  50
  51   auto const norm = nfkc(dom);
  52
  53   // idn2_to_ascii_8z() converts (ASCII) to lower case
  54
  55   char* ptr  = nullptr;
  56   auto  code = idn2_to_ascii_8z(norm.c_str(), &ptr, IDN2_TRANSITIONAL);
  57   if (code != IDN2_OK)
  58     return false;
  59   std::string ascii(ptr);
  60   idn2_free(ptr);
  61
  62   ptr  = nullptr;
  63   code = idn2_to_unicode_8z8z(ascii.c_str(), &ptr, IDN2_TRANSITIONAL);
  64   if (code != IDN2_OK)
  65     return false;
  66   idn2_free(ptr);
  67
  68   // FIXME: check syntax is dot-string?
  69
  70   return true;
  71 }
  72
  73 void Domain::set(std::string_view dom)
  74 {
  75   if (dom.length() > max_length) {
  76     throw std::invalid_argument("domain name too long");
  77   }
  78
  79   // Handle "bare" IP addresses, without the brackets.
  80   if (IP::is_address(dom)) {
  81     ascii_              = IP::to_address_literal(dom);
  82     utf8_               = ascii_;
  83     is_address_literal_ = true;
  84     return;
  85   }
  86
  87   if (IP::is_address_literal(dom)) {
  88     ascii_              = std::string(dom.data(), dom.length());
  89     utf8_               = ascii_;
  90     is_address_literal_ = true;
  91     return;
  92   }
  93
  94   is_address_literal_ = false;
  95
  96   // Since all Domains are fully qualified and not just some bag of
  97   // labels, the trailing dot provides no real information and will
  98   // mess up name matching on certs and stuff.
  99
 100   dom = remove_trailing_dot(dom);
 101
 102   auto const norm = nfkc(dom);
 103
 104   // idn2_to_ascii_8z() converts (ASCII) to lower case
 105
 106   char* ptr  = nullptr;
 107   auto  code = idn2_to_ascii_8z(norm.c_str(), &ptr, IDN2_TRANSITIONAL);
 108   if (code != IDN2_OK)
 109     throw std::invalid_argument(idn2_strerror(code));
 110   ascii_ = ptr;
 111   idn2_free(ptr);
 112
 113   ptr  = nullptr;
 114   code = idn2_to_unicode_8z8z(ascii_.c_str(), &ptr, IDN2_TRANSITIONAL);
 115   if (code != IDN2_OK)
 116     throw std::invalid_argument(idn2_strerror(code));
 117   utf8_ = ptr;
 118   idn2_free(ptr);
 119 }