Surround ipv6 with [], and properly cleanup otherwise invalid domains

fix #2694 (yeah, it was closed, but it was badly fixed)

2 files changed, 77 insertions, 17 deletions

diff --git a/louloulibs/xmpp/jid.cpp b/louloulibs/xmpp/jid.cpp
index 7b62f3e..d17d4e9 100644
--- a/louloulibs/xmpp/jid.cpp
+++ b/louloulibs/xmpp/jid.cpp
@@ -6,6 +6,9 @@
 #include <louloulibs.h>
 #ifdef LIBIDN_FOUND
  #include <stringprep.h>
+ #include <netdb.h>
+ #include <utils/scopeguard.hpp>
+ #include <set>
 #endif
 
 #include <logger/logger.hpp>
@@ -58,19 +61,68 @@ std::string jidprep(const std::string& original)
 
   char domain[max_jid_part_len] = {};
   memcpy(domain, jid.domain.data(), std::min(max_jid_part_len, jid.domain.size()));
-  rc = static_cast<Stringprep_rc>(::stringprep(domain, max_jid_part_len,
-       static_cast<Stringprep_profile_flags>(0), stringprep_nameprep));
-  if (rc != STRINGPREP_OK)
+
   {
-    log_error(error_msg + stringprep_strerror(rc));
-    return "";
+    // Using getaddrinfo, check if the domain part is a valid IPv4 (then use
+    // it as is), or IPv6 (surround it with []), or a domain name (run
+    // nameprep)
+    struct addrinfo hints;
+    memset(&hints, 0, sizeof(hints));
+    hints.ai_flags = AI_NUMERICHOST;
+    hints.ai_family = AF_UNSPEC;
+
+    struct addrinfo* res = nullptr;
+    auto addrinfo_deleter = utils::make_scope_guard([res]() { if (res) freeaddrinfo(res); });
+    if (::getaddrinfo(domain, nullptr, &hints, &res) || !res ||
+        (res->ai_family != AF_INET && res->ai_family != AF_INET6))
+      { // Not an IP, run nameprep on it
+        rc = static_cast<Stringprep_rc>(::stringprep(domain, max_jid_part_len,
+                                                     static_cast<Stringprep_profile_flags>(0), stringprep_nameprep));
+        if (rc != STRINGPREP_OK)
+          {
+            log_error(error_msg + stringprep_strerror(rc));
+            return "";
+          }
+
+        // Make sure it contains only allowed characters
+        using std::begin;
+        using std::end;
+        char* domain_end = domain + ::strlen(domain);
+        std::replace_if(std::begin(domain), domain + ::strlen(domain),
+                        [](const char c) -> bool
+                        {
+                          return !((c >= 'a' && c <= 'z') || c == '-' ||
+                                   (c >= '0' && c <= '9') || c == '.');
+                        }, '-');
+        // Make sure there are no doubled - or .
+        std::set<char> special_chars{'-', '.'};
+        domain_end = std::unique(begin(domain), domain + ::strlen(domain), [&special_chars](const char& a, const char& b) -> bool
+        {
+          return special_chars.count(a) && special_chars.count(b);
+        });
+        // remove leading and trailing -. if any
+        if (domain_end != domain && special_chars.count(*(domain_end - 1)))
+          --domain_end;
+        if (domain_end != domain && special_chars.count(domain[0]))
+          {
+            std::memmove(domain, domain + 1, domain_end - domain + 1);
+            --domain_end;
+          }
+        // And if the final result is an empty string, return a dummy hostname
+        if (domain_end == domain)
+          ::strcpy(domain, "empty");
+        else
+          *domain_end = '\0';
+      }
+    else if (res->ai_family == AF_INET6)
+      { // IPv6, surround it with []. The length is always enough:
+        // the longest possible IPv6 is way shorter than max_jid_part_len
+        ::memmove(domain + 1, domain, jid.domain.size());
+        domain[0] = '[';
+        domain[jid.domain.size() + 1] = ']';
+      }
   }
-  std::replace_if(std::begin(domain), domain + ::strlen(domain),
-                  [](const char c) -> bool
-                  {
-                    return !((c >= 'a' && c <= 'z') || c == '-' ||
-                      (c >= '0' && c <= '9') || c == '.');
-                  }, '-');
+
 
   // If there is no resource, stop here
   if (jid.resource.empty())
diff --git a/tests/jid.cpp b/tests/jid.cpp
index 9015afd..089f797 100644
--- a/tests/jid.cpp
+++ b/tests/jid.cpp
@@ -15,7 +15,10 @@ TEST_CASE("Jid")
   CHECK(jid2.local == "");
   CHECK(jid2.domain == "ツ.coucou");
   CHECK(jid2.resource == "coucou@coucou/coucou");
+}
 
+TEST_CASE("jidprep")
+{
   // Jidprep
   const std::string badjid("~zigougou™@EpiK-7D9D1FDE.poez.io/Boujour/coucou/slt™");
   const std::string correctjid = jidprep(badjid);
@@ -26,13 +29,18 @@ TEST_CASE("Jid")
   CHECK(jidprep(badjid) == "~zigougoutm@epik-7d9d1fde.poez.io/Boujour/coucou/sltTM");
   CHECK(jidprep(badjid) == "~zigougoutm@epik-7d9d1fde.poez.io/Boujour/coucou/sltTM");
 
-  const std::string badjid2("Zigougou@poez.io");
-  const std::string correctjid2 = jidprep(badjid2);
-  CHECK(correctjid2 == "zigougou@poez.io");
+  CHECK(jidprep("Zigougou@poez.io") == "zigougou@poez.io");
+
+  CHECK(jidprep("~Bisous@88.123.43.45") == "~bisous@88.123.43.45");
+
+  CHECK(jidprep("~Bisous@::ffff:42.156.139.46") == "~bisous@[::ffff:42.156.139.46]");
+
+  CHECK(jidprep("louiz!6bf74289@2001:bc8:38e7::") == "louiz!6bf74289@[2001:bc8:38e7::]");
 
-  const std::string crappy("~Bisous@7ea8beb1:c5fd2849:da9a048e:ip");
-  const std::string fixed_crappy = jidprep(crappy);
-  CHECK(fixed_crappy == "~bisous@7ea8beb1-c5fd2849-da9a048e-ip");
+  CHECK(jidprep("louiz@+:::::----coucou.com78--.") == "louiz@coucou.com78");
+  CHECK(jidprep("louiz@coucou.com78--.") == "louiz@coucou.com78");
+  CHECK(jidprep("louiz@+:::::----coucou.com78") == "louiz@coucou.com78");
+  CHECK(jidprep("louiz@:::::") == "louiz@empty");
 #else // Without libidn, jidprep always returns an empty string
   CHECK(jidprep(badjid) == "");
 #endif