Add initial JID checks

The rexmpp_jid_check function now ensures that JID parts are valid
UTF-8 strings, and that only allowed code points (per RFC 8265) are
used in those. Though there is a few more checks to perform still.

7 files changed, 128 insertions, 3 deletions

diff --git a/README b/README
index a5a149c..6f317ef 100644
--- a/README
+++ b/README
@@ -15,7 +15,7 @@ of implementing additional XEPs on top of it, and should try to make
 it easy to implement a decent client application using it.
 
 Current dependencies: libunbound, libxml2, gnutls, gnutls-dane, gsasl,
-gpgme.
+gpgme, libicu.
 
 
 A rough roadmap:
diff --git a/configure.ac b/configure.ac
index 02ea44c..c175e74 100644
--- a/configure.ac
+++ b/configure.ac
@@ -38,6 +38,10 @@ AC_SUBST(UNBOUND_LIBS)
 
 AM_PATH_GPGME
 
+PKG_CHECK_MODULES([ICU_I18N], [icu-i18n])
+AC_SUBST(ICU_I18N_CFLAGS)
+AC_SUBST(ICU_I18N_LIBS)
+
 # Checks for header files.
 AC_CHECK_HEADERS([arpa/inet.h netdb.h netinet/in.h sys/socket.h syslog.h])
 
diff --git a/examples/basic.c b/examples/basic.c
index 123225a..404b281 100644
--- a/examples/basic.c
+++ b/examples/basic.c
@@ -106,6 +106,8 @@ main (int argc, char **argv) {
   /*                                        GNUTLS_X509_FMT_PEM); */
   s.roster_cache_file = "roster.xml";
 
+  /* rexmpp_openpgp_set_home_dir(&s, "./pgp/"); */
+
   /* Once the main structure is initialised and everything is
      sufficiently configured, we are ready to run the main loop and
      call rexmpp from it. */
diff --git a/src/Makefile.am b/src/Makefile.am
index ac34f06..52d6c4a 100644
--- a/src/Makefile.am
+++ b/src/Makefile.am
@@ -21,5 +21,6 @@ include_HEADERS = rexmpp_roster.h rexmpp_tcp.h rexmpp_socks.h rexmpp.h \
 	rexmpp_dns.h rexmpp_jid.h rexmpp_openpgp.h rexmpp_console.h
 librexmpp_la_CFLAGS = $(AM_CFLAGS) $(LIBXML_CFLAGS) $(GNUTLS_CFLAGS) \
 	$(LIBDANE_CFLAGS) $(GSASL_CFLAGS) $(UNBOUND_CFLAGS) $(GPGME_CFLAGS)
+	$(ICU_I18N_CFLAGS)
 librexmpp_la_LIBADD = $(LIBXML_LIBS) $(GNUTLS_LIBS) $(LIBDANE_LIBS) \
-	$(GSASL_LIBS) $(UNBOUND_LIBS) $(GPGME_LIBS)
+	$(GSASL_LIBS) $(UNBOUND_LIBS) $(GPGME_LIBS) $(ICU_I18N_LIBS)
diff --git a/src/rexmpp.c b/src/rexmpp.c
index 03e666e..3064f3f 100644
--- a/src/rexmpp.c
+++ b/src/rexmpp.c
@@ -441,6 +441,10 @@ rexmpp_err_t rexmpp_init (rexmpp_t *s, const char *jid)
     rexmpp_log(s, LOG_CRIT, "Failed to parse the initial JID.");
     return REXMPP_E_JID;
   }
+  if (! rexmpp_jid_check(&s->initial_jid)) {
+    rexmpp_log(s, LOG_CRIT, "An invalid initial JID is provided.");
+    return REXMPP_E_JID;
+  }
 
   s->xml_parser = xmlCreatePushParserCtxt(&sax, s, "", 0, NULL);
 
diff --git a/src/rexmpp_jid.c b/src/rexmpp_jid.c
index f19bd7c..ee7b326 100644
--- a/src/rexmpp_jid.c
+++ b/src/rexmpp_jid.c
@@ -8,10 +8,14 @@
 
 #include <stddef.h>
 #include <string.h>
+#include <stdio.h>
+#include <unicode/ustring.h>
+#include <unicode/uset.h>
+#include <unicode/uspoof.h>
 #include "rexmpp_jid.h"
 
 int rexmpp_jid_parse (const char *str, struct rexmpp_jid *jid) {
-  const char *resource = NULL, *domain = NULL;
+  const char *resource = NULL, *domain = str;
   size_t i;
   size_t resource_len = 0, local_len = 0;
   size_t domain_len, bare_len, full_len = strlen(str);
@@ -63,3 +67,112 @@ int rexmpp_jid_parse (const char *str, struct rexmpp_jid *jid) {
 
   return 0;
 }
+
+/* <https://tools.ietf.org/html/rfc7622#section-3>,
+   <https://tools.ietf.org/html/rfc8265#section-3.3> */
+int rexmpp_jid_check (struct rexmpp_jid *jid) {
+  UErrorCode err = U_ZERO_ERROR;
+  UChar local[1023], domain[1023], resource[1023];
+  int32_t local_len = 0, domain_len = 0, resource_len = 0;
+
+  /* Initial length checks are performed on parsing. */
+
+  /* Ensure that it's all valid UTF-8. */
+
+  u_strFromUTF8(local, 1023, &local_len, jid->local, -1, &err);
+  if (U_FAILURE(err)) {
+    return 0;
+  }
+  /* TODO: IDNA2008 on domain part. */
+  u_strFromUTF8(domain, 1023, &domain_len, jid->domain, -1, &err);
+  if (U_FAILURE(err)) {
+    return 0;
+  }
+  u_strFromUTF8(resource, 1023, &resource_len, jid->resource, -1, &err);
+  if (U_FAILURE(err)) {
+    return 0;
+  }
+
+  /* TODO: width mapping. */
+
+
+  /* Check character classes */
+  USpoofChecker *sc;
+  int32_t spoof;
+
+  /* IdentifierClass: {Ll, Lu, Lo, Nd, Lm, Mn, Mc} + 0x21 to 0x7e */
+  USet *identifier_chars = uset_openEmpty();
+  uset_applyIntPropertyValue(identifier_chars, UCHAR_GENERAL_CATEGORY_MASK,
+                             U_GC_LL_MASK | U_GC_LU_MASK | U_GC_LO_MASK |
+                             U_GC_ND_MASK | U_GC_LM_MASK | U_GC_MN_MASK |
+                             U_GC_MC_MASK,
+                             &err);
+  if (U_FAILURE(err)) {
+    return 0;
+  }
+  uset_addRange(identifier_chars, 0x21, 0x7e);
+
+  sc = uspoof_open(&err);
+  if (U_FAILURE(err)) {
+    uset_close(identifier_chars);
+    return 0;
+  }
+  uspoof_setChecks(sc, USPOOF_CHAR_LIMIT, &err);
+  if (U_FAILURE(err)) {
+    uset_close(identifier_chars);
+    uspoof_close(sc);
+    return 0;
+  }
+  uspoof_setAllowedChars(sc, identifier_chars, &err);
+  if (U_FAILURE(err)) {
+    uset_close(identifier_chars);
+    uspoof_close(sc);
+    return 0;
+  }
+  spoof = uspoof_check(sc, local, local_len, NULL, &err);
+  if (U_FAILURE(err) || spoof) {
+    uset_close(identifier_chars);
+    uspoof_close(sc);
+    return 0;
+  }
+
+  /* FreeformClass: Zs, {Sm, Sc, Sk, So}, {Pc, Pd, Ps, Pe, Pi, Pf,
+     Po}, toNFKC(cp) != cp ones, {Lt, Nl, No, Me}, and IdentifierClass
+     ones. */
+
+  /* TODO: https://tools.ietf.org/html/rfc8264#section-9.17 */
+  USet *freeform_chars = uset_openEmpty();
+  uset_applyIntPropertyValue(freeform_chars, UCHAR_GENERAL_CATEGORY_MASK,
+                             U_GC_ZS_MASK | U_GC_SM_MASK | U_GC_SC_MASK |
+                             U_GC_SK_MASK | U_GC_SO_MASK | U_GC_PC_MASK |
+                             U_GC_PD_MASK | U_GC_PS_MASK | U_GC_PE_MASK |
+                             U_GC_PI_MASK | U_GC_PF_MASK | U_GC_PO_MASK |
+                             U_GC_LT_MASK | U_GC_NL_MASK | U_GC_NO_MASK |
+                             U_GC_ME_MASK,
+                             &err);
+  if (U_FAILURE(err)) {
+    uset_close(freeform_chars);
+    uset_close(identifier_chars);
+    uspoof_close(sc);
+    return 0;
+  }
+  uset_addAll(freeform_chars, identifier_chars);
+  uset_close(identifier_chars);
+
+  spoof = uspoof_check(sc, resource, resource_len, NULL, &err);
+  if (U_FAILURE(err) || spoof) {
+    uset_close(freeform_chars);
+    uspoof_close(sc);
+    return 0;
+  }
+  uset_close(freeform_chars);
+  uspoof_close(sc);
+
+  /* TODO: case mapping, u_strToLower */
+
+  /* TODO: normalization, unorm2_normalize */
+
+  /* TODO: directionality */
+
+  return 1;
+}
diff --git a/src/rexmpp_jid.h b/src/rexmpp_jid.h
index 8613e3d..bfeedcb 100644
--- a/src/rexmpp_jid.h
+++ b/src/rexmpp_jid.h
@@ -20,5 +20,6 @@ struct rexmpp_jid {
 };
 
 int rexmpp_jid_parse (const char *str, struct rexmpp_jid *jid);
+int rexmpp_jid_check (struct rexmpp_jid *jid);
 
 #endif