Add a consensus method in which md families get canonicalized.

Implements prop298. Closes ticket 28266.

3 files changed, 55 insertions, 4 deletions

diff --git a/src/feature/dirauth/dirvote.c b/src/feature/dirauth/dirvote.c
index 066a9e6e8a..6c4e12cd51 100644
--- a/src/feature/dirauth/dirvote.c
+++ b/src/feature/dirauth/dirvote.c
@@ -28,6 +28,7 @@
 #include "feature/nodelist/fmt_routerstatus.h"
 #include "feature/nodelist/microdesc.h"
 #include "feature/nodelist/networkstatus.h"
+#include "feature/nodelist/nodefamily.h"
 #include "feature/nodelist/nodelist.h"
 #include "feature/nodelist/routerlist.h"
 #include "feature/relay/router.h"
@@ -3799,8 +3800,16 @@ dirvote_create_microdescriptor(const routerinfo_t *ri, int consensus_method)
     smartlist_add_asprintf(chunks, "a %s\n",
                            fmt_addrport(&ri->ipv6_addr, ri->ipv6_orport));
 
-  if (family)
-    smartlist_add_asprintf(chunks, "family %s\n", family);
+  if (family) {
+    if (consensus_method < MIN_METHOD_FOR_CANONICAL_FAMILIES_IN_MICRODESCS) {
+      smartlist_add_asprintf(chunks, "family %s\n", family);
+    } else {
+      const uint8_t *id = (const uint8_t *)ri->cache_info.identity_digest;
+      char *canonical_family = nodefamily_canonicalize(family, id, 0);
+      smartlist_add_asprintf(chunks, "family %s\n", canonical_family);
+      tor_free(canonical_family);
+    }
+  }
 
   if (summary && strcmp(summary, "reject 1-65535"))
     smartlist_add_asprintf(chunks, "p %s\n", summary);
@@ -3898,7 +3907,10 @@ static const struct consensus_method_range_t {
   int high;
 } microdesc_consensus_methods[] = {
   {MIN_SUPPORTED_CONSENSUS_METHOD, MIN_METHOD_FOR_NO_A_LINES_IN_MICRODESC - 1},
-  {MIN_METHOD_FOR_NO_A_LINES_IN_MICRODESC, MAX_SUPPORTED_CONSENSUS_METHOD},
+  {MIN_METHOD_FOR_NO_A_LINES_IN_MICRODESC,
+   MIN_METHOD_FOR_CANONICAL_FAMILIES_IN_MICRODESCS - 1},
+  {MIN_METHOD_FOR_CANONICAL_FAMILIES_IN_MICRODESCS,
+   MAX_SUPPORTED_CONSENSUS_METHOD},
   {-1, -1}
 };
 
diff --git a/src/feature/dirauth/dirvote.h b/src/feature/dirauth/dirvote.h
index a21e9f3455..6afb6047ff 100644
--- a/src/feature/dirauth/dirvote.h
+++ b/src/feature/dirauth/dirvote.h
@@ -57,7 +57,7 @@
 #define MIN_SUPPORTED_CONSENSUS_METHOD 25
 
 /** The highest consensus method that we currently support. */
-#define MAX_SUPPORTED_CONSENSUS_METHOD 28
+#define MAX_SUPPORTED_CONSENSUS_METHOD 29
 
 /** Lowest consensus method where authorities vote on required/recommended
  * protocols. */
@@ -79,6 +79,12 @@
  * addresses. See #23828 and #20916. */
 #define MIN_METHOD_FOR_NO_A_LINES_IN_MICRODESC 28
 
+/**
+ * Lowest consensus method where microdescriptor lines are put in canonical
+ * form for improved compressibility and ease of storage. See proposal 298.
+ **/
+#define MIN_METHOD_FOR_CANONICAL_FAMILIES_IN_MICRODESCS 29
+
 /** Default bandwidth to clip unmeasured bandwidths to using method >=
  * MIN_METHOD_TO_CLIP_UNMEASURED_BW.  (This is not a consensus method; do not
  * get confused with the above macros.) */
diff --git a/src/test/test_microdesc.c b/src/test/test_microdesc.c
index debb11155a..fd79aee6be 100644
--- a/src/test/test_microdesc.c
+++ b/src/test/test_microdesc.c
@@ -421,6 +421,28 @@ static const char test_md2_21[] =
   "ntor-onion-key hbxdRnfVUJJY7+KcT4E3Rs7/zuClbN3hJrjSBiEGMgI=\n"
   "id ed25519 wqfLzgfCtRfYNg88LsL1QpzxS0itapJ1aj6TbnByx/Q\n";
 
+static const char test_md2_withfamily_28[] =
+  "onion-key\n"
+  "-----BEGIN RSA PUBLIC KEY-----\n"
+  "MIGJAoGBAL2R8EfubUcahxha4u02P4VAR0llQIMwFAmrHPjzcK7apcQgDOf2ovOA\n"
+  "+YQnJFxlpBmCoCZC6ssCi+9G0mqo650lFuTMP5I90BdtjotfzESfTykHLiChyvhd\n"
+  "l0dlqclb2SU/GKem/fLRXH16aNi72CdSUu/1slKs/70ILi34QixRAgMBAAE=\n"
+  "-----END RSA PUBLIC KEY-----\n"
+  "ntor-onion-key hbxdRnfVUJJY7+KcT4E3Rs7/zuClbN3hJrjSBiEGMgI=\n"
+  "family OtherNode !Strange\n"
+  "id ed25519 wqfLzgfCtRfYNg88LsL1QpzxS0itapJ1aj6TbnByx/Q\n";
+
+static const char test_md2_withfamily_29[] =
+  "onion-key\n"
+  "-----BEGIN RSA PUBLIC KEY-----\n"
+  "MIGJAoGBAL2R8EfubUcahxha4u02P4VAR0llQIMwFAmrHPjzcK7apcQgDOf2ovOA\n"
+  "+YQnJFxlpBmCoCZC6ssCi+9G0mqo650lFuTMP5I90BdtjotfzESfTykHLiChyvhd\n"
+  "l0dlqclb2SU/GKem/fLRXH16aNi72CdSUu/1slKs/70ILi34QixRAgMBAAE=\n"
+  "-----END RSA PUBLIC KEY-----\n"
+  "ntor-onion-key hbxdRnfVUJJY7+KcT4E3Rs7/zuClbN3hJrjSBiEGMgI=\n"
+  "family !Strange $B7E27F104213C36F13E7E9829182845E495997A0 othernode\n"
+  "id ed25519 wqfLzgfCtRfYNg88LsL1QpzxS0itapJ1aj6TbnByx/Q\n";
+
 static void
 test_md_generate(void *arg)
 {
@@ -451,6 +473,17 @@ test_md_generate(void *arg)
   tt_assert(ed25519_pubkey_eq(md->ed25519_identity_pkey,
                               &ri->cache_info.signing_key_cert->signing_key));
 
+  // Try family encoding.
+  microdesc_free(md);
+  ri->declared_family = smartlist_new();
+  smartlist_add_strdup(ri->declared_family, "OtherNode !Strange");
+  md = dirvote_create_microdescriptor(ri, 28);
+  tt_str_op(md->body, OP_EQ, test_md2_withfamily_28);
+
+  microdesc_free(md);
+  md = dirvote_create_microdescriptor(ri, 29);
+  tt_str_op(md->body, OP_EQ, test_md2_withfamily_29);
+
  done:
   microdesc_free(md);
   routerinfo_free(ri);