[GDPR] DB schema rework for MAM MUC cassandra

esl · May 30, 2019 · 71dab2b · 71dab2b
1 parent 2f1fb1a
commit 71dab2b
Show file tree

Hide file tree

Showing 4 changed files with 35 additions and 7 deletions.
diff --git a/big_tests/tests/gdpr_SUITE.erl b/big_tests/tests/gdpr_SUITE.erl
@@ -131,7 +131,7 @@ groups() ->
      {retrieve_personal_data_mam_rdbms, [], mam_testcases()},
      {retrieve_personal_data_mam_riak, [], mam_testcases()},
      {retrieve_personal_data_mam_cassandra, [], mam_testcases()},
-     {retrieve_personal_data_mam_elasticsearch, [], [retrieve_mam_pm]},
+     {retrieve_personal_data_mam_elasticsearch, [], mam_testcases()},
      {remove_personal_data, [], removal_testcases()},
      {remove_personal_data_with_mods_disabled, [], removal_testcases()},
      {remove_personal_data_inbox, [], [remove_inbox, remove_inbox_muclight, remove_inbox_muc]}].

diff --git a/priv/elasticsearch/muc.json b/priv/elasticsearch/muc.json
@@ -8,6 +8,9 @@
 				"room": {
 					"type": "keyword"
 				},
+				"from_jid" : {
+					"type": "keyword"
+				},
 				"source_jid": {
 					"type": "keyword"
 				},

diff --git a/src/mam/mod_mam_muc.erl b/src/mam/mod_mam_muc.erl
@@ -242,7 +242,7 @@ archive_room_packet(Packet, FromNick, FromJID=#jid{}, RoomJID=#jid{}, Role, Affi
             MessID = generate_message_id(),
             Packet1 = replace_x_user_element(FromJID, Role, Affiliation, Packet),
             Result = archive_message(Host, MessID, ArcID,
-                                     RoomJID, SrcJID, SrcJID, incoming, Packet1),
+                                     RoomJID, mod_mam_utils:bare_jid(FromJID), SrcJID, incoming, Packet1),
             %% Packet2 goes to archive, Packet to other users
             case Result of
                 ok ->

diff --git a/src/mam/mod_mam_muc_elasticsearch_arch.erl b/src/mam/mod_mam_muc_elasticsearch_arch.erl
@@ -31,6 +31,9 @@
 -export([remove_archive/4]).
 -export([archive_size/4]).
 
+%gdpr
+-export([get_mam_muc_gdpr_data/2]).
+
 -include("mongoose.hrl").
 -include("mongoose_rsm.hrl").
 -include("mod_mam.hrl").
@@ -56,12 +59,28 @@ stop(Host) ->
 %%-------------------------------------------------------------------
 %% ejabberd_gen_mam_archive callbacks
 %%-------------------------------------------------------------------
+-spec get_mam_muc_gdpr_data(jid:username(), jid:server()) ->
+    {ok, ejabberd_gen_mam_archive:mam_muc_gdpr_data()}.
+get_mam_muc_gdpr_data(User, Host) ->
+    Source = jid:make(User, Host, <<"">>),
+    BinSource = mod_mam_utils:bare_jid(Source),
+    Filter = #{term => #{from_jid => BinSource}},
+    Sorting = #{mam_id => #{order => asc}},
+    SearchQuery = #{query => #{bool => #{filter => Filter}},
+                    sort => Sorting},
+    case mongoose_elasticsearch:search(?INDEX_NAME, ?TYPE_NAME, SearchQuery) of
+        {ok, #{<<"hits">> := #{<<"hits">> := Hits}}} ->
+            Messages = lists:map(fun hit_to_gdpr_mam_message/1, Hits),
+            {ok, Messages};
+        {error, _} ->
+            {ok, []}
+    end.
 
-archive_message(_Result, Host, MessageId, _UserId, RoomJid, _SourceJid, SourceJid, _Dir, Packet) ->
+archive_message(_Result, Host, MessageId, _UserId, RoomJid, FromJID, SourceJid, _Dir, Packet) ->
     Room = mod_mam_utils:bare_jid(RoomJid),
     SourceBinJid = mod_mam_utils:full_jid(SourceJid),
     DocId = make_document_id(Room, MessageId),
-    Doc = make_document(MessageId, Room, SourceBinJid, Packet),
+    Doc = make_document(MessageId, Room, SourceBinJid, Packet, FromJID),
     case mongoose_elasticsearch:insert_document(?INDEX_NAME, ?TYPE_NAME, DocId, Doc) of
         {ok, _} ->
             ok;
@@ -144,10 +163,11 @@ hooks(Host) ->
 make_document_id(Room, MessageId) ->
     <<Room/binary, $$, (integer_to_binary(MessageId))/binary>>.
 
--spec make_document(mod_mam:message_id(), binary(), binary(), exml:element()) ->
-    map().
-make_document(MessageId, Room, SourceBinJid, Packet) ->
+-spec make_document(mod_mam:message_id(), binary(), binary(), exml:element(),
+                    binary()) -> map().
+make_document(MessageId, Room, SourceBinJid, Packet, FromJID) ->
     #{mam_id     => MessageId,
+      from_jid   => FromJID,
       room       => Room,
       source_jid => SourceBinJid,
       message    => exml:to_binary(Packet),
@@ -257,6 +277,11 @@ hit_to_mam_message(#{<<"_source">> := JSON}) ->
     {ok, Stanza} = exml:parse(Packet),
     {MessageId, jid:from_binary(SourceJid), Stanza}.
 
+hit_to_gdpr_mam_message(#{<<"_source">> := JSON}) ->
+    MessageId = maps:get(<<"mam_id">>, JSON),
+    Packet = maps:get(<<"message">>, JSON),
+    {integer_to_binary(MessageId), Packet}.
+
 %% Usage of RSM affects the `"total"' value returned by ElasticSearch. Per RSM spec, the count
 %% returned by the query should represent the size of the whole result set, which in case of MAM
 %% is bound only by the MAM filters.