Merge pull request #3711 from esl/graphql-gdpr-resolver

Adding gdpr resolver

big_tests/default.spec

@@ -38,6 +38,7 @@
 {suites, "tests", graphql_session_SUITE}.
 {suites, "tests", graphql_stanza_SUITE}.
 {suites, "tests", graphql_stats_SUITE}.
+{suites, "tests", graphql_gdpr_SUITE}.
 {suites, "tests", graphql_vcard_SUITE}.
 {suites, "tests", graphql_http_upload_SUITE}.
 {suites, "tests", graphql_metric_SUITE}.

big_tests/dynamic_domains.spec

@@ -55,6 +55,7 @@
 {suites, "tests", graphql_vcard_SUITE}.
 {suites, "tests", graphql_offline_SUITE}.
 {suites, "tests", graphql_stats_SUITE}.
+{suites, "tests", graphql_gdpr_SUITE}.
 {suites, "tests", graphql_http_upload_SUITE}.
 {suites, "tests", graphql_metric_SUITE}.
 

big_tests/tests/gdpr_SUITE.erl

@@ -1583,7 +1583,7 @@ get_personal_data_via_rpc(Client, ExpectedKeys) ->
     ClientU = escalus_utils:jid_to_lower(escalus_client:username(Client)),
     ClientS = escalus_utils:jid_to_lower(escalus_client:server(Client)),
     AllPersonalData = mongoose_helper:successful_rpc(
-                        service_admin_extra_gdpr, get_data_from_modules, [ClientU, ClientS]),
+                        gdpr_api, get_data_from_modules, [ClientU, ClientS]),
     %% We don't use lists:filter/2 because this line also ensures order
     [ lists:keyfind(Key, 1, AllPersonalData) || Key <- ExpectedKeys ].
 

big_tests/tests/graphql_gdpr_SUITE.erl

@@ -0,0 +1,96 @@
+-module(graphql_gdpr_SUITE).
+
+-compile([export_all, nowarn_export_all]).
+
+-import(distributed_helper, [require_rpc_nodes/1]).
+-import(domain_helper, [host_type/0, domain/0]).
+-import(graphql_helper, [execute_user/3, execute_auth/2, user_to_bin/1]).
+-import(distributed_helper, [mim/0, rpc/4]).
+
+-include_lib("common_test/include/ct.hrl").
+-include_lib("eunit/include/eunit.hrl").
+
+suite() ->
+    require_rpc_nodes([mim]) ++ escalus:suite().
+
+all() ->
+    [{group, admin_gdpr}].
+
+groups() ->
+    [{admin_gdpr, [], admin_stats_handler()}].
+
+admin_stats_handler() ->
+    [admin_gdpr_test,
+     admin_gdpr_no_user_test].
+
+init_per_suite(Config) ->
+    escalus:init_per_suite(Config).
+
+end_per_suite(Config) ->
+    escalus:end_per_suite(Config).
+
+init_per_group(_, Config) ->
+    graphql_helper:init_admin_handler(Config).
+
+end_per_group(_, _Config) ->
+    escalus_fresh:clean().
+
+init_per_testcase(CaseName, Config) ->
+    escalus:init_per_testcase(CaseName, Config).
+
+end_per_testcase(CaseName, Config) ->
+    escalus_fresh:clean(),
+    escalus:end_per_testcase(CaseName, Config).
+
+% Admin test cases
+
+admin_gdpr_test(Config) ->
+    escalus:fresh_story_with_config(Config, [{alice, 1}], fun admin_gdpr_test/2).
+
+admin_gdpr_test(Config, Alice) ->
+    Filename = random_filename(Config),
+    Vars = #{<<"username">> => escalus_client:username(Alice),
+             <<"domain">> => escalus_client:server(Alice),
+             <<"resultFilepath">> => list_to_binary(Filename)},
+    Result = admin_retrieve_personal_data(Config, Vars),
+    ParsedResult = ok_result(<<"gdpr">>, <<"retrievePersonalData">>, Result),
+    ?assertEqual(<<"Data retrieved">>, ParsedResult),
+    FullPath = get_mim_cwd() ++ "/" ++ Filename,
+    Dir = make_dir_name(Filename, escalus_client:username(Alice)),
+    ct:log("extracting logs ~s", [Dir]),
+    ?assertMatch({ok, _}, zip:extract(FullPath, [{cwd, Dir}])).
+
+admin_gdpr_no_user_test(Config) ->
+    Vars = #{<<"username">> => <<"AAAA">>, <<"domain">> => domain(),
+             <<"resultFilepath">> => <<"AAA">>},
+    Result = admin_retrieve_personal_data(Config, Vars),
+    ParsedResult = error_result(<<"extensions">>, <<"code">>, Result),
+    ?assertEqual(<<"user_does_not_exist_error">>, ParsedResult).
+
+% Helpers
+
+admin_retrieve_personal_data(Config, Vars) ->
+    Query = <<"query Q1($username: String!, $domain: String!, $resultFilepath: String!)
+                   {gdpr{retrievePersonalData(username: $username, domain: $domain,
+                                              resultFilepath: $resultFilepath)}}">>,
+    Body = #{query => Query, operationName => <<"Q1">>, variables => Vars},
+    execute_auth(Body, Config).
+
+error_result(What1, What2, {{<<"200">>, <<"OK">>}, #{<<"errors">> := [Data]}}) ->
+    maps:get(What2, maps:get(What1, Data)).
+
+ok_result(What1, What2, {{<<"200">>, <<"OK">>}, #{<<"data">> := Data}}) ->
+    maps:get(What2, maps:get(What1, Data)).
+
+random_filename(Config) ->
+    TCName = atom_to_list(?config(tc_name, Config)),
+    TCName ++ "." ++ integer_to_list(erlang:system_time()) ++ ".zip".
+
+get_mim_cwd() ->
+    {ok, Cwd} = rpc(mim(), file, get_cwd, []),
+    Cwd.
+
+make_dir_name(Filename, User) when is_binary(User) ->
+    make_dir_name(Filename, binary_to_list(User));
+make_dir_name(Filename, User) when is_list(User) ->
+    Filename ++ "." ++ User ++ ".unzipped".

priv/graphql/schemas/admin/admin_schema.gql

@@ -34,6 +34,8 @@ type AdminQuery{
   metric: MetricAdminQuery
   "Statistics"
   stats: StatsAdminQuery
+  "Personal data management according to GDPR"
+  gdpr: GdprAdminQuery
 }
 
 """

priv/graphql/schemas/admin/gdpr.gql

@@ -0,0 +1,5 @@
+"Retrieve user's presonal data"
+type GdprAdminQuery {
+    "Retrieves all personal data from MongooseIM for a given user"
+    retrievePersonalData(username: String!, domain: String!, resultFilepath: String!): String
+}

src/admin_extra/service_admin_extra_gdpr.erl

@@ -4,140 +4,17 @@
 -include("mongoose_logger.hrl").
 -include("jlib.hrl").
 
--export([commands/0, retrieve_all/3]).
-
-% Exported for RPC call
--export([retrieve_logs/2, get_data_from_modules/2]).
-
--ignore_xref([commands/0, retrieve_all/3, retrieve_logs/2, get_data_from_modules/2]).
-
--define(CMD_TIMEOUT, 300000).
+-export([commands/0]).
+-ignore_xref([commands/0]).
 
 -spec commands() -> [ejabberd_commands:cmd()].
 commands() -> [
         #ejabberd_commands{name = retrieve_personal_data, tags = [gdpr],
             desc = "Retrieve user's presonal data.",
             longdesc = "Retrieves all personal data from MongooseIM for a given user. Example:\n"
                        " mongooseimctl retrieve_personal_data alice localhost /home/mim/alice.smith.zip ",
-            module = ?MODULE,
+            module = gdpr_api,
             function = retrieve_all,
             args = [{username, binary}, {domain, binary}, {path, binary}],
             result = {res, rescode}}
     ].
-
--spec retrieve_all(jid:user(), jid:server(), Path :: binary()) -> ok | {error, Reason :: any()}.
-retrieve_all(Username, Domain, ResultFilePath) ->
-    JID = jid:make(Username, Domain, <<>>),
-    case user_exists(JID) of
-        true ->
-            DataFromModules = get_data_from_modules(JID),
-            % The contract is that we create personal data files only when there are any items
-            % returned for the data group.
-            DataToWrite = lists:filter(fun({_, _, Items}) -> Items /= [] end, DataFromModules),
-
-            TmpDir = make_tmp_dir(),
-
-            CsvFiles = lists:map(
-                fun({DataGroup, Schema, Entries}) ->
-                    BinDataGroup = atom_to_binary(DataGroup, utf8),
-                    FileName = <<BinDataGroup/binary, ".csv">>,
-                    to_csv_file(FileName, Schema, Entries, TmpDir),
-                    binary_to_list(FileName)
-                end,
-                DataToWrite),
-
-            LogFiles = get_all_logs(Username, Domain, TmpDir),
-
-            ZipFile = binary_to_list(ResultFilePath),
-            {ok, ZipFile} = zip:create(ZipFile, CsvFiles ++ LogFiles, [{cwd, TmpDir}]),
-            remove_tmp_dir(TmpDir),
-            ok;
-        false ->
-            {error, "User does not exist"}
-    end.
-
-%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
-%%%                       Private funs
-%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
-
--spec get_data_from_modules(jid:user(), jid:server()) -> gdpr:personal_data().
-get_data_from_modules(Username, Domain) ->
-    JID = jid:make(Username, Domain, <<>>),
-    get_data_from_modules(JID).
-
--spec get_data_from_modules(jid:jid()) -> gdpr:personal_data().
-get_data_from_modules(JID) ->
-    {ok, HostType} = mongoose_domain_api:get_domain_host_type(JID#jid.lserver),
-    mongoose_hooks:get_personal_data(HostType, JID).
-
--spec to_csv_file(file:name_all(), gdpr:schema(), gdpr:entries(), file:name()) -> ok.
-to_csv_file(Filename, DataSchema, DataRows, TmpDir) ->
-    FilePath = <<(list_to_binary(TmpDir))/binary, "/", Filename/binary>>,
-    {ok, File} = file:open(FilePath, [write]),
-    Encoded = erl_csv:encode([DataSchema | DataRows]),
-    file:write(File, Encoded),
-    file:close(File).
-
--spec user_exists(jid:jid()) -> boolean().
-user_exists(JID) ->
-    ejabberd_auth:does_user_exist(JID).
-
--spec make_tmp_dir() -> file:name().
-make_tmp_dir() ->
-    TmpDirName = lists:flatten(io_lib:format("/tmp/gdpr-~4.36.0b", [rand:uniform(36#zzzz)])),
-    case file:make_dir(TmpDirName) of
-        ok -> TmpDirName;
-        {error, eexist} -> make_tmp_dir();
-        {error, Error} -> {error, Error}
-    end.
-
--spec remove_tmp_dir(file:name()) -> ok.
-remove_tmp_dir(TmpDir) ->
-    {ok, FileNames} = file:list_dir(TmpDir),
-    [file:delete(TmpDir ++ "/" ++ File) || File <- FileNames],
-    file:del_dir(TmpDir).
-
--type cmd() :: string() | binary().
--spec run(cmd(), [cmd()], timeout()) -> non_neg_integer() | timeout.
-run(Cmd, Args, Timeout) ->
-    Port = erlang:open_port({spawn_executable, Cmd}, [exit_status, {args, Args}]),
-    receive
-        {Port, {exit_status, ExitStatus}} -> ExitStatus
-    after Timeout ->
-        timeout
-    end.
-
-%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
-%%%                       Logs retrieval
-%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
-
--spec retrieve_logs(gdpr:username(), gdpr:domain()) -> {ok, ZippedLogs :: binary()}.
-retrieve_logs(Username, Domain) ->
-    TmpDir = make_tmp_dir(),
-    LogFile = get_logs(Username, Domain, TmpDir),
-    {ok, {_, ZippedLogs}} = zip:create("archive.zip", [LogFile], [memory, {cwd, TmpDir}]),
-    remove_tmp_dir(TmpDir),
-    {ok, ZippedLogs}.
-
--spec get_all_logs(gdpr:username(), gdpr:domain(), file:name()) -> [file:name()].
-get_all_logs(Username, Domain, TmpDir) ->
-    OtherNodes = mongoose_cluster:other_cluster_nodes(),
-    LogFile = get_logs(Username, Domain, TmpDir),
-    LogFilesFromOtherNodes = [get_logs_from_node(Node, Username, Domain, TmpDir) || Node <- OtherNodes],
-    [LogFile | LogFilesFromOtherNodes].
-
--spec get_logs(gdpr:username(), gdpr:domain(), file:name()) -> file:name().
-get_logs(Username, Domain, TmpDir) ->
-    FileList = [filename:absname(F) || F <- mongoose_logs:get_log_files()],
-    Cmd = code:priv_dir(mongooseim) ++ "/parse_logs.sh",
-    FileName = "logs-" ++ atom_to_list(node()) ++ ".txt",
-    FilePath = TmpDir ++ "/" ++ FileName,
-    Args = [FilePath, Username, Domain | FileList],
-    0 = run(Cmd, Args, ?CMD_TIMEOUT),
-    FileName.
-
--spec get_logs_from_node(node(), gdpr:username(), gdpr:domain(), file:name()) -> file:name().
-get_logs_from_node(Node, Username, Domain, TmpDir) ->
-    {ok, ZippedData} = rpc:call(Node, ?MODULE, retrieve_logs, [Username, Domain]),
-    {ok, [File]} = zip:unzip(ZippedData, [{cwd, TmpDir}]),
-    filename:basename(File).