Merge pull request #3239 from rabbitmq/some-sort-of-unicode-support-in-env-files

Do not crash on badarg when env file has certain Unicode characters

(cherry picked from commit 1596f03)

Author: michaelklishin

deps/rabbit_common/src/rabbit_env.erl

@@ -1653,14 +1653,17 @@ parse_conf_env_file_output2([Line | Lines], Vars) ->
     end.
 
 is_sh_set_x_output(Line) ->
-    re:run(Line, "^\\++ ", [{capture, none}]) =:= match.
+    UnicodeLine = unicode:characters_to_binary(Line),
+    re:run(UnicodeLine, "^\\++ ", [{capture, none}]) =:= match.
 
 is_sh_function(_, []) ->
     false;
 is_sh_function(Line, Lines) ->
-    re:run(Line, "\\s\\(\\)\\s*$", [{capture, none}]) =:= match
+    UnicodeLine1 = unicode:characters_to_binary(Line),
+    UnicodeLine2 = unicode:characters_to_binary(hd(Lines)),
+    re:run(UnicodeLine1, "\\s\\(\\)\\s*$", [{capture, none}]) =:= match
     andalso
-    re:run(hd(Lines), "^\\s*\\{\\s*$", [{capture, none}]) =:= match.
+    re:run(UnicodeLine2, "^\\s*\\{\\s*$", [{capture, none}]) =:= match.
 
 parse_sh_literal("'" ++ SingleQuoted, Lines, Literal) ->
     parse_single_quoted_literal(SingleQuoted, Lines, Literal);

deps/rabbit_common/test/rabbit_env_SUITE.erl

@@ -1057,11 +1057,15 @@ check_parse_conf_env_file_output(_) ->
         )),
     ?assertEqual(
        #{"UNQUOTED" => "a",
+         "UNICODE" => [43, 43, 32, 1550, 32],
          "SINGLE_QUOTED" => "b",
          "DOUBLE_QUOTED" => "c",
          "SINGLE_DOLLAR" => "d"},
        rabbit_env:parse_conf_env_file_output2(
-         ["UNQUOTED=a",
+         %% a relatively rarely used Unicode character
+         ["++ ؎ ",
+          "UNQUOTED=a",
+          "UNICODE='++ ؎ '",
           "SINGLE_QUOTED='b'",
           "DOUBLE_QUOTED=\"c\"",
           "SINGLE_DOLLAR=$'d'"],