More fixes found by QC and some QC improvements

src/trunc_io.erl

@@ -35,7 +35,7 @@
 
 format(String, Args, Max) ->
     Parts = re:split(String,
-        "(~(?:-??\\d+\\.|\\*\\.|)(?:-??\d+\\.|\\*\\.|)(?:-??\\d+|\\*|)(?:t|)(?:[cfegswpWPBX#bx+ni]))",
+        "(~(?:-??\\d+\\.|\\*\\.|\\.|)(?:-??\\d+\\.|\\*\\.|\\.|)(?:-??\\d+|\\*|)(?:t|)(?:[cfegswpWPBX#bx+ni~]))",
         [{return, list}, trim]),
     Maxlen = Max - length(String),
     format(Parts, Args, Maxlen, [], []).
@@ -46,25 +46,37 @@ format([], _Args, Max, Acc, ArgAcc) ->
 format([[] | T], Args, Max, Acc, ArgAcc) ->
     % discard the null list generated by split
     format(T, Args, Max, Acc, ArgAcc);
+format(["~~" | T], Args, Max, Acc, ArgAcc) ->
+    format(T, Args, Max+1, ["~~" | Acc], ArgAcc);
 format(["~n" | T], Args, Max, Acc, ArgAcc) ->
     % ignore newlines for the purposes of argument indexing
     format(T, Args, Max+1, ["~n" | Acc], ArgAcc);
 format(["~i" | T], [AH | AT], Max, Acc, ArgAcc) ->
     % ~i means ignore this argument, but we'll just pass it through
     format(T, AT, Max+2, ["~i" | Acc], [AH | ArgAcc]);
-format([[$~|H]| T], [AH1, AH2 | AT], Max, Acc, ArgAcc) when H == "W"; H == "P"; H == "X" ->
-    % these crazy ones consume TWO arguments, just pass them through
-    % because W and P implicitly limit size so we trust the user knows
-    % what they're doing. Unfortunately we can't change Max at all because
-    % depth based limits are not of known length.
-    format(T, AT, Max, [[$~ | H] | Acc], [AH2, AH1 | ArgAcc]);
+format([[$~|H]| T], [AH1, AH2 | AT], Max, Acc, ArgAcc) when H == "X"; H == "x" ->
+    %% ~X consumes 2 arguments. It only prints integers so we can leave it alone
+    format(T, AT, Max, ["~X" | Acc], [AH2, AH1 | ArgAcc]);
+format([[$~|H]| T], [AH1, _AH2 | AT], Max, Acc, ArgAcc) when H == "W"; H == "P" ->
+    %% ~P and ~W consume 2 arguments, the second one being a depth limiter.
+    %% trunc_io isn't (yet) depth aware, so we can't honor this format string
+    %% safely at the moment, so just treat it like a regular ~p
+    %% TODO support for depth limiting
+    case print(AH1, Max + 2) of
+        {_Res, Max} ->
+            % this isn't the last argument, but it consumed all available space
+            % delay calculating the print size until the end
+            format(T, AT, Max + 2, ["~s" | Acc], [{future, AH1} | ArgAcc]);
+        {String, Length} ->
+            format(T, AT, Max + 2 - Length, ["~s" | Acc], [String | ArgAcc])
+    end;
 format([[$~|H]| T], [AH | AT], Max, Acc, ArgAcc) when length(H) == 1 ->
     % single character format specifier, relatively simple
     case H of
         _ when H == "p"; H == "w"; H == "s" ->
             %okay, these are prime candidates for rewriting
             case print(AH, Max + 2) of
-                {_Res, Max} when AT /= [] ->
+                {_Res, Max} ->
                     % this isn't the last argument, but it consumed all available space
                     % delay calculating the print size until the end
                     format(T, AT, Max + 2, ["~s" | Acc], [{future, AH} | ArgAcc]);
@@ -80,16 +92,18 @@ format([[$~|H]| T], [AH | AT], Max, Acc, ArgAcc) when length(H) == 1 ->
             end;
         _ ->
             % whatever, just pass them on through
-            format(T, AT, Max, [[$~ | H], Acc], [AH | ArgAcc])
+            format(T, AT, Max, [[$~ | H] | Acc], [AH | ArgAcc])
     end;
 format([[$~|H]| T], [AH | AT], Max, Acc, ArgAcc) ->
-    %% TODO
     % complicated format specifier, gonna have to parse it..
-    case re:run(H, "^(?:-??(\\d+|\\*)\\.|)(?:-??(\\d+|\\*)\\.|)(-??\\d+|\\*|)(t|)([cfegswpWPBX#bx+ni])$", [{capture, all_but_first, list}]) of
-        {match, [_F, _P, _Pad, _Mod, C]} when C == "p"; C=="w"; C=="s" ->
+    %case re:run(H, "^(?:-??(\\d+|\\*)\\.|\\.|)(?:-??(\\d+|\\*)\\.|\\.|)(-??.|)(t|)([cfegswpWPBX#bx+ni])$", [{capture, all_but_first, list}]) of
+    %% its actually simpler to just look at the last character in the string
+    case lists:nth(length(H), H) of
+        C when C == $p; C == $w; C == $s ->
+        %{match, [_F, _P, _Pad, _Mod, C]} when C == "p"; C=="w"; C=="s" ->
             %okay, these are prime candidates for rewriting
             case print(AH, Max + length(H) + 1) of
-                {_Res, Max} when AT /= [] ->
+                {_Res, Max} ->
                     % this isn't the last argument, but it consumed all available space
                     % delay calculating the print size until the end
                     format(T, AT, Max + length(H) + 1, ["~s" | Acc], [{future, AH} | ArgAcc]);
@@ -103,18 +117,31 @@ format([[$~|H]| T], [AH | AT], Max, Acc, ArgAcc) ->
                     end,
                     format(T, AT, Max + length(H) + 1 - RealLen, ["~s" | Acc], [Value | ArgAcc])
             end;
-        {match, [_F, _P, _Pad, _Mod, C]} when C == "P"; C=="W" ->
-            % these crazy ones consume TWO arguments, just pass them through
-            % because W and P implicitly limit size so we trust the user knows
-            % what they're doing. Unfortunately we can't change Max at all because
-            % depth based limits are not of known length.
+        C when C == $P; C == $W ->
+        %{match, [_F, _P, _Pad, _Mod, C]} when C == "P"; C == "W" ->
+            %% ~P and ~W consume 2 arguments, the second one being a depth limiter.
+            %% trunc_io isn't (yet) depth aware, so we can't honor this format string
+            %% safely at the moment, so just treat it like a regular ~p
+            %% TODO support for depth limiting
+            [_ | AT2] = AT,
+            case print(AH, Max + 2) of
+                {_Res, Max} ->
+                    % this isn't the last argument, but it consumed all available space
+                    % delay calculating the print size until the end
+                    format(T, AT2, Max + 2, ["~s" | Acc], [{future, AH} | ArgAcc]);
+                {String, Length} ->
+                    format(T, AT2, Max + 2 - Length, ["~s" | Acc], [String | ArgAcc])
+            end;
+        C when C == $X; C == $x ->
+        %{match, [_F, _P, _Pad, _Mod, C]} when C == "X"; C == "x" ->
+            %% ~X consumes 2 arguments. It only prints integers so we can leave it alone
             [AH2 | AT2] = AT,
             format(T, AT2, Max, [[$~|H]|Acc], [AH2, AH |ArgAcc]);
-        {match, _} ->
-            format(T, AT, Max, [H | Acc], [AH|ArgAcc]);
-        nomatch ->
-            io:format("unable to match format pattern ~~~p~n", [H]),
-            format(T, AT, Max, [H | Acc], [AH|ArgAcc])
+        _ ->
+            format(T, AT, Max, [[$~|H] | Acc], [AH|ArgAcc])
+        %nomatch ->
+            %io:format("unable to match format pattern ~~~p~n", [H]),
+            %format(T, AT, Max, [[$~|H] | Acc], [AH|ArgAcc])
     end;
 format([H | T], Args, Max, Acc, ArgAcc) ->
     format(T, Args, Max, [H | Acc], ArgAcc).

test/trunc_io_eqc.erl

@@ -60,9 +60,29 @@ check() ->
 
 gen_fmt_args() ->
     list(oneof([gen_print_str(),
+                "~~",
                 {"~p", gen_any(5)},
                 {"~w", gen_any(5)},
-                {"~s", gen_print_str()}
+                {"~s", gen_print_str()},
+                {"~P", gen_any(5), 4},
+                {"~W", gen_any(5), 4},
+                {"~i", gen_any(5)},
+                {"~B", nat()},
+                {"~b", nat()},
+                {"~X", nat(), "0x"},
+                {"~x", nat(), "0x"},
+                {"~.10#", nat()},
+                {"~.10+", nat()},
+                {"~.36B", nat()},
+                {"~62P", gen_any(5), 4},
+                {"~c", gen_char()},
+                {"~tc", gen_char()}
+                %{"~f", real()}, %% floats like to make the fudge limit fail, so don't enable them
+                %{"~10.f", real()},
+                %{"~g", real()},
+                %{"~10.g", real()},
+                %{"~e", real()},
+                %{"~10.e", real()}
                ])).
              
 
@@ -74,6 +94,7 @@ gen_any(MaxDepth) ->
     oneof([largeint(),
            gen_atom(),
            nat(),
+           %real(),
            binary(),
            gen_pid(),
            gen_port(),
@@ -106,6 +127,9 @@ gen_ref() ->
 
 gen_fun() ->
     ?LAZY(fun() -> ok end).
+
+gen_char() ->
+    oneof(lists:seq($A, $z)).
                
 %%====================================================================
 %% Property
@@ -115,7 +139,7 @@ gen_fun() ->
 prop_format() ->
     ?FORALL({FmtArgs, MaxLen}, {gen_fmt_args(), gen_max_len()},
             begin
-                FudgeLen = 31, %% trunc_io does not correctly calc safe size of pid/port/numbers/funs
+                FudgeLen = 50, %% trunc_io does not correctly calc safe size of pid/port/numbers/funs
                 {FmtStr, Args} = build_fmt_args(FmtArgs),
                 try
                     Str = lists:flatten(trunc_io:format(FmtStr, Args, MaxLen)),
@@ -151,6 +175,8 @@ prop_format() ->
 build_fmt_args(FmtArgs) ->
     F = fun({Fmt, Arg}, {FmtStr0, Args0}) ->
                 {FmtStr0 ++ Fmt, Args0 ++ [Arg]};
+           ({Fmt, Arg1, Arg2}, {FmtStr0, Args0}) ->
+                {FmtStr0 ++ Fmt, Args0 ++ [Arg1, Arg2]};
            (Str, {FmtStr0, Args0}) ->
                 {FmtStr0 ++ Str, Args0}
         end,