Handle token/parse errors more carefully

src/hz_sophia.erl

@@ -24,8 +24,8 @@ parse_literal2(Result, Pos, String) ->
     case next_token(Pos, String) of
         {ok, {{eof, _, _, _, _}, _, _}} ->
             {ok, Result};
-        {ok, {{_, S, _, Row, Start, End}, _, _}} ->
+        {ok, {Token, _, _}} ->
-            {error, {unexpected_token, S, Row, Start, End}};
+            unexpected_token(Token);
         {error, Reason} ->
             {error, Reason}
     end.
@@ -235,8 +235,10 @@ parse_expression2(Type, Pos, String, {character, "{", _, Row, Start, _}) ->
     parse_record_or_map(Type, Pos, String, Row, Start);
 parse_expression2(Type, Pos, String, {alphanum, S, _, Row, Start, End}) ->
     parse_alphanum(Type, Pos, String, S, Row, Start, End);
-parse_expression2(_, _, _, {_, S, _, Row, Start, End}) ->
+parse_expression2(_, _, _, {eof, _, _, _, _, _}) ->
-    {error, {unexpected_token, S, Row, Start, End}}.
+    {error, unexpected_end_of_file};
+parse_expression2(_, _, _, Token) ->
+    unexpected_token(Token).
 
 unknown_type() ->
     {unknown_type, already_normalized, unknown_type}.
@@ -247,10 +249,24 @@ expect_tokens([Str | Rest], Pos, String) ->
     case next_token(Pos, String) of
         {ok, {{_, Str, _, _, _, _}, NewPos, NewString}} ->
             expect_tokens(Rest, NewPos, NewString);
-        {ok, {{_, Actual, _, Row, Start, End}, _, _}} ->
+        {ok, {Token, _, _}} ->
-            {error, {unexpected_token, Actual, Row, Start, End}}
+            unexpected_token(Token, Str);
+        {error, Reason} ->
+            {error, Reason}
     end.
 
+unexpected_token(Token, _Expected) ->
+    % I don't know if this is a good idea, but sometimes there are only one or
+    % two tokens that could have worked, which might make for simple
+    % non-technical error messages. I don't know how to format that yet,
+    % though.
+    unexpected_token(Token).
+
+unexpected_token({eof, _, _, _, _, _}) ->
+    {error, expression_incomplete};
+unexpected_token({_, S, _, Row, Start, End}) ->
+    {error, {unexpected_token, S, Row, Start, End}}.
+
 %%% Ambiguous Chain Object vs Identifier Parsing
 
 parse_alphanum(Type, Pos, String, [C | _] = S, Row, Start, End) when ?IS_LATIN_UPPER(C) ->
@@ -328,7 +344,9 @@ parse_list_loop(Inner, Pos, String, CloseChar, Row, Start, Acc) ->
         {ok, {{character, CloseChar, _, _, _, _}, NewPos, NewString}} ->
             {ok, {lists:reverse(Acc), NewPos, NewString}};
         {ok, {Token, NewPos, NewString}} ->
-            parse_list_loop2(Inner, NewPos, NewString, CloseChar, Row, Start, Acc, Token)
+            parse_list_loop2(Inner, NewPos, NewString, CloseChar, Row, Start, Acc, Token);
+        {error, Reason} ->
+            {error, Reason}
     end.
 
 parse_list_loop2(Inner, Pos, String, CloseChar, Row, Start, Acc, Token) ->
@@ -348,6 +366,8 @@ parse_list_loop3(Inner, Pos, String, CloseChar, Row, Start, Acc) ->
             {ok, {lists:reverse(Acc), NewPos, NewString}};
         {ok, {{character, ",", _, _, _, _}, NewPos, NewString}} ->
             parse_list_loop(Inner, NewPos, NewString, CloseChar, Row, Start, Acc);
+        {ok, {Token, _, _}} ->
+            unexpected_token(Token, CloseChar);
         {error, Reason} ->
             {error, Reason}
     end.
@@ -472,9 +492,9 @@ parse_tuple_tails(O, N, ExcessCount, HeadTerm, Tails, Pos, String) ->
         {ok, {{character, ",", _, _, _, _}, NewPos, NewString}} ->
             % It is a real tuple! Try the normal logic, then.
             parse_tuple_tails2(O, N, ExcessCount, HeadTerm, Tails, NewPos, NewString);
-        {ok, {{_, Actual, _, Row, Start, End}, _, _}} ->
+        {ok, {Token, _, _}} ->
             % Anything else is just a boring parse error we can complain about.
-            {error, {unexpected_token, Actual, Row, Start, End}};
+            unexpected_token(Token, ")");
         {error, Reason} ->
             {error, Reason}
     end.
@@ -509,7 +529,9 @@ parse_multivalue(ElemTypes, Pos, String, Row, Start, Acc) ->
         {ok, {{character, ")", _, Row2, Start2, _}, NewPos, NewString}} ->
             check_multivalue_long_enough(ElemTypes, NewPos, NewString, Row2, Start2, Acc);
         {ok, {Token, NewPos, NewString}} ->
-            parse_multivalue2(ElemTypes, NewPos, NewString, Row, Start, Acc, Token)
+            parse_multivalue2(ElemTypes, NewPos, NewString, Row, Start, Acc, Token);
+        {error, Reason} ->
+            {error, Reason}
     end.
 
 parse_multivalue2([Next | Rest], Pos, String, Row, Start, Acc, Token) ->
@@ -524,8 +546,8 @@ parse_multivalue2([Next | Rest], Pos, String, Row, Start, Acc, Token) ->
     end;
 parse_multivalue2([], Pos, String, _, _, Acc, {character, ")", _, _, _, _}) ->
     {ok, {lists:reverse(Acc), Pos, String}};
-parse_multivalue2([], _, _, _, _, _, {_, S, _, Row, Start, End}) ->
+parse_multivalue2([], _, _, _, _, _, Token) ->
-    {error, {unexpected_token, S, Row, Start, End}}.
+    unexpected_token(Token, ")").
 
 parse_multivalue3(ElemTypes, Pos, String, Row, Start, Acc) ->
     case next_token(Pos, String) of
@@ -533,8 +555,8 @@ parse_multivalue3(ElemTypes, Pos, String, Row, Start, Acc) ->
             check_multivalue_long_enough(ElemTypes, NewPos, NewString, Row2, Start2, Acc);
         {ok, {{character, ",", _, _, _, _}, NewPos, NewString}} ->
             parse_multivalue(ElemTypes, NewPos, NewString, Row, Start, Acc);
-        {ok, {{_, Actual, _, Row, Start, End}, _, _}} ->
+        {ok, {Token, _, _}} ->
-            {error, {unexpected_token, Actual, Row, Start, End}};
+            unexpected_token(Token, ")");
         {error, Reason} ->
             {error, Reason}
     end.
@@ -578,8 +600,10 @@ parse_variant3(Arities, Tag, ElemTypes, Pos, String) ->
     case next_token(Pos, String) of
         {ok, {{character, "(", _, Row, Start, _}, NewPos, NewString}} ->
             parse_variant4(Arities, Tag, ElemTypes, NewPos, NewString, Row, Start);
-        {ok, {{_, Actual, _, Row, Start, End}, _, _}} ->
+        {ok, {Token, _, _}} ->
-            {error, {unexpected_token, Actual, Row, Start, End}}
+            unexpected_token(Token, "(");
+        {error, Reason} ->
+            {error, Reason}
     end.
 
 parse_variant4(Arities, Tag, ElemTypes, Pos, String, Row, Start) ->
@@ -612,8 +636,10 @@ parse_record_or_map({_, _, unknown_type}, Pos, String, _, _) ->
             parse_map2(unknown_type(), unknown_type(), NewPos, NewString, #{});
         {ok, {{alphanum, _, _, Row, Start, End}, _, _}} ->
             {error, {unresolved_record, Row, Start, End}};
-        {ok, {{_, S, _, Row, Start, End}, _, _}} ->
+        {ok, {Token, _, _}} ->
-            {error, {unexpected_token, S, Row, Start, End}}
+            unexpected_token(Token, "}");
+        {error, Reason} ->
+            {error, Reason}
     end;
 parse_record_or_map({O, N, _}, _, _, Row, Start) ->
     {error, {wrong_type, O, N, map, Row, Start, Start}}.
@@ -624,8 +650,8 @@ parse_record(Fields, Pos, String, Acc) ->
             parse_record2(Fields, NewPos, NewString, Acc, Ident, Row, Start, End);
         {ok, {{character, "}", _, Row, Start, End}, NewPos, NewString}} ->
             parse_record_end(Fields, NewPos, NewString, Acc, Row, Start, End);
-        {ok, {{_, S, _, Row, Start, End}, _, _}} ->
+        {ok, {Token, _, _}} ->
-            {error, {unexpected_token, S, Row, Start, End}};
+            unexpected_token(Token, "}");
         {error, Reason} ->
             {error, Reason}
     end.
@@ -669,8 +695,8 @@ parse_record6(Fields, Pos, String, Acc) ->
             parse_record(Fields, NewPos, NewString, Acc);
         {ok, {{character, "}", _, Row, Start, End}, NewPos, NewString}} ->
             parse_record_end(Fields, NewPos, NewString, Acc, Row, Start, End);
-        {ok, {{_, S, _, Row, Start, End}, _, _}} ->
+        {ok, {Token, _, _}} ->
-            {error, {unexpected_token, S, Row, Start, End}};
+            unexpected_token(Token, "}");
         {error, Reason} ->
             {error, Reason}
     end.
@@ -708,8 +734,10 @@ parse_map(KeyType, ValueType, Pos, String, Acc) ->
             parse_map2(KeyType, ValueType, NewPos, NewString, Acc);
         {ok, {{character, "}", _, _, _, _}, NewPos, NewString}} ->
             {ok, {Acc, NewPos, NewString}};
-        {ok, {{_, S, _, Row, Start, End}, _, _}} ->
+        {ok, {Token, _, _}} ->
-            {error, {unexpected_token, S, Row, Start, End}}
+            unexpected_token(Token, "}");
+        {error, Reason} ->
+            {error, Reason}
     end.
 
 parse_map2(KeyType, ValueType, Pos, String, Acc) ->
@@ -743,8 +771,10 @@ parse_map5(KeyType, ValueType, Pos, String, Acc) ->
             parse_map(KeyType, ValueType, NewPos, NewString, Acc);
         {ok, {{character, "}", _, _, _, _}, NewPos, NewString}} ->
             {ok, {Acc, NewPos, NewString}};
-        {ok, {{_, S, _, Row, Start, End}, _, _}} ->
+        {ok, {Token, _, _}} ->
-            {error, {unexpected_token, S, Row, Start, End}}
+            unexpected_token(Token, "}");
+        {error, Reason} ->
+            {error, Reason}
     end.
 
 % TODO