Parser should not fail for unrecognised tags, store them aside

2 files changed, 35 insertions, 14 deletions

diff --git a/src/dmarc_lexer.xrl b/src/dmarc_lexer.xrl
index 29bc778..77dadde 100644
--- a/src/dmarc_lexer.xrl
+++ b/src/dmarc_lexer.xrl
@@ -8,14 +8,14 @@ LOST_MAILTO = mailto
 
 Rules.
 
-{KEY}          : {token, {key, TokenLine, list_to_atom(TokenChars)}}.
-{MAILTO_EMAIL} : {token, {mailto, TokenLine, TokenChars}}.
-=              : {token, {equals, TokenLine}}.
-:              : {token, {colon, TokenLine}}.
-;              : {token, {semicolon, TokenLine}}.
-,              : {token, {comma, TokenLine}}.
-{LOST_MAILTO}  : {error, invalid_mailto}.
-{VALUE}        : {token, {string, TokenLine, TokenChars}}.
-{WHITESPACE}+  : skip_token.
+{KEY}            : {token, {key, TokenLine, list_to_atom(TokenChars)}}.
+{MAILTO_EMAIL}   : {token, {mailto, TokenLine, TokenChars}}.
+=                : {token, {equals, TokenLine}}.
+:                : {token, {colon, TokenLine}}.
+;                : {token, {semicolon, TokenLine}}.
+,                : {token, {comma, TokenLine}}.
+{LOST_MAILTO}    : {error, invalid_mailto}.
+{VALUE}          : {token, {string, TokenLine, TokenChars}}.
+{WHITESPACE}+    : skip_token.
 
 Erlang code.
diff --git a/src/dmarc_parser.yrl b/src/dmarc_parser.yrl
index c1f4e1d..14c375b 100644
--- a/src/dmarc_parser.yrl
+++ b/src/dmarc_parser.yrl
@@ -1,12 +1,14 @@
 Nonterminals
-    record property_list property optional_semicolon value.
+    record property_list property optional_semicolon value invalid_property.
 
 Terminals
     key equals semicolon string comma colon mailto.
 
 Rootsymbol record.
 
-record -> property_list optional_semicolon : '$1'.
+record -> property_list optional_semicolon : 
+    {Valid, Invalid} = separate_properties('$1'),
+    #{valid => maps:from_list(Valid), invalid => maps:from_list(Invalid)}.
 
 optional_semicolon -> semicolon : [].
 optional_semicolon -> '$empty' : [].
@@ -14,8 +16,13 @@ optional_semicolon -> '$empty' : [].
 property_list -> property : ['$1'].
 property_list -> property property_list : ['$1' | '$2'].
 
-property -> key equals value semicolon : {value('$1'), process_value(value('$1'), value_to_list('$3'))}.
-property -> key equals value : {value('$1'), process_value(value('$1'), value_to_list('$3'))}.
+property -> key equals value semicolon : {valid, {value('$1'), process_value(value('$1'), value_to_list('$3'))}}.
+property -> key equals value : {valid, {value('$1'), process_value(value('$1'), value_to_list('$3'))}}.
+
+property -> invalid_property : {invalid, '$1'}.
+
+invalid_property -> string equals value semicolon : {value('$1'), value_to_list('$3')}.
+invalid_property -> string equals value : {value('$1'), value_to_list('$3')}.
 
 value -> mailto : '$1'.
 value -> mailto comma value : join_values('$1', '$3').
@@ -23,7 +30,7 @@ value -> mailto comma value : join_values('$1', '$3').
 value -> string : '$1'.
 value -> string colon value : join_values('$1', '$3').
 
-Expect 1.
+Expect 2.
 
 Erlang code.
 
@@ -45,3 +52,17 @@ process_value(_Key, [Value]) ->
     value(Value);
 process_value(_Key, Values) ->
     [value(V) || V <- Values].
+
+separate_properties(Properties) ->
+    separate_properties(Properties, [], []).
+
+separate_properties([], ValidAcc, InvalidAcc) ->
+    {lists:reverse(ValidAcc), lists:reverse(InvalidAcc)};
+separate_properties([{valid, Property} | Rest], ValidAcc, InvalidAcc) ->
+    separate_properties(Rest, [Property | ValidAcc], InvalidAcc);
+separate_properties([{invalid, {Key, Values}} | Rest], ValidAcc, InvalidAcc) ->
+    ProcessedValues = case Values of
+        [SingleValue] -> value(SingleValue);
+        MultipleValues -> [value(V) || V <- MultipleValues]
+    end,
+    separate_properties(Rest, ValidAcc, [{Key, ProcessedValues} | InvalidAcc]).