yamerl_parser: Fix an infinite loop

... with document/directive end markers not terminated by a whitespace
or a newline.

Thanks to Barry Allard for the bug report and testcases.

Fixes #18.

src/yamerl_parser.erl

@@ -723,6 +723,9 @@ determine_token_type([$-, $-, $-, C | _] = Chars, Line, 1 = Col, Delta,
   when ?IS_NEWLINE(C) orelse ?IS_SPACE(C) orelse
   (Version == {1,1} andalso ?IS_NEWLINE_11(C)) ->
     parse_document_sep(Chars, Line, Col, Delta, Parser, directives_end);
+determine_token_type([$-, $-, $-] = Chars, Line, 1 = Col, Delta,
+  #yamerl_parser{raw_eos = true} = Parser) ->
+    parse_document_sep(Chars, Line, Col, Delta, Parser, directives_end);
 
 %% Document end indicator.
 determine_token_type([$., $., $., C | _] = Chars, Line, 1 = Col, Delta,
@@ -730,6 +733,9 @@ determine_token_type([$., $., $., C | _] = Chars, Line, 1 = Col, Delta,
   when ?IS_NEWLINE(C) orelse ?IS_SPACE(C) orelse
   (Version == {1,1} andalso ?IS_NEWLINE_11(C)) ->
     parse_document_sep(Chars, Line, Col, Delta, Parser, document_end);
+determine_token_type([$., $., $.] = Chars, Line, 1 = Col, Delta,
+  #yamerl_parser{raw_eos = true} = Parser) ->
+    parse_document_sep(Chars, Line, Col, Delta, Parser, document_end);
 
 %% Directive indicator.
 determine_token_type([$% | _] = Chars, Line, 1 = Col, Delta,

test/construction/document_delimiters.erl

@@ -0,0 +1,62 @@
+-module(document_delimiters).
+
+-include_lib("eunit/include/eunit.hrl").
+
+%% This file contains testcases provided by Barry Allard when he
+%% reported #18. Before the bugfix, they would trigger an infinite loop
+%% in the parser.
+%%
+%% FIXME: It's still unclear if the returned values are correct for the
+%% given documents.
+
+setup() ->
+    application:start(yamerl).
+
+parse_blank_string_test_() ->
+    {setup,
+     fun setup/0,
+     [
+      ?_assertEqual([], yamerl_constr:string(""))
+     ]
+    }.
+
+parse_empty_json_map_test_() ->
+    {setup,
+     fun setup/0,
+     [
+      ?_assertEqual([[]], yamerl_constr:string("{}"))
+     ]
+    }.
+
+parse_dos_vuln_string_test_() ->
+    %% TODO: Consider fuzzing using QuickCheck Mini, PropEr, etc.
+    {setup,
+     fun setup/0,
+     [
+      ?_assertEqual([null], yamerl_constr:string("---")),
+      ?_assertEqual([], yamerl_constr:string("...")),
+      ?_assertEqual([null], yamerl_constr:string("\n---")),
+      ?_assertEqual([null], yamerl_constr:string("\r---")),
+      ?_assertEqual([null], yamerl_constr:string("\r\n---")),
+      ?_assertEqual([null, null], yamerl_constr:string("\n---\n---")),
+      ?_assertEqual([null, null], yamerl_constr:string(" \n---\n---")),
+      ?_assertEqual([null, null], yamerl_constr:string("\n---\n---\n...")),
+      ?_assertEqual(["anything", null], yamerl_constr:string("anything\n---"))
+     ]
+    }.
+
+parse_dos_vuln_bad_string_1_test_() ->
+    {setup,
+     fun setup/0,
+     [
+      ?_assertEqual(["--"], yamerl_constr:string("--"))
+     ]
+    }.
+
+parse_dos_vuln_bad_string_2_test_() ->
+    {setup,
+     fun setup/0,
+     [
+      ?_assertEqual(["----"], yamerl_constr:string("----"))
+     ]
+    }.

test/parsing/directives_end_alone.erl

@@ -0,0 +1,55 @@
+-module('directives_end_alone').
+
+-include_lib("eunit/include/eunit.hrl").
+
+single_test_() ->
+    ?_assertMatch(
+      {yamerl_parser,
+        string,
+        [],
+        <<>>,
+        3,
+        true,
+        [],
+        0,
+        4,
+        1,
+        4,
+        false,
+        1,
+        4,
+        utf8,
+        false,
+        undefined,
+        _,
+        _,
+        [],
+        {bcoll,root,0,-1,1,1,-1,1,1},
+        false,
+        false,
+        false,
+        [{impl_key,false,false,undefined,undefined,1,1}],
+        false,
+        false,
+        _,
+        [],
+        0,
+        6,
+        5,
+        undefined,
+        undefined,
+        _,
+        false,
+        [],
+        [
+          {yamerl_stream_end,1,4},
+          {yamerl_doc_end,1,4},
+          {yamerl_scalar,1,4,
+            {yamerl_tag,1,4,{non_specific,"?"}},
+            flow,plain,""},
+          {yamerl_doc_start,1,1,{1,2},_},
+          {yamerl_stream_start,1,1,utf8}
+        ]
+      },
+      yamerl_parser:string("---")
+    ).

test/parsing/document_end_alone.erl

@@ -0,0 +1,50 @@
+-module('document_end_alone').
+
+-include_lib("eunit/include/eunit.hrl").
+
+single_test_() ->
+    ?_assertMatch(
+      {yamerl_parser,
+        string,
+        [],
+        <<>>,
+        3,
+        true,
+        [],
+        0,
+        4,
+        1,
+        4,
+        false,
+        1,
+        4,
+        utf8,
+        false,
+        undefined,
+        _,
+        _,
+        [],
+        {bcoll,root,0,-1,1,1,-1,1,1},
+        false,
+        false,
+        false,
+        [{impl_key,false,false,undefined,undefined,1,1}],
+        false,
+        false,
+        _,
+        [],
+        0,
+        3,
+        2,
+        undefined,
+        undefined,
+        _,
+        false,
+        [],
+        [
+          {yamerl_stream_end,1,4},
+          {yamerl_stream_start,1,1,utf8}
+        ]
+      },
+      yamerl_parser:string("...")
+    ).