fix(src,include,test): fixed multiple cases where a bad yaml was acce…

…pted.

Signed-off-by: Federico Di Pierro <[email protected]>

include/yaml-cpp/exceptions.h

@@ -48,6 +48,8 @@ const char* const UNKNOWN_TOKEN = "unknown token";
 const char* const DOC_IN_SCALAR = "illegal document indicator in scalar";
 const char* const EOF_IN_SCALAR = "illegal EOF in scalar";
 const char* const CHAR_IN_SCALAR = "illegal character in scalar";
+const char* const UNEXPECTED_SCALAR = "unexpected scalar";
+const char* const UNEXPECTED_FLOW = "plain value cannot start with flow indicator character";
 const char* const TAB_IN_INDENTATION =
     "illegal tab when looking for indentation";
 const char* const FLOW_END = "illegal flow end";

src/scanner.cpp

@@ -13,6 +13,7 @@ Scanner::Scanner(std::istream& in)
       m_startedStream(false),
       m_endedStream(false),
       m_simpleKeyAllowed(false),
+      m_scalarValueAllowed(false),
       m_canBeJSONFlow(false),
       m_simpleKeys{},
       m_indents{},
@@ -127,6 +128,17 @@ void Scanner::ScanNextToken() {
   }
 
   if (INPUT.peek() == Keys::FlowEntry) {
+    // values starting with `,` are not allowed.
+    // eg: reject `,foo`
+    if (INPUT.column() == 0) {
+      throw ParserException(INPUT.mark(), ErrorMsg::UNEXPECTED_FLOW);
+    }
+    // if we already parsed a quoted scalar value and we are not in a flow,
+    // then `,` is not a valid character.
+    // eg: reject `"foo",`
+    if (!m_scalarValueAllowed) {
+      throw ParserException(INPUT.mark(), ErrorMsg::UNEXPECTED_SCALAR);
+    }
     return ScanFlowEntry();
   }
 
@@ -159,6 +171,13 @@ void Scanner::ScanNextToken() {
     return ScanBlockScalar();
   }
 
+  // if we already parsed a quoted scalar value in this line,
+  // another scalar value is an error.
+  // eg: reject `"foo" "bar"`
+  if (!m_scalarValueAllowed) {
+    throw ParserException(INPUT.mark(), ErrorMsg::UNEXPECTED_SCALAR);
+  }
+
   if (INPUT.peek() == '\'' || INPUT.peek() == '\"') {
     return ScanQuotedScalar();
   }
@@ -203,6 +222,9 @@ void Scanner::ScanToNextToken() {
     // oh yeah, and let's get rid of that simple key
     InvalidateSimpleKey();
 
+    // new line - we accept a scalar value now
+    m_scalarValueAllowed = true;
+
     // new line - we may be able to accept a simple key now
     if (InBlockContext()) {
       m_simpleKeyAllowed = true;
@@ -245,6 +267,7 @@ const RegEx& Scanner::GetValueRegex() const {
 void Scanner::StartStream() {
   m_startedStream = true;
   m_simpleKeyAllowed = true;
+  m_scalarValueAllowed = true;
   std::unique_ptr<IndentMarker> pIndent(
       new IndentMarker(-1, IndentMarker::NONE));
   m_indentRefs.push_back(std::move(pIndent));
@@ -261,6 +284,7 @@ void Scanner::EndStream() {
   PopAllSimpleKeys();
 
   m_simpleKeyAllowed = false;
+  m_scalarValueAllowed = false;
   m_endedStream = true;
 }
 

src/scanner.h

@@ -177,6 +177,7 @@ class Scanner {
   // state info
   bool m_startedStream, m_endedStream;
   bool m_simpleKeyAllowed;
+  bool m_scalarValueAllowed;
   bool m_canBeJSONFlow;
   std::stack<SimpleKey> m_simpleKeys;
   std::stack<IndentMarker *> m_indents;

src/scantoken.cpp

@@ -211,6 +211,9 @@ void Scanner::ScanValue() {
     m_simpleKeyAllowed = InBlockContext();
   }
 
+  // we are parsing a `key: value` pair; scalars are always allowed
+  m_scalarValueAllowed = true;
+
   // eat
   Mark mark = INPUT.mark();
   INPUT.eat(1);
@@ -360,6 +363,10 @@ void Scanner::ScanQuotedScalar() {
   // and scan
   scalar = ScanScalar(INPUT, params);
   m_simpleKeyAllowed = false;
+  // we just scanned a quoted scalar;
+  // we can only have another scalar in this line
+  // if we are in a flow, eg: `[ "foo", "bar" ]` is ok, but `"foo", "bar"` isn't.
+  m_scalarValueAllowed = InFlowContext();
   m_canBeJSONFlow = true;
 
   Token token(Token::NON_PLAIN_SCALAR, mark);

test/integration/load_node_test.cpp

@@ -334,6 +334,69 @@ TEST(NodeTest, LoadQuotedNull) {
   EXPECT_EQ(node.as<std::string>(), "null");
 }
 
+TEST(NodeTest, LoadNonClosedQuotedString) {
+  EXPECT_THROW(Load(R"("foo)"), ParserException);
+}
+
+TEST(NodeTest, LoadWrongQuotedString) {
+  EXPECT_THROW(Load(R"("foo" [)"), ParserException);
+  EXPECT_THROW(Load(R"("foo", [)"), ParserException);
+}
+
+TEST(NodeTest, LoadUnquotedQuotedStrings) {
+  Node node = Load(R"(foo,"bar")");
+  EXPECT_EQ(node.as<std::string>(), "foo,\"bar\"");
+
+  node = Load(R"(foo,bar)");
+  EXPECT_EQ(node.as<std::string>(), "foo,bar");
+
+  node = Load(R"(foo,)");
+  EXPECT_EQ(node.as<std::string>(), "foo,");
+
+  node = Load(R"(foo "bar")");
+  EXPECT_EQ(node.as<std::string>(), "foo \"bar\"");
+}
+
+TEST(NodeTest, LoadCommaSeparatedStrings) {
+  EXPECT_THROW(Load(R"("foo","bar")"), ParserException);
+  EXPECT_THROW(Load(R"("foo",bar)"), ParserException);
+  EXPECT_THROW(Load(R"(,)"), ParserException);
+  EXPECT_THROW(Load(R"("foo",)"), ParserException);
+  EXPECT_THROW(Load(R"("foo","")"), ParserException);
+  EXPECT_THROW(Load(R"("foo",)"), ParserException);
+  EXPECT_THROW(Load(R"(,"foo")"), ParserException);
+  EXPECT_THROW(Load(R"(,foo)"), ParserException);
+
+  const char *doc_ok = R"(
+top
+, aa
+)";
+  EXPECT_NO_THROW(Load(doc_ok));
+
+  const char *doc_ok_2 = R"(
+top
+, "aa"
+)";
+  EXPECT_NO_THROW(Load(doc_ok_2));
+
+  const char *doc_fail = R"(
+"top"
+, "aa"
+)";
+  EXPECT_THROW(Load(doc_fail), ParserException);
+
+  const char *doc_fail_2 = R"(
+"top"
+, aa
+)";
+  EXPECT_THROW(Load(doc_fail_2), ParserException);
+}
+
+TEST(NodeTest, LoadSameLineStrings) {
+  EXPECT_THROW(Load(R"("foo" "bar")"), ParserException);
+  EXPECT_THROW(Load(R"("foo" bar)"), ParserException);
+}
+
 TEST(NodeTest, LoadTagWithParenthesis) {
     Node node = Load("!Complex(Tag) foo");
     EXPECT_EQ(node.Tag(), "!Complex(Tag)");