Feat/fix faulty string (#1406)

This fixes the example "Example 5.13 Escaped Characters" from the YAML
specification.

The example demands that `\_` is being translated to unicode `\u00a0`.
The unicode codepoint `\u00A0` encoded for utf-8 in hex is `\xC2\xA0`.

Fixing this test case will cause the unit test to not pass, since the
codepoint `\u00A0` is not handled correctly.
(Failing unittest is expected).

README.md

@@ -68,7 +68,7 @@ target_link_libraries(YOUR_LIBRARY PUBLIC yaml-cpp::yaml-cpp) # The library or e
 
 ## Recent Releases
 
-[yaml-cpp 0.8.0](https://github.com/jbeder/yaml-cpp/releases/tag/0.8.0) released!
+[yaml-cpp 0.9.0](https://github.com/jbeder/yaml-cpp/releases/tag/yaml-cpp-0.9.0) released!
 
 [yaml-cpp 0.3.0](https://github.com/jbeder/yaml-cpp/releases/tag/release-0.3.0) is still available if you want the old API.
 

include/yaml-cpp/exceptions.h

@@ -89,6 +89,7 @@ const char* const INVALID_ANCHOR = "invalid anchor";
 const char* const INVALID_ALIAS = "invalid alias";
 const char* const INVALID_TAG = "invalid tag";
 const char* const BAD_FILE = "bad file";
+const char* const UNEXPECTED_TOKEN_AFTER_DOC = "unexpected token after end of document";
 
 template <typename T>
 inline const std::string KEY_NOT_FOUND_WITH_KEY(

src/binary.cpp

@@ -74,7 +74,8 @@ std::vector<unsigned char> DecodeBase64(const std::string &input) {
   unsigned char *out = &ret[0];
 
   unsigned value = 0;
-  for (std::size_t i = 0, cnt = 0; i < input.size(); i++) {
+  std::size_t cnt = 0;
+  for (std::size_t i = 0; i < input.size(); i++) {
     if (std::isspace(static_cast<unsigned char>(input[i]))) {
       // skip newlines
       continue;
@@ -84,14 +85,20 @@ std::vector<unsigned char> DecodeBase64(const std::string &input) {
       return ret_type();
 
     value = (value << 6) | d;
-    if (cnt % 4 == 3) {
+    if (cnt == 3) {
       *out++ = value >> 16;
       if (i > 0 && input[i - 1] != '=')
         *out++ = value >> 8;
       if (input[i] != '=')
         *out++ = value;
+      cnt = 0;
+    } else {
+      ++cnt;
     }
-    ++cnt;
+  }
+  if (cnt != 0) {
+    // An invalid number of characters were encountered.
+    return ret_type();
   }
 
   ret.resize(out - &ret[0]);

src/exp.cpp

@@ -117,11 +117,11 @@ std::string Escape(Stream& in) {
     case 'N':
       return "\x85";
     case '_':
-      return "\xA0";
+      return "\xC2\xA0";      // NBSP (U+00A0)
     case 'L':
-      return "\xE2\x80\xA8";  // LS (#x2028)
+      return "\xE2\x80\xA8";  // LS (U+2028)
     case 'P':
-      return "\xE2\x80\xA9";  // PS (#x2029)
+      return "\xE2\x80\xA9";  // PS (U+2029)
     case 'x':
       return Escape(in, 2);
     case 'u':

src/fptostring.cpp

@@ -28,7 +28,7 @@ namespace fp_formatting {
  * assert(buffer[1] == '2');
  * assert(buffer[2] == '3');
  */
-int ConvertToChars(char* begin, char* end, size_t value, int width=1) {
+int ConvertToChars(char* begin, char* end, uint64_t value, int width=1) {
   // precondition of this function (will trigger in debug build)
   assert(width >= 1);
   assert(end >= begin);       // end must be after begin

src/parse.cpp

@@ -53,7 +53,7 @@ std::vector<Node> LoadAll(std::istream& input) {
   Parser parser(input);
   while (true) {
     NodeBuilder builder;
-    if (!parser.HandleNextDocument(builder) || builder.Root().IsNull()) {
+    if (!parser.HandleNextDocument(builder)) {
       break;
     }
     docs.push_back(builder.Root());

src/parser.cpp

@@ -33,9 +33,24 @@ bool Parser::HandleNextDocument(EventHandler& eventHandler) {
     return false;
   }
 
+  auto oldPos = m_pScanner->peek().mark.pos;
+
   SingleDocParser sdp(*m_pScanner, *m_pDirectives);
   sdp.HandleDocument(eventHandler);
-  return true;
+
+  // checks if progress was made
+  // 1. if scanner has no more tokens, progress was made
+  if (m_pScanner->empty()) {
+    return true;
+  }
+
+  // 2. if token position has changed, progress was made
+  auto newPos = m_pScanner->peek().mark.pos;
+  if (newPos != oldPos) {
+    return true;
+  }
+  // No progress was made, no further processing
+  return false;
 }
 
 void Parser::ParseDirectives() {

src/singledocparser.cpp

@@ -41,8 +41,13 @@ void SingleDocParser::HandleDocument(EventHandler& eventHandler) {
 
   eventHandler.OnDocumentEnd();
 
+  // check if any tokens left after the text
+  if (!m_scanner.empty() && m_scanner.peek().type != Token::DOC_END
+      && m_scanner.peek().type != Token::DOC_START)
+    throw ParserException(m_scanner.mark(), ErrorMsg::UNEXPECTED_TOKEN_AFTER_DOC);
+
   // and finally eat any doc ends we see
-  while (!m_scanner.empty() && m_scanner.peek().type == Token::DOC_END)
+  if (!m_scanner.empty() && m_scanner.peek().type == Token::DOC_END)
     m_scanner.pop();
 }
 

test/binary_test.cpp

@@ -12,3 +12,9 @@ TEST(BinaryTest, DecodingNoCrashOnNegative) {
   const std::vector<unsigned char> &result = YAML::DecodeBase64(input);
   EXPECT_TRUE(result.empty());
 }
+
+TEST(BinaryTest, DecodingTooShort) {
+  std::string input{90, 71, 86, 104, 90, 71, 74, 108, 90, 87, 89};
+  const std::vector<unsigned char> &result = YAML::DecodeBase64(input);
+  EXPECT_TRUE(result.empty());
+}

test/cmake/CMakeLists.txt

@@ -1,4 +1,4 @@
-cmake_minimum_required(VERSION 3.5)
+cmake_minimum_required(VERSION 3.5...3.30)
 project(yaml-cpp-consumer LANGUAGES CXX)
 
 find_package(yaml-cpp CONFIG REQUIRED)

test/integration/handler_spec_test.cpp

@@ -751,7 +751,7 @@ TEST_F(HandlerSpecTest, Ex5_13_EscapedCharacters) {
       OnScalar(_, "!", 0,
                "Fun with \x5C \x22 \x07 \x08 \x1B \x0C \x0A \x0D \x09 \x0B " +
                    std::string("\x00", 1) +
-                   " \x20 \xA0 \x85 \xe2\x80\xa8 \xe2\x80\xa9 A A A"));
+                   " \x20 \xC2\xA0 \x85 \xe2\x80\xa8 \xe2\x80\xa9 A A A"));
   EXPECT_CALL(handler, OnDocumentEnd());
   Parse(ex5_13);
 }

test/integration/load_node_test.cpp

@@ -368,13 +368,28 @@ TEST(NodeTest, LoadCommaSeparatedStrings) {
   EXPECT_THROW(Load(R"(,foo)"), ParserException);
 }
 
-TEST(NodeSpecTest, InfiniteLoopNodes) {
+TEST(NodeTest, InfiniteLoopNodes) {
   // Until yaml-cpp <= 0.8.0 this caused an infinite loop;
   // After, it triggers an exception (but LoadAll is smart enough to avoid
   // the infinite loop in any case).
   EXPECT_THROW(LoadAll(R"(,)"), ParserException);
 }
 
+TEST(NodeTest, MultipleDocumentsBeginning) {
+  std::vector<Node> docs = LoadAll("\n---\n---\nA\n");
+  EXPECT_EQ(docs.size(), 2);
+}
+
+TEST(NodeTest, MultipleDocumentsEnds) {
+  std::vector<Node> docs = LoadAll("\n...\nA\n...\n");
+  EXPECT_EQ(docs.size(), 2);
+}
+
+TEST(NodeTest, MultipleDocumentsEndsWithEmptyDocs) {
+  std::vector<Node> docs = LoadAll("\n...\nA\n...\n...\nB\n...");
+  EXPECT_EQ(docs.size(), 4);
+}
+
 struct NewLineStringsTestCase {
   std::string input;
   std::string expected_content;
@@ -446,6 +461,10 @@ TEST(LoadNodeTest, BlockCREncoded) {
   EXPECT_EQ(1, node["followup"].as<int>());
 }
 
+TEST(LoadNodeTest, IncorrectSeqEnd) {
+  EXPECT_THROW(Load("[foo]_bar"), ParserException);
+}
+
 
 }  // namespace
 }  // namespace YAML

test/integration/node_spec_test.cpp

@@ -467,7 +467,7 @@ TEST(NodeSpecTest, Ex5_13_EscapedCharacters) {
   EXPECT_TRUE(doc.as<std::string>() ==
               "Fun with \x5C \x22 \x07 \x08 \x1B \x0C \x0A \x0D \x09 \x0B " +
                   std::string("\x00", 1) +
-                  " \x20 \xA0 \x85 \xe2\x80\xa8 \xe2\x80\xa9 A A A");
+                  " \x20 \xC2\xA0 \x85 \xe2\x80\xa8 \xe2\x80\xa9 A A A");
 }
 
 TEST(NodeSpecTest, Ex5_14_InvalidEscapedCharacters) {