fix json/yaml issues with illegal format (alibaba#1054)

Author: zhenjiaseu

ecosystem/simple_dom.cpp

@@ -177,9 +177,9 @@ struct JHandler : public BaseReaderHandler<UTF8<>, JHandler> {
     vector<vector<JNode>> _nodes{1};
     str _key;
     JNode* _root;
-    JHandler(const char* text, bool text_ownership) {
+    JHandler(const char* text) {
         assert(_nodes.size() == 1);
-        _root = new JNode(text, text_ownership);
+        _root = new JNode(text, false);
     }
     ~JHandler() {
         delete _root;
@@ -277,15 +277,26 @@ static bool fix_trail(estring_view text, char header, char trailer) {
 }
 
 static NodeImpl* parse_json(char* text, size_t size, int flags) {
+    // without kParseStopWhenDoneFlag, rapidjson will expect the last character of
+    // the JSON string to be '\0'.
     const auto kFlags = kParseNumbersAsStringsFlag | kParseBoolsAsStringFlag |
-             kParseInsituFlag | kParseCommentsFlag | kParseTrailingCommasFlag;
+                        kParseInsituFlag | kParseCommentsFlag |
+                        kParseTrailingCommasFlag | kParseStopWhenDoneFlag;
     if (!fix_trail({text, size}, '{', '}')) return nullptr;
-    JHandler h(text, flags & DOC_FREE_TEXT_ON_DESTRUCTION);
+    JHandler h(text);
     using Encoding = UTF8<>;
     GenericInsituStringStream<Encoding> s(text);
     GenericReader<Encoding, Encoding> reader;
-    reader.Parse<kFlags>(s, h);
-    return h.get_root();
+    auto res = reader.Parse<kFlags>(s, h);
+    if (!res) {
+        LOG_ERROR("json parsing failed with error ` at `, origin size `",
+                  res.Code(), res.Offset(), size);
+        return nullptr;
+    }
+    auto root = h.get_root();
+    if (flags & DOC_FREE_TEXT_ON_DESTRUCTION)
+        root->_flags |= NodeImpl::FLAG_TEXT_OWNERSHIP;
+    return root;
 }
 
 using namespace rapidxml;
@@ -359,6 +370,12 @@ class YAMLNode : public DocNode<YAMLNode> {
 };
 
 static NodeImpl* parse_yaml(char* text, size_t size, int flags) {
+    ryml::s_default_callbacks.m_error = [](const char* msg, size_t msg_len,
+                                           ryml::Location location,
+                                           void* user_data) {
+        LOG_ERROR("yaml parsing failed with `", std::string_view(msg, msg_len));
+        throw std::runtime_error("yaml parsing failed");
+    };
     auto yaml = ryml::parse_in_place({text, size});
     auto root = make_unique<YAMLNode>(text, flags & DOC_FREE_TEXT_ON_DESTRUCTION);
     assert(root);
@@ -451,7 +468,7 @@ Node parse(char* text, size_t size, int flags) {
     constexpr static Parser parsers[] = {&parse_json, &parse_xml,
                                          &parse_yaml, &parse_ini};
     auto i = flags & DOC_TYPE_MASK;
-    if ((size_t) i > LEN(parsers)) {
+    if ((size_t) i >= LEN(parsers)) {
         if (flags & DOC_FREE_TEXT_IF_PARSING_FAILED) free(text);
         LOG_ERROR_RETURN(EINVAL, nullptr, "invalid document type ", HEX(i));
     }

ecosystem/test/test_simple_dom.cpp

@@ -218,6 +218,34 @@ TEST(simple_dom, json) {
     expect_eq_vals(doc["a"], {"1", "2", "3", "4"});
 }
 
+TEST(simple_dom, illegal_formats) {
+    auto ck_parse_copy = [](const std::vector<string>& formats, int type) {
+        for (const auto& s : formats) {
+            auto doc = parse_copy((const char*)s.data(), s.size(), type);
+            EXPECT_FALSE(doc);
+        }
+    };
+    auto ck_parse = [](std::vector<string>& formats, int type) {
+        for (auto& s : formats) {
+            auto doc = parse((char*)s.data(), s.size(), type);
+            EXPECT_FALSE(doc);
+        }
+    };
+
+    std::vector<std::string> xmls = {"xml1", "<<xml2>>", "4<xml>"};
+    ck_parse_copy(xmls, DOC_XML);
+    ck_parse(xmls, DOC_XML);
+    std::vector<std::string> jsons = {"json1", "33{{json22}}22", "{3{json}}"};
+    ck_parse_copy(jsons, DOC_JSON);
+    ck_parse(jsons, DOC_JSON);
+    std::vector<std::string> yamls = {"[1,2,3,4,5", "{{a:1,b:}", "{yaml3}}"};
+    ck_parse_copy(yamls, DOC_YAML);
+    ck_parse(yamls, DOC_YAML);
+    std::vector<std::string> inis = {"ini1:", ":ini2", "ini3"};
+    ck_parse_copy(inis, DOC_INI);
+    ck_parse(inis, DOC_INI);
+}
+
 TEST(simple_dom, fix_trail) {
     static char s0[] = R"({"a":"b"}})";
     static char s1[] = R"({"a":"b")";