JSON: detect duplicate dictionary keys while parsing

Jay Berkenbilt
1 parent 6d4e3ba8
Showing 6 changed files with 44 additions and 1 deletions
include/qpdf/JSON.hh
libqpdf/JSON.cc
libtests/libtests.testcov
libtests/qtest/json_parse.test
libtests/qtest/json_parse/bad-40.json
libtests/qtest/json_parse/bad-40.out
@@ -42,6 +42,7 @@
 #include <list>
 #include <map>
 #include <memory>
+#include <set>
 #include <string>
 #include <vector>
@@ -149,6 +150,14 @@ class JSON
     QPDF_DLL
     bool isDictionary() const;
+    // If the key is already in the dictionary, return true.
+    // Otherwise, mark it has seen and return false. This is primarily
+    // intended to used by the parser to detect duplicate keys when
+    // the reactor blocks them from being added to the final
+    // dictionary.
+    QPDF_DLL
+    bool checkDictionaryKeySeen(std::string const& key);
+
     // Accessors. Accessor behavior:
     //
     // - If argument is wrong type, including null, return false
@@ -314,6 +323,7 @@ class JSON
         virtual ~JSON_dictionary() = default;
         virtual void write(Pipeline*, size_t depth) const;
         std::map<std::string, std::shared_ptr<JSON_value>> members;
+        std::set<std::string> parsed_keys;
     };
     struct JSON_array: public JSON_value
     {
@@ -274,6 +274,21 @@ JSON::addDictionaryMember(std::string const&amp; key, JSON const&amp; val)
     return obj->members[encode_string(key)];
 }
+bool
+JSON::checkDictionaryKeySeen(std::string const& key)
+{
+    JSON_dictionary* obj = dynamic_cast<JSON_dictionary*>(this->m->value.get());
+    if (0 == obj) {
+        throw std::logic_error(
+            "JSON::checkDictionaryKey called on non-dictionary");
+    }
+    if (obj->parsed_keys.count(key)) {
+        return true;
+    }
+    obj->parsed_keys.insert(key);
+    return false;
+}
+
 JSON
 JSON::makeArray()
 {
@@ -565,7 +580,8 @@ namespace
             u_count(0),
             offset(0),
             done(false),
-            parser_state(ps_top)
+            parser_state(ps_top),
+            dict_key_offset(0)
         {
         }
@@ -625,6 +641,7 @@ namespace
         std::vector<std::shared_ptr<JSON>> stack;
         std::vector<parser_state_e> ps_stack;
         std::string dict_key;
+        size_t dict_key_offset;
     };
 } // namespace
@@ -1201,11 +1218,18 @@ JSONParser::handleToken()
         case ps_dict_begin:
         case ps_dict_after_comma:
             this->dict_key = s_value;
+            this->dict_key_offset = item->getStart();
             item = nullptr;
             next_state = ps_dict_after_key;
             break;
         case ps_dict_after_colon:
+            if (tos->checkDictionaryKeySeen(dict_key)) {
+                QTC::TC("libtests", "JSON parse duplicate key");
+                throw std::runtime_error(
+                    "JSON: offset " + QUtil::uint_to_string(dict_key_offset) +
+                    ": duplicated dictionary key");
+            }
             if (!reactor || !reactor->dictionaryItem(dict_key, *item)) {
                 tos->addDictionaryMember(dict_key, *item);
             }
@@ -92,3 +92,4 @@ JSON optional key 0
 JSON 16 high high 0
 JSON 16 low not after high 0
 JSON 16 dangling high 0
+JSON parse duplicate key 0
@@ -120,6 +120,7 @@ my @bad = (
     "stray low surrogate",      # 37
     "high high surrogate",      # 38
     "dangling high surrogate",  # 39
+    "duplicate dictionary key", # 40
     );
 my $i = 0;
+{
+  "one": 1,
+  "two": 2,
+  "one": 3,
+  "four": 4
+}
+exception: bad-40.json: JSON: offset 28: duplicated dictionary key