issue #1409: fix NUL-byte handling

arangod/Aql/Ast.cpp

@@ -1012,7 +1012,7 @@ AstNode* Ast::createNodeValueString (char const* value,
     THROW_ARANGO_EXCEPTION(TRI_ERROR_OUT_OF_MEMORY);
   }
 
-  if (*value == '\0') {
+  if (length == 0) {
     // performance optimization:
     // return a pointer to the singleton empty string node
     // note: these nodes are never registered nor freed

js/common/tests/shell-attributes.js

@@ -485,6 +485,87 @@ function AttributesSuite () {
       assertEqual("foo\u0001bar\u0000baz", result[0].abc);
       assertEqual("\u0000test\u0000test", result[0].def);
       assertEqual("abc\u0000", result[0]["123"]);
+    },
+
+////////////////////////////////////////////////////////////////////////////////
+/// @brief test values containing NULL bytes
+////////////////////////////////////////////////////////////////////////////////
+
+    testNullBytesValues : function () {
+      var docs = [
+        { value: "\u0000zero\u0000", _key: "first" },
+        { value: "\u0000\r\nnew line", _key: "second" },
+        { value: "\u0000zero", _key: "third" },
+        { value: "0\r\nxxx", _key: "fourth" }
+      ];
+
+      docs.forEach(function(doc) {
+        c.insert(doc);
+      });
+
+      // test return value of document()
+      docs.forEach(function(doc) {
+        var d = c.document(doc._key);
+        assertEqual(doc.value, d.value);
+      });
+
+      // test return value of AQL, using literals
+      docs.forEach(function(doc) {
+        var result = db._query("FOR doc IN @@collection FILTER doc.value == " + 
+                               JSON.stringify(doc.value) + " RETURN doc", { 
+          "@collection" : c.name()
+        }).toArray();
+
+        assertEqual(1, result.length, doc);
+        assertEqual(doc.value, result[0].value);
+        assertEqual(doc._key, result[0]._key);
+      });
+
+      // test return value of AQL, using bind parameters
+      docs.forEach(function(doc) {
+        var result = db._query("FOR doc IN @@collection FILTER doc.value == @value RETURN doc", { 
+          "@collection" : c.name(),
+          value: doc.value 
+        }).toArray();
+
+        assertEqual(1, result.length, doc);
+        assertEqual(doc.value, result[0].value);
+        assertEqual(doc._key, result[0]._key);
+      });
+
+      // test return value of AQL
+      var result;
+      result = db._query("FOR doc IN @@collection FILTER SUBSTRING(doc.value, 0, 5) == @value RETURN doc._key", { 
+        "@collection" : c.name(),
+        value: "\u0000zero" 
+      }).toArray().sort();
+
+      assertEqual(2, result.length);
+      assertEqual([ "first", "third" ], result);
+
+      result = db._query("FOR doc IN @@collection FILTER SUBSTRING(doc.value, 0, 6) == @value RETURN doc._key", { 
+        "@collection" : c.name(),
+        value: "\u0000zero\u0000" 
+      }).toArray();
+
+      assertEqual(1, result.length);
+      assertEqual([ "first" ], result);
+
+      result = db._query("FOR doc IN @@collection FILTER SUBSTRING(doc.value, 0, 1) == " + 
+                         JSON.stringify("\0") + " RETURN doc._key", { 
+        "@collection" : c.name()
+      }).toArray().sort();
+
+      assertEqual(3, result.length);
+      assertEqual([ "first", "second", "third" ], result);
+
+      result = db._query("FOR doc IN @@collection FILTER SUBSTRING(doc.value, 0, 1) == @value RETURN doc._key", { 
+        "@collection" : c.name(),
+        value: "\0",
+      }).toArray().sort();
+
+      assertEqual(3, result.length);
+      assertEqual([ "first", "second", "third" ], result);
     }
 
   };

lib/Basics/json-utilities.cpp

@@ -244,14 +244,18 @@ int TRI_CompareValuesJson (TRI_json_t const* lhs,
     case TRI_JSON_STRING_REFERENCE: {
       // same for STRING and STRING_REFERENCE
       int res;
+      size_t const nl = lhs->_value._string.length - 1;   
+      size_t const nr = rhs->_value._string.length - 1;   
       if (useUTF8) {
         res = TRI_compare_utf8(lhs->_value._string.data,
-                               lhs->_value._string.length - 1, 
+                               nl,
                                rhs->_value._string.data,
-                               rhs->_value._string.length - 1);
+                               nr);
       }
       else {
-        res = strcmp(lhs->_value._string.data, rhs->_value._string.data);
+        // beware of strings containing NUL bytes
+        size_t len = nl < nr ? nl : nr;
+        res = memcmp(lhs->_value._string.data, rhs->_value._string.data, len);
       }
       if (res < 0) {
         return -1;
@@ -259,8 +263,13 @@ int TRI_CompareValuesJson (TRI_json_t const* lhs,
       else if (res > 0) {
         return 1;
       }
+      // res == 0
+      if (nl == nr) {
         return 0;
       }
+      // res == 0, but different string lengths
+      return nl < nr ? -1 : 1;
+    }
 
     case TRI_JSON_ARRAY: {
       size_t const nl = TRI_LengthVector(&lhs->_value._objects);