Fixed parsing errors

Author: GrigoriyPA

ydb/core/fq/libs/row_dispatcher/format_handler/parsers/json_parser.cpp

@@ -414,23 +414,23 @@ class TJsonParser : public TTopicParserBase {
 
         simdjson::ondemand::document_stream documents;
         CHECK_JSON_ERROR(Parser.iterate_many(values, size, simdjson::ondemand::DEFAULT_BATCH_SIZE).get(documents)) {
-            return TStatus::Fail(EStatusId::BAD_REQUEST, TStringBuilder() << "Failed to parse message batch from offset " << Buffer.Offsets.front() << ", json documents was corrupted: " << simdjson::error_message(error));
+            return TStatus::Fail(EStatusId::BAD_REQUEST, TStringBuilder() << "Failed to parse message batch from offset " << Buffer.Offsets.front() << ", json documents was corrupted: " << simdjson::error_message(error) << " Current data batch:\n" << TruncateString(std::string_view(values, size)));
         }
 
         size_t rowId = 0;
         for (auto document : documents) {
             if (Y_UNLIKELY(rowId >= Buffer.NumberValues)) {
-                return TStatus::Fail(EStatusId::INTERNAL_ERROR, TStringBuilder() << "Failed to parse json messages, expected " << Buffer.NumberValues << " json rows from offset " << Buffer.Offsets.front() << " but got " << rowId + 1);
+                return TStatus::Fail(EStatusId::INTERNAL_ERROR, TStringBuilder() << "Failed to parse json messages, expected " << Buffer.NumberValues << " json rows from offset " << Buffer.Offsets.front() << " but got " << rowId + 1 << " (expected one json row for each offset from topic API in json each row format, maybe initial data was corrupted or messages is not in json format), current data batch:\n" << TruncateString(std::string_view(values, size)));
             }
 
             const ui64 offset = Buffer.Offsets[rowId];
             CHECK_JSON_ERROR(document.error()) {
-                return TStatus::Fail(EStatusId::BAD_REQUEST, TStringBuilder() << "Failed to parse json message for offset " << offset << ", json document was corrupted: " << simdjson::error_message(error));
+                return TStatus::Fail(EStatusId::BAD_REQUEST, TStringBuilder() << "Failed to parse json message for offset " << offset << ", json document was corrupted: " << simdjson::error_message(error) << " Current data batch:\n" << TruncateString(std::string_view(values, size)));
             }
 
             for (auto item : document.get_object()) {
                 CHECK_JSON_ERROR(item.error()) {
-                    return TStatus::Fail(EStatusId::BAD_REQUEST, TStringBuilder() << "Failed to parse json message for offset " << offset << ", json item was corrupted: " << simdjson::error_message(error));
+                    return TStatus::Fail(EStatusId::BAD_REQUEST, TStringBuilder() << "Failed to parse json message for offset " << offset << ", json item was corrupted: " << simdjson::error_message(error) << " Current data batch:\n" << TruncateString(std::string_view(values, size)));
                 }
 
                 const auto it = ColumnsIndex.find(item.escaped_key().value());
@@ -445,7 +445,7 @@ class TJsonParser : public TTopicParserBase {
         }
 
         if (Y_UNLIKELY(rowId != Buffer.NumberValues)) {
-            return TStatus::Fail(EStatusId::INTERNAL_ERROR, TStringBuilder() << "Failed to parse json messages, expected " << Buffer.NumberValues << " json rows from offset " << Buffer.Offsets.front() << " but got " << rowId);
+            return TStatus::Fail(EStatusId::INTERNAL_ERROR, TStringBuilder() << "Failed to parse json messages, expected " << Buffer.NumberValues << " json rows from offset " << Buffer.Offsets.front() << " but got " << rowId << " (expected one json row for each offset from topic API in json each row format, maybe initial data was corrupted or messages is not in json format), current data batch:\n" << TruncateString(std::string_view(values, size)));
         }
 
         const ui64 firstOffset = Buffer.Offsets.front();

ydb/core/fq/libs/row_dispatcher/format_handler/ut/topic_parser_ut.cpp

@@ -414,9 +414,9 @@ Y_UNIT_TEST_SUITE(TestJsonParser) {
     Y_UNIT_TEST_F(JsonStructureValidation, TJsonParserFixture) {
         CheckSuccess(MakeParser({{"a1", "[OptionalType; [DataType; String]]"}}));
         CheckColumnError(R"({"a1": Yelse})", 0, EStatusId::BAD_REQUEST, TStringBuilder() << "Failed to parse json string at offset " << FIRST_OFFSET << ", got parsing error for column 'a1' with type [OptionalType; [DataType; String]] subissue: { <main>: Error: Failed to determine json value type, current token: 'Yelse', error: TAPE_ERROR: The JSON document has an improper structure: missing or superfluous commas, braces, missing keys, etc. }");
-        CheckBatchError(R"({"a1": "st""r"})", EStatusId::BAD_REQUEST, TStringBuilder() << "Failed to parse json message for offset " << FIRST_OFFSET + 1 << ", json item was corrupted: TAPE_ERROR: The JSON document has an improper structure: missing or superfluous commas, braces, missing keys, etc.");
-        CheckBatchError(R"({"a1": "x"} {"a1": "y"})", EStatusId::INTERNAL_ERROR, TStringBuilder() << "Failed to parse json messages, expected 1 json rows from offset " << FIRST_OFFSET + 2 << " but got 2");
-        CheckBatchError(R"({)", EStatusId::INTERNAL_ERROR, TStringBuilder() << "Failed to parse json messages, expected 1 json rows from offset " << FIRST_OFFSET + 3 << " but got 0");
+        CheckBatchError(R"({"a1": "st""r"})", EStatusId::BAD_REQUEST, TStringBuilder() << "Failed to parse json message for offset " << FIRST_OFFSET + 1 << ", json item was corrupted: TAPE_ERROR: The JSON document has an improper structure: missing or superfluous commas, braces, missing keys, etc. Current data batch: {\"a1\": \"st\"\"r\"}");
+        CheckBatchError(R"({"a1": "x"} {"a1": "y"})", EStatusId::INTERNAL_ERROR, TStringBuilder() << "Failed to parse json messages, expected 1 json rows from offset " << FIRST_OFFSET + 2 << " but got 2 (expected one json row for each offset from topic API in json each row format, maybe initial data was corrupted or messages is not in json format), current data batch: {\"a1\": \"x\"} {\"a1\": \"y\"}");
+        CheckBatchError(R"({)", EStatusId::INTERNAL_ERROR, TStringBuilder() << "Failed to parse json messages, expected 1 json rows from offset " << FIRST_OFFSET + 3 << " but got 0 (expected one json row for each offset from topic API in json each row format, maybe initial data was corrupted or messages is not in json format), current data batch: {");
     }
 }