fix row key not in standard format (#7901)

dbms/src/Storages/DeltaMerge/RowKeyRange.h

@@ -17,6 +17,7 @@
 #include <Core/Types.h>
 #include <Functions/FunctionHelpers.h>
 #include <IO/WriteHelpers.h>
+#include <Poco/Logger.h>
 #include <Storages/DeltaMerge/DeltaMergeDefines.h>
 #include <Storages/DeltaMerge/DeltaMergeHelpers.h>
 #include <Storages/Transaction/DatumCodec.h>
@@ -25,6 +26,7 @@
 #include <Storages/Transaction/TiKVKeyValue.h>
 #include <Storages/Transaction/TiKVRecordFormat.h>
 #include <Storages/Transaction/Types.h>
+#include <common/logger_useful.h>
 
 namespace DB::DM
 {
@@ -83,6 +85,46 @@ struct RowKeyValue
         {
             size_t cursor = 0;
             int_value = DB::DecodeInt64(cursor, *value);
+            if (unlikely(value->size() != sizeof(Int64)))
+            {
+                // For int type handle, the standard key enconding format should be t{table_id}_r{handle_value}.
+                // But TiKV may generate region range keys which are not strictly following the standard format.
+                // More concretely, the key may be t{table_id}_r{handle_value} + some other bytes.
+                // We need to adapt the key to the standard format.
+                // For example, the key may be t100_r1000 + 0x00, we need to adapt it to t100_r1001.
+                // This is ok, because
+                //  1) if the key is the start range, then [t100_r1000 + 0x00, xxx) has the same semantics with [t100_r1001, xxx)
+                //  2) if the key is the end range, then [xxx, t100_r1000 + 0x00) also has the same semantics with [xxx, t100_r1001)
+                //
+                // Note if the `int_value` is Int64::max_value,
+                // it is a value generated by tiflash itself to means +inf()(which is RowKeyValue::INT_HANDLE_MAX_KEY).
+                // So we can just ignore it.
+                if (value->size() != sizeof(UInt64) + 1 || value->back() != 0x00)
+                {
+                    LOG_FMT_WARNING(
+                        &Poco::Logger::get("RowKeyValue"),
+                        "Meet rowkey {} with unexpected encoding format",
+                        Redact::keyToDebugString(value->data(), value->size()));
+                }
+                else
+                {
+                    if (int_value < std::numeric_limits<Int64>::max())
+                    {
+                        LOG_FMT_WARNING(
+                            &Poco::Logger::get("RowKeyValue"),
+                            "Meet rowkey {} which has an extra zero suffix",
+                            Redact::keyToDebugString(value->data(), value->size()));
+                        int_value = int_value + 1;
+                        WriteBufferFromOwnString ss;
+                        DB::EncodeInt64(int_value, ss);
+                        value = std::make_shared<String>(ss.releaseStr());
+                    }
+                    else
+                    {
+                        // ignore RowKeyValue::INT_HANDLE_MAX_KEY
+                    }
+                }
+            }
         }
     }
 

dbms/src/Storages/DeltaMerge/tests/gtest_key_range.cpp

@@ -98,6 +98,22 @@ TEST(RowKeyRange_test, RedactRangeFromCommonHandle)
     Redact::setRedactLog(false); // restore flags
 }
 
+TEST(RowKey, DecodeKeyWithExtraZeroSuffix)
+{
+    // Note: {20,00} will be regarded as Key=21 in RowKeyRange::fromRegionRange.
+    auto key_end = RecordKVFormat::genRawKey(1, 20);
+    key_end.push_back(0);
+    auto tikv_key_end = RecordKVFormat::encodeAsTiKVKey(key_end);
+    const auto range_keys = std::make_shared<RegionRangeKeys>(
+        RecordKVFormat::genKey(1, 0),
+        std::move(tikv_key_end));
+    const auto range = RowKeyRange::fromRegionRange(
+        range_keys,
+        /* table_id */ 1,
+        /* is_common_handle */ false,
+        /* row_key_column_size */ 1);
+    EXPECT_EQ(0, compare(RowKeyValue::fromHandle(21).toRowKeyValueRef(), range.getEnd()));
+}
 } // namespace tests
 } // namespace DM
 } // namespace DB