From a4e0bbd34f1bbafe96503ad9d6b5b84576fb3509 Mon Sep 17 00:00:00 2001
From: Alex Duan <417921451@qq.com>
Date: Wed, 14 Dec 2022 20:29:58 +0800
Subject: [PATCH 01/15] fix(query): fixed avg function sum over range

---
 source/libs/function/inc/builtinsimpl.h       |   1 +
 .../libs/function/src/detail/tavgfunction.c   | 102 ++++++++++++++++--
 2 files changed, 93 insertions(+), 10 deletions(-)

diff --git a/source/libs/function/inc/builtinsimpl.h b/source/libs/function/inc/builtinsimpl.h
index 307a82e256..a7ac53fab6 100644
--- a/source/libs/function/inc/builtinsimpl.h
+++ b/source/libs/function/inc/builtinsimpl.h
@@ -32,6 +32,7 @@ typedef struct SSumRes {
   int16_t type;
   int64_t prevTs;
   bool    isPrevTsSet;
+  bool    overflow; // if overflow is true, dsum to be used for any type;
 } SSumRes;
 
 typedef struct SMinmaxResInfo {
diff --git a/source/libs/function/src/detail/tavgfunction.c b/source/libs/function/src/detail/tavgfunction.c
index 60bf30d8ed..98ed7f818d 100644
--- a/source/libs/function/src/detail/tavgfunction.c
+++ b/source/libs/function/src/detail/tavgfunction.c
@@ -373,7 +373,16 @@ static int32_t doAddNumericVector(SColumnInfoData* pCol, int32_t type, SInputCol
 
         numOfElems += 1;
         pRes->count += 1;
-        pRes->sum.isum += plist[i];
+
+        // check overflow for int
+        if(pRes->sum.isum > 0 && plist[i] > 0 && INT64_MAX - pRes->sum.isum <= plist[i] ||
+            pRes->sum.isum < 0 && plist[i] < 0 && INT64_MIN - pRes->sum.isum >= plist[i] ) {
+            double dsum = (double) pAvgRes->sum.isum;
+            pRes->sum.overflow = true;
+            pRes->sum.dsum = dsum + plist[i];
+        } else {
+          pRes->sum.isum += plist[i];
+        }
       }
 
       break;
@@ -388,7 +397,16 @@ static int32_t doAddNumericVector(SColumnInfoData* pCol, int32_t type, SInputCol
 
         numOfElems += 1;
         pRes->count += 1;
-        pRes->sum.isum += plist[i];
+        
+        // check overflow
+        if(pRes->sum.isum > 0 && plist[i] > 0 && INT64_MAX - pRes->sum.isum <= plist[i] ||
+            pRes->sum.isum < 0 && plist[i] < 0 && INT64_MIN - pRes->sum.isum >= plist[i] ) {
+            double dsum = (double) pAvgRes->sum.isum;
+            pRes->sum.overflow = true;
+            pRes->sum.dsum = dsum + plist[i];
+        } else {
+          pRes->sum.isum += plist[i];
+        }
       }
       break;
     }
@@ -431,7 +449,16 @@ static int32_t doAddNumericVector(SColumnInfoData* pCol, int32_t type, SInputCol
 
         numOfElems += 1;
         pRes->count += 1;
-        pRes->sum.usum += plist[i];
+    
+        // check overflow
+        if(UINT64_MAX - pRes->sum.usum <= plist[i] ) {
+            double dsum = (double) pRes->sum.isum;
+            pRes->sum.overflow = true;
+            pRes->sum.dsum = dsum + plist[i];
+        } else {
+            pRes->sum.usum += (uint32_t)plist[i];
+        }
+
       }
 
       break;
@@ -446,7 +473,15 @@ static int32_t doAddNumericVector(SColumnInfoData* pCol, int32_t type, SInputCol
 
         numOfElems += 1;
         pRes->count += 1;
-        pRes->sum.usum += plist[i];
+        
+        // check overflow
+        if(UINT64_MAX - pRes->sum.usum <= plist[i] ) {
+            double dsum = (double) pRes->sum.isum;
+            pRes->sum.overflow = true;
+            pRes->sum.dsum = dsum + plist[i];
+        } else {
+            pRes->sum.usum += (uint64_t)plist[i];
+        }
       }
       break;
     }
@@ -565,9 +600,24 @@ int32_t avgFunction(SqlFunctionCtx* pCtx) {
         } else {
           for (int32_t i = pInput->startRowIndex; i < pInput->numOfRows + pInput->startRowIndex; ++i) {
             if (type == TSDB_DATA_TYPE_INT) {
-              pAvgRes->sum.isum += plist[i];
+              // check overflow for int
+              if(pAvgRes->sum.isum > 0 && plist[i] > 0 && INT64_MAX - pAvgRes->sum.isum <= plist[i] ||
+                 pAvgRes->sum.isum < 0 && plist[i] < 0 && INT64_MIN - pAvgRes->sum.isum >= plist[i] ) {
+                  double dsum = (double) pAvgRes->sum.isum;
+                  pAvgRes->sum.overflow = true;
+                  pAvgRes->sum.dsum = dsum + plist[i];
+              } else {
+                pAvgRes->sum.isum += plist[i];
+              }
             } else {
-              pAvgRes->sum.usum += (uint32_t)plist[i];
+              // check overflow for uint
+              if(UINT64_MAX - pAvgRes->sum.usum <= plist[i] ) {
+                  double dsum = (double) pAvgRes->sum.isum;
+                  pAvgRes->sum.overflow = true;
+                  pAvgRes->sum.dsum = dsum + plist[i];
+              } else {
+                  pAvgRes->sum.usum += (uint32_t)plist[i];
+              }
             }
           }
         }
@@ -584,9 +634,24 @@ int32_t avgFunction(SqlFunctionCtx* pCtx) {
         } else {
           for (int32_t i = pInput->startRowIndex; i < pInput->numOfRows + pInput->startRowIndex; ++i) {
             if (type == TSDB_DATA_TYPE_BIGINT) {
-              pAvgRes->sum.isum += plist[i];
+              // check overflow for int
+              if(pAvgRes->sum.isum > 0 && plist[i] > 0 && INT64_MAX - pAvgRes->sum.isum <= plist[i] ||
+                 pAvgRes->sum.isum < 0 && plist[i] < 0 && INT64_MIN - pAvgRes->sum.isum >= plist[i] ) {
+                  double dsum = (double) pAvgRes->sum.isum;
+                  pAvgRes->sum.overflow = true;
+                  pAvgRes->sum.dsum = dsum + plist[i];
+              } else {
+                pAvgRes->sum.isum += plist[i];
+              }
             } else {
-              pAvgRes->sum.usum += (uint64_t)plist[i];
+              // check overflow for uint
+              if(UINT64_MAX - pAvgRes->sum.usum <= plist[i] ) {
+                  double dsum = (double) pAvgRes->sum.isum;
+                  pAvgRes->sum.overflow = true;
+                  pAvgRes->sum.dsum = dsum + plist[i];
+              } else {
+                  pAvgRes->sum.usum += (uint64_t)plist[i];
+              }
             }
           }
         }
@@ -639,9 +704,26 @@ static void avgTransferInfo(SAvgRes* pInput, SAvgRes* pOutput) {
 
   pOutput->type = pInput->type;
   if (IS_SIGNED_NUMERIC_TYPE(pOutput->type)) {
-    pOutput->sum.isum += pInput->sum.isum;
+    int64_t val = pInput->sum.isum;
+    // check overflow
+    if(pOutput->sum.isum > 0 && val > 0 && INT64_MAX - pOutput->sum.isum <= val ||
+        pOutput->sum.isum < 0 && val < 0 && INT64_MIN - pOutput->sum.isum >= val ) {
+        double dsum = (double) pOutput->sum.isum;
+        pOutput->sum.overflow = true;
+        pOutput->sum.dsum = dsum + val;
+    } else {
+      pOutput->sum.isum += val;
+    }
   } else if (IS_UNSIGNED_NUMERIC_TYPE(pOutput->type)) {
-    pOutput->sum.usum += pInput->sum.usum;
+    // check overflow
+    uint64_t val = pInput->sum.usum;
+    if(UINT64_MAX - pOutput->sum.usum <= val ) {
+        double dsum = (double) pOutput->sum.isum;
+        pOutput->sum.overflow = true;
+        pOutput->sum.dsum = dsum + val;
+    } else {
+        pOutput->sum.usum += val;
+    }
   } else {
     pOutput->sum.dsum += pInput->sum.dsum;
   }

From 74cf77643909f47f394cb7522e3d50e52c05cb35 Mon Sep 17 00:00:00 2001
From: Alex Duan <417921451@qq.com>
Date: Wed, 14 Dec 2022 21:34:50 +0800
Subject: [PATCH 02/15] fix(query): avg function sum overflow

---
 source/libs/function/src/detail/tavgfunction.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/source/libs/function/src/detail/tavgfunction.c b/source/libs/function/src/detail/tavgfunction.c
index 98ed7f818d..193922e218 100644
--- a/source/libs/function/src/detail/tavgfunction.c
+++ b/source/libs/function/src/detail/tavgfunction.c
@@ -377,7 +377,7 @@ static int32_t doAddNumericVector(SColumnInfoData* pCol, int32_t type, SInputCol
         // check overflow for int
         if(pRes->sum.isum > 0 && plist[i] > 0 && INT64_MAX - pRes->sum.isum <= plist[i] ||
             pRes->sum.isum < 0 && plist[i] < 0 && INT64_MIN - pRes->sum.isum >= plist[i] ) {
-            double dsum = (double) pAvgRes->sum.isum;
+            double dsum = (double) pRes->sum.isum;
             pRes->sum.overflow = true;
             pRes->sum.dsum = dsum + plist[i];
         } else {

From 500ee304a6fff8f0197262baaf8d06b5d8e5d92a Mon Sep 17 00:00:00 2001
From: Alex Duan <417921451@qq.com>
Date: Thu, 15 Dec 2022 12:43:28 +0800
Subject: [PATCH 03/15] fix(query): avg sum over range replace with macro

---
 .../libs/function/src/detail/tavgfunction.c   | 154 ++++++------------
 1 file changed, 52 insertions(+), 102 deletions(-)

diff --git a/source/libs/function/src/detail/tavgfunction.c b/source/libs/function/src/detail/tavgfunction.c
index 193922e218..f3bfa39798 100644
--- a/source/libs/function/src/detail/tavgfunction.c
+++ b/source/libs/function/src/detail/tavgfunction.c
@@ -41,6 +41,31 @@
     }                                                                     \
   } while (0)
 
+// define signed number sum with check overflow
+#define CHECK_OVERFLOW_SUM_SIGNED(out, val)                                      \
+  if (out->sum.overflow) {                                                       \
+    out->sum.dsum += val;                                                        \
+  } else if (out->sum.isum > 0 && val > 0 && INT64_MAX - out->sum.isum <= val || \
+             out->sum.isum < 0 && val < 0 && INT64_MIN - out->sum.isum >= val) { \
+    double dsum = (double)out->sum.isum;                                         \
+    out->sum.overflow = true;                                                    \
+    out->sum.dsum = dsum + val;                                                  \
+  } else {                                                                       \
+    out->sum.isum += val;                                                        \
+  }
+
+// define unsigned number sum with check overflow
+#define CHECK_OVERFLOW_SUM_UNSIGNED (out, val)                \
+  if (out->sum.overflow) {                                    \
+    out->sum.dsum += val;                                     \
+  } else if (UINT64_MAX - out->sum.usum <= val) {             \
+    double dsum = (double)out->sum.usum;                      \
+    out->sum.overflow = true;                                 \
+    out->sum.dsum = dsum + val;                               \
+  } else {                                                    \
+    out->sum.usum += val;                                     \
+  }
+
 typedef struct SAvgRes {
   double  result;
   SSumRes sum;
@@ -344,7 +369,7 @@ static int32_t doAddNumericVector(SColumnInfoData* pCol, int32_t type, SInputCol
 
         numOfElems += 1;
         pRes->count += 1;
-        pRes->sum.isum += plist[i];
+        CHECK_OVERFLOW_SUM_SIGNED(pRes, plist[i])
       }
 
       break;
@@ -359,7 +384,7 @@ static int32_t doAddNumericVector(SColumnInfoData* pCol, int32_t type, SInputCol
 
         numOfElems += 1;
         pRes->count += 1;
-        pRes->sum.isum += plist[i];
+        CHECK_OVERFLOW_SUM_SIGNED(pRes, plist[i])
       }
       break;
     }
@@ -373,16 +398,7 @@ static int32_t doAddNumericVector(SColumnInfoData* pCol, int32_t type, SInputCol
 
         numOfElems += 1;
         pRes->count += 1;
-
-        // check overflow for int
-        if(pRes->sum.isum > 0 && plist[i] > 0 && INT64_MAX - pRes->sum.isum <= plist[i] ||
-            pRes->sum.isum < 0 && plist[i] < 0 && INT64_MIN - pRes->sum.isum >= plist[i] ) {
-            double dsum = (double) pRes->sum.isum;
-            pRes->sum.overflow = true;
-            pRes->sum.dsum = dsum + plist[i];
-        } else {
-          pRes->sum.isum += plist[i];
-        }
+        CHECK_OVERFLOW_SUM_SIGNED(pRes, plist[i])
       }
 
       break;
@@ -397,16 +413,7 @@ static int32_t doAddNumericVector(SColumnInfoData* pCol, int32_t type, SInputCol
 
         numOfElems += 1;
         pRes->count += 1;
-        
-        // check overflow
-        if(pRes->sum.isum > 0 && plist[i] > 0 && INT64_MAX - pRes->sum.isum <= plist[i] ||
-            pRes->sum.isum < 0 && plist[i] < 0 && INT64_MIN - pRes->sum.isum >= plist[i] ) {
-            double dsum = (double) pAvgRes->sum.isum;
-            pRes->sum.overflow = true;
-            pRes->sum.dsum = dsum + plist[i];
-        } else {
-          pRes->sum.isum += plist[i];
-        }
+        CHECK_OVERFLOW_SUM_SIGNED(pRes, plist[i])
       }
       break;
     }
@@ -420,7 +427,7 @@ static int32_t doAddNumericVector(SColumnInfoData* pCol, int32_t type, SInputCol
 
         numOfElems += 1;
         pRes->count += 1;
-        pRes->sum.usum += plist[i];
+        CHECK_OVERFLOW_SUM_UNSIGNED(pRes, plist[i])
       }
 
       break;
@@ -435,7 +442,7 @@ static int32_t doAddNumericVector(SColumnInfoData* pCol, int32_t type, SInputCol
 
         numOfElems += 1;
         pRes->count += 1;
-        pRes->sum.usum += plist[i];
+        CHECK_OVERFLOW_SUM_UNSIGNED(pRes, plist[i])
       }
       break;
     }
@@ -449,16 +456,7 @@ static int32_t doAddNumericVector(SColumnInfoData* pCol, int32_t type, SInputCol
 
         numOfElems += 1;
         pRes->count += 1;
-    
-        // check overflow
-        if(UINT64_MAX - pRes->sum.usum <= plist[i] ) {
-            double dsum = (double) pRes->sum.isum;
-            pRes->sum.overflow = true;
-            pRes->sum.dsum = dsum + plist[i];
-        } else {
-            pRes->sum.usum += (uint32_t)plist[i];
-        }
-
+        CHECK_OVERFLOW_SUM_UNSIGNED(pRes, plist[i])
       }
 
       break;
@@ -473,15 +471,8 @@ static int32_t doAddNumericVector(SColumnInfoData* pCol, int32_t type, SInputCol
 
         numOfElems += 1;
         pRes->count += 1;
+        CHECK_OVERFLOW_SUM_UNSIGNED(pRes, plist[i])
         
-        // check overflow
-        if(UINT64_MAX - pRes->sum.usum <= plist[i] ) {
-            double dsum = (double) pRes->sum.isum;
-            pRes->sum.overflow = true;
-            pRes->sum.dsum = dsum + plist[i];
-        } else {
-            pRes->sum.usum += (uint64_t)plist[i];
-        }
       }
       break;
     }
@@ -563,8 +554,9 @@ int32_t avgFunction(SqlFunctionCtx* pCtx) {
           for (int32_t i = pInput->startRowIndex; i < pInput->numOfRows + pInput->startRowIndex; ++i) {
             if (type == TSDB_DATA_TYPE_TINYINT) {
               pAvgRes->sum.isum += plist[i];
+              CHECK_OVERFLOW_SUM_SIGNED(pAvgRes, plist[i])
             } else {
-              pAvgRes->sum.usum += (uint8_t)plist[i];
+              CHECK_OVERFLOW_SUM_UNSIGNED(pAvgRes, (uint8_t)plist[i])
             }
           }
         }
@@ -581,9 +573,9 @@ int32_t avgFunction(SqlFunctionCtx* pCtx) {
         } else {
           for (int32_t i = pInput->startRowIndex; i < pInput->numOfRows + pInput->startRowIndex; ++i) {
             if (type == TSDB_DATA_TYPE_SMALLINT) {
-              pAvgRes->sum.isum += plist[i];
+              CHECK_OVERFLOW_SUM_SIGNED(pAvgRes, plist[i])
             } else {
-              pAvgRes->sum.usum += (uint16_t)plist[i];
+              CHECK_OVERFLOW_SUM_UNSIGNED(pAvgRes, (uint16_t)plist[i])
             }
           }
         }
@@ -600,24 +592,9 @@ int32_t avgFunction(SqlFunctionCtx* pCtx) {
         } else {
           for (int32_t i = pInput->startRowIndex; i < pInput->numOfRows + pInput->startRowIndex; ++i) {
             if (type == TSDB_DATA_TYPE_INT) {
-              // check overflow for int
-              if(pAvgRes->sum.isum > 0 && plist[i] > 0 && INT64_MAX - pAvgRes->sum.isum <= plist[i] ||
-                 pAvgRes->sum.isum < 0 && plist[i] < 0 && INT64_MIN - pAvgRes->sum.isum >= plist[i] ) {
-                  double dsum = (double) pAvgRes->sum.isum;
-                  pAvgRes->sum.overflow = true;
-                  pAvgRes->sum.dsum = dsum + plist[i];
-              } else {
-                pAvgRes->sum.isum += plist[i];
-              }
+              CHECK_OVERFLOW_SUM_SIGNED(pAvgRes, plist[i])
             } else {
-              // check overflow for uint
-              if(UINT64_MAX - pAvgRes->sum.usum <= plist[i] ) {
-                  double dsum = (double) pAvgRes->sum.isum;
-                  pAvgRes->sum.overflow = true;
-                  pAvgRes->sum.dsum = dsum + plist[i];
-              } else {
-                  pAvgRes->sum.usum += (uint32_t)plist[i];
-              }
+              CHECK_OVERFLOW_SUM_UNSIGNED(pAvgRes, (uint32_t)plist[i])
             }
           }
         }
@@ -634,24 +611,9 @@ int32_t avgFunction(SqlFunctionCtx* pCtx) {
         } else {
           for (int32_t i = pInput->startRowIndex; i < pInput->numOfRows + pInput->startRowIndex; ++i) {
             if (type == TSDB_DATA_TYPE_BIGINT) {
-              // check overflow for int
-              if(pAvgRes->sum.isum > 0 && plist[i] > 0 && INT64_MAX - pAvgRes->sum.isum <= plist[i] ||
-                 pAvgRes->sum.isum < 0 && plist[i] < 0 && INT64_MIN - pAvgRes->sum.isum >= plist[i] ) {
-                  double dsum = (double) pAvgRes->sum.isum;
-                  pAvgRes->sum.overflow = true;
-                  pAvgRes->sum.dsum = dsum + plist[i];
-              } else {
-                pAvgRes->sum.isum += plist[i];
-              }
-            } else {
-              // check overflow for uint
-              if(UINT64_MAX - pAvgRes->sum.usum <= plist[i] ) {
-                  double dsum = (double) pAvgRes->sum.isum;
-                  pAvgRes->sum.overflow = true;
-                  pAvgRes->sum.dsum = dsum + plist[i];
-              } else {
-                  pAvgRes->sum.usum += (uint64_t)plist[i];
-              }
+              CHECK_OVERFLOW_SUM_SIGNED(pAvgRes, plist[i])
+\            } else {
+              CHECK_OVERFLOW_SUM_UNSIGNED(pAvgRes, (uint64_t)plist[i])
             }
           }
         }
@@ -697,6 +659,8 @@ _over:
   return TSDB_CODE_SUCCESS;
 }
 
+
+
 static void avgTransferInfo(SAvgRes* pInput, SAvgRes* pOutput) {
   if (IS_NULL_TYPE(pInput->type)) {
     return;
@@ -704,26 +668,9 @@ static void avgTransferInfo(SAvgRes* pInput, SAvgRes* pOutput) {
 
   pOutput->type = pInput->type;
   if (IS_SIGNED_NUMERIC_TYPE(pOutput->type)) {
-    int64_t val = pInput->sum.isum;
-    // check overflow
-    if(pOutput->sum.isum > 0 && val > 0 && INT64_MAX - pOutput->sum.isum <= val ||
-        pOutput->sum.isum < 0 && val < 0 && INT64_MIN - pOutput->sum.isum >= val ) {
-        double dsum = (double) pOutput->sum.isum;
-        pOutput->sum.overflow = true;
-        pOutput->sum.dsum = dsum + val;
-    } else {
-      pOutput->sum.isum += val;
-    }
+    CHECK_OVERFLOW_SUM_SIGNED(pOutput, pInput->sum.isum)
   } else if (IS_UNSIGNED_NUMERIC_TYPE(pOutput->type)) {
-    // check overflow
-    uint64_t val = pInput->sum.usum;
-    if(UINT64_MAX - pOutput->sum.usum <= val ) {
-        double dsum = (double) pOutput->sum.isum;
-        pOutput->sum.overflow = true;
-        pOutput->sum.dsum = dsum + val;
-    } else {
-        pOutput->sum.usum += val;
-    }
+    CHECK_OVERFLOW_SUM_UNSIGNED(pOutput, pInput->sum.usum)
   } else {
     pOutput->sum.dsum += pInput->sum.dsum;
   }
@@ -823,9 +770,9 @@ int32_t avgCombine(SqlFunctionCtx* pDestCtx, SqlFunctionCtx* pSourceCtx) {
   int16_t              type = pDBuf->type == TSDB_DATA_TYPE_NULL ? pSBuf->type : pDBuf->type;
 
   if (IS_SIGNED_NUMERIC_TYPE(type)) {
-    pDBuf->sum.isum += pSBuf->sum.isum;
+    CHECK_OVERFLOW_SUM_SIGNED(pDBuf, pSBuf->sum.isum)
   } else if (IS_UNSIGNED_NUMERIC_TYPE(type)) {
-    pDBuf->sum.usum += pSBuf->sum.usum;
+    CHECK_OVERFLOW_SUM_UNSIGNED(pDBuf, pSBuf->sum.usum)
   } else {
     pDBuf->sum.dsum += pSBuf->sum.dsum;
   }
@@ -841,7 +788,10 @@ int32_t avgFinalize(SqlFunctionCtx* pCtx, SSDataBlock* pBlock) {
   int32_t  type = pRes->type;
 
   if (pRes->count > 0) {
-    if (IS_SIGNED_NUMERIC_TYPE(type)) {
+    if(pRes->sum.overflow) {
+      // overflow flag set , use dsum
+      pRes->result = pRes->sum.dsum / ((double)pRes->count);
+    }else if (IS_SIGNED_NUMERIC_TYPE(type)) {
       pRes->result = pRes->sum.isum / ((double)pRes->count);
     } else if (IS_UNSIGNED_NUMERIC_TYPE(type)) {
       pRes->result = pRes->sum.usum / ((double)pRes->count);

From e9f7572a5491a0fc823a45ba5aeb91ae7f4d35e2 Mon Sep 17 00:00:00 2001
From: Alex Duan <417921451@qq.com>
Date: Thu, 15 Dec 2022 14:52:53 +0800
Subject: [PATCH 04/15] fix(query): avg sum operation check range

---
 source/libs/function/src/detail/tavgfunction.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/source/libs/function/src/detail/tavgfunction.c b/source/libs/function/src/detail/tavgfunction.c
index f3bfa39798..e31db0f360 100644
--- a/source/libs/function/src/detail/tavgfunction.c
+++ b/source/libs/function/src/detail/tavgfunction.c
@@ -55,7 +55,7 @@
   }
 
 // define unsigned number sum with check overflow
-#define CHECK_OVERFLOW_SUM_UNSIGNED (out, val)                \
+#define CHECK_OVERFLOW_SUM_UNSIGNED(out, val)                 \
   if (out->sum.overflow) {                                    \
     out->sum.dsum += val;                                     \
   } else if (UINT64_MAX - out->sum.usum <= val) {             \

From 1ac428ff61c755694233478097fa52e041a44109 Mon Sep 17 00:00:00 2001
From: Alex Duan <417921451@qq.com>
Date: Thu, 15 Dec 2022 14:54:22 +0800
Subject: [PATCH 05/15] fix(query): avg sum operation check range build error

---
 source/libs/function/src/detail/tavgfunction.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/source/libs/function/src/detail/tavgfunction.c b/source/libs/function/src/detail/tavgfunction.c
index e31db0f360..657320f7ff 100644
--- a/source/libs/function/src/detail/tavgfunction.c
+++ b/source/libs/function/src/detail/tavgfunction.c
@@ -612,7 +612,7 @@ int32_t avgFunction(SqlFunctionCtx* pCtx) {
           for (int32_t i = pInput->startRowIndex; i < pInput->numOfRows + pInput->startRowIndex; ++i) {
             if (type == TSDB_DATA_TYPE_BIGINT) {
               CHECK_OVERFLOW_SUM_SIGNED(pAvgRes, plist[i])
-\            } else {
+            } else {
               CHECK_OVERFLOW_SUM_UNSIGNED(pAvgRes, (uint64_t)plist[i])
             }
           }

From 01309501704226bddf57b84fb84f1ba1b4d88ea0 Mon Sep 17 00:00:00 2001
From: Alex Duan <417921451@qq.com>
Date: Thu, 15 Dec 2022 15:32:53 +0800
Subject: [PATCH 06/15] fix(query): avg sum operation check range merge two
 SAvgRes pointer

---
 source/libs/function/src/detail/tavgfunction.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/source/libs/function/src/detail/tavgfunction.c b/source/libs/function/src/detail/tavgfunction.c
index 657320f7ff..04f5c99d47 100644
--- a/source/libs/function/src/detail/tavgfunction.c
+++ b/source/libs/function/src/detail/tavgfunction.c
@@ -668,9 +668,9 @@ static void avgTransferInfo(SAvgRes* pInput, SAvgRes* pOutput) {
 
   pOutput->type = pInput->type;
   if (IS_SIGNED_NUMERIC_TYPE(pOutput->type)) {
-    CHECK_OVERFLOW_SUM_SIGNED(pOutput, pInput->sum.isum)
+    CHECK_OVERFLOW_SUM_SIGNED(pOutput, pInput->sum.overflow ? pInput->sum.dsum : pInput->sum.isum)
   } else if (IS_UNSIGNED_NUMERIC_TYPE(pOutput->type)) {
-    CHECK_OVERFLOW_SUM_UNSIGNED(pOutput, pInput->sum.usum)
+    CHECK_OVERFLOW_SUM_UNSIGNED(pOutput, pInput->sum.overflow ? pInput->sum.dsum : pInput->sum.usum)
   } else {
     pOutput->sum.dsum += pInput->sum.dsum;
   }

From 7fa0a2e0b59639bdd2f7c52e2243fe5906867ee4 Mon Sep 17 00:00:00 2001
From: Alex Duan <417921451@qq.com>
Date: Thu, 15 Dec 2022 17:10:53 +0800
Subject: [PATCH 07/15] fix(query): avg function check overflow range add case

---
 tests/system-test/2-query/avg.py | 30 ++++++++++++++++++++++++++++++
 1 file changed, 30 insertions(+)

diff --git a/tests/system-test/2-query/avg.py b/tests/system-test/2-query/avg.py
index 1d4d9a2494..4cad127111 100644
--- a/tests/system-test/2-query/avg.py
+++ b/tests/system-test/2-query/avg.py
@@ -425,7 +425,35 @@ class TDTestCase:
         tdSql.checkData(0,4,2.9142856660228804e+37)
         tdSql.checkData(0,5,None)
 
+    #
+    # test bigint to check overflow
+    #
+    def avg_check_overflow(self):
+        # create db
+        tdSql.execute(f"drop database if exists db")
+        tdSql.execute(f"create database if not exists db")
+        time.sleep(3)
+        tdSql.execute(f"use db")
+        tdSql.execute(f"create table st(ts timestamp, ibv bigint, ubv bigint unsigned) tags(area int)")
+        # insert t1 data
+        tdSql.execute(f"insert into t1 using st tags(1) values(now,9223372036854775801,18446744073709551611)")
+        tdSql.execute(f"insert into t1 using st tags(1) values(now,8223372036854775801,17446744073709551611)")
+        tdSql.execute(f"insert into t1 using st tags(1) values(now,7223372036854775801,16446744073709551611)")
+        # insert t2 data
+        tdSql.execute(f"insert into t2 using st tags(2) values(now,9223372036854775801,18446744073709551611)")
+        tdSql.execute(f"insert into t2 using st tags(2) values(now,8223372036854775801,17446744073709551611)")
+        tdSql.execute(f"insert into t2 using st tags(2) values(now,7223372036854775801,16446744073709551611)")
+        
+        # check single table answer
+        tdSql.query(f"select avg(ibv), avg(ubv) from t1")
+        tdSql.checkRows(1)
+        tdSql.checkData(0, 0,8.223372036854776e+18)
+        tdSql.checkData(0, 1,1.744674407370955e+19)
 
+        tdSql.query(f"select avg(ibv), avg(ubv) from st")
+        tdSql.checkRows(1)
+        tdSql.checkData(0, 0,8.223372036854776e+18)
+        tdSql.checkData(0, 1,1.744674407370955e+19)
 
     def run(self):  # sourcery skip: extract-duplicate-method, remove-redundant-fstring
         tdSql.prepare()
@@ -455,6 +483,8 @@ class TDTestCase:
         self.avg_func_filter()
         self.avg_check_unsigned()
 
+        # check avg overflow
+        self.avg_check_overflow()
     def stop(self):
         tdSql.close()
         tdLog.success(f"{__file__} successfully executed")

From 3fb64bf078d565a694a21153f35f4265b15d1e4a Mon Sep 17 00:00:00 2001
From: Alex Duan <417921451@qq.com>
Date: Thu, 15 Dec 2022 17:58:33 +0800
Subject: [PATCH 08/15] fix(query): avg function check overflow case

---
 tests/system-test/2-query/avg.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/tests/system-test/2-query/avg.py b/tests/system-test/2-query/avg.py
index 4cad127111..29603e63f7 100644
--- a/tests/system-test/2-query/avg.py
+++ b/tests/system-test/2-query/avg.py
@@ -409,7 +409,7 @@ class TDTestCase:
         tdSql.query(f"select avg(c1) ,avg(c2) , avg(c3) , avg(c4), avg(c5), avg(c6) from {dbname}.sub1_bound ")
         tdSql.checkRows(1)
         tdSql.checkData(0,0,920350133.571428537)
-        tdSql.checkData(0,1,1.3176245766935393e+18)
+        tdSql.checkData(0,1,3.952873730080618e+18)
         tdSql.checkData(0,2,14042.142857143)
         tdSql.checkData(0,3,53.571428571)
         tdSql.checkData(0,4,5.828571332045761e+37)

From ba31ce80701ac408ebea95a46f21f0dd577bb51e Mon Sep 17 00:00:00 2001
From: Alex Duan <417921451@qq.com>
Date: Sat, 17 Dec 2022 10:46:13 +0800
Subject: [PATCH 09/15] feat(query): avg function check range overflow

---
 source/libs/function/src/detail/tavgfunction.c | 1 -
 1 file changed, 1 deletion(-)

diff --git a/source/libs/function/src/detail/tavgfunction.c b/source/libs/function/src/detail/tavgfunction.c
index 04f5c99d47..1473a89bd7 100644
--- a/source/libs/function/src/detail/tavgfunction.c
+++ b/source/libs/function/src/detail/tavgfunction.c
@@ -553,7 +553,6 @@ int32_t avgFunction(SqlFunctionCtx* pCtx) {
         } else {
           for (int32_t i = pInput->startRowIndex; i < pInput->numOfRows + pInput->startRowIndex; ++i) {
             if (type == TSDB_DATA_TYPE_TINYINT) {
-              pAvgRes->sum.isum += plist[i];
               CHECK_OVERFLOW_SUM_SIGNED(pAvgRes, plist[i])
             } else {
               CHECK_OVERFLOW_SUM_UNSIGNED(pAvgRes, (uint8_t)plist[i])

From 110ed6b4b37ea088e3b1417d5bbb33fd2109e7cd Mon Sep 17 00:00:00 2001
From: Alex Duan <417921451@qq.com>
Date: Sat, 17 Dec 2022 11:40:00 +0800
Subject: [PATCH 10/15] fix(query): avg funciton add case

---
 tests/system-test/2-query/avg.py | 17 ++++++++++++++++-
 1 file changed, 16 insertions(+), 1 deletion(-)

diff --git a/tests/system-test/2-query/avg.py b/tests/system-test/2-query/avg.py
index 29603e63f7..3e670232f3 100644
--- a/tests/system-test/2-query/avg.py
+++ b/tests/system-test/2-query/avg.py
@@ -419,7 +419,7 @@ class TDTestCase:
         # check  + - * / in functions
         tdSql.query(f" select avg(c1+1) ,avg(c2) , avg(c3*1) , avg(c4/2), avg(c5)/2, avg(c6) from {dbname}.sub1_bound ")
         tdSql.checkData(0,0,920350134.5714285)
-        tdSql.checkData(0,1,1.3176245766935393e+18)
+        tdSql.checkData(0,1,3.952873730080618e+18)
         tdSql.checkData(0,2,14042.142857143)
         tdSql.checkData(0,3,26.785714286)
         tdSql.checkData(0,4,2.9142856660228804e+37)
@@ -450,11 +450,26 @@ class TDTestCase:
         tdSql.checkData(0, 0,8.223372036854776e+18)
         tdSql.checkData(0, 1,1.744674407370955e+19)
 
+        # check super table
         tdSql.query(f"select avg(ibv), avg(ubv) from st")
         tdSql.checkRows(1)
         tdSql.checkData(0, 0,8.223372036854776e+18)
         tdSql.checkData(0, 1,1.744674407370955e+19)
 
+        # check child query
+        tdSql.query(f"select avg(ibv), avg(ubv) from (select * from st)")
+        tdSql.checkRows(1)
+        tdSql.checkData(0, 0,8.223372036854776e+18)
+        tdSql.checkData(0, 1,1.744674407370955e+19)
+
+        # check group by
+        tdSql.query(f"select avg(ibv), avg(ubv) from st group by tbname")
+        tdSql.checkRows(2)
+        tdSql.checkData(0, 0,8.223372036854776e+18)
+        tdSql.checkData(0, 1,1.744674407370955e+19)
+        tdSql.checkData(1, 0,8.223372036854776e+18)
+        tdSql.checkData(1, 1,1.744674407370955e+19)
+
     def run(self):  # sourcery skip: extract-duplicate-method, remove-redundant-fstring
         tdSql.prepare()
 

From 680b7aa090aec54cadef245f74e1104d036948a1 Mon Sep 17 00:00:00 2001
From: Alex Duan <417921451@qq.com>
Date: Sat, 17 Dec 2022 16:59:25 +0800
Subject: [PATCH 11/15] fix(query): avg fix case error

---
 tests/system-test/2-query/avg.py | 22 +++++++++++-----------
 1 file changed, 11 insertions(+), 11 deletions(-)

diff --git a/tests/system-test/2-query/avg.py b/tests/system-test/2-query/avg.py
index 3e670232f3..dd8b38f9ea 100644
--- a/tests/system-test/2-query/avg.py
+++ b/tests/system-test/2-query/avg.py
@@ -434,36 +434,36 @@ class TDTestCase:
         tdSql.execute(f"create database if not exists db")
         time.sleep(3)
         tdSql.execute(f"use db")
-        tdSql.execute(f"create table st(ts timestamp, ibv bigint, ubv bigint unsigned) tags(area int)")
+        tdSql.execute(f"create table db.st(ts timestamp, ibv bigint, ubv bigint unsigned) tags(area int)")
         # insert t1 data
-        tdSql.execute(f"insert into t1 using st tags(1) values(now,9223372036854775801,18446744073709551611)")
-        tdSql.execute(f"insert into t1 using st tags(1) values(now,8223372036854775801,17446744073709551611)")
-        tdSql.execute(f"insert into t1 using st tags(1) values(now,7223372036854775801,16446744073709551611)")
+        tdSql.execute(f"insert into db.t1 using db.st tags(1) values(now,9223372036854775801,18446744073709551611)")
+        tdSql.execute(f"insert into db.t1 using db.st tags(1) values(now,8223372036854775801,17446744073709551611)")
+        tdSql.execute(f"insert into db.t1 using db.st tags(1) values(now,7223372036854775801,16446744073709551611)")
         # insert t2 data
-        tdSql.execute(f"insert into t2 using st tags(2) values(now,9223372036854775801,18446744073709551611)")
-        tdSql.execute(f"insert into t2 using st tags(2) values(now,8223372036854775801,17446744073709551611)")
-        tdSql.execute(f"insert into t2 using st tags(2) values(now,7223372036854775801,16446744073709551611)")
+        tdSql.execute(f"insert into db.t2 using db.st tags(2) values(now,9223372036854775801,18446744073709551611)")
+        tdSql.execute(f"insert into db.t2 using db.st tags(2) values(now,8223372036854775801,17446744073709551611)")
+        tdSql.execute(f"insert into db.t2 using db.st tags(2) values(now,7223372036854775801,16446744073709551611)")
         
         # check single table answer
-        tdSql.query(f"select avg(ibv), avg(ubv) from t1")
+        tdSql.query(f"select avg(ibv), avg(ubv) from db.t1")
         tdSql.checkRows(1)
         tdSql.checkData(0, 0,8.223372036854776e+18)
         tdSql.checkData(0, 1,1.744674407370955e+19)
 
         # check super table
-        tdSql.query(f"select avg(ibv), avg(ubv) from st")
+        tdSql.query(f"select avg(ibv), avg(ubv) from db.st")
         tdSql.checkRows(1)
         tdSql.checkData(0, 0,8.223372036854776e+18)
         tdSql.checkData(0, 1,1.744674407370955e+19)
 
         # check child query
-        tdSql.query(f"select avg(ibv), avg(ubv) from (select * from st)")
+        tdSql.query(f"select avg(ibv), avg(ubv) from (select * from db.st)")
         tdSql.checkRows(1)
         tdSql.checkData(0, 0,8.223372036854776e+18)
         tdSql.checkData(0, 1,1.744674407370955e+19)
 
         # check group by
-        tdSql.query(f"select avg(ibv), avg(ubv) from st group by tbname")
+        tdSql.query(f"select avg(ibv), avg(ubv) from db.st group by tbname")
         tdSql.checkRows(2)
         tdSql.checkData(0, 0,8.223372036854776e+18)
         tdSql.checkData(0, 1,1.744674407370955e+19)

From ea5bf3d31e8f7c607cfb9ae56762333497e4a153 Mon Sep 17 00:00:00 2001
From: Alex Duan <417921451@qq.com>
Date: Sat, 17 Dec 2022 17:15:44 +0800
Subject: [PATCH 12/15] fix(query): fix(query): avg check overflow

---
 source/libs/function/src/detail/tavgfunction.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/source/libs/function/src/detail/tavgfunction.c b/source/libs/function/src/detail/tavgfunction.c
index 1473a89bd7..5f1a4675ea 100644
--- a/source/libs/function/src/detail/tavgfunction.c
+++ b/source/libs/function/src/detail/tavgfunction.c
@@ -344,9 +344,9 @@ static int32_t calculateAvgBySMAInfo(SAvgRes* pRes, int32_t numOfRows, int32_t t
 
   pRes->count += numOfElem;
   if (IS_SIGNED_NUMERIC_TYPE(type)) {
-    pRes->sum.isum += pAgg->sum;
+    CHECK_OVERFLOW_SUM_SIGNED(pRes, pAgg->sum);
   } else if (IS_UNSIGNED_NUMERIC_TYPE(type)) {
-    pRes->sum.usum += pAgg->sum;
+    CHECK_OVERFLOW_SUM_UNSIGNED(pRes, pAgg->sum);
   } else if (IS_FLOAT_TYPE(type)) {
     pRes->sum.dsum += GET_DOUBLE_VAL((const char*)&(pAgg->sum));
   }

From 1b8199d98d8cab140f15f958c7a3d2e7c249f6d5 Mon Sep 17 00:00:00 2001
From: Alex Duan <417921451@qq.com>
Date: Sat, 17 Dec 2022 20:38:03 +0800
Subject: [PATCH 13/15] fix(query): avg function check overflow mac build

---
 source/libs/function/src/detail/tavgfunction.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/source/libs/function/src/detail/tavgfunction.c b/source/libs/function/src/detail/tavgfunction.c
index 5f1a4675ea..adc7aae559 100644
--- a/source/libs/function/src/detail/tavgfunction.c
+++ b/source/libs/function/src/detail/tavgfunction.c
@@ -667,9 +667,9 @@ static void avgTransferInfo(SAvgRes* pInput, SAvgRes* pOutput) {
 
   pOutput->type = pInput->type;
   if (IS_SIGNED_NUMERIC_TYPE(pOutput->type)) {
-    CHECK_OVERFLOW_SUM_SIGNED(pOutput, pInput->sum.overflow ? pInput->sum.dsum : pInput->sum.isum)
+    CHECK_OVERFLOW_SUM_SIGNED(pOutput, (pInput->sum.overflow ? pInput->sum.dsum : pInput->sum.isum))
   } else if (IS_UNSIGNED_NUMERIC_TYPE(pOutput->type)) {
-    CHECK_OVERFLOW_SUM_UNSIGNED(pOutput, pInput->sum.overflow ? pInput->sum.dsum : pInput->sum.usum)
+    CHECK_OVERFLOW_SUM_UNSIGNED(pOutput, (pInput->sum.overflow ? pInput->sum.dsum : pInput->sum.usum))
   } else {
     pOutput->sum.dsum += pInput->sum.dsum;
   }

From 1a75e9bb58cb1cb473d8446bd498f3026939e01c Mon Sep 17 00:00:00 2001
From: Alex Duan <417921451@qq.com>
Date: Sun, 18 Dec 2022 15:15:13 +0800
Subject: [PATCH 14/15] feat(query): avg function overflow add big check

---
 .../libs/function/src/detail/tavgfunction.c   | 34 ++++++++++++++++---
 1 file changed, 30 insertions(+), 4 deletions(-)

diff --git a/source/libs/function/src/detail/tavgfunction.c b/source/libs/function/src/detail/tavgfunction.c
index adc7aae559..b124678559 100644
--- a/source/libs/function/src/detail/tavgfunction.c
+++ b/source/libs/function/src/detail/tavgfunction.c
@@ -54,6 +54,20 @@
     out->sum.isum += val;                                                        \
   }
 
+// val is big than INT64_MAX, val come from merge
+#define CHECK_OVERFLOW_SUM_SIGNED_BIG(out, val, big)                             \
+  if (out->sum.overflow) {                                                       \
+    out->sum.dsum += val;                                                        \
+  } else if (out->sum.isum > 0 && val > 0 && INT64_MAX - out->sum.isum <= val || \
+             out->sum.isum < 0 && val < 0 && INT64_MIN - out->sum.isum >= val || \
+             big) {                                                              \
+    double dsum = (double)out->sum.isum;                                         \
+    out->sum.overflow = true;                                                    \
+    out->sum.dsum = dsum + val;                                                  \
+  } else {                                                                       \
+    out->sum.isum += val;                                                        \
+  }
+
 // define unsigned number sum with check overflow
 #define CHECK_OVERFLOW_SUM_UNSIGNED(out, val)                 \
   if (out->sum.overflow) {                                    \
@@ -66,6 +80,18 @@
     out->sum.usum += val;                                     \
   }
 
+// val is big than UINT64_MAX, val come from merge
+#define CHECK_OVERFLOW_SUM_UNSIGNED_BIG(out, val, big)        \
+  if (out->sum.overflow) {                                    \
+    out->sum.dsum += val;                                     \
+  } else if (UINT64_MAX - out->sum.usum <= val || big) {      \
+    double dsum = (double)out->sum.usum;                      \
+    out->sum.overflow = true;                                 \
+    out->sum.dsum = dsum + val;                               \
+  } else {                                                    \
+    out->sum.usum += val;                                     \
+  }
+
 typedef struct SAvgRes {
   double  result;
   SSumRes sum;
@@ -658,8 +684,6 @@ _over:
   return TSDB_CODE_SUCCESS;
 }
 
-
-
 static void avgTransferInfo(SAvgRes* pInput, SAvgRes* pOutput) {
   if (IS_NULL_TYPE(pInput->type)) {
     return;
@@ -667,9 +691,11 @@ static void avgTransferInfo(SAvgRes* pInput, SAvgRes* pOutput) {
 
   pOutput->type = pInput->type;
   if (IS_SIGNED_NUMERIC_TYPE(pOutput->type)) {
-    CHECK_OVERFLOW_SUM_SIGNED(pOutput, (pInput->sum.overflow ? pInput->sum.dsum : pInput->sum.isum))
+    bool overflow = pInput->sum.overflow;
+    CHECK_OVERFLOW_SUM_SIGNED_BIG(pOutput, overflow ? pInput->sum.dsum : pInput->sum.isum, overflow);
   } else if (IS_UNSIGNED_NUMERIC_TYPE(pOutput->type)) {
-    CHECK_OVERFLOW_SUM_UNSIGNED(pOutput, (pInput->sum.overflow ? pInput->sum.dsum : pInput->sum.usum))
+    bool overflow = pInput->sum.overflow;
+    CHECK_OVERFLOW_SUM_UNSIGNED_BIG(pOutput, overflow ? pInput->sum.dsum : pInput->sum.usum, overflow);
   } else {
     pOutput->sum.dsum += pInput->sum.dsum;
   }

From 0a5fd77b17e950014d246c3278387d44b4d24729 Mon Sep 17 00:00:00 2001
From: Alex Duan <417921451@qq.com>
Date: Sun, 18 Dec 2022 19:04:38 +0800
Subject: [PATCH 15/15] feat(query): avg function overflow mac build

---
 source/libs/function/src/detail/tavgfunction.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/source/libs/function/src/detail/tavgfunction.c b/source/libs/function/src/detail/tavgfunction.c
index b124678559..8e010181d1 100644
--- a/source/libs/function/src/detail/tavgfunction.c
+++ b/source/libs/function/src/detail/tavgfunction.c
@@ -692,10 +692,10 @@ static void avgTransferInfo(SAvgRes* pInput, SAvgRes* pOutput) {
   pOutput->type = pInput->type;
   if (IS_SIGNED_NUMERIC_TYPE(pOutput->type)) {
     bool overflow = pInput->sum.overflow;
-    CHECK_OVERFLOW_SUM_SIGNED_BIG(pOutput, overflow ? pInput->sum.dsum : pInput->sum.isum, overflow);
+    CHECK_OVERFLOW_SUM_SIGNED_BIG(pOutput, (overflow ? pInput->sum.dsum : pInput->sum.isum), overflow);
   } else if (IS_UNSIGNED_NUMERIC_TYPE(pOutput->type)) {
     bool overflow = pInput->sum.overflow;
-    CHECK_OVERFLOW_SUM_UNSIGNED_BIG(pOutput, overflow ? pInput->sum.dsum : pInput->sum.usum, overflow);
+    CHECK_OVERFLOW_SUM_UNSIGNED_BIG(pOutput, (overflow ? pInput->sum.dsum : pInput->sum.usum), overflow);
   } else {
     pOutput->sum.dsum += pInput->sum.dsum;
   }