From 375847cad65c8db0216da94ff0084f093052226e Mon Sep 17 00:00:00 2001
From: wpan <wpan@taosdata.com>
Date: Fri, 16 Jul 2021 08:13:06 +0800
Subject: [PATCH] support nchar filter

---
 src/query/src/qFilter.c                   | 91 ++++++++++++++++++++++-
 tests/script/general/parser/condition.sim | 39 ++++++++++
 2 files changed, 127 insertions(+), 3 deletions(-)

diff --git a/src/query/src/qFilter.c b/src/query/src/qFilter.c
index 8f1530551c..fa26d652fe 100644
--- a/src/query/src/qFilter.c
+++ b/src/query/src/qFilter.c
@@ -1470,7 +1470,7 @@ _err_return:
   return TSDB_CODE_SUCCESS;
 }
 
-int32_t filterGenerateGroupFromArray(SFilterInfo *info, SArray* group) {
+int32_t filterConvertGroupFromArray(SFilterInfo *info, SArray* group) {
   size_t groupSize = taosArrayGetSize(group);
 
   info->groupNum = (uint16_t)groupSize;
@@ -1488,6 +1488,89 @@ int32_t filterGenerateGroupFromArray(SFilterInfo *info, SArray* group) {
   return TSDB_CODE_SUCCESS;
 }
 
+int32_t filterCheckRangeCoverage(SFilterInfo *info, SFilterGroupCtx* gctx, SFilterGroupCtx** uctx) {
+  if (gctx->num == 0) {
+    return TSDB_CODE_SUCCESS;
+  }
+
+  SFilterInfo oinfo = *info;
+  uint16_t gNum = 0;
+  SArray* group = taosArrayInit(FILTER_DEFAULT_GROUP_SIZE, sizeof(SFilterGroup));
+
+  memset(info, 0, sizeof(*info));
+
+  filterInitUnitsFields(info);
+  
+  for (uint16_t i = 0; i < oinfo.groupNum; ++i) {
+    SFilterGroup* g = oinfo.groups + i;
+    SFilterGroup ng = {0};
+    uint16_t unitNum = (uctx && uctx[i]) ? uctx[i]->num : g->unitNum;
+
+    if (unitNum == 0) {
+      continue;
+    }
+
+    ++gNum;
+
+    if ((uctx == NULL) || (uctx[i] == NULL)) {
+      for (uint16_t n = 0; n < g->unitNum; ++n) {
+        SFilterUnit* u = FILTER_GROUP_UNIT(&oinfo, g, n);
+        filterAddUnitFromUnit(info, &oinfo, u);      
+        filterAddUnitToGroup(&ng, info->unitNum - 1);
+      }
+      
+      taosArrayPush(group, &ng);
+
+      continue;
+    }
+    
+    SFilterGroupCtx* ctx = uctx[i];
+    for (uint16_t n = 0; n < g->unitNum; ++n) {
+      SFilterUnit* u = FILTER_GROUP_UNIT(&oinfo, g, n);
+      int32_t type = FILTER_UNIT_DATA_TYPE(u);
+      if (FILTER_NO_MERGE_DATA_TYPE(type)) {
+        filterAddUnitFromUnit(info, &oinfo, u);
+        filterAddUnitToGroup(&ng, info->unitNum - 1);
+        continue;
+      }
+
+      uint16_t cidx = FILTER_UNIT_COL_IDX(u);
+
+      assert(ctx->col[cidx] > 0 || ctx->col[cidx] == -1);
+      
+      if (ctx->col[cidx] != -1) {
+        filterAddUnitFromUnit(info, &oinfo, u);        
+        filterAddUnitToGroup(&ng, info->unitNum - 1);
+      }
+    }
+
+    if (ctx->colRange && taosArrayGetSize(ctx->colRange) > 0) {
+      int32_t size = (int32_t)taosArrayGetSize(ctx->colRange);
+      for (int32_t m = 0; m < size; ++m) {
+        SFilterColRange *cra = taosArrayGet(ctx->colRange, m);
+        filterAddGroupUnitFromRange(info, &oinfo, cra, &ng, TSDB_RELATION_AND, NULL);
+      }
+    }
+
+    taosArrayPush(group, &ng);
+  }
+
+  if (gctx->colRange && taosArrayGetSize(gctx->colRange) > 0) {
+    int32_t size = (int32_t)taosArrayGetSize(gctx->colRange);
+    for (int32_t i = 0; i < size; ++i) {
+      SFilterColRange *cra = taosArrayGet(gctx->colRange, i);
+      filterAddGroupUnitFromRange(info, &oinfo, cra, NULL, TSDB_RELATION_OR, group);
+    }
+  }
+
+  filterConvertGroupFromArray(info, group);
+
+  taosArrayDestroy(group);
+
+  return TSDB_CODE_SUCCESS;
+}
+
+
 
 int32_t filterRewrite(SFilterInfo *info, SFilterGroupCtx* gctx, SFilterGroupCtx** uctx) {
   if (gctx->num == 0) {
@@ -1564,7 +1647,7 @@ int32_t filterRewrite(SFilterInfo *info, SFilterGroupCtx* gctx, SFilterGroupCtx*
     }
   }
 
-  filterGenerateGroupFromArray(info, group);
+  filterConvertGroupFromArray(info, group);
 
   taosArrayDestroy(group);
 
@@ -1590,6 +1673,8 @@ int32_t filterPreprocess(SFilterInfo *info) {
     return TSDB_CODE_SUCCESS;
   }  
 
+  filterCheckRangeCoverage(info, &groupRes, unitsRes);
+
   //TODO GET COLUMN RANGE
 
   filterRewrite(info, &groupRes, unitsRes);
@@ -1699,7 +1784,7 @@ int32_t filterInitFromTree(tExprNode* tree, SFilterInfo **pinfo, uint32_t option
 
   ERR_JRET(code);
 
-  filterGenerateGroupFromArray(info, group);
+  filterConvertGroupFromArray(info, group);
 
   ERR_JRET(filterInitValFieldData(info));
 
diff --git a/tests/script/general/parser/condition.sim b/tests/script/general/parser/condition.sim
index d414cb1554..8048ccc192 100644
--- a/tests/script/general/parser/condition.sim
+++ b/tests/script/general/parser/condition.sim
@@ -93,6 +93,12 @@ sql_error select ts,c1,c7 from stb1 where c7 > false
 sql_error select * from stb1 where c1 > NULL;
 sql_error select * from stb1 where c1 = NULL;
 sql_error select * from stb1 where c1 LIKE '%1';
+sql_error select * from stb1 where c2 LIKE '%1';
+sql_error select * from stb1 where c3 LIKE '%1';
+sql_error select * from stb1 where c4 LIKE '%1';
+sql_error select * from stb1 where c5 LIKE '%1';
+sql_error select * from stb1 where c6 LIKE '%1';
+sql_error select * from stb1 where c7 LIKE '%1';
 sql_error select * from stb1 where c1 = 'NULL';
 sql_error select * from stb1 where c2 > 'NULL';
 sql_error select * from stb1 where c3 <> 'NULL';
@@ -1225,6 +1231,39 @@ if $data20 != @21-05-05 18:19:18.000@ then
   return -1
 endi
 
+sql select * from stb1 where c1 in (11,21,31,41) and c1 in (11,42);
+if $rows != 1 then
+  return -1
+endi
+if $data00 != @21-05-05 18:19:04.000@ then
+  return -1
+endi
+
+sql select * from stb1 where c8 in ('11','21','31','41') and c8 in ('11','42');
+if $rows != 1 then
+  return -1
+endi
+if $data00 != @21-05-05 18:19:04.000@ then
+  return -1
+endi
+
+sql select * from stb1 where (c1 > 60 and c2 > 40) or (c1 > 62 and c2 > 50);
+if $rows != 4 then
+  return -1
+endi
+if $data00 != @21-05-05 18:19:24.000@ then
+  return -1
+endi
+if $data10 != @21-05-05 18:19:25.000@ then
+  return -1
+endi
+if $data20 != @21-05-05 18:19:26.000@ then
+  return -1
+endi
+if $data30 != @21-05-05 18:19:27.000@ then
+  return -1
+endi
+
 sql select * from stb1 where c1 = 3 or c1 = 5 or c1 >= 44 and c1 <= 52;
 if $rows != 4 then
   return -1