优化器直方图

chusql-server/src/ChuSQL/Server/Backend.hs

@@ -319,7 +319,7 @@ exprTables expr = case expr of
     -- | 两侧都取表名
     both a b = exprTables a ++ exprTables b
 
--- | 一张表的线上信息（内存实现：没有索引，统计现算）
+-- | 一张表的线上信息（内存实现：没有索引，统计与直方图现算）
 tableInfoOf :: (String, Table) -> TableInfo
 tableInfoOf (name, tbl) =
     TableInfo
@@ -328,10 +328,14 @@ tableInfoOf (name, tbl) =
         , tiRows = length (tableRows tbl)
         , tiIndexes = []
         , tiStats = [(c, distinctOf c, False) | (c, _) <- tableCols tbl]
+        , tiHistograms = [(c, h) | (c, _) <- tableCols tbl, Just h <- [histogramOf (columnValues c)]]
         }
   where
+    -- | 一列的所有值
+    columnValues c = [v | r <- tableRows tbl, Just v <- [lookup c r]]
+
     -- | 一列的不同值个数
-    distinctOf c = length (nub [v | r <- tableRows tbl, Just v <- [lookup c r]])
+    distinctOf c = length (nub (columnValues c))
 
 -- | 本地列类型转线上字符串（和 IPC 那边同一套写法）
 wireType :: Column -> String

chusql-server/test/Spec.hs

@@ -6,13 +6,14 @@ module Main (main) where
 import ChuSQL.Core.Engine.Storage.IPC (
     Account (..),
     Request (ReqAccountReset),
+    TableInfo (..),
     closeConnection,
     localStorageLink,
     setStorageLink,
  )
 import ChuSQL.Core.Engine.Syntax.AST (Statement)
 import ChuSQL.Core.Engine.Syntax.Parser (parseStatement)
-import ChuSQL.Core.Model (Database, Row, Table (..), Value (..), pattern TInt, pattern TStr)
+import ChuSQL.Core.Model (Database, Histogram (..), Row, Table (..), Value (..), pattern TInt, pattern TStr)
 import ChuSQL.Interface.Auth (hashPasswordWith)
 import ChuSQL.Interface.Protocol (ClientRequest (..), Grant (..), RoleView (..), ServerResponse (..), decodeRequest, encodeResponse)
 import ChuSQL.Interface.RateLimit (newRateLimiter)
@@ -307,6 +308,21 @@ tcpSpec = do
             asBool (at "truncated" value) `shouldBe` False
             asText (at "database" value) `shouldBe` "test"
 
+    describe "memory backend: catalog statistics" $ do
+        it "reports a histogram for a numeric column" $ do
+            db <- newMVar testDb
+            let backend = memoryBackend testDatabaseName db
+            infos <- beCatalog backend
+            case infos of
+                Left err -> expectationFailure err
+                Right tables -> case [i | i <- tables, tiTable i == "users"] of
+                    [] -> expectationFailure "the catalog is missing the users table"
+                    (info : _) -> do
+                        fmap histLow (lookup "id" (tiHistograms info)) `shouldBe` Just 1
+                        fmap histHigh (lookup "id" (tiHistograms info)) `shouldBe` Just 5
+                        fmap (sum . histBuckets) (lookup "id" (tiHistograms info)) `shouldBe` Just 5
+                        lookup "name" (tiHistograms info) `shouldBe` Nothing
+
     describe "server config ([server] in chusql.toml)" $ do
         it "falls back to the defaults when the file has no server section" $ do
             path <- tcpConfigPath "server-default"