Merge pull request h2database#4070 from katzyn/disk_space

Add DB_OBJECT_APPROXIMATE_SIZE and DB_OBJECT_APPROXIMATE_TOTAL_SIZE
andreitokar · Jul 12, 2024 · b728c42 · b728c42
2 parents 575302d + 6375771
commit b728c42
Show file tree

Hide file tree

Showing 17 changed files with 138 additions and 34 deletions.
diff --git a/h2/src/main/org/h2/command/Parser.java b/h2/src/main/org/h2/command/Parser.java
@@ -4331,6 +4331,10 @@ private Expression readBuiltinFunctionIf(String upperName) {
             return readDbObjectFunction(DBObjectFunction.DB_OBJECT_SIZE);
         case "DB_OBJECT_TOTAL_SIZE":
             return readDbObjectFunction(DBObjectFunction.DB_OBJECT_TOTAL_SIZE);
+        case "DB_OBJECT_APPROXIMATE_SIZE":
+            return readDbObjectFunction(DBObjectFunction.DB_OBJECT_APPROXIMATE_SIZE);
+        case "DB_OBJECT_APPROXIMATE_TOTAL_SIZE":
+            return readDbObjectFunction(DBObjectFunction.DB_OBJECT_APPROXIMATE_TOTAL_SIZE);
         case "CSVWRITE":
             return readParameters(new CSVWriteFunction());
         case "SIGNAL":

diff --git a/h2/src/main/org/h2/expression/function/DBObjectFunction.java b/h2/src/main/org/h2/expression/function/DBObjectFunction.java
@@ -46,8 +46,19 @@ public final class DBObjectFunction extends FunctionN {
      */
     public static final int DB_OBJECT_TOTAL_SIZE = DB_OBJECT_SIZE + 1;
 
+    /**
+     * DB_OBJECT_APPROXIMATE_SIZE() (non-standard).
+     */
+    public static final int DB_OBJECT_APPROXIMATE_SIZE = DB_OBJECT_TOTAL_SIZE + 1;
+
+    /**
+     * DB_OBJECT_APPROXIMATE_TOTAL_SIZE() (non-standard).
+     */
+    public static final int DB_OBJECT_APPROXIMATE_TOTAL_SIZE = DB_OBJECT_APPROXIMATE_SIZE + 1;
+
     private static final String[] NAMES = { //
-            "DB_OBJECT_ID", "DB_OBJECT_SQL", "DB_OBJECT_SIZE", "DB_OBJECT_TOTAL_SIZE" //
+            "DB_OBJECT_ID", "DB_OBJECT_SQL", "DB_OBJECT_SIZE", "DB_OBJECT_TOTAL_SIZE", //
+            "DB_OBJECT_APPROXIMATE_SIZE", "DB_OBJECT_APPROXIMATE_TOTAL_SIZE" //
     };
 
     private final int function;
@@ -130,20 +141,28 @@ public Value getValue(SessionLocal session, Value v1, Value v2, Value v3) {
             return sql != null ? ValueVarchar.get(sql, session) : ValueNull.INSTANCE;
         }
         case DB_OBJECT_SIZE:
-        case DB_OBJECT_TOTAL_SIZE: {
-            long size = 0L;
-            if (object instanceof Table) {
-                size = ((Table) object).getDiskSpaceUsed(function == DB_OBJECT_TOTAL_SIZE);
-            } else if (object instanceof Index) {
-                size = ((Index) object).getDiskSpaceUsed();
-            }
-            return ValueBigint.get(size);
-        }
+            return getDbObjectSize(object, false, false);
+        case DB_OBJECT_TOTAL_SIZE:
+            return getDbObjectSize(object, true, false);
+        case DB_OBJECT_APPROXIMATE_SIZE:
+            return getDbObjectSize(object, false, true);
+        case DB_OBJECT_APPROXIMATE_TOTAL_SIZE:
+            return getDbObjectSize(object, true, true);
         default:
             throw DbException.getInternalError("function=" + function);
         }
     }
 
+    private static Value getDbObjectSize(DbObject object, boolean total, boolean appproximate) {
+        long size = 0L;
+        if (object instanceof Table) {
+            size = ((Table) object).getDiskSpaceUsed(total, appproximate);
+        } else if (object instanceof Index) {
+            size = ((Index) object).getDiskSpaceUsed(appproximate);
+        }
+        return ValueBigint.get(size);
+    }
+
     @Override
     public Expression optimize(SessionLocal session) {
         optimizeArguments(session, false);
@@ -156,6 +175,8 @@ public Expression optimize(SessionLocal session) {
             break;
         case DB_OBJECT_SIZE:
         case DB_OBJECT_TOTAL_SIZE:
+        case DB_OBJECT_APPROXIMATE_SIZE:
+        case DB_OBJECT_APPROXIMATE_TOTAL_SIZE:
             type = TypeInfo.TYPE_BIGINT;
             break;
         default:

diff --git a/h2/src/main/org/h2/expression/function/TableInfoFunction.java b/h2/src/main/org/h2/expression/function/TableInfoFunction.java
@@ -52,7 +52,7 @@ public Value getValue(SessionLocal session, Value v1, Value v2) {
         Table table = new Parser(session).parseTableName(v1.getString());
         l: switch (function) {
         case DISK_SPACE_USED:
-            v1 = ValueBigint.get(table.getDiskSpaceUsed(false));
+            v1 = ValueBigint.get(table.getDiskSpaceUsed(false, false));
             break;
         case ESTIMATED_ENVELOPE: {
             Column column = table.getColumn(v2.getString());

diff --git a/h2/src/main/org/h2/index/Index.java b/h2/src/main/org/h2/index/Index.java
@@ -352,9 +352,12 @@ public Cursor findFirstOrLast(SessionLocal session, boolean first) {
     /**
      * Get the used disk space for this index.
      *
+     * @param approximate
+     *            {@code true} to return quick approximation
+     *
      * @return the estimated number of bytes
      */
-    public long getDiskSpaceUsed() {
+    public long getDiskSpaceUsed(boolean approximate) {
         return 0L;
     }
 

diff --git a/h2/src/main/org/h2/index/MetaIndex.java b/h2/src/main/org/h2/index/MetaIndex.java
@@ -118,8 +118,8 @@ public long getRowCountApproximation(SessionLocal session) {
     }
 
     @Override
-    public long getDiskSpaceUsed() {
-        return meta.getDiskSpaceUsed(false);
+    public long getDiskSpaceUsed(boolean approximate) {
+        return meta.getDiskSpaceUsed(false, approximate);
     }
 
     @Override

diff --git a/h2/src/main/org/h2/jdbc/JdbcDatabaseMetaData.java b/h2/src/main/org/h2/jdbc/JdbcDatabaseMetaData.java
@@ -246,7 +246,7 @@ public ResultSet getColumns(String catalog, String schemaPattern, String tableNa
      *            (uppercase for unquoted names)
      * @param table table name (must be specified)
      * @param unique only unique indexes
-     * @param approximate if true, return fast, but approximate CARDINALITY
+     * @param approximate if true, return fast, but approximate CARDINALITY and PAGES
      * @return the list of indexes and columns
      * @throws SQLException if the connection is closed
      */

diff --git a/h2/src/main/org/h2/jdbc/meta/DatabaseMetaLocal.java b/h2/src/main/org/h2/jdbc/meta/DatabaseMetaLocal.java
@@ -1288,7 +1288,7 @@ private void getIndexInfo(Value catalogValue, Value schemaValue, Table table, bo
                                     ? index.getRowCountApproximation(session)
                                     : index.getRowCount(session)),
                             // PAGES
-                            ValueBigint.get(index.getDiskSpaceUsed() / db.getPageSize()),
+                            ValueBigint.get(index.getDiskSpaceUsed(approximate) / db.getPageSize()),
                             // FILTER_CONDITION
                             ValueNull.INSTANCE);
                 }

diff --git a/h2/src/main/org/h2/mode/FunctionsPostgreSQL.java b/h2/src/main/org/h2/mode/FunctionsPostgreSQL.java
@@ -384,7 +384,7 @@ private static Value relationSize(SessionLocal session, Value tableOidOrName, bo
         } else {
             t = new Parser(session).parseTableName(tableOidOrName.getString());
         }
-        return ValueBigint.get(t.getDiskSpaceUsed(total));
+        return ValueBigint.get(t.getDiskSpaceUsed(total, false));
     }
 
 }
diff --git a/h2/src/main/org/h2/mvstore/Page.java b/h2/src/main/org/h2/mvstore/Page.java
@@ -828,17 +828,58 @@ public final int getMemory() {
     /**
      * Amount of used disk space in persistent case including child pages.
      *
+     * @param approximate
+     *            {@code true} to return quick approximation
+     *
      * @return amount of used disk space in persistent case
      */
-    public long getDiskSpaceUsed() {
-        long r = 0;
-        if (isPersistent()) {
-            r += diskSpaceUsed;
-            if (!isLeaf()) {
-                for (int i = 0; i < getRawChildPageCount(); i++) {
+    public final long getDiskSpaceUsed(boolean approximate) {
+        return isPersistent() //
+                ? approximate ? getDiskSpaceUsedApproximation(3, false) : getDiskSpaceUsedAccurate()
+                : 0L;
+    }
+
+    private long getDiskSpaceUsedAccurate() {
+        long r = diskSpaceUsed;
+        if (!isLeaf()) {
+            for (int i = 0, l = getRawChildPageCount(); i < l; i++) {
+                long pos = getChildPagePos(i);
+                if (pos != 0) {
+                    r += getChildPage(i).getDiskSpaceUsedAccurate();
+                }
+            }
+        }
+        return r;
+    }
+
+    private long getDiskSpaceUsedApproximation(int maxLevel, boolean f) {
+        long r = diskSpaceUsed;
+        if (!isLeaf()) {
+            int l = getRawChildPageCount();
+            if (--maxLevel == 0 && l > 4) {
+                if (f) {
+                    for (int i = 0; i < l; i++) {
+                        long pos = getChildPagePos(i);
+                        if (pos != 0) {
+                            r += getChildPage(i).getDiskSpaceUsedApproximation(maxLevel, f) * l;
+                            break;
+                        }
+                    }
+                } else {
+                    for (int i = l; --i >= 0;) {
+                        long pos = getChildPagePos(i);
+                        if (pos != 0) {
+                            r += getChildPage(i).getDiskSpaceUsedApproximation(maxLevel, f) * l;
+                            break;
+                        }
+                    }
+                }
+            } else {
+                for (int i = 0; i < l; i++) {
                     long pos = getChildPagePos(i);
                     if (pos != 0) {
-                        r += getChildPage(i).getDiskSpaceUsed();
+                        r += getChildPage(i).getDiskSpaceUsedApproximation(maxLevel, f);
+                        f = !f;
                     }
                 }
             }

diff --git a/h2/src/main/org/h2/mvstore/db/MVIndex.java b/h2/src/main/org/h2/mvstore/db/MVIndex.java
@@ -45,8 +45,8 @@ protected MVIndex(Table newTable, int id, String name, IndexColumn[] newIndexCol
     public abstract MVMap<K,VersionedValue<V>> getMVMap();
 
     @Override
-    public long getDiskSpaceUsed() {
-        return getMVMap().getRootPage().getDiskSpaceUsed();
+    public long getDiskSpaceUsed(boolean approximate) {
+        return getMVMap().getRootPage().getDiskSpaceUsed(approximate);
     }
 
 }
diff --git a/h2/src/main/org/h2/mvstore/db/MVTable.java b/h2/src/main/org/h2/mvstore/db/MVTable.java
@@ -622,17 +622,17 @@ public long getRowCountApproximation(SessionLocal session) {
     }
 
     @Override
-    public long getDiskSpaceUsed(boolean total) {
+    public long getDiskSpaceUsed(boolean total, boolean approximate) {
         if (total) {
             long size = 0L;
             for (Index index : getIndexes()) {
                 if (!(index instanceof MVDelegateIndex)) {
-                    size += index.getDiskSpaceUsed();
+                    size += index.getDiskSpaceUsed(approximate);
                 }
             }
             return size;
         } else {
-            return primaryIndex.getDiskSpaceUsed();
+            return primaryIndex.getDiskSpaceUsed(approximate);
         }
     }
 

diff --git a/h2/src/main/org/h2/res/help.csv b/h2/src/main/org/h2/res/help.csv
@@ -6517,8 +6517,11 @@ CALL DB_OBJECT_SQL('TABLE', 'PUBLIC', 'MY_TABLE');
 @h2@ DB_OBJECT_SIZE({'INDEX'|'TABLE'}, schemaNameString, objectNameString)
 ","
 Returns the approximate amount of space used by the specified table (excluding its indexes) or index.
+Only size of version used by the current transaction is estimated.
 Size of large LOBs currently is not included into estimation.
 This function may be expensive since it has to load every page in the table or index.
+Use [DB_OBJECT_APPROXIMATE_SIZE](https://h2database.com/html/functions.html#db_object_approximate_size)
+for a faster coarse approximation.
 ","
 CALL DB_OBJECT_SIZE('TABLE', 'PUBLIC', 'MY_TABLE');
 "
@@ -6527,12 +6530,35 @@ CALL DB_OBJECT_SIZE('TABLE', 'PUBLIC', 'MY_TABLE');
 @h2@ DB_OBJECT_TOTAL_SIZE('TABLE', schemaNameString, objectNameString)
 ","
 Returns the approximate amount of space used by the specified table and all its indexes.
+Only size of version used by the current transaction is estimated.
 Size of large LOBs currently is not included into estimation.
 This function may be expensive since it has to load every page in the table and its indexes.
+Use [DB_OBJECT_APPROXIMATE_TOTAL_SIZE](https://h2database.com/html/functions.html#db_object_approximate_total_size)
+for a faster coarse approximation.
 ","
 CALL DB_OBJECT_TOTAL_SIZE('TABLE', 'PUBLIC', 'MY_TABLE');
 "
 
+"Functions (System)","DB_OBJECT_APPROXIMATE_SIZE","
+@h2@ DB_OBJECT_APPROXIMATE_SIZE({'INDEX'|'TABLE'}, schemaNameString, objectNameString)
+","
+Returns the coarse approximate amount of space used by the specified table (excluding its indexes) or index.
+Only size of version used by the current transaction is estimated.
+Size of large LOBs currently is not included into estimation.
+","
+CALL DB_OBJECT_APPROXIMATE_SIZE('TABLE', 'PUBLIC', 'MY_TABLE');
+"
+
+"Functions (System)","DB_OBJECT_APPROXIMATE_TOTAL_SIZE","
+@h2@ DB_OBJECT_APPROXIMATE_TOTAL_SIZE('TABLE', schemaNameString, objectNameString)
+","
+Returns the coarse approximate amount of space used by the specified table and all its indexes.
+Only size of version used by the current transaction is estimated.
+Size of large LOBs currently is not included into estimation.
+","
+CALL DB_OBJECT_APPROXIMATE_TOTAL_SIZE('TABLE', 'PUBLIC', 'MY_TABLE');
+"
+
 "Functions (System)","DECODE","
 @c@ DECODE(value, whenValue, thenValue [,...])
 ","
@@ -6548,6 +6574,7 @@ CALL DECODE(RAND()>0.5, 0, 'Red', 1, 'Black');
 @c@ DISK_SPACE_USED(tableNameString)
 ","
 Returns the approximate amount of space used by the table specified.
+Only size of version used by the current transaction is estimated.
 Does not currently take into account indexes or LOB's.
 This function may be expensive since it has to load every page in the table.
 This function is deprecated,

diff --git a/h2/src/main/org/h2/table/Table.java b/h2/src/main/org/h2/table/Table.java
@@ -388,7 +388,7 @@ public boolean canReference() {
      */
     public abstract long getRowCountApproximation(SessionLocal session);
 
-    public long getDiskSpaceUsed(boolean total) {
+    public long getDiskSpaceUsed(boolean total, boolean approximate) {
         return 0L;
     }
 

diff --git a/h2/src/test/org/h2/test/db/TestOpenClose.java b/h2/src/test/org/h2/test/db/TestOpenClose.java
@@ -152,7 +152,8 @@ private void testBackupWithYoungDeadChunks() throws SQLException {
             stat.execute(String.format("INSERT INTO BIG_TABLE VALUES (%s, %s, TIMESTAMP '2024-06-14 10:00:00.000'," +
                     "'y4TMhpkNcw566aUxHtQGL8Hj6rEK8NNyDxZ2hV6HjNbJEHXKwszmyVVi4VI=', 1)", i++, i % 10));
         }
-        stat.execute("CREATE INDEX IDX_BIG_TABLE_HASH_GROUP ON BIG_TABLE(HASH_DATA NULLS FIRST, GROUP_ID NULLS FIRST)");
+        stat.execute("CREATE INDEX IDX_BIG_TABLE_HASH_GROUP ON BIG_TABLE("
+                + "HASH_DATA NULLS FIRST, GROUP_ID NULLS FIRST)");
         stat.execute("CREATE INDEX IDX_BIG_TABLE_CREATION_DATA ON BIG_TABLE(CREATION_DATE NULLS FIRST)");
 
         try {

diff --git a/h2/src/test/org/h2/test/db/TestTableEngines.java b/h2/src/test/org/h2/test/db/TestTableEngines.java
@@ -544,8 +544,8 @@ public long getRowCountApproximation(SessionLocal session) {
                 }
 
                 @Override
-                public long getDiskSpaceUsed() {
-                    return table.getDiskSpaceUsed(false);
+                public long getDiskSpaceUsed(boolean approximate) {
+                    return table.getDiskSpaceUsed(false, approximate);
                 }
 
                 @Override

diff --git a/h2/src/test/org/h2/test/scripts/functions/system/db_object.sql b/h2/src/test/org/h2/test/scripts/functions/system/db_object.sql
@@ -302,5 +302,12 @@ SELECT DB_OBJECT_TOTAL_SIZE('TABLE', 'PUBLIC', 'T_B') =
     DB_OBJECT_SIZE('TABLE', 'PUBLIC', 'T_B') + DB_OBJECT_SIZE('INDEX', 'PUBLIC', 'I_B');
 >> TRUE
 
+SELECT DB_OBJECT_APPROXIMATE_TOTAL_SIZE('TABLE', 'PUBLIC', 'T_B') >= DB_OBJECT_APPROXIMATE_SIZE('TABLE', 'PUBLIC', 'T_B');
+>> TRUE
+
+SELECT DB_OBJECT_APPROXIMATE_TOTAL_SIZE('TABLE', 'PUBLIC', 'T_B') =
+    DB_OBJECT_APPROXIMATE_SIZE('TABLE', 'PUBLIC', 'T_B') + DB_OBJECT_APPROXIMATE_SIZE('INDEX', 'PUBLIC', 'I_B');
+>> TRUE
+
 DROP TABLE T_B;
 > ok
diff --git a/h2/src/tools/org/h2/build/doc/dictionary.txt b/h2/src/tools/org/h2/build/doc/dictionary.txt
@@ -856,4 +856,4 @@ bck clo cur hwm materializedview udca vol connectionpooldatasource xadatasource
 ampm sssssff sstzh tzs yyyysssss newsequentialid solidus openjdk furthermore ssff secons nashorn fractions
 btrim underscores ffl decomposed decomposition subfield infinities retryable salted establish
 hatchet fis loom birthdate penrosed eve graalvm roberto polyglot truffle scriptengine unstarted conversation
-recompilations normalizer tablename tablenames
+recompilations normalizer tablename tablenames coarse