CSV reading is now faster.

h2/src/docsrc/html/changelog.html

@@ -18,7 +18,8 @@ Change Log
 <h1>Change Log</h1>
 
 <h2>Next Version (unreleased)</h2>
-<ul><li>-
+<ul><li>SimpleResultSet.newInstance(SimpleRowSource rs) did not work.
+</li><li>CSV reading is now faster.
 </li></ul>
 
 <h2>Version 1.1.116 (2009-07-18)</h2>

h2/src/main/org/h2/bnf/Bnf.java

@@ -6,7 +6,6 @@
  */
 package org.h2.bnf;
 
-import java.io.BufferedReader;
 import java.io.ByteArrayInputStream;
 import java.io.IOException;
 import java.io.InputStreamReader;
@@ -81,7 +80,6 @@ public class Bnf {
     }
 
     private void parse(Reader csv) throws SQLException, IOException {
-        csv = new BufferedReader(csv);
         Rule functions = null;
         statements = New.arrayList();
         ResultSet rs = Csv.getInstance().read(csv, null);

h2/src/main/org/h2/table/MetaTable.java

@@ -6,7 +6,6 @@
  */
 package org.h2.table;
 
-import java.io.BufferedReader;
 import java.io.ByteArrayInputStream;
 import java.io.IOException;
 import java.io.InputStreamReader;
@@ -920,7 +919,7 @@ public class MetaTable extends Table {
             String resource = "/org/h2/res/help.csv";
             try {
                 byte[] data = Resources.get(resource);
-                Reader reader = new BufferedReader(new InputStreamReader(new ByteArrayInputStream(data)));
+                Reader reader = new InputStreamReader(new ByteArrayInputStream(data));
                 ResultSet rs = Csv.getInstance().read(reader, null);
                 for (int i = 0; rs.next(); i++) {
                     add(rows, new String[] {

h2/src/main/org/h2/tools/Recover.java

@@ -732,7 +732,7 @@ public class Recover extends Tool implements DataHandler {
             } catch (Exception e) {
                 writeError(writer, e);
             }
-            DataPage s = DataPage.create(this, 128);
+            Data s = Data.create(this, 128);
             store.seek(0);
             store.readFully(s.getBytes(), 0, 128);
             s.setPos(48);
@@ -747,7 +747,7 @@ public class Recover extends Tool implements DataHandler {
                 writer.println("-- ERROR: page size; using " + pageSize);
             }
             int pageCount = (int) (length / pageSize);
-            s = DataPage.create(this, pageSize);
+            s = Data.create(this, pageSize);
             int logFirstTrunkPage = 0, logFirstDataPage = 0;
             for (int i = 1;; i++) {
                 if (i == 3) {
@@ -778,10 +778,10 @@ public class Recover extends Tool implements DataHandler {
             writer.println("-- firstTrunkPage: " + logFirstTrunkPage +
                     " firstDataPage: " + logFirstDataPage);
 
-            s = DataPage.create(this, pageSize);
+            s = Data.create(this, pageSize);
             int free = 0;
             for (long page = 3; page < pageCount; page++) {
-                s = DataPage.create(this, pageSize);
+                s = Data.create(this, pageSize);
                 store.seek(page * pageSize);
                 store.readFully(s.getBytes(), 0, pageSize);
                 int parentPageId = s.readInt();
@@ -814,7 +814,7 @@ public class Recover extends Tool implements DataHandler {
                     break;
                 }
                 case Page.TYPE_BTREE_NODE:
-                    writer.println("-- page " + page + ": btree node" + (last ? "(last)" : ""));
+                    writer.println("-- page " + page + ": b-tree node" + (last ? "(last)" : ""));
                     if (trace) {
                         dumpPageBtreeNode(writer, s, !last);
                     }
@@ -822,7 +822,7 @@ public class Recover extends Tool implements DataHandler {
                 case Page.TYPE_BTREE_LEAF: {
                     setStorage(s.readInt());
                     int entries = s.readShortInt();
-                    writer.println("-- page " + page + ": btree leaf " + (last ? "(last)" : "") + " table: " + storageId + " entries: " + entries);
+                    writer.println("-- page " + page + ": b-tree leaf " + (last ? "(last)" : "") + " table: " + storageId + " entries: " + entries);
                     if (trace) {
                         dumpPageBtreeLeaf(writer, s, entries, !last);
                     }
@@ -1030,16 +1030,21 @@ public class Recover extends Tool implements DataHandler {
         }
     }
 
-    private void dumpPageBtreeNode(PrintWriter writer, DataPage s, boolean positionOnly) {
+    private void dumpPageBtreeNode(PrintWriter writer, Data s, boolean positionOnly) {
         int entryCount = s.readShortInt();
         int rowCount = s.readInt();
         int[] children = new int[entryCount + 1];
         int[] offsets = new int[entryCount];
         children[entryCount] = s.readInt();
+        int empty = Integer.MAX_VALUE;
         for (int i = 0; i < entryCount; i++) {
             children[i] = s.readInt();
-            offsets[i] = s.readInt();
+            int off = s.readInt();
+            empty = Math.min(off, empty);
+            offsets[i] = off;
         }
+        empty = empty - s.length();
+        writer.println("--   empty: " + empty);
         for (int i = 0; i < entryCount; i++) {
             int off = offsets[i];
             s.setPos(off);
@@ -1060,7 +1065,7 @@ public class Recover extends Tool implements DataHandler {
         writer.println("-- [" + entryCount + "] child: " + children[entryCount] + " rowCount: " + rowCount);
     }
 
-    private int dumpPageFreeList(PrintWriter writer, DataPage s, int pageSize, long pageId, long pageCount) {
+    private int dumpPageFreeList(PrintWriter writer, Data s, int pageSize, long pageId, long pageCount) {
         int pagesAddressed = PageFreeList.getPagesAddressed(pageSize);
         BitField used = new BitField();
         for (int i = 0; i < pagesAddressed; i += 8) {
@@ -1087,11 +1092,16 @@ public class Recover extends Tool implements DataHandler {
         return free;
     }
 
-    private void dumpPageBtreeLeaf(PrintWriter writer, DataPage s, int entryCount, boolean positionOnly) {
+    private void dumpPageBtreeLeaf(PrintWriter writer, Data s, int entryCount, boolean positionOnly) {
         int[] offsets = new int[entryCount];
+        int empty = Integer.MAX_VALUE;
         for (int i = 0; i < entryCount; i++) {
-            offsets[i] = s.readShortInt();
+            int off = s.readShortInt();
+            empty = Math.min(off, empty);
+            offsets[i] = off;
         }
+        empty = empty - s.length();
+        writer.println("--   empty: " + empty);
         for (int i = 0; i < entryCount; i++) {
             int off = offsets[i];
             s.setPos(off);
@@ -1111,17 +1121,22 @@ public class Recover extends Tool implements DataHandler {
         }
     }
 
-    private void dumpPageDataLeaf(FileStore store, int pageSize, PrintWriter writer, DataPage s, boolean last, long pageId, int entryCount) throws SQLException {
+    private void dumpPageDataLeaf(FileStore store, int pageSize, PrintWriter writer, Data s, boolean last, long pageId, int entryCount) throws SQLException {
         int[] keys = new int[entryCount];
         int[] offsets = new int[entryCount];
         long next = 0;
         if (!last) {
             next = s.readInt();
         }
+        int empty = Integer.MAX_VALUE;
         for (int i = 0; i < entryCount; i++) {
             keys[i] = s.readInt();
-            offsets[i] = s.readShortInt();
+            int off = s.readShortInt();
+            empty = Math.min(off, empty);
+            offsets[i] = off;
         }
+        empty = empty - s.length();
+        writer.println("--   empty: " + empty);
         if (!last) {
             DataPage s2 = DataPage.create(this, pageSize);
             s.setPos(pageSize);

h2/src/test/org/h2/test/db/TestCsv.java

@@ -92,18 +92,18 @@ public class TestCsv extends TestBase {
         csv.setNullString("\\N");
         ResultSet rs = csv.read(f.getPath(), null, "UTF8");
         ResultSetMetaData meta = rs.getMetaData();
-        assertEquals(meta.getColumnCount(), 4);
-        assertEquals(meta.getColumnLabel(1), "A");
-        assertEquals(meta.getColumnLabel(2), "B");
-        assertEquals(meta.getColumnLabel(3), "C");
-        assertEquals(meta.getColumnLabel(4), "D");
+        assertEquals(4, meta.getColumnCount());
+        assertEquals("A", meta.getColumnLabel(1));
+        assertEquals("B", meta.getColumnLabel(2));
+        assertEquals("C", meta.getColumnLabel(3));
+        assertEquals("D", meta.getColumnLabel(4));
         assertTrue(rs.next());
-        assertEquals(rs.getString(1), null);
-        assertEquals(rs.getString(2), "");
+        assertEquals(null, rs.getString(1));
+        assertEquals("", rs.getString(2));
         // null is never quoted
-        assertEquals(rs.getString(3), "\\N");
+        assertEquals("\\N", rs.getString(3));
         // an empty string is always parsed as null
-        assertEquals(rs.getString(4), null);
+        assertEquals(null, rs.getString(4));
         assertFalse(rs.next());
 
         Connection conn = getConnection("csv");
@@ -177,12 +177,12 @@ public class TestCsv extends TestBase {
         assertEquals("ID|NAME 1|Hello", text);
         ResultSet rs = stat.executeQuery("select * from csvread('" + baseDir + "/test.csv', null, null, '|', '')");
         ResultSetMetaData meta = rs.getMetaData();
-        assertEquals(meta.getColumnCount(), 2);
-        assertEquals(meta.getColumnLabel(1), "ID");
-        assertEquals(meta.getColumnLabel(2), "NAME");
+        assertEquals(2, meta.getColumnCount());
+        assertEquals("ID", meta.getColumnLabel(1));
+        assertEquals("NAME", meta.getColumnLabel(2));
         assertTrue(rs.next());
-        assertEquals(rs.getString(1), "1");
-        assertEquals(rs.getString(2), "Hello");
+        assertEquals("1", rs.getString(1));
+        assertEquals("Hello", rs.getString(2));
         assertFalse(rs.next());
         conn.close();
         FileUtils.delete(baseDir + "/test.csv");
@@ -198,22 +198,22 @@ public class TestCsv extends TestBase {
         Statement stat = conn.createStatement();
         ResultSet rs = stat.executeQuery("select * from csvread('" + baseDir + "/test.csv', null, null, ';', '''', '\\')");
         ResultSetMetaData meta = rs.getMetaData();
-        assertEquals(meta.getColumnCount(), 2);
-        assertEquals(meta.getColumnLabel(1), "A");
-        assertEquals(meta.getColumnLabel(2), "B");
+        assertEquals(2, meta.getColumnCount());
+        assertEquals("A", meta.getColumnLabel(1));
+        assertEquals("B", meta.getColumnLabel(2));
         assertTrue(rs.next());
-        assertEquals(rs.getString(1), "It's nice");
-        assertEquals(rs.getString(2), "\nHello*\n");
+        assertEquals("It's nice", rs.getString(1));
+        assertEquals("\nHello*\n", rs.getString(2));
         assertFalse(rs.next());
         stat.execute("call csvwrite('" + baseDir + "/test2.csv', 'select * from csvread(''" + baseDir + "/test.csv'', null, null, '';'', '''''''', ''\\'')', null, '+', '*', '#')");
         rs = stat.executeQuery("select * from csvread('" + baseDir + "/test2.csv', null, null, '+', '*', '#')");
         meta = rs.getMetaData();
-        assertEquals(meta.getColumnCount(), 2);
-        assertEquals(meta.getColumnLabel(1), "A");
-        assertEquals(meta.getColumnLabel(2), "B");
+        assertEquals(2, meta.getColumnCount());
+        assertEquals("A", meta.getColumnLabel(1));
+        assertEquals("B", meta.getColumnLabel(2));
         assertTrue(rs.next());
-        assertEquals(rs.getString(1), "It's nice");
-        assertEquals(rs.getString(2), "\nHello*\n");
+        assertEquals("It's nice", rs.getString(1));
+        assertEquals("\nHello*\n", rs.getString(2));
         assertFalse(rs.next());
         conn.close();
         FileUtils.delete(baseDir + "/test.csv");
@@ -227,8 +227,8 @@ public class TestCsv extends TestBase {
         stat.execute("call csvwrite('" + baseDir + "/test.csv', 'select 1 id, ''Hello'' name', 'utf-8', '|')");
         ResultSet rs = stat.executeQuery("select * from csvread('" + baseDir + "/test.csv', null, 'utf-8', '|')");
         assertTrue(rs.next());
-        assertEquals(rs.getInt(1), 1);
-        assertEquals(rs.getString(2), "Hello");
+        assertEquals(1, rs.getInt(1));
+        assertEquals("Hello", rs.getString(2));
         assertFalse(rs.next());
         new File(baseDir + "/test.csv").delete();
 
@@ -250,12 +250,12 @@ public class TestCsv extends TestBase {
         stat.execute("call csvwrite('" + baseDir + "/test.csv', 'select 1 id, ''Hello'' name')");
         ResultSet rs = stat.executeQuery("select name from csvread('" + baseDir + "/test.csv')");
         assertTrue(rs.next());
-        assertEquals(rs.getString(1), "Hello");
+        assertEquals("Hello", rs.getString(1));
         assertFalse(rs.next());
         rs = stat.executeQuery("call csvread('" + baseDir + "/test.csv')");
         assertTrue(rs.next());
-        assertEquals(rs.getInt(1), 1);
-        assertEquals(rs.getString(2), "Hello");
+        assertEquals(1, rs.getInt(1));
+        assertEquals("Hello", rs.getString(2));
         assertFalse(rs.next());
         new File(baseDir + "/test.csv").delete();
         conn.close();
@@ -269,31 +269,31 @@ public class TestCsv extends TestBase {
         file.close();
         ResultSet rs = Csv.getInstance().read(baseDir + "/test.csv", null, "UTF8");
         ResultSetMetaData meta = rs.getMetaData();
-        assertEquals(meta.getColumnCount(), 4);
-        assertEquals(meta.getColumnLabel(1), "a");
-        assertEquals(meta.getColumnLabel(2), "b");
-        assertEquals(meta.getColumnLabel(3), "c");
-        assertEquals(meta.getColumnLabel(4), "d");
+        assertEquals(4, meta.getColumnCount());
+        assertEquals("a", meta.getColumnLabel(1));
+        assertEquals("b", meta.getColumnLabel(2));
+        assertEquals("c", meta.getColumnLabel(3));
+        assertEquals("d", meta.getColumnLabel(4));
         assertTrue(rs.next());
-        assertEquals(rs.getString(1), "201");
-        assertEquals(rs.getString(2), "-2");
-        assertEquals(rs.getString(3), "0");
-        assertEquals(rs.getString(4), "18");
+        assertEquals("201", rs.getString(1));
+        assertEquals("-2", rs.getString(2));
+        assertEquals("0", rs.getString(3));
+        assertEquals("18", rs.getString(4));
         assertTrue(rs.next());
-        assertEquals(rs.getString(1), null);
-        assertEquals(rs.getString(2), "abc\"");
-        assertEquals(rs.getString(3), null);
-        assertEquals(rs.getString(4), "");
+        assertEquals(null, rs.getString(1));
+        assertEquals("abc\"", rs.getString(2));
+        assertEquals(null, rs.getString(3));
+        assertEquals("", rs.getString(4));
         assertTrue(rs.next());
-        assertEquals(rs.getString(1), "1");
-        assertEquals(rs.getString(2), "2");
-        assertEquals(rs.getString(3), "3");
-        assertEquals(rs.getString(4), "4");
+        assertEquals("1", rs.getString(1));
+        assertEquals("2", rs.getString(2));
+        assertEquals("3", rs.getString(3));
+        assertEquals("4", rs.getString(4));
         assertTrue(rs.next());
-        assertEquals(rs.getString(1), "5");
-        assertEquals(rs.getString(2), "6");
-        assertEquals(rs.getString(3), "7");
-        assertEquals(rs.getString(4), "8");
+        assertEquals("5", rs.getString(1));
+        assertEquals("6", rs.getString(2));
+        assertEquals("7", rs.getString(3));
+        assertEquals("8", rs.getString(4));
         assertFalse(rs.next());
 
         // a,b,c,d
@@ -324,8 +324,8 @@ public class TestCsv extends TestBase {
         assertEquals(2, meta.getColumnCount());
         for (int i = 0; i < len; i++) {
             rs.next();
-            assertEquals(rs.getString("ID"), "" + (i + 1));
-            assertEquals(rs.getString("NAME"), "Ruebezahl");
+            assertEquals("" + (i + 1), rs.getString("ID"));
+            assertEquals("Ruebezahl", rs.getString("NAME"));
         }
         assertFalse(rs.next());
         rs.close();