All Downloads are FREE. Search and download functionalities are using the official Maven repository.
Please wait. This can take some minutes ...
Many resources are needed to download a project. Please understand that we have to compensate our server costs. Thank you in advance.
Project price only 1 $
You can buy this project and download/modify it how often you want.
org.apache.hadoop.hive.ql.parse.ColumnStatsSemanticAnalyzer Maven / Gradle / Ivy
/*
* Licensed to the Apache Software Foundation (ASF) under one
* or more contributor license agreements. See the NOTICE file
* distributed with this work for additional information
* regarding copyright ownership. The ASF licenses this file
* to you under the Apache License, Version 2.0 (the
* "License"); you may not use this file except in compliance
* with the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package org.apache.hadoop.hive.ql.parse;
import java.io.IOException;
import java.util.ArrayList;
import java.util.List;
import java.util.Map;
import org.apache.hadoop.hive.common.HiveStatsUtils;
import org.apache.hadoop.hive.conf.HiveConf;
import org.apache.hadoop.hive.conf.HiveConf.ConfVars;
import org.apache.hadoop.hive.conf.HiveVariableSource;
import org.apache.hadoop.hive.conf.VariableSubstitution;
import org.apache.hadoop.hive.metastore.api.FieldSchema;
import org.apache.hadoop.hive.ql.Context;
import org.apache.hadoop.hive.ql.ErrorMsg;
import org.apache.hadoop.hive.ql.QueryState;
import org.apache.hadoop.hive.ql.exec.Utilities;
import org.apache.hadoop.hive.ql.metadata.HiveException;
import org.apache.hadoop.hive.ql.metadata.Table;
import org.apache.hadoop.hive.ql.plan.HiveOperation;
import org.apache.hadoop.hive.ql.session.SessionState;
import org.apache.hadoop.hive.ql.session.SessionState.LogHelper;
import org.apache.hadoop.hive.serde.serdeConstants;
import org.apache.hadoop.hive.serde2.objectinspector.ObjectInspector;
import org.apache.hadoop.hive.serde2.typeinfo.TypeInfo;
import org.apache.hadoop.hive.serde2.typeinfo.TypeInfoUtils;
import com.facebook.presto.hive.$internal.org.slf4j.Logger;
import com.facebook.presto.hive.$internal.org.slf4j.LoggerFactory;
/**
* ColumnStatsSemanticAnalyzer.
* Handles semantic analysis and rewrite for gathering column statistics both at the level of a
* partition and a table. Note that table statistics are implemented in SemanticAnalyzer.
*
*/
public class ColumnStatsSemanticAnalyzer extends SemanticAnalyzer {
private static final Logger LOG = LoggerFactory
.getLogger(ColumnStatsSemanticAnalyzer.class);
private static final LogHelper CONSOLE = new LogHelper(LOG);
private ASTNode originalTree;
private ASTNode rewrittenTree;
private String rewrittenQuery;
private Context ctx;
private boolean isRewritten;
private boolean isTableLevel;
private List colNames;
private List colType;
private Table tbl;
public ColumnStatsSemanticAnalyzer(QueryState queryState) throws SemanticException {
super(queryState);
}
private boolean shouldRewrite(ASTNode tree) {
boolean rwt = false;
if (tree.getChildCount() > 1) {
ASTNode child0 = (ASTNode) tree.getChild(0);
ASTNode child1;
if (child0.getToken().getType() == HiveParser.TOK_TAB) {
child0 = (ASTNode) child0.getChild(0);
if (child0.getToken().getType() == HiveParser.TOK_TABNAME) {
child1 = (ASTNode) tree.getChild(1);
if (child1.getToken().getType() == HiveParser.KW_COLUMNS) {
rwt = true;
}
}
}
}
return rwt;
}
private List getColumnName(ASTNode tree) throws SemanticException {
switch (tree.getChildCount()) {
case 2:
return Utilities.getColumnNamesFromFieldSchema(tbl.getCols());
case 3:
int numCols = tree.getChild(2).getChildCount();
List colName = new ArrayList(numCols);
for (int i = 0; i < numCols; i++) {
colName.add(getUnescapedName((ASTNode) tree.getChild(2).getChild(i)));
}
return colName;
default:
throw new SemanticException("Internal error. Expected number of children of ASTNode to be"
+ " either 2 or 3. Found : " + tree.getChildCount());
}
}
private void handlePartialPartitionSpec(Map partSpec, ColumnStatsAutoGatherContext context) throws
SemanticException {
// If user has fully specified partition, validate that partition exists
int partValsSpecified = 0;
for (String partKey : partSpec.keySet()) {
partValsSpecified += partSpec.get(partKey) == null ? 0 : 1;
}
try {
// for static partition, it may not exist when HIVESTATSCOLAUTOGATHER is
// set to true
if (context == null) {
if ((partValsSpecified == tbl.getPartitionKeys().size())
&& (db.getPartition(tbl, partSpec, false, null, false) == null)) {
throw new SemanticException(ErrorMsg.COLUMNSTATSCOLLECTOR_INVALID_PARTITION.getMsg()
+ " : " + partSpec);
}
}
} catch (HiveException he) {
throw new SemanticException(ErrorMsg.COLUMNSTATSCOLLECTOR_INVALID_PARTITION.getMsg() + " : "
+ partSpec);
}
// User might have only specified partial list of partition keys, in which case add other partition keys in partSpec
List partKeys = Utilities.getColumnNamesFromFieldSchema(tbl.getPartitionKeys());
for (String partKey : partKeys) {
if (!partSpec.containsKey(partKey)) {
partSpec.put(partKey, null);
}
}
// Check if user have erroneously specified non-existent partitioning columns
for (String partKey : partSpec.keySet()) {
if (!partKeys.contains(partKey)) {
throw new SemanticException(ErrorMsg.COLUMNSTATSCOLLECTOR_INVALID_PART_KEY.getMsg() + " : " + partKey);
}
}
}
private StringBuilder genPartitionClause(Map partSpec) throws SemanticException {
StringBuilder whereClause = new StringBuilder(" where ");
boolean predPresent = false;
StringBuilder groupByClause = new StringBuilder(" group by ");
boolean aggPresent = false;
for (String partKey : partSpec.keySet()) {
String value;
if ((value = partSpec.get(partKey)) != null) {
if (!predPresent) {
predPresent = true;
} else {
whereClause.append(" and ");
}
whereClause.append("`").append(partKey).append("` = ").append(genPartValueString(partKey, value));
}
}
for (FieldSchema fs : tbl.getPartitionKeys()) {
if (!aggPresent) {
aggPresent = true;
} else {
groupByClause.append(',');
}
groupByClause.append("`" + fs.getName() + "`");
}
// attach the predicate and group by to the return clause
return predPresent ? whereClause.append(groupByClause) : groupByClause;
}
private String genPartValueString(String partKey, String partVal) throws SemanticException {
String returnVal = partVal;
String partColType = getColTypeOf(partKey);
if (partColType.equals(serdeConstants.STRING_TYPE_NAME) ||
partColType.contains(serdeConstants.VARCHAR_TYPE_NAME) ||
partColType.contains(serdeConstants.CHAR_TYPE_NAME)) {
returnVal = "'" + escapeSQLString(partVal) + "'";
} else if (partColType.equals(serdeConstants.TINYINT_TYPE_NAME)) {
returnVal = partVal + "Y";
} else if (partColType.equals(serdeConstants.SMALLINT_TYPE_NAME)) {
returnVal = partVal + "S";
} else if (partColType.equals(serdeConstants.INT_TYPE_NAME)) {
returnVal = partVal;
} else if (partColType.equals(serdeConstants.BIGINT_TYPE_NAME)) {
returnVal = partVal + "L";
} else if (partColType.contains(serdeConstants.DECIMAL_TYPE_NAME)) {
returnVal = partVal + "BD";
} else if (partColType.equals(serdeConstants.DATE_TYPE_NAME) ||
partColType.equals(serdeConstants.TIMESTAMP_TYPE_NAME)) {
returnVal = partColType + " '" + escapeSQLString(partVal) + "'";
} else {
//for other usually not used types, just quote the value
returnVal = "'" + escapeSQLString(partVal) + "'";
}
return returnVal;
}
private String getColTypeOf(String partKey) throws SemanticException{
for (FieldSchema fs : tbl.getPartitionKeys()) {
if (partKey.equalsIgnoreCase(fs.getName())) {
return fs.getType().toLowerCase();
}
}
throw new SemanticException("Unknown partition key : " + partKey);
}
private List getColumnTypes(List colNames)
throws SemanticException{
List colTypes = new ArrayList();
List cols = tbl.getCols();
List copyColNames = new ArrayList<>(colNames);
for (String colName : copyColNames) {
for (FieldSchema col : cols) {
if (colName.equalsIgnoreCase(col.getName())) {
String type = col.getType();
TypeInfo typeInfo = TypeInfoUtils.getTypeInfoFromTypeString(type);
if (typeInfo.getCategory() != ObjectInspector.Category.PRIMITIVE) {
logTypeWarning(colName, type);
colNames.remove(colName);
} else {
colTypes.add(type);
}
}
}
}
return colTypes;
}
private String escapeBackTicks(String colName) {
return colName.replaceAll("`", "``");
}
private String genRewrittenQuery(List colNames, HiveConf conf, Map partSpec,
boolean isPartitionStats) throws SemanticException{
StringBuilder rewrittenQueryBuilder = new StringBuilder("select ");
for (int i = 0; i < colNames.size(); i++) {
if (i > 0) {
rewrittenQueryBuilder.append(" , ");
}
String func = HiveConf.getVar(conf, HiveConf.ConfVars.HIVE_STATS_NDV_ALGO).toLowerCase();
rewrittenQueryBuilder.append("compute_stats(`");
rewrittenQueryBuilder.append(escapeBackTicks(colNames.get(i)));
rewrittenQueryBuilder.append("`, '" + func + "'");
if ("fm".equals(func)) {
int numBitVectors = 0;
try {
numBitVectors = HiveStatsUtils.getNumBitVectorsForNDVEstimation(conf);
} catch (Exception e) {
throw new SemanticException(e.getMessage());
}
rewrittenQueryBuilder.append(", " + numBitVectors);
}
rewrittenQueryBuilder.append(')');
}
if (isPartitionStats) {
for (FieldSchema fs : tbl.getPartCols()) {
rewrittenQueryBuilder.append(" , `" + fs.getName() + "`");
}
}
rewrittenQueryBuilder.append(" from `");
rewrittenQueryBuilder.append(tbl.getDbName());
rewrittenQueryBuilder.append("`.");
rewrittenQueryBuilder.append("`" + tbl.getTableName() + "`");
isRewritten = true;
// If partition level statistics is requested, add predicate and group by as needed to rewritten
// query
if (isPartitionStats) {
rewrittenQueryBuilder.append(genPartitionClause(partSpec));
}
String rewrittenQuery = rewrittenQueryBuilder.toString();
rewrittenQuery = new VariableSubstitution(new HiveVariableSource() {
@Override
public Map getHiveVariable() {
return SessionState.get().getHiveVariables();
}
}).substitute(conf, rewrittenQuery);
return rewrittenQuery;
}
private ASTNode genRewrittenTree(String rewrittenQuery) throws SemanticException {
// Parse the rewritten query string
try {
ctx = new Context(conf);
} catch (IOException e) {
throw new SemanticException(ErrorMsg.COLUMNSTATSCOLLECTOR_IO_ERROR.getMsg());
}
ctx.setCmd(rewrittenQuery);
try {
return ParseUtils.parse(rewrittenQuery, ctx);
} catch (ParseException e) {
throw new SemanticException(ErrorMsg.COLUMNSTATSCOLLECTOR_PARSE_ERROR.getMsg());
}
}
// fail early if the columns specified for column statistics are not valid
private void validateSpecifiedColumnNames(List specifiedCols)
throws SemanticException {
List tableCols = Utilities.getColumnNamesFromFieldSchema(tbl.getCols());
for (String sc : specifiedCols) {
if (!tableCols.contains(sc.toLowerCase())) {
String msg = "'" + sc + "' (possible columns are " + tableCols.toString() + ")";
throw new SemanticException(ErrorMsg.INVALID_COLUMN.getMsg(msg));
}
}
}
private void checkForPartitionColumns(List specifiedCols, List partCols)
throws SemanticException {
// Raise error if user has specified partition column for stats
for (String pc : partCols) {
for (String sc : specifiedCols) {
if (pc.equalsIgnoreCase(sc)) {
throw new SemanticException(ErrorMsg.COLUMNSTATSCOLLECTOR_INVALID_COLUMN.getMsg()
+ " [Try removing column '" + sc + "' from column list]");
}
}
}
}
private void logTypeWarning(String colName, String colType) {
String warning = "Only primitive type arguments are accepted but " + colType
+ " is passed for " + colName + ".";
warning = "WARNING: " + warning;
CONSOLE.printInfo(warning);
}
@Override
public void analyze(ASTNode ast, Context origCtx) throws SemanticException {
QB qb;
QBParseInfo qbp;
// initialize QB
init(true);
// check if it is no scan. grammar prevents coexit noscan/columns
super.processNoScanCommand(ast);
/* Rewrite only analyze table <> column <> compute statistics; Don't rewrite analyze table
* command - table stats are collected by the table scan operator and is not rewritten to
* an aggregation.
*/
if (shouldRewrite(ast)) {
tbl = AnalyzeCommandUtils.getTable(ast, this);
colNames = getColumnName(ast);
// Save away the original AST
originalTree = ast;
boolean isPartitionStats = AnalyzeCommandUtils.isPartitionLevelStats(ast);
Map partSpec = null;
checkForPartitionColumns(
colNames, Utilities.getColumnNamesFromFieldSchema(tbl.getPartitionKeys()));
validateSpecifiedColumnNames(colNames);
if (conf.getBoolVar(ConfVars.HIVE_STATS_COLLECT_PART_LEVEL_STATS) && tbl.isPartitioned()) {
isPartitionStats = true;
}
if (isPartitionStats) {
isTableLevel = false;
partSpec = AnalyzeCommandUtils.getPartKeyValuePairsFromAST(tbl, ast, conf);
handlePartialPartitionSpec(partSpec, null);
} else {
isTableLevel = true;
}
colType = getColumnTypes(colNames);
rewrittenQuery = genRewrittenQuery(colNames, conf, partSpec, isPartitionStats);
rewrittenTree = genRewrittenTree(rewrittenQuery);
} else {
// Not an analyze table column compute statistics statement - don't do any rewrites
originalTree = rewrittenTree = ast;
rewrittenQuery = null;
isRewritten = false;
}
// Setup the necessary metadata if originating from analyze rewrite
if (isRewritten) {
qb = getQB();
qb.setAnalyzeRewrite(true);
qbp = qb.getParseInfo();
analyzeRewrite = new AnalyzeRewriteContext();
analyzeRewrite.setTableName(tbl.getFullyQualifiedName());
analyzeRewrite.setTblLvl(isTableLevel);
analyzeRewrite.setColName(colNames);
analyzeRewrite.setColType(colType);
qbp.setAnalyzeRewrite(analyzeRewrite);
initCtx(ctx);
ctx.setExplainConfig(origCtx.getExplainConfig());
LOG.info("Invoking analyze on rewritten query");
analyzeInternal(rewrittenTree);
// After analyzeInternal() Hiveop get set as Query
// since we are passing in AST for select query, so reset it.
this.queryState.setCommandType(HiveOperation.ANALYZE_TABLE);
} else {
initCtx(origCtx);
LOG.info("Invoking analyze on original query");
analyzeInternal(originalTree);
}
}
/**
* @param ast
* is the original analyze ast
* @param qb
* is the qb that calls this function
* @param sem
* is the semantic analyzer that calls this function
* @return
* @throws SemanticException
*/
public ASTNode rewriteAST(ASTNode ast, ColumnStatsAutoGatherContext context)
throws SemanticException {
tbl = AnalyzeCommandUtils.getTable(ast, this);
colNames = getColumnName(ast);
// Save away the original AST
originalTree = ast;
boolean isPartitionStats = AnalyzeCommandUtils.isPartitionLevelStats(ast);
Map partSpec = null;
checkForPartitionColumns(colNames,
Utilities.getColumnNamesFromFieldSchema(tbl.getPartitionKeys()));
validateSpecifiedColumnNames(colNames);
if (conf.getBoolVar(ConfVars.HIVE_STATS_COLLECT_PART_LEVEL_STATS) && tbl.isPartitioned()) {
isPartitionStats = true;
}
if (isPartitionStats) {
isTableLevel = false;
partSpec = AnalyzeCommandUtils.getPartKeyValuePairsFromAST(tbl, ast, conf);
handlePartialPartitionSpec(partSpec, context);
} else {
isTableLevel = true;
}
colType = getColumnTypes(colNames);
rewrittenQuery = genRewrittenQuery(colNames, conf, partSpec, isPartitionStats);
rewrittenTree = genRewrittenTree(rewrittenQuery);
context.analyzeRewrite = new AnalyzeRewriteContext();
context.analyzeRewrite.setTableName(tbl.getFullyQualifiedName());
context.analyzeRewrite.setTblLvl(isTableLevel);
context.analyzeRewrite.setColName(colNames);
context.analyzeRewrite.setColType(colType);
return rewrittenTree;
}
}