Examples of QBParseInfo

Examples of org.apache.hadoop.hive.ql.parse.QBParseInfo

61
62
63
64
65
66
67
68
69
    aliasToTabs = new HashMap<String, String>();
    aliasToSubq = new HashMap<String, QBExpr>();
    if (alias != null) {
      alias = alias.toLowerCase();
    }
    qbp = new QBParseInfo(alias, isSubQ);
    qbm = new QBMetaData();
    this.id = (outer_id == null ? alias : outer_id + ":" + alias);
  }
View Full Code Here

Examples of org.apache.hadoop.hive.ql.parse.QBParseInfo

66
67
68
69
70
71
72
73
74
    aliasToTabs = new HashMap<String, String>();
    aliasToSubq = new HashMap<String, QBExpr>();
    if (alias != null) {
      alias = alias.toLowerCase();
    }
    qbp = new QBParseInfo(alias, isSubQ);
    qbm = new QBMetaData();
    this.id = (outer_id == null ? alias : outer_id + ":" + alias);
  }
View Full Code Here

Examples of org.apache.hadoop.hive.ql.parse.QBParseInfo

77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
      if (currOp == op) {
        String currAliasId = alias;
        ctx.setCurrAliasId(currAliasId);
        mapCurrCtx.put(op, new GenMapRedCtx(currTask, currTopOp, currAliasId));

        QBParseInfo parseInfo = parseCtx.getQB().getParseInfo();
        if (parseInfo.isAnalyzeCommand()) {

          //   ANALYZE TABLE T [PARTITION (...)] COMPUTE STATISTICS;
          // The plan consists of a simple MapRedTask followed by a StatsTask.
          // The MR task is just a simple TableScanOperator

          StatsWork statsWork = new StatsWork(parseCtx.getQB().getParseInfo().getTableSpec());
          statsWork.setAggKey(op.getConf().getStatsAggPrefix());
          statsWork.setStatsReliable(
            parseCtx.getConf().getBoolVar(HiveConf.ConfVars.HIVE_STATS_RELIABLE));
          Task<StatsWork> statsTask = TaskFactory.get(statsWork, parseCtx.getConf());
          currTask.addDependentTask(statsTask);
          if (!ctx.getRootTasks().contains(currTask)) {
            ctx.getRootTasks().add(currTask);
          }
          currWork.setGatheringStats(true);
          // NOTE: here we should use the new partition predicate pushdown API to get a list of pruned list,
          // and pass it to setTaskPlan as the last parameter
          Set<Partition> confirmedPartns = new HashSet<Partition>();
          tableSpec tblSpec = parseInfo.getTableSpec();
          if (tblSpec.specType == tableSpec.SpecType.STATIC_PARTITION) {
            // static partition
            confirmedPartns.add(tblSpec.partHandle);
          } else if (tblSpec.specType == tableSpec.SpecType.DYNAMIC_PARTITION) {
            // dynamic partition
View Full Code Here

Examples of org.apache.hadoop.hive.ql.parse.QBParseInfo

64
65
66
67
68
69
70
71
72
    aliasToTabs = new HashMap<String, String>();
    aliasToSubq = new HashMap<String, QBExpr>();
    if (alias != null) {
      alias = alias.toLowerCase();
    }
    qbp = new QBParseInfo(alias, isSubQ);
    qbm = new QBMetaData();
    this.id = (outer_id == null ? alias : outer_id + ":" + alias);
  }
View Full Code Here

Examples of org.apache.hadoop.hive.ql.parse.QBParseInfo

74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
      if (currOp == op) {
        String currAliasId = alias;
        ctx.setCurrAliasId(currAliasId);
        mapCurrCtx.put(op, new GenMapRedCtx(currTask, currTopOp, currAliasId));

        QBParseInfo parseInfo = parseCtx.getQB().getParseInfo();
        if (parseInfo.isAnalyzeCommand()) {

          //   ANALYZE TABLE T [PARTITION (...)] COMPUTE STATISTICS;
          // The plan consists of a simple MapRedTask followed by a StatsTask.
          // The MR task is just a simple TableScanOperator

          StatsWork statsWork = new StatsWork(parseCtx.getQB().getParseInfo().getTableSpec());
          statsWork.setAggKey(op.getConf().getStatsAggPrefix());
          Task<StatsWork> statsTask = TaskFactory.get(statsWork, parseCtx.getConf());
          currTask.addDependentTask(statsTask);
          ctx.getRootTasks().add(currTask);
          currWork.setGatheringStats(true);
          // NOTE: here we should use the new partition predicate pushdown API to get a list of pruned list,
          // and pass it to setTaskPlan as the last parameter
          Set<Partition> confirmedPartns = new HashSet<Partition>();
          tableSpec tblSpec = parseInfo.getTableSpec();
          if (tblSpec.specType == tableSpec.SpecType.STATIC_PARTITION) {
            // static partition
            confirmedPartns.add(tblSpec.partHandle);
          } else if (tblSpec.specType == tableSpec.SpecType.DYNAMIC_PARTITION) {
            // dynamic partition
View Full Code Here

Examples of org.apache.hadoop.hive.ql.parse.QBParseInfo

62
63
64
65
66
67
68
69
70
    aliasToTabs = new HashMap<String, String>();
    aliasToSubq = new HashMap<String, QBExpr>();
    if (alias != null) {
      alias = alias.toLowerCase();
    }
    qbp = new QBParseInfo(alias, isSubQ);
    qbm = new QBMetaData();
    this.id = (outer_id == null ? alias : outer_id + ":" + alias);
  }
View Full Code Here

Examples of org.apache.hadoop.hive.ql.parse.QBParseInfo

74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
      if (currOp == op) {
        String currAliasId = alias;
        ctx.setCurrAliasId(currAliasId);
        mapCurrCtx.put(op, new GenMapRedCtx(currTask, currTopOp, currAliasId));

        QBParseInfo parseInfo = parseCtx.getQB().getParseInfo();
        if (parseInfo.isAnalyzeCommand()) {

          //   ANALYZE TABLE T [PARTITION (...)] COMPUTE STATISTICS;
          // The plan consists of a simple MapRedTask followed by a StatsTask.
          // The MR task is just a simple TableScanOperator

          StatsWork statsWork = new StatsWork(parseCtx.getQB().getParseInfo().getTableSpec());
          statsWork.setAggKey(op.getConf().getStatsAggPrefix());
          Task<StatsWork> statsTask = TaskFactory.get(statsWork, parseCtx.getConf());
          currTask.addDependentTask(statsTask);
          if (!ctx.getRootTasks().contains(currTask)) {
            ctx.getRootTasks().add(currTask);
          }
          currWork.setGatheringStats(true);
          // NOTE: here we should use the new partition predicate pushdown API to get a list of pruned list,
          // and pass it to setTaskPlan as the last parameter
          Set<Partition> confirmedPartns = new HashSet<Partition>();
          tableSpec tblSpec = parseInfo.getTableSpec();
          if (tblSpec.specType == tableSpec.SpecType.STATIC_PARTITION) {
            // static partition
            confirmedPartns.add(tblSpec.partHandle);
          } else if (tblSpec.specType == tableSpec.SpecType.DYNAMIC_PARTITION) {
            // dynamic partition
View Full Code Here

Examples of org.apache.hadoop.hive.ql.parse.QBParseInfo

64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
    Map<String, SplitSample> nameToSplitSample = pctx.getNameToSplitSample();
    Map<TableScanOperator, Table> topToTable = pctx.getTopToTable();

    QB qb = pctx.getQB();
    HiveConf conf = pctx.getConf();
    QBParseInfo qbParseInfo = qb.getParseInfo();

    // determine the query qualifies reduce input size for LIMIT
    // The query only qualifies when there are only one top operator
    // and there is no transformer or UDTF and no block sampling
    // is used.
    if (ctx.getTryCount() == 0 && topOps.size() == 1
        && !globalLimitCtx.ifHasTransformOrUDTF() &&
        nameToSplitSample.isEmpty()) {

      // Here we recursively check:
      // 1. whether there are exact one LIMIT in the query
      // 2. whether there is no aggregation, group-by, distinct, sort by,
      //    distributed by, or table sampling in any of the sub-query.
      // The query only qualifies if both conditions are satisfied.
      //
      // Example qualified queries:
      //    CREATE TABLE ... AS SELECT col1, col2 FROM tbl LIMIT ..
      //    INSERT OVERWRITE TABLE ... SELECT col1, hash(col2), split(col1)
      //                               FROM ... LIMIT...
      //    SELECT * FROM (SELECT col1 as col2 (SELECT * FROM ...) t1 LIMIT ...) t2);
      //
      Integer tempGlobalLimit = checkQbpForGlobalLimit(qb);

      // query qualify for the optimization
      if (tempGlobalLimit != null && tempGlobalLimit != 0) {
        TableScanOperator ts = (TableScanOperator) topOps.values().toArray()[0];
        Table tab = topToTable.get(ts);

        if (!tab.isPartitioned()) {
          if (qbParseInfo.getDestToWhereExpr().isEmpty()) {
            globalLimitCtx.enableOpt(tempGlobalLimit);
          }
        } else {
          // check if the pruner only contains partition columns
          if (PartitionPruner.onlyContainsPartnCols(tab,
View Full Code Here

Examples of org.apache.hadoop.hive.ql.parse.QBParseInfo

138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
   * @return if there is one and only one limit for all subqueries, return the limit
   *         if there is no limit, return 0
   *         otherwise, return null
   */
  private Integer checkQbpForGlobalLimit(QB localQb) {
    QBParseInfo qbParseInfo = localQb.getParseInfo();
    if (localQb.getNumSelDi() == 0 && qbParseInfo.getDestToClusterBy().isEmpty()
        && qbParseInfo.getDestToDistributeBy().isEmpty()
        && qbParseInfo.getDestToOrderBy().isEmpty()
        && qbParseInfo.getDestToSortBy().isEmpty()
        && qbParseInfo.getDestToAggregationExprs().size() <= 1
        && qbParseInfo.getDestToDistinctFuncExprs().size() <= 1
        && qbParseInfo.getNameToSample().isEmpty()) {
      if ((qbParseInfo.getDestToAggregationExprs().size() < 1 ||
          qbParseInfo.getDestToAggregationExprs().values().iterator().next().isEmpty()) &&
          (qbParseInfo.getDestToDistinctFuncExprs().size() < 1 ||
              qbParseInfo.getDestToDistinctFuncExprs().values().iterator().next().isEmpty())
          && qbParseInfo.getDestToLimit().size() <= 1) {
        Integer retValue;
        if (qbParseInfo.getDestToLimit().size() == 0) {
          retValue = 0;
        } else {
          retValue = qbParseInfo.getDestToLimit().values().iterator().next();
        }

        for (String alias : localQb.getSubqAliases()) {
          Integer limit = checkQbpForGlobalLimit(localQb.getSubqForAlias(alias).getQB());
          if (limit == null) {
View Full Code Here

Examples of org.apache.hadoop.hive.ql.parse.QBParseInfo

86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
      if (currOp == op) {
        String currAliasId = alias;
        ctx.setCurrAliasId(currAliasId);
        mapCurrCtx.put(op, new GenMapRedCtx(currTask, currAliasId));

        QBParseInfo parseInfo = parseCtx.getQB().getParseInfo();
        if (parseInfo.isAnalyzeCommand()) {
          boolean partialScan = parseInfo.isPartialScanAnalyzeCommand();
          boolean noScan = parseInfo.isNoScanAnalyzeCommand();
          if (inputFormat.equals(OrcInputFormat.class) && (noScan || partialScan)) {

            // ANALYZE TABLE T [PARTITION (...)] COMPUTE STATISTICS partialscan;
            // ANALYZE TABLE T [PARTITION (...)] COMPUTE STATISTICS noscan;
            // There will not be any MR or Tez job above this task
            StatsNoJobWork snjWork = new StatsNoJobWork(parseCtx.getQB().getParseInfo().getTableSpec());
            snjWork.setStatsReliable(parseCtx.getConf().getBoolVar(
                HiveConf.ConfVars.HIVE_STATS_RELIABLE));
            Task<StatsNoJobWork> snjTask = TaskFactory.get(snjWork, parseCtx.getConf());
            ctx.setCurrTask(snjTask);
            ctx.setCurrTopOp(null);
            ctx.getRootTasks().clear();
            ctx.getRootTasks().add(snjTask);
          } else {
            // ANALYZE TABLE T [PARTITION (...)] COMPUTE STATISTICS;
            // The plan consists of a simple MapRedTask followed by a StatsTask.
            // The MR task is just a simple TableScanOperator

            StatsWork statsWork = new StatsWork(parseCtx.getQB().getParseInfo().getTableSpec());
            statsWork.setAggKey(op.getConf().getStatsAggPrefix());
            statsWork.setSourceTask(currTask);
            statsWork.setStatsReliable(parseCtx.getConf().getBoolVar(
                HiveConf.ConfVars.HIVE_STATS_RELIABLE));
            Task<StatsWork> statsTask = TaskFactory.get(statsWork, parseCtx.getConf());
            currTask.addDependentTask(statsTask);
            if (!ctx.getRootTasks().contains(currTask)) {
              ctx.getRootTasks().add(currTask);
            }

            // ANALYZE TABLE T [PARTITION (...)] COMPUTE STATISTICS noscan;
            // The plan consists of a StatsTask only.
            if (parseInfo.isNoScanAnalyzeCommand()) {
              statsTask.setParentTasks(null);
              statsWork.setNoScanAnalyzeCommand(true);
              ctx.getRootTasks().remove(currTask);
              ctx.getRootTasks().add(statsTask);
            }

            // ANALYZE TABLE T [PARTITION (...)] COMPUTE STATISTICS partialscan;
            if (parseInfo.isPartialScanAnalyzeCommand()) {
              handlePartialScanCommand(op, ctx, parseCtx, currTask, parseInfo, statsWork, statsTask);
            }

            currWork.getMapWork().setGatheringStats(true);
            if (currWork.getReduceWork() != null) {
View Full Code Here
TOP
Copyright © 2018 www.massapi.com. All rights reserved.
All source code are property of their respective owners. Java is a trademark of Sun Microsystems, Inc and owned by ORACLE Inc. Contact coftware#gmail.com.