java/pig-galago/src/com/yahoo/pig/data/DataBag.java

   1 /*
   2  * Copyright (c) 2007 Yahoo! Inc.  All rights reserved.
   3  * See accompanying LICENSE file.
   4  */
   5 package com.yahoo.pig.data;
   6
   7 import java.io.IOException;
   8 import java.util.ArrayList;
   9 import java.util.Collections;
  10 import java.util.Comparator;
  11 import java.util.Iterator;
  12 import java.util.List;
  13
  14 import com.yahoo.pig.impl.eval.EvalItem;
  15 import com.yahoo.pig.impl.eval.EvalItemList;
  16 import com.yahoo.pig.impl.eval.StarEvalItem;
  17
  18 /**
  19  * A collection of Tuples
  20  */
  21 public class DataBag extends DataCollector implements Datum{
  22     protected List<Tuple> content;
  23
  24     public DataBag() {
  25         content = new ArrayList<Tuple>();
  26     }
  27
  28     public DataBag(List<Tuple> c) {
  29         content = c;
  30     }
  31
  32     public DataBag(Tuple t) {
  33         content = new ArrayList<Tuple>();
  34         content.add(t);
  35     }
  36
  37     public int cardinality() {
  38         return content.size();
  39     }
  40
  41     public boolean isEmpty() {
  42         return content.size() == 0;
  43     }
  44
  45     public int compareTo(Object other) {
  46         if (other instanceof DataAtom) return +1;
  47         if (other instanceof Tuple) return -1;
  48         if (other instanceof DataBag){
  49                 DataBag bOther = (DataBag) other;
  50                 if (this.cardinality() != bOther.cardinality()) {
  51                     return (this.cardinality() - bOther.cardinality());
  52                 }
  53
  54                 // same cardinality, so compare tuple by tuple ...
  55                 this.sort();
  56                 bOther.sort();
  57
  58                 Iterator<Tuple> thisIt = this.content();
  59                 Iterator<Tuple> otherIt = bOther.content();
  60                 while (thisIt.hasNext() && otherIt.hasNext()) {
  61                     Tuple thisT = thisIt.next();
  62                     Tuple otherT = otherIt.next();
  63
  64                     int c = thisT.compareTo(otherT);
  65                     if (c != 0) return c;
  66                 }
  67
  68                 return 0;   // if we got this far, they must be equal
  69         }else{
  70                 return -1;
  71         }
  72
  73     }
  74
  75     public boolean equals(Datum other) {
  76         return (compareTo(other) == 0);
  77     }
  78
  79     public void sort() {
  80         EvalItem item = new StarEvalItem(null);
  81         EvalItemList itemList = new EvalItemList(null);
  82         itemList.add(item);
  83
  84         sort(itemList);
  85     }
  86
  87     public void sort(EvalItemList spec) {
  88         Collections.sort(content, spec.getComparator());
  89     }
  90
  91     public void arrange(EvalItemList spec) {
  92         sort(spec);
  93     }
  94
  95     public void distinct(EvalItemList spec) {
  96         Comparator<Tuple> comparator = spec.getComparator();
  97
  98         Collections.sort(content, comparator);
  99
 100         Tuple lastTup = null;
 101         for (Iterator<Tuple> it = content.iterator(); it.hasNext(); ) {
 102             Tuple thisTup = it.next();
 103
 104             if (lastTup == null) {
 105                 lastTup = thisTup;
 106                 continue;
 107             }
 108
 109             if (comparator.compare(thisTup, lastTup) == 0) {
 110                 it.remove();
 111             } else {
 112                 lastTup = thisTup;
 113             }
 114         }
 115     }
 116
 117     public Iterator<Tuple> content() {
 118         return content.iterator();
 119     }
 120
 121
 122     public void add(Tuple t) {
 123         //if (t!=null)
 124                 content.add(t);
 125     }
 126
 127     public void addAll(DataBag b) {
 128         if (b instanceof BigDataBag) {
 129             Iterator<Tuple> it = b.content();
 130             while (it.hasNext()) {
 131                 content.add(it.next());
 132             }
 133         } else {
 134             content.addAll(b.content);
 135         }
 136     }
 137
 138     public void remove(Tuple d) {
 139         content.remove(d);
 140     }
 141
 142     /**
 143      * Returns the value of field i. Since there may be more than one tuple in the bag, this
 144      * function throws an exception if it is not the case that all tuples agree on this field
 145      */
 146     public DataAtom getField(int i) throws IOException {
 147         DataAtom val = null;
 148
 149         for (Iterator<Tuple> it = content(); it.hasNext();) {
 150             DataAtom currentVal = it.next().getAtomField(i);
 151
 152             if (val == null) {
 153                 val = currentVal;
 154             } else {
 155                 if (!val.strval().equals(currentVal.strval()))
 156                     throw new IOException("Cannot call getField on a databag unless all tuples agree.");
 157             }
 158         }
 159
 160         if (val == null)
 161             throw new IOException("Cannot call getField on an empty databag.");
 162
 163         return val;
 164     }
 165
 166     public void clear(){
 167         content = new ArrayList<Tuple>();
 168         Runtime.getRuntime().gc();
 169     }
 170
 171     public String toString() {
 172         StringBuffer sb = new StringBuffer();
 173         sb.append('{');
 174         for (Iterator it = content(); it.hasNext();) {
 175             sb.append(it.next().toString());
 176             if (it.hasNext())
 177                 sb.append(", ");
 178         }
 179         sb.append('}');
 180         return sb.toString();
 181     }
 182 }