mcs/tools/monkeydoc/Monkeydoc/SearchableIndex.cs

   1 //
   2 //
   3 // SearchableIndex.cs: Index that uses Lucene to search through the docs
   4 //
   5 // Author: Mario Sopena
   6 //
   7
   8 using System;
   9 using System.IO;
  10 using System.Collections;
  11 // Lucene imports
  12 using Lucene.Net.Index;
  13 using Lucene.Net.Documents;
  14 using Lucene.Net.Analysis;
  15 using Lucene.Net.Analysis.Standard;
  16 using Lucene.Net.Search;
  17 using Lucene.Net.QueryParsers;
  18 using Lucene.Net.Store;
  19
  20 namespace MonkeyDoc
  21 {
  22         public class SearchableIndex
  23         {
  24                 const int maxSearchCount = 30;
  25
  26                 IndexSearcher searcher;
  27                 string dir;
  28                 public string Dir {
  29                         get {
  30                                 if (dir == null) dir = "search_index";
  31                                 return dir;
  32                         }
  33                         set { dir = value; }
  34                 }
  35                 public ArrayList Results;
  36
  37                 public static SearchableIndex Load (string dir) {
  38                         SearchableIndex s = new SearchableIndex ();
  39                         s.dir = dir;
  40                         s.Results = new ArrayList (20);
  41                         try {
  42                                 //s.searcher = new IndexSearcher (dir);
  43                                 // TODO: parametrize that depending if we run on the desktop (low footprint) or the server (use RAMDirectory for instance)
  44                                 s.searcher = new IndexSearcher (FSDirectory.Open (dir));
  45                         } catch (IOException) {
  46                                 Console.WriteLine ("Index nonexistent or in bad format");
  47                                 return null;
  48                         }
  49                         return s;
  50                 }
  51
  52                 //
  53                 // Search the index with term
  54                 //
  55
  56                 public Result Search (string term)
  57                 {
  58                         return Search (term, maxSearchCount);
  59                 }
  60
  61                 public Result Search (string term, int count)
  62                 {
  63                         return Search (term, count, 0);
  64                 }
  65
  66                 public Result Search (string term, int count, int start) {
  67                         try {
  68                                 term = term.ToLower ();
  69                                 Term htTerm = new Term ("hottext", term);
  70                                 Query qq1 = new FuzzyQuery (htTerm);
  71                                 Query qq2 = new TermQuery (htTerm);
  72                                 qq2.Boost = 10f;
  73                                 Query qq3 = new PrefixQuery (htTerm);
  74                                 qq3.Boost = 10f;
  75                                 DisjunctionMaxQuery q1 = new DisjunctionMaxQuery (0f);
  76                                 q1.Add (qq1);
  77                                 q1.Add (qq2);
  78                                 q1.Add (qq3);
  79                                 Query q2 = new TermQuery (new Term ("text", term));
  80                                 q2.Boost = 3f;
  81                                 Query q3 = new TermQuery (new Term ("examples", term));
  82                                 q3.Boost = 3f;
  83                                 DisjunctionMaxQuery q = new DisjunctionMaxQuery (0f);
  84
  85                                 q.Add (q1);
  86                                 q.Add (q2);
  87                                 q.Add (q3);
  88
  89                                 TopDocs top = SearchInternal (q, count, start);
  90                                 Result r = new Result (term, searcher, top.ScoreDocs);
  91                                 Results.Add (r);
  92                                 return r;
  93                         } catch (IOException) {
  94                                 Console.WriteLine ("No index in {0}", dir);
  95                                 return null;
  96                         }
  97                 }
  98
  99                 TopDocs SearchInternal (Query q, int count, int start)
 100                 {
 101                         // Easy path that doesn't involve creating a Collector ourselves
 102                         // watch for Lucene.NET improvement on that (like searcher.SearchAfter)
 103                         if (start == 0)
 104                                 return searcher.Search (q, count);
 105
 106                         var weight = searcher.CreateWeight (q); // TODO: reuse weight instead of query
 107                         var collector = TopScoreDocCollector.Create (start + count + 1, false);
 108                         searcher.Search (q, collector);
 109
 110                         return collector.TopDocs (start, count);
 111                 }
 112
 113                 public Result FastSearch (string term, int number)
 114                 {
 115                         try {
 116                                 term = term.ToLower ();
 117                                 Query q1 = new TermQuery (new Term ("hottext", term));
 118                                 Query q2 = new PrefixQuery (new Term ("hottext", term));
 119                                 q2.Boost = 0.5f;
 120                                 DisjunctionMaxQuery q = new DisjunctionMaxQuery (0f);
 121                                 q.Add (q1);
 122                                 q.Add (q2);
 123                                 TopDocs top = searcher.Search (q, number);
 124                                 return new Result (term, searcher, top.ScoreDocs);
 125                         } catch (IOException) {
 126                                 Console.WriteLine ("No index in {0}", dir);
 127                                 return null;
 128                         }
 129                 }
 130
 131                 Query Parse (string term, string field, bool fuzzy)
 132                 {
 133                         QueryParser parser = new QueryParser (Lucene.Net.Util.Version.LUCENE_CURRENT,
 134                                                               field,
 135                                                               new StandardAnalyzer (Lucene.Net.Util.Version.LUCENE_CURRENT));
 136                         return parser.Parse (term);
 137                 }
 138         }
 139         //
 140         // An object representing the search term with the results
 141         //
 142         public class Result {
 143                 string term;
 144                 Searcher searcher;
 145                 ScoreDoc[] docs;
 146
 147                 public string Term {
 148                         get { return term;}
 149                 }
 150
 151                 public int Count {
 152                         get { return docs.Length; }
 153                 }
 154
 155                 public Document this [int i] {
 156                         get { return searcher.Doc (docs[i].Doc); }
 157                 }
 158
 159                 public string GetTitle (int i)
 160                 {
 161                         Document d = this[i];
 162                         return d == null ? string.Empty : d.Get ("title");
 163                 }
 164
 165                 public string GetUrl (int i)
 166                 {
 167                         Document d = this[i];
 168                         return d == null ? string.Empty : d.Get ("url");
 169                 }
 170
 171                 public string GetFullTitle (int i)
 172                 {
 173                         Document d = this[i];
 174                         return d == null ? string.Empty : d.Get ("fulltitle");
 175                 }
 176
 177                 public float Score (int i)
 178                 {
 179                         return docs[i].Score;
 180                 }
 181
 182                 public Result (string Term, Searcher searcher, ScoreDoc[] docs)
 183                 {
 184                         this.term = Term;
 185                         this.searcher = searcher;
 186                         this.docs = docs;
 187                 }
 188         }
 189 }
 190