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