xapian-trec.cc

   1 /*
   2 **  xapian-trec.cc
   3 **
   4 **  AM 8/11/2006
   5 **  An example TREC search mechanism using Xapian
   6 **
   7 */
   8
   9 #include <xapian.h>
  10 #include <iostream>
  11 #include <fstream>
  12 #include <string>
  13 #include <vector>
  14 #include "split.h"
  15
  16 using namespace Xapian;
  17 using namespace std;
  18
  19 int main (int argc, char *argv[]) {
  20
  21   // There must be four command line arguments passed in the order:
  22   // database_name, query file name, results file name, db type, run type,
  23   if (argc < 5) {
  24     cerr << "ERROR: Insufficient arguments passed to program\n";
  25     cerr << "USAGE: xapian-trec <database> <query file> <results file>  <run id>  \n";
  26     exit(0);
  27   } // END if
  28
  29   // Catch any Error exceptions thrown
  30   try {
  31
  32     /* set up xapian search and term handling */
  33     Database db(argv[1]);
  34     Enquire enquire(db);
  35     Stem stemmer("english");
  36
  37     /* open files for reading and writing, and items to use/record during search */
  38     std::ifstream query_file( argv[2]);
  39     if(!query_file) cerr << "main) Can't open the query file" << argv[2] << '\n';
  40     std::ofstream results_file( argv[3]);
  41     if(!results_file) cerr << "main) Can't open the results file" << argv[3] << '\n';
  42     string run_id( argv[4] );
  43
  44     // iterate through query file until none left
  45     while(!query_file.eof()) {
  46       string topic_no;           // the topic number of a given query
  47
  48       /* get the line */
  49       char ch;
  50       string line;
  51       while( query_file.good() )
  52                                                  line += (char) query_file.get();
  53
  54       /* get the topic no, and terms from the line */
  55       vector<string> terms;
  56       split(line, ' ', terms );
  57
  58       /*get the topic number */
  59       topic_no = terms[0];
  60                         vector<string>::const_iterator termiter = terms.begin();
  61       termiter++;
  62
  63       /*get the terms for querying */
  64       vector<string> stemmed_terms;
  65       for( ; termiter != terms.end(); termiter++)
  66                                          stemmed_terms.push_back(stemmer(*termiter));
  67
  68       /* create the query and get the results for it */
  69       Query query(Query::OP_OR, stemmed_terms.begin(), stemmed_terms.end());
  70       enquire.set_query(query);
  71       MSet matches = enquire.get_mset(0, 1000 );
  72
  73       // get the results and save them to the file
  74                         int next=0;
  75       for ( MSetIterator i = matches.begin(); i != matches.end(); ++i ) {
  76                                         results_file << topic_no << "Q0" << i.get_document().get_data() << next << i.get_weight() << run_id << '\n';
  77                                         next++;
  78                         } // END for
  79
  80     } /* END while */
  81
  82   } catch (const Error &error) {
  83     cout << "Exception: "  << error.get_msg() << endl;
  84     exit(1);
  85   } // END try/catch
  86
  87 } // END main