Changeset 12388

Timestamp:

2006-08-02T17:34:50+12:00 (18 years ago)

Author:

mdewsnip

Message:

Fuzzy searching with Lucene, many thanks to John Thompson and DL Consulting Ltd.

Location:

trunk/gsdl/src

Files:

• colservr/lucenesearch.cpp (modified) (2 diffs)
• colservr/queryfilter.cpp (modified) (3 diffs)
• colservr/queryinfo.cpp (modified) (4 diffs)
• colservr/queryinfo.h (modified) (1 diff)
• recpt/queryaction.cpp (modified) (1 diff)
• recpt/querytools.cpp (modified) (1 diff)

trunk/gsdl/src/colservr/lucenesearch.cpp

@@ -105 +105 @@
     defaultStemMethod |= 2;
   }
-  
+
   text_t utf8querystring = to_utf8(queryparams.querystring);
   cerr << "**** query string = " << utf8querystring << endl;
@@ -124 +124 @@
     cmd += " -sort \"" + queryparams.sortfield + "\"";
   }
+  if (queryparams.fuzzysearch)
+    {
+      cmd += " -fuzzy";
+    }
 
   // New code to support configuration of the default conjuction operator

trunk/gsdl/src/colservr/queryfilter.cpp

@@ -90 +90 @@
   query.maxdocs = filterOptions["Maxdocs"].defaultValue.getint();
   query.level = filterOptions["Level"].defaultValue;
+  // Lucene specific
   query.sortfield = filterOptions["SortField"].defaultValue;
+  query.fuzzysearch = (filterOptions["FuzzySearch"].defaultValue == "true");
   query.maxnumeric = maxnumeric;
   OptionValue_tarray::const_iterator options_here = request.filterOptions.begin();
@@ -117 +119 @@
       query.stemming = (filterOptions["Stem"].defaultValue == "true");
       query.level = filterOptions["Level"].defaultValue;
+      // Lucene specific
       query.sortfield = filterOptions["SortField"].defaultValue;
+      query.fuzzysearch = (filterOptions["FuzzySearch"].defaultValue == "true");
       query.maxnumeric = maxnumeric;
       // "all", needed when combining queries where the document results are needed
@@ -152 +156 @@
     } else if ((*options_here).name == "SortField") {
       query.sortfield = (*options_here).value;
+    } else if ((*options_here).name == "FuzzySearch") {
+      query.fuzzysearch = ((*options_here).value == "true");
     } else {
       logout << text_t2ascii

trunk/gsdl/src/colservr/queryinfo.cpp

@@ -48 +48 @@
   maxnumeric = 4;  // must default to the same value as mg_passes
   sortfield.clear();
+  fuzzysearch = 0; // 0 = not fuzzy, 1 = fuzzy
 }
 
@@ -66 +67 @@
   maxnumeric = q.maxnumeric;
   sortfield = q.sortfield;
-
+  fuzzysearch = q.fuzzysearch;
   return *this;
 }
@@ -85 +86 @@
       (x.maxdocs == y.maxdocs) &&
       (x.maxnumeric == y.maxnumeric) &&
-          (x.sortfield == y.sortfield));
+          (x.sortfield == y.sortfield) &&
+          (x.fuzzysearch == y.fuzzysearch));
 }
 
@@ -111 +113 @@
   outs << " maxnumeric = \"" << q.maxnumeric << "\"\n";
   outs << " sortfield = \"" << q.sortfield << "\"\n";
+  outs << " fuzzysearch = \"" << q.fuzzysearch << "\"\n";
   outs << "\n";
 

trunk/gsdl/src/colservr/queryinfo.h

@@ -67 +67 @@
   int maxnumeric;
   text_t sortfield; // Field to use for sorting result set (currently used by lucene)
+  int fuzzysearch; // Should search be fuzzy (only used by Lucene)
 
   queryparamclass ();

trunk/gsdl/src/recpt/queryaction.cpp

@@ -436 +436 @@
   arg_ainfo.defaultstatus = cgiarginfo::weak;
   arg_ainfo.argdefault = g_EmptyText;
+  arg_ainfo.savedarginfo = cgiarginfo::must;
+  argsinfo.addarginfo (NULL, arg_ainfo);
+
+  // "fuzzy" controls whether the search is fuzzy logic or not
+  // (only implemented for Lucene collection... umm.. ever).
+  arg_ainfo.shortname = "fuzzy";
+  arg_ainfo.longname = "is this search be fuzzy";
+  arg_ainfo.multiplechar = false;
+  arg_ainfo.defaultstatus = cgiarginfo::weak;
+  arg_ainfo.argdefault = "0";
   arg_ainfo.savedarginfo = cgiarginfo::must;
   argsinfo.addarginfo (NULL, arg_ainfo);

trunk/gsdl/src/recpt/querytools.cpp

@@ -146 +146 @@
     request.filterOptions.push_back (option);
   }
+
+  // sort field for lucene
+  option.name = "FuzzySearch";
+  option.value = (args.getintarg("fuzzy")) ? "true" : "false";
+  request.filterOptions.push_back (option);
 
   set_more_queryfilter_options (request, args);