Changeset 12380

Timestamp:

2006-08-02T16:30:23+12:00 (18 years ago)

Author:

mdewsnip

Message:

Now shows the stopwords removed by Lucene, many thanks to Me and DL Consulting Ltd.

Location:

trunk/gsdl

Files:

• macros/english.dm (modified) (1 diff)
• src/colservr/expat_resultset.cpp (modified) (1 diff)
• src/colservr/lucenequeryfilter.cpp (modified) (2 diffs)
• src/colservr/queryinfo.h (modified) (1 diff)
• src/recpt/comtypes.cpp (modified) (2 diffs)
• src/recpt/comtypes.h (modified) (1 diff)
• src/recpt/queryaction.cpp (modified) (3 diffs)

trunk/gsdl/macros/english.dm

@@ -316 +316 @@
 )}
 _textinvalidquery_ {Invalid query syntax}
+_textstopwordsmsg_ {The following words are too common and were ignored: }
 
 _textmorethan_ {More than }

trunk/gsdl/src/colservr/expat_resultset.cpp

@@ -51 +51 @@
     termfreqobj.termfreq = atoi(term_frequency_str);
     queryresult_ptr->orgterms.push_back(termfreqobj);
+  }
+
+  if (element_name == "StopWord") {
+    char* stop_word_value_str = get_attribute(attributes, "value");
+    queryresult_ptr->stopwords.insert((text_t) stop_word_value_str);
   }
 

trunk/gsdl/src/colservr/lucenequeryfilter.cpp

@@ -238 +238 @@
       ++terms_here;
     }
+
+    // add the stop words
+    text_tset::iterator stopwords_here = queryresults.stopwords.begin();
+    text_tset::iterator stopwords_end = queryresults.stopwords.end();
+    while (stopwords_here != stopwords_end) {
+      response.stopwords.insert(*stopwords_here);
+      ++stopwords_here;
+    }
   }
 
@@ -367 +375 @@
     ++termvar_here;
       }
+
+      // add the stop words
+      text_tset::iterator stopwords_here = thisqueryresults.stopwords.begin();
+      text_tset::iterator stopwords_end = thisqueryresults.stopwords.end();
+      while (stopwords_here != stopwords_end) {
+    multiresults.stopwords.insert(*stopwords_here);
+    ++stopwords_here;
+      }
     }
    

trunk/gsdl/src/colservr/queryinfo.h

@@ -176 +176 @@
   termfreqclassarray terms;
   text_tset termvariants;
-  
+  text_tset stopwords;
+
   void clear ();
   queryresultsclass &operator=(const queryresultsclass &q);

trunk/gsdl/src/recpt/comtypes.cpp

@@ -267 +267 @@
   termInfo.erase (termInfo.begin(), termInfo.end());
   docInfo.erase (docInfo.begin(), docInfo.end());
+  stopwords.clear();
 }
 
@@ -274 +275 @@
   termInfo = x.termInfo;
   docInfo = x.docInfo;
+  stopwords = x.stopwords;
   return *this;
 }

trunk/gsdl/src/recpt/comtypes.h

@@ -332 +332 @@
   TermInfo_tarray termInfo; // empty if not requested
   ResultDocInfo_tarray docInfo; // empty if not requested
+  text_tset stopwords;
 
   FilterResponse_t &operator=(const FilterResponse_t &x);

trunk/gsdl/src/recpt/queryaction.cpp

@@ -1225 +1225 @@
       ++this_term;
     }
-    
+
     if (response.numDocs > 0) {
       numdocs += response.numDocs;
@@ -1578 +1578 @@
 // define_single_query_macros sets the extra macros for search_single_coll 
 // that couldn't be set until the query had been done. Those macros are 
-// _freqmsg_,
+// _freqmsg_ and _stopwordsmsg_
 void queryaction::define_single_query_macros (cgiargsclass &args, 
                           displayclass &disp, 
                           const FilterResponse_t &response) {
-  // set up _freqmsg_ macros
+  // set up _freqmsg_ and _stopwordsmsg_ macros
 
   text_t freqmsg = "";
@@ -1594 +1594 @@
     ++this_term;
   }
-  
   disp.setmacro ("freqmsg", "query", freqmsg);
+
+  text_t stopwordsmsg = "_textstopwordsmsg_ ";
+  text_tset::const_iterator stopwords_iterator = response.stopwords.begin();
+  while (stopwords_iterator != response.stopwords.end()) {
+    if (stopwordsmsg != "_textstopwordsmsg_ ") {
+      stopwordsmsg += ", ";
+    }
+    stopwordsmsg += (*stopwords_iterator);
+    ++stopwords_iterator;
+  }
+  disp.setmacro("stopwordsmsg", "query", stopwordsmsg);
 }