Changeset 766 for trunk/gsdl

Timestamp:

1999-11-02T11:06:06+13:00 (25 years ago)

Author:

sjboddie

Message:

Added filter option to remove documents not matching a phrase match.
This used to be done in the receptionist.

Location:

trunk/gsdl/src/colservr

Files:

• queryfilter.cpp (modified) (9 diffs)
• queryfilter.h (modified) (2 diffs)

trunk/gsdl/src/colservr/queryfilter.cpp

@@ -28 +28 @@
 /*
    $Log$
+   Revision 1.20  1999/11/01 22:06:06  sjboddie
+   Added filter option to remove documents not matching a phrase match.
+   This used to be done in the receptionist.
+
    Revision 1.19  1999/10/19 03:23:40  davidb
    Collection building support through web pages
@@ -243 +247 @@
     get_phrase_terms (queryparams.querystring, queryresults.orgterms, phrases);
 
-    if (phrases.size() > 0) {
+    num_phrases = phrases.size();
+    if (num_phrases > 0) {
 
       // get the long version of the index
@@ -276 +281 @@
 void queryfilterclass::parse_query_params (const FilterRequest_t &request,
                        vector<queryparamclass> &query_params,
-                       int &startresults,
-                       int &endresults,
-                       ostream &logout) {
+                       int &startresults, int &endresults,
+                       text_t &phrasematch, ostream &logout) {
   outconvertclass text_t2ascii;
 
@@ -285 +289 @@
   startresults = filterOptions["StartResults"].defaultValue.getint();
   endresults = filterOptions["EndResults"].defaultValue.getint();
+  phrasematch = filterOptions["PhraseMatch"].defaultValue;
 
   // set defaults for query parameters
@@ -352 +357 @@
     } else if ((*options_here).name == "Maxdocs") {
       query.maxdocs = (*options_here).value.getint();
+    } else if ((*options_here).name == "PhraseMatch") {
+      phrasematch = (*options_here).value;
     } else {
       logout << text_t2ascii
@@ -464 +471 @@
   gdbmptr = NULL;
   mgsearchptr = NULL;
+  num_phrases = 0;
 
   FilterOption_t filtopt;
@@ -576 +584 @@
   filtopt.validValues.push_back("1000");
   filterOptions["Maxdocs"] = filtopt;
+
+  // -- onePerQuery  PhraseMatch  enumerated
+  filtopt.clear();
+  filtopt.name = "PhraseMatch";
+  filtopt.type = FilterOption_t::enumeratedt;
+  filtopt.repeatable = FilterOption_t::onePerQuery;
+  filtopt.defaultValue = "some_phrases";
+  filtopt.validValues.push_back ("all_phrases");
+  filtopt.validValues.push_back ("some_phrases");
+  filtopt.validValues.push_back ("all_docs");
+  filterOptions["PhraseMatch"] = filtopt;
 }
 
@@ -678 +697 @@
   int startresults = filterOptions["StartResults"].defaultValue.getint();
   int endresults = filterOptions["EndResults"].defaultValue.getint();
+  text_t phrasematch = filterOptions["PhraseMatch"].defaultValue;
+
   vector<queryparamclass> queryfilterparams;
-  parse_query_params (request, queryfilterparams, startresults, endresults, logout);  
+  parse_query_params (request, queryfilterparams, startresults, 
+              endresults, phrasematch, logout);  
   
   // do query
@@ -710 +732 @@
     docresultmap::iterator docset_here = queryresults.docs.docset.find (*docorder_here);
     
+    // documents containing matching phrases will be sorted to the top so
+    // we can break out once we're past those that match the PhraseMatch
+    // option -- "all_phrases" = return only those documents containing all
+    //                       phrases in query string
+    //           "some_phrases" = return only those documents containing 
+    //                            at least 1 of the phrases in the document
+    //           "all_docs" = return all documents regardless
+    if (num_phrases > 0) {
+      if ((phrasematch == "all_phrases") && ((*docset_here).second.num_phrase_match < num_phrases))
+        break;
+      if ((phrasematch == "some_phrases") && ((*docset_here).second.num_phrase_match < 1))
+        break;
+    }
+
     // see if there is a result for this number,
     // if it is in the request set (or the request set is empty)

trunk/gsdl/src/colservr/queryfilter.h

@@ -66 +66 @@
   
   mgsearchclass *mgsearchptr;
-
+  
+  int num_phrases;
 
   void get_phrase_terms (const text_t &querystring, 
@@ -79 +80 @@
   void parse_query_params (const FilterRequest_t &request,
                vector<queryparamclass> &query_params,
-               int &startresults,
-               int &endresults,
-               ostream &logout);
+               int &startresults, int &endresults,
+               text_t &phrasematch, ostream &logout);
 
   // do query that might involve multiple sub queries