Changeset 403 for trunk/gsdl/src/recpt/queryaction.cpp

Timestamp:

1999-07-19T12:16:59+12:00 (25 years ago)

Author:

sjboddie

Message:

no longer display documents that don't match all phrases in query string

File:

• trunk/gsdl/src/recpt/queryaction.cpp (modified) (7 diffs)

trunk/gsdl/src/recpt/queryaction.cpp

@@ -12 +12 @@
 /*
    $Log$
+   Revision 1.16  1999/07/19 00:16:58  sjboddie
+   no longer display documents that don't match all phrases in query string
+
    Revision 1.15  1999/07/16 08:33:36  rjmcnab
    Changed the logic for getting the results string slightly
@@ -77 +80 @@
 
 queryaction::queryaction () {
+
+  num_phrases = 0;
+
   // this action uses cgi variable "a"
   cgiarginfo arg_ainfo;
@@ -294 +300 @@
                       recptproto */*collectproto*/, ostream &/*logout*/) {
 
-  // define_internal_macros doesn't sets the following macros.
+  // define_internal_macros sets the following macros:
 
   // _quotedquery_  the part of the query string that was quoted for post-processing
@@ -315 +321 @@
 
 
-
   // get the quoted bits of the query string and set _quotedquery_
-  text_t &arg_q = args["q"];
-  if (!arg_q.empty()) {
-
-    text_t::const_iterator end = arg_q.end();
-    text_t::const_iterator here = findchar (arg_q.begin(), end, '\"');
-    if (here != end) {
-      text_t quotedquery, tmptext;
-      bool first = true;
-      bool foundquote = false;
-      while (here != end) {
-    if (*here == '\"') {
-      if (foundquote) {
-        if (!tmptext.empty()) {
-          if (!first) quotedquery += " and ";
-          else first = false;
-          quotedquery += "\"" + tmptext + "\"";
-          tmptext.clear();
-        }
-        foundquote = false;
-      } else foundquote = true;
-    } else {
-      if (foundquote) tmptext.push_back (*here);
-    }
-    here ++;
-      }
-      disp.setmacro ("quotedquery", "query", quotedquery);
-    }
-  }
+  text_tarray phrases;
+  get_phrases (args["q"], phrases);
+  text_tarray::const_iterator phere = phrases.begin();
+  text_tarray::const_iterator pend = phrases.end();
+  bool first = true;
+  text_t quotedquery;
+  while (phere != pend) {
+    if (!first)
+      if ((phere +1) == pend) quotedquery += " and ";
+      else quotedquery += ", ";
+    
+    quotedquery += "\"" + *phere + "\"";
+    first = false;
+    phere ++;
+  }
+  if (args.getintarg("s")) quotedquery += "_textstemon_";
+  disp.setmacro ("quotedquery", "query", quotedquery);
+
+  // we'll also set num_phrases here so we don't have to parse the 
+  // querystring again later (we need to know this before outputting 
+  // results so we don't include results for documents not containing 
+  // all requested phrases).
+  num_phrases = phrases.size();
+
 }
 
@@ -470 +471 @@
 
     while (this_doc != end_doc) {
+      // don't include docs that didn't match phrases (if there were any)
+      // those that did match will have been sorted to the top
+      if ((*this_doc).num_phrase_match < num_phrases) break;
       textout << outconvert << disp << "<tr>\n"
-
+    
     //            << "<td valign=top nowrap>r: " << (*this_doc).ranking
     //            << " t: " << (*this_doc).num_terms_matched << " p: "
@@ -478 +482 @@
           << get_formatted_string (*this_doc, formatlistptr) << "\n"
           << "</tr>\n";
+
       this_doc ++;
     }
@@ -515 +520 @@
   int maxdocs = args.getintarg("m");
   int numdocs = response.numDocs;
-  if (response.isApprox == MoreThan && numdocs > maxdocs) numdocs = maxdocs;
-
-  if (response.isApprox == Approximate) resline = "_textapprox_";
-  else if (response.isApprox == MoreThan) resline = "_textmorethan_";
+  isapprox isApprox = response.isApprox;
+
+  // if there were phrases (post-processing) we're not going to include 
+  // those documents that didn't match 
+  if (num_phrases > 0) {
+    numdocs = 0;
+    isApprox = Exact;
+    ResultDocInfo_tarray::const_iterator this_doc = response.docInfo.begin();
+    ResultDocInfo_tarray::const_iterator end_doc = response.docInfo.end();
+    while (this_doc != end_doc) {
+      if ((*this_doc).num_phrase_match == num_phrases) numdocs ++;
+      else break; // we can bail here as matching docs are sorted to top
+      this_doc++;
+    }
+  }
+
+  if (isApprox == MoreThan && numdocs > maxdocs) numdocs = maxdocs;
+
+  if (isApprox == Approximate) resline = "_textapprox_";
+  else if (isApprox == MoreThan) resline = "_textmorethan_";
   
   if (numdocs == 0) resline = "_textnodocs_";