Changeset 8357

Timestamp:

2004-10-18T12:30:06+13:00 (20 years ago)

Author:

kjdon

Message:

added a partial fix to plain searching with mgg - if you are searching in a field, it used to put [ ]:TI around the whole query string. now if you are simple 'and' searching, then it does this, otherwise it individually tags words, like [q1]:TI [q2]:TI.

Location:

trunk/gsdl/src/recpt

Files:

• queryaction.cpp (modified) (3 diffs)
• querytools.cpp (modified) (1 diff)
• querytools.h (modified) (1 diff)

trunk/gsdl/src/recpt/queryaction.cpp

@@ -1312 +1312 @@
   get_formatted_query_string(formattedstring, segment, args, disp, logout);
 
-
   if (!formattedstring.empty()) { // do the query
     // note! formattedstring is in unicode! mg and mgpp must convert!
@@ -1397 +1396 @@
     // remove & | ! for simple search,do segmentation if necessary
     format_querystring (formattedstring, args.getintarg("b"), segment);
-    if (args["ct"]=="1") { // mgpp - we need to add in the field info
-      mgpp_format_field_info(formattedstring, args["fqf"]);
-    }
-    else if (args["ct"]=="2") { // lucene - we need to add in the field info
-      lucene_format_field_info(formattedstring, args["fqf"]);
+    if (args["ct"]!=0) { // mgpp and lucene - need to add in tag info if appropriate
+      format_field_info(formattedstring, args);
     }
 
@@ -1408 +1404 @@
           args.getintarg("ct"));
     args["q"] = formattedstring;
+    
   }
   else if (args["qt"]=="1" || args["qto"]=="2"){ // form search

trunk/gsdl/src/recpt/querytools.cpp

@@ -608 +608 @@
 
 
-void mgpp_format_field_info(text_t & querystring, const text_t &tag) {
-
+void add_field_info(text_t &querystring, const text_t &tag, int type) {
+
+  if (type == 1) { //mgpp
+    querystring = "["+querystring+"]:"+tag;
+  } else if (type == 2) { // lucene
+    querystring = tag+":("+querystring+")";
+  }
+    
+}
+
+
+void format_field_info(text_t &querystring, cgiargsclass &args) {
+  
+  text_t tag = args["fqf"];
   if (tag == "ZZ" || tag == "") {
     return; // do nothing
   }
-
-  querystring = "["+querystring+"]:"+tag;
-}
-
-void lucene_format_field_info(text_t & querystring, const text_t &tag) {
-
-  if (tag == "ZZ" || tag == "") {
-    return; // do nothing
-  }
-
-  querystring = tag+":("+querystring+")";
-}
+  
+  int argct = args.getintarg("ct");
+  int argt = args.getintarg("t");// t=0 -and, t=1 - or
+  int argb = args.getintarg("b"); // b=0 simple, b=1 advanced
+
+  if (argb==0 && argt==0) {
+    // simple 'and' search - just put tag info round whole query string
+    add_field_info(querystring, tag, argct);
+    return;
+  }
+  
+  // we need to individually tag words
+  text_t outtext;
+  text_t word;
+  //unsigned short c;                                                            
+  text_t::const_iterator here = querystring.begin();
+  text_t::const_iterator end = querystring.end();
+
+  while (here !=end) {
+
+    if (is_unicode_letdig(*here)|| *here == '#' || *here == '/' ) {
+      // include term modifiers in a word just in case
+      // not word boundary
+      word.push_back(*here);
+      here++;    
+    }
+    else {
+      // found word boundary   
+      if (!word.empty() ) {
+    add_field_info(word, tag, argct);
+    outtext += word;
+    word.clear();
+      } 
+      // everything else, we add into the query string
+      outtext.push_back(*here);
+      here++;
+    }
+  }
+   
+  // get last word
+  if (!word.empty()) {
+    add_field_info(word, tag, argct);
+    outtext += word;
+  }
+  
+  querystring =  outtext;
+}
+

trunk/gsdl/src/recpt/querytools.h

@@ -68 +68 @@
 text_t addstemcase(const text_t &terms, const text_t &stem, const text_t &fold);
 text_t formatelem(text_t &text);
-void mgpp_format_field_info(text_t &querystring, const text_t &tag);
-void lucene_format_field_info(text_t &querystring, const text_t &tag);
-
+void format_field_info(text_t &querystring, cgiargsclass &args);
 #endif