Changeset 11765

Timestamp:

2006-05-03T16:06:16+12:00 (18 years ago)

Author:

kjdon

Message:

made mgpp query parsing better for plain searching. now handles bool operators and NEAR/WITHIN as part of the query. also, plain advanced searching, default operator is now always OR

File:

• trunk/gsdl/src/recpt/querytools.cpp (modified) (5 diffs)

trunk/gsdl/src/recpt/querytools.cpp

@@ -46 +46 @@
 
   option.name = "MatchMode";
-  option.value = (args.getintarg("t")) ? "some" : "all";
+  // mgpp in advanced mode, always use some query
+  if (args.getintarg("ct") !=0 && args.getintarg("b") == 1) {
+    option.value = "some";
+  } else {
+    option.value = (args.getintarg("t")) ? "some" : "all";
+  }
   request.filterOptions.push_back (option);
 
@@ -285 +290 @@
 }
 
-// some query form parsing functions for use with mgpp
+// some query form parsing functions for use with mgpp & lucene
 
 void parse_reg_query_form(text_t &querystring, cgiargsclass &args)
@@ -606 +611 @@
 }
 
-void format_field_info(text_t &querystring, cgiargsclass &args) {
-  
+void format_field_info_lucene(text_t &querystring, cgiargsclass &args) {
   text_t tag = args["fqf"];
   if (tag == "ZZ") tag = ""; // ZZ is a special tag meaning no tag (all fields)
-
-  int argct = args.getintarg("ct");
-  bool mgpp = (argct == 1);
-  bool lucene = (argct == 2);
-
-  if (mgpp && tag == "") {
-    return; // no field specifier: do nothing
-  }
-  
+  int type = 2; //lucene
   int argt = args.getintarg("t");// t=0 -and, t=1 - or
   int argb = args.getintarg("b"); // b=0 simple, b=1 advanced
 
-  bool simple_AND_search = (argb==0 && argt==0); 
-  bool simple_OR_search = (argb==0 && argt==1);
-
-  if (mgpp && simple_AND_search) {
-    // mgpp, simple AND search, tag the whole query string
-    add_field_info(querystring, tag, argct);
+  // lucene simple OR - the string stays as is, but may need field tag
+  if (argb==0 && argt == 1) {
+    // just tag the entire thing
+    if (tag != "") {
+      add_field_info(querystring, tag, type);
+    }
     return;
   }
-  // resulting mgpp case - we need to tag each individual term or phrase
-  // TODO - allow AND. OR in query string and don't tag these words
-
-  if (lucene && (simple_OR_search || argb == 1)) {
-      // OR search or advanced search (here we assume that the user has added their term mods - don't need to add term mods
-    if (tag != "") {
-      // tag the whole string
-      add_field_info(querystring, tag, argct);
-    }
-    return;
-  }
-  
-
-  // if we have got here, we need to add in combiners (lucene) or
-  // we need to tag each individual word (mgpp OR search - mgpp can't do OR inside a field)
-
-  text_t combine = ((lucene)? "+" : "");
-  
-  text_t processed_querystring = "";
-  text_t queryelement = "";
-  
   bool in_phrase = false;
+  
+  text_t queryelem = "";
+  text_t finalquery = "";
+  
+  // only add in + for simple AND search
+  text_t combine = ((argb==0)? "+" : "");
+
+  // for lucene, we need to change & to && and | to || if advanced search
+  // we need to tag the entire string, if we have a field
+  // if we are simple and search, then we put && in between words
+  
   text_t::const_iterator here = querystring.begin();
   text_t::const_iterator end = querystring.end();
   while (here != end) {
-    if (is_unicode_letdig(*here) || is_special_character(argct, *here)) {
-      queryelement.push_back(*here);
+    if (is_unicode_letdig(*here) || is_special_character(type, *here)) {
+      queryelem.push_back(*here);
     }
 
     // Detect phrase starts/finishes
     else if (*here == '"') {
-      queryelement.push_back(*here);
+      queryelem.push_back(*here);
       if (in_phrase == false) in_phrase = true;
       else {
-    if (mgpp) {add_field_info(queryelement, tag, argct);}
-    processed_querystring += combine + queryelement;
-    queryelement.clear();
+    finalquery += combine + queryelem;
+    queryelem.clear();
     in_phrase = false;
       }
@@ -673 +658 @@
     // Found word boundary, in a phrase
     else if (in_phrase) {
-      queryelement.push_back(*here);
+      queryelem.push_back(*here);
     }
     // Word boundary, but not in a phrase
     else {
-      if (!queryelement.empty()) {
-    if (mgpp) {add_field_info(queryelement, tag, argct);}
-    processed_querystring += combine + queryelement;
-    queryelement.clear();
-      }
-      processed_querystring.push_back(*here);
+      if (*here == '&') {
+    queryelem.push_back('&');
+    queryelem.push_back('&');
+      } else if (*here == '|') {
+    queryelem.push_back('|');
+    queryelem.push_back('|');
+      } else {
+    if (!queryelem.empty()) {
+      finalquery += combine + queryelem;
+      queryelem.clear();
+    }
+    finalquery.push_back(*here);
+      }
     }
 
@@ -689 +681 @@
 
   // Get last element
-  if (!queryelement.empty()) {
-    if (mgpp) {add_field_info(queryelement, tag, argct);}
-    processed_querystring += combine + queryelement;
-  }
-
-  querystring = processed_querystring;
+  if (!queryelem.empty()) {
+    finalquery += combine + queryelem;
+  }
+
+  add_field_info(finalquery, tag, type);
+  querystring = finalquery;
+ cerr << "final query = "<<finalquery<<endl;
+}
+
+void format_field_info_mgpp(text_t &querystring, cgiargsclass &args) {
+  
+  text_t tag = args["fqf"];
+  if (tag == "ZZ") tag = ""; // ZZ is a special tag meaning no tag (all fields)
  
-  if (lucene) {
-    // tag the whole query string
-    add_field_info(querystring, tag, argct);
-  }
-}
-
-
+  int argt = args.getintarg("t");// t=0 -and, t=1 - or
+  int argb = args.getintarg("b"); // b=0 simple, b=1 advanced
+
+  if (tag == "" && argb ==1) {
+    return; // no field specifier, advanced mode, the query stays as written
+  }
+
+  int type = 1; // mgpp
+
+  bool simple_and = (argb==0 && argt==0); 
+  text_t finalquery = "";
+  text_t fieldpart ="";
+  text_t queryelem = "";
+  bool in_phrase = false;
+  bool in_field = false;
+
+  text_t::const_iterator here = querystring.begin();
+  text_t::const_iterator end = querystring.end();
+  while (here != end) {
+    if (is_unicode_letdig(*here)  || *here == '&' || is_special_character(type, *here)) {
+      queryelem.push_back(*here);
+    }
+    else if (*here == '|') {
+      in_field = false;
+    }
+    else if (*here == '!' || *here == '(' || *here == ')') {
+      if (!in_phrase) { // ignore these if in_phrase
+    // output field, then output operator
+    in_field = false;
+    if (!queryelem.empty()) {
+      if (!simple_and && !fieldpart.empty()) {
+        add_field_info(fieldpart, tag, type);
+        finalquery += fieldpart;
+        finalquery.push_back(' ');
+        fieldpart.clear();
+      }
+      fieldpart += queryelem;
+    }
+    if (!fieldpart.empty()) {
+      add_field_info(fieldpart, tag, type);
+      finalquery += fieldpart;
+      finalquery.push_back(' ');
+    }
+    fieldpart.clear();
+    queryelem.clear();
+    finalquery.push_back(*here);
+    finalquery.push_back(' ');
+      }
+    }
+    else if (*here == '"') {
+      queryelem.push_back(*here);
+      if (in_phrase == false) in_phrase = true;
+      else {
+    in_phrase = false;
+      }
+    }
+
+    // Found word boundary, in a phrase
+    else if (in_phrase) {
+      queryelem.push_back(*here);
+    }
+    // Found a word boundary
+    else {
+      if (!queryelem.empty()) {
+    if (queryelem == "&") {
+      in_field = true;
+      queryelem.clear();
+    }
+    else if (starts_with(queryelem, "NEAR") || starts_with(queryelem, "WITHIN")) {
+      
+      if (argb==1) {
+        // simple search, these not allowed
+        in_field = true;
+        fieldpart += queryelem;
+        fieldpart.push_back(' ');
+      }
+      queryelem.clear();
+      
+    } 
+    else {
+      if (!simple_and && !in_field) {
+        if (!fieldpart.empty()) {
+          add_field_info(fieldpart, tag, type);
+          finalquery += fieldpart;
+          finalquery.push_back(' ');
+          fieldpart.clear();
+        } 
+      }
+      
+      fieldpart += queryelem;
+      fieldpart.push_back(' ');
+      queryelem.clear();
+    }
+      }
+    }
+    ++here;
+  }
+  // at the end
+  if (!queryelem.empty()) {
+    if (!simple_and && !in_field && !fieldpart.empty()) {
+      add_field_info(fieldpart, tag, type);
+      finalquery += fieldpart;
+      finalquery.push_back(' ');
+      fieldpart.clear();
+    }
+    fieldpart += queryelem;
+  }
+  if (!fieldpart.empty()) {
+    add_field_info(fieldpart, tag, type);
+    finalquery += fieldpart;
+    fieldpart.clear();
+    finalquery.push_back(' ');
+  }
+
+  querystring  = finalquery;
+  cerr << "final query = "<<finalquery<<endl;
+}
+
+void format_field_info(text_t &querystring, cgiargsclass &args) {
+  int argct = args.getintarg("ct");
+  if (argct == 1) {
+    format_field_info_mgpp(querystring, args);
+  } else if (argct == 2) {
+    format_field_info_lucene(querystring, args);
+  }
+}
+