Changeset 10995 for trunk/gsdl

Timestamp:

2005-12-15T16:57:58+13:00 (18 years ago)

Author:

kjdon

Message:

fixed up mgpp and lucene plain search query string formatting. mgpp now handles phrases inside fields, and * inside fields. lucene doesn't need to tag individual words for an OR query - it can do OR and AND inside a field

File:

• trunk/gsdl/src/recpt/querytools.cpp (modified) (2 diffs)

trunk/gsdl/src/recpt/querytools.cpp

@@ -612 +612 @@
 void add_field_info(text_t &querystring, const text_t &tag, int type) {
 
+  if (tag == "") return; // do nothing
   if (type == 1) { //mgpp
     querystring = "["+querystring+"]:"+tag;
@@ -620 +621 @@
 }
 
+bool is_special_character(int indexer_type, unsigned short character) {
+  // mgpp
+  if (indexer_type == 1) {
+    return (character == '#' || character == '/' || character == '*');
+  }
+  // lucene
+  else if (indexer_type ==2) {
+    return (character == '?' || character == '*' || character == '~' || 
+        character == '^');
+  }
+  return false;
+}
 
 void format_field_info(text_t &querystring, cgiargsclass &args) {
   
   text_t tag = args["fqf"];
-  if (tag == "ZZ" || tag == "") {
-    return; // do nothing
-  }
-  
+  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 argt = args.getintarg("t");// t=0 -and, t=1 - or
   int argb = args.getintarg("b"); // b=0 simple, b=1 advanced
 
-  // Special code for Lucene
-  // The default operator for Lucene is "or", so we need to add "+" symbols when t == 0
-  // Also, we need to be careful not to mess up phrase searches
-  if (argct == 2) {
-    text_t processed_querystring = "";
-    text_t queryelement = "";
-    text_t combine = ((argt == 0) ? "+" : "");
-    bool in_phrase = false;
-    text_t::const_iterator here = querystring.begin();
-    text_t::const_iterator end = querystring.end();
-    while (here != end) {
-      if (is_unicode_letdig(*here)) {
-    queryelement.push_back(*here);
-      }
-
-      // Detect phrase starts/finishes
-      else if (*here == '"') {
-    queryelement.push_back(*here);
-    if (in_phrase == false) in_phrase = true;
-    else if (in_phrase == true) {
-      add_field_info(queryelement, tag, argct);
-      processed_querystring += combine + queryelement;
-      queryelement.clear();
-      in_phrase = false;
-    }
-      }
-
-      // Found word boundary
-      else if (in_phrase) {
-    queryelement.push_back(*here);
-      }
-      else {
-    if (!queryelement.empty()) {
-      add_field_info(queryelement, tag, argct);
-      processed_querystring += combine + queryelement;
-      queryelement.clear();
-    }
-    processed_querystring.push_back(*here);
-      }
-
-      ++here;
-    }
-
-    // Get last element
-    if (!queryelement.empty()) {
-      add_field_info(queryelement, tag, argct);
-      processed_querystring += combine + queryelement;
-    }
-
-    querystring = processed_querystring;
-    return;
-  }
-
-  if (argb==0 && argt==0) {
-    // simple 'and' search - just put tag info round whole query string
+  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);
     return;
   }
-  
-  // we need to individually tag words
-  text_t outtext;
-  text_t word;
-  //unsigned short c;                                                            
+  // 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::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;    
-    }
+  while (here != end) {
+    if (is_unicode_letdig(*here) || is_special_character(argct, *here)) {
+      queryelement.push_back(*here);
+    }
+
+    // Detect phrase starts/finishes
+    else if (*here == '"') {
+      queryelement.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();
+    in_phrase = false;
+      }
+    }
+
+    // Found word boundary, in a phrase
+    else if (in_phrase) {
+      queryelement.push_back(*here);
+    }
+    // Word boundary, but not in a phrase
     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;
-}
-
+      if (!queryelement.empty()) {
+    if (mgpp) {add_field_info(queryelement, tag, argct);}
+    processed_querystring += combine + queryelement;
+    queryelement.clear();
+      }
+      processed_querystring.push_back(*here);
+    }
+
+    ++here;
+  }
+
+  // Get last element
+  if (!queryelement.empty()) {
+    if (mgpp) {add_field_info(queryelement, tag, argct);}
+    processed_querystring += combine + queryelement;
+  }
+
+  querystring = processed_querystring;
+ 
+  if (lucene) {
+    // tag the whole query string
+    add_field_info(querystring, tag, argct);
+  }
+}
+
+