Changeset 2134

Timestamp:

2001-03-08T14:11:48+13:00 (23 years ago)

Author:

sjboddie

Message:

mg phrase searching now always sets maxdocs to -1 (all) - this means that
a phrase search is guaranteed always to hit any document that contains the
phrase but also means that bad phrase searches (like "and the") will take
a very long time, especially on a large collection.

also added a bit of a hack to handle program number indexes for various
bbc collections.

File:

• trunk/gsdl/src/colservr/mgqueryfilter.cpp (modified) (5 diffs)

trunk/gsdl/src/colservr/mgqueryfilter.cpp

@@ -446 +446 @@
   while (query_here != query_end) {
 
-    // if we're doing a phrase search we want to maximise hits by making it a boolean
-    // search on the index with the finest granularity
+    // if we're doing a phrase search we want to maximise hits by making it
+    // a boolean search on the index with the finest granularity - we'll
+    // also set maxdocs to "all" (realizing that this will cause searches
+    // like "and the" on a large collection to take a very very long time).
+
     // we're deciding it's a phrase search based on if the querystring 
     // contains at least 2 double quotes (not very scientific but 
@@ -453 +456 @@
     if (countchar ((*query_here).querystring.begin(), (*query_here).querystring.end(), '"') > 1) {
       (*query_here).search_type = 0;
+
+      // set maxdocs to "all"
+      (*query_here).maxdocs = -1;
 
       // Get the long version of the index and test to see if any indexes with
@@ -467 +473 @@
     text_t shortindex;
     if (indexmap.fromexists ("paragraph:" + indextype)) {
-      //      logout << text_t2ascii << "changing index from " << longindex << " to " << ("paragraph:" + indextype) << "\n";
       indexmap.from2to ("paragraph:" + indextype, shortindex);
       (*query_here).index = shortindex;
@@ -473 +478 @@
     }
     if (!found && granularity == "document" && indexmap.fromexists ("section:" + indextype)) {
-      //      logout << text_t2ascii << "changing index from " << longindex << " to " << ("section:" + indextype) << "\n";
       indexmap.from2to ("section:" + indextype, shortindex);
       (*query_here).index = shortindex;
@@ -480 +484 @@
     }
 
+#ifdef GSDL_BBC_COLLECTION
+    // This is a special hack for the BBC collection's ProgNumber index
+
+    // if we're searching a ProgNumber index we want to:
+    // 1. Remove all non-alphanumeric characters from the query string
+    // 2. Make it a boolean search
+    // 3. Turn off case-folding
+    text_t longindex; text_tarray splitindex;
+    indexmap.to2from ((*query_here).index, longindex);
+    splitchar (longindex.begin(), longindex.end(), ':', splitindex);
+    text_t &indextype = splitindex[1];
+    if (indextype == "ProgNumber") {
+      (*query_here).search_type = 0;
+      (*query_here).casefolding = 0;
+      text_t new_querystring;
+      text_t::const_iterator here = (*query_here).querystring.begin();
+      text_t::const_iterator end = (*query_here).querystring.end();
+      while (here != end) {
+        if ((*here >= 'a' && *here <= 'z') || (*here >= 'A' && *here <= 'Z') ||
+            (*here >= '0' && *here <= '9')) {
+          new_querystring.push_back (*here);
+        }
+        here ++;
+      }
+      (*query_here).querystring = new_querystring;
+    }
+#endif
+
     query_here ++;
   }