Changeset 20602

Timestamp:

2009-09-14T15:17:01+12:00 (14 years ago)

Author:

kjdon

Message:

get_plain_query_terms: first pass through to remove TI:(...) and [...]:TI, and AND,OR,NOT for lucene, then remove term modifiers etc

Location:

gsdl/trunk/runtime-src/src/recpt

Files:

• querytools.cpp (modified) (2 diffs)
• querytools.h (modified) (1 diff)

gsdl/trunk/runtime-src/src/recpt/querytools.cpp

@@ -334 +334 @@
 // turn query string into terms separated by spaces.
 // still working on this...
-text_t get_plain_query_terms(const text_t &querystring) {
-
+text_t get_plain_query_terms(const text_t &querystring, const text_t &arg_ct) {
   text_t::const_iterator here = querystring.begin();
   text_t::const_iterator end = querystring.end();
+
+  // lets look for [] and () first - these are a pain.
+  text_t::const_iterator bracket;
+  text_t query_no_brackets = "";
+
+  // mgpp brackets: [xxx]:TI
+  if (findchar(here, end, '[') != end) {
+    while ((bracket = findchar(here, end, '[')) != end) {
+      // get the first bit
+      query_no_brackets += substr(here, bracket);
+      bracket++;
+      here = bracket;
+      // get the end bracket
+      bracket = findchar(here, end, ']');
+      query_no_brackets += substr(here, bracket);
+      // skip the :TI bits
+      while (*bracket != ' ' && bracket != end) { bracket++;}
+      here = bracket;
+    }
+    if (here != end) {
+      query_no_brackets += substr(here,end);
+    }
+  } else if (findchar(here, end, '(') != end) {
+    // lucene brackets TI:(xxx)
+    while ((bracket = findchar(here, end, '(')) != end) {
+      // back up the field name
+      text_t::const_iterator old_bracket = bracket;
+      while (*bracket != ' ' && bracket != here) {
+    --bracket;
+      }
+      if (bracket != here) {
+    // get the first bit
+    query_no_brackets += substr(here, bracket+1);
+      }
+      here = old_bracket +1;
+      // get the end bracket
+      bracket = findchar(here, end, ')');
+      query_no_brackets += substr(here, bracket);
+      if (bracket != end) {
+    here = bracket+1;
+      }
+    }
+    if (here != end) {
+      query_no_brackets += substr(here,end);
+    }
+  } else {
+    // was no brackets
+    query_no_brackets = querystring;
+  }
+  
+  
+  if (arg_ct == "2") { // lucene
+    // look for AND OR NOT and remove
+    here = query_no_brackets.begin();
+    end = query_no_brackets.end();
+    text_tlist terms;
+    splitword(here, end, "AND", terms);
+    joinchar(terms, ' ', query_no_brackets);
+    here = query_no_brackets.begin();
+    end = query_no_brackets.end();
+    splitword(here, end, "OR", terms);
+    joinchar(terms, ' ', query_no_brackets);
+    here = query_no_brackets.begin();
+    end = query_no_brackets.end();
+    splitword(here, end, "NOT", terms);
+    joinchar(terms, ' ', query_no_brackets);
+    
+  }
   text_t terms = "";
   bool space = false;
+  here = query_no_brackets.begin();
+  end = query_no_brackets.end();
+  
   while (here != end) {
     if (*here ==  '#' || *here == '/') {
@@ -348 +418 @@
       }
       if (here == end) break;
-    }
-    if (*here == '[') {
-      // get the text out and ignore the :TI after
     }
     if (is_unicode_letdig(*here)) {

gsdl/trunk/runtime-src/src/recpt/querytools.h

@@ -49 +49 @@
 
 // get a plain version of the query string - terms, separated by space
-text_t get_plain_query_terms(const text_t &querystring);
+text_t get_plain_query_terms(const text_t &querystring, const text_t &arg_ct);
 
 // search history tool