Changeset 22046 for main/trunk/greenstone2/runtime-src/src/recpt/querytools.cpp

Timestamp:

2010-05-06T12:55:03+12:00 (14 years ago)

Author:

davidb

Message:

Changes necessary to support new sql-query action

File:

• main/trunk/greenstone2/runtime-src/src/recpt/querytools.cpp (modified) (9 diffs)

main/trunk/greenstone2/runtime-src/src/recpt/querytools.cpp

@@ -82 +82 @@
     args.setintarg("qto", type);
   } 
-  
+
   if (args["qt"].empty()) {
     int arg_qto = args.getintarg("qto");
@@ -91 +91 @@
     }
   }
+
+
+  // decide if sqlqto should be set or not
+  unsigned int sql_type = 0;
+  text_t infodb_type = cinfo->infodbType;
+  if ((infodb_type == "sqlite") || (infodb_type == "mssql")) {
+    if (findword(search_types.begin(), search_types.end(), "sqlform") != search_types.end()) {
+      sql_type = 1;
+    }
+  }
+
+  if (sql_type) {
+    args["sqlqto"] = "1";
+  }
+  else {
+    args["sqlqto"] = "0";
+  }
+
+
 }
 
@@ -109 +128 @@
 }
 
+
+
+void set_basequeryfilter_options (FilterRequest_t &request, 
+                  cgiargsclass &args) 
+{
+
+  OptionValue_t option;
+  int arg_m = args.getintarg("m");
+  
+  option.name = "Maxdocs";
+  option.value = arg_m;
+  request.filterOptions.push_back (option);
+
+  //  option.name = "StartResults";
+  //  option.value = args["r"];
+  //  request.filterOptions.push_back (option);
+
+  //  option.name = "EndResults";
+  //  int endresults = args.getintarg("o") + (args.getintarg("r") - 1);
+  //  if ((endresults > arg_m) && (arg_m != -1)) endresults = arg_m;
+  //  option.value = endresults;
+  //  request.filterOptions.push_back (option);
+}
+
+
 // request.filterResultOptions and request.fields (if required) should
 // be set from the calling code
-void set_queryfilter_options (FilterRequest_t &request, 
-                  const text_t &querystring,
-                  cgiargsclass &args) {
+void set_fulltext_queryfilter_options (FilterRequest_t &request, 
+                       const text_t &querystring,
+                       cgiargsclass &args) 
+{
+  // better if this function, and the two-query companion function
+  // was implemented in queryaction.cpp
+  // Has to be done here to documentaction.cpp can call it directly
 
   request.filterName = "QueryFilter";
@@ -190 +238 @@
   }
 
-  set_more_queryfilter_options (request, args);
-}
-
-void set_queryfilter_options (FilterRequest_t &request, 
-                  const text_t &querystring1,
-                  const text_t &querystring2, cgiargsclass &args) {
-
-  set_queryfilter_options (request, querystring1, args);
+  set_basequeryfilter_options(request, args);
+}
+
+
+
+void set_fulltext_queryfilter_options (FilterRequest_t &request, 
+                       const text_t &querystring1,
+                       const text_t &querystring2, 
+                       cgiargsclass &args) 
+{
+
+  set_fulltext_queryfilter_options (request, querystring1, args);
 
   // fill in the second query if needed
@@ -245 +297 @@
     }
   }
-  set_more_queryfilter_options (request, args);
-}
-
-void set_more_queryfilter_options (FilterRequest_t &request, 
-                   cgiargsclass &args) {
-
-  OptionValue_t option;
-  int arg_m = args.getintarg("m");
-  
-  option.name = "Maxdocs";
-  option.value = arg_m;
-  request.filterOptions.push_back (option);
-
-  //  option.name = "StartResults";
-  //  option.value = args["r"];
-  //  request.filterOptions.push_back (option);
-
-  //  option.name = "EndResults";
-  //  int endresults = args.getintarg("o") + (args.getintarg("r") - 1);
-  //  if ((endresults > arg_m) && (arg_m != -1)) endresults = arg_m;
-  //  option.value = endresults;
-  //  request.filterOptions.push_back (option);
-}
+
+  // this is probably redundant, as first line to this method will have
+  // already caused it to invoke set_basequeryfilter_options
+
+  set_basequeryfilter_options(request, args);
+}
+
+
+
+// request.filterResultOptions and request.fields (if required) should
+// be set from the calling code
+void set_sql_queryfilter_options (FilterRequest_t &request, 
+                  cgiargsclass &args) 
+{
+  if (!args["sqlsf"].empty()) { // sort field for lucene
+    OptionValue_t option;
+
+    option.name = "SortField";
+    option.value = args["sqlsf"];
+    request.filterOptions.push_back (option);
+  }
+
+  set_basequeryfilter_options(request, args);
+}
+
 
 bool is_special_character(int indexer_type, unsigned short character) {
@@ -551 +605 @@
     if (!values[i].empty()) {
       text_t this_value = values[i];
+
       // remove operators for simple search, segments text if necessary
       format_querystring(this_value, argb, segment); 
+           
       // add tag info for this field (and other processing)
       format_field_info(this_value, fields[i], argct, argt, argb);
+
       // add into query string
       if (argct == 2) {
@@ -644 +701 @@
   }
 }
+
+
+// SQL versions for parsing query form
+
+void parse_sqlreg_query_form(text_t &querystring, cgiargsclass &args, bool segment)
+{
+  querystring.clear();
+
+  int argt = args.getintarg("t");// t=0 -and, t=1 - or
+  int argb = args.getintarg("b");
+    
+  text_t combine;
+
+  if (argt == 0) combine = "AND";
+  else combine = "OR";
+  
+  text_t field = args["sqlfqf"];
+  if (field.empty()) return; // no query
+  text_tarray fields;
+  splitchar(field.begin(), field.end(), ',', fields); 
+
+  text_t sqlcomb = args["sqlfqc"];
+  if (sqlcomb.empty()) return; //somethings wrong
+  text_tarray sqlcombs;
+  splitchar(sqlcomb.begin(), sqlcomb.end(), ',', sqlcombs);
+  
+  text_t value = args["fqv"];
+  if (value.empty()) return; // somethings wrong
+  text_tarray values;
+  splitchar(value.begin(), value.end(), ',', values);
+
+
+  for (int i=0; i< values.size(); ++i) {
+    if (!values[i].empty()) {
+      text_t this_value = values[i];
+
+      // remove operators for simple search, segments text if necessary
+      format_querystring(this_value, argb, segment); 
+           
+      // add tag info for this field (and other processing)
+      format_field_info_sql(this_value, fields[i], sqlcombs[i], argt, argb);
+
+      const text_t DISTINCT_SELECT_WHERE
+    = "SELECT DISTINCT docOID FROM document_metadata WHERE ";
+
+      if (querystring.empty()) {
+    // first query term
+    querystring = DISTINCT_SELECT_WHERE + this_value;
+      }
+      else {
+    this_value = DISTINCT_SELECT_WHERE + this_value;
+
+    if (combine=="AND") {   
+      // INNER JOIN to restrict to only matching docOIDs
+      querystring = "SELECT docOID FROM (" + querystring + ")"
+        + " INNER JOIN (" + this_value +") USING (docOID)";
+    }
+    else if (combine=="OR") {
+      // Union to allow union of the two
+      querystring = querystring + " UNION " + this_value;
+    }
+      }
+    }
+  }
+}
+
+
+void parse_sqladv_query_form(text_t &querystring, cgiargsclass &args, 
+                 bool segment)
+{
+  querystring.clear();
+
+  int argt = 0; // set it to 0 = AND, by default
+  int argb = args.getintarg("b");
+  text_t combine = "AND";
+
+  text_t field = args["sqlfqf"];
+
+  if (field.empty()) return; // no query
+  text_tarray fields;
+  splitchar(field.begin(), field.end(), ',', fields); 
+  
+  text_t sqlcomb = args["sqlfqc"];
+  if (sqlcomb.empty()) return; //somethings wrong
+  text_tarray sqlcombs;
+  splitchar(sqlcomb.begin(), sqlcomb.end(), ',', sqlcombs);
+
+  text_t value = args["fqv"];
+  if (value.empty()) return; // somethings wrong
+  text_tarray values;
+  splitchar(value.begin(), value.end(), ',', values);
+
+  text_t comb = args["fqc"];
+  if (comb.empty()) return; //somethings wrong
+  text_tarray combs;
+  splitchar(comb.begin(), comb.end(), ',', combs);
+
+  for(int i=0; i< values.size(); ++i) {
+    if (!values[i].empty()) {
+      if (i>0) {
+    if (combs[i-1]=="and") { combine = "AND"; }
+    else if (combs[i-1]=="or") { combine = "OR"; }
+    else if (combs[i-1]=="not") { combine = "NOT"; }
+      }
+      text_t this_value = values[i];
+
+      // remove operators for simple search, segments text if necessary
+      format_querystring(this_value, argb, segment); 
+
+      // add tag info for this field (and other processing)
+      format_field_info_sql(this_value, fields[i], sqlcombs[i], argt, argb);
+
+      // add into query string
+
+      const text_t DISTINCT_SELECT_WHERE
+    = "SELECT DISTINCT docOID FROM document_metadata WHERE ";
+
+      if (querystring.empty()) {
+    // first query term
+    querystring = DISTINCT_SELECT_WHERE + this_value;
+      }
+      else {
+    this_value = DISTINCT_SELECT_WHERE + this_value;
+
+    if (combine=="AND") {   
+      // INNER JOIN to restrict to only matching docOIDs
+      querystring = "SELECT docOID FROM (" + querystring + ")"
+        + " INNER JOIN (" + this_value +") USING (docOID)";
+    }
+    else if (combine=="OR") {
+      // Union to allow union of the two
+      querystring = querystring + " UNION " + this_value;
+    }
+    else {
+      cerr << "Unsupported combination operation: " << combine << endl;
+    }
+      }
+      
+    }
+  }  
+}
+
+
+
 
 // Extended addqueryelem for Human Info project
@@ -738 +939 @@
     querystring = tag+":("+querystring+")";
   }
+    
+}
+
+
+void add_field_info_sql(text_t &querystring, const text_t &tagseq, 
+            const text_t& sqlcomb) 
+{
+
+  if (tagseq == "") return; // do nothing
+
+  text_t element_in = "(element IN (";
+
+  text_tlist mdterms;
+
+  splitword(tagseq.begin(), tagseq.end(), "/", mdterms);
+
+  text_t tags_in = "";
+
+  while (!mdterms.empty()) {
+    text_t tag = mdterms.front();
+    mdterms.pop_front();
+
+    if (!tag.empty()) {
+
+      if (tag.size()>3 && (substr(tag.begin(), tag.begin()+3) == "ex.")) {
+    tag = substr (tag.begin()+3, tag.end());
+      }
+
+      if (!tags_in.empty()) {
+    tags_in += ",";
+      }
+      
+      tags_in += "'" + tag + "'";
+    }
+  }
+
+  element_in += tags_in + ") AND (";
+
+  if (sqlcomb == "=") {
+    // override what it means to do equality, to make it more like full text
+    // searching
+
+    text_t orterms = "";
+    text_t term = "";
+    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)) {
+    term.push_back(*here);
+      }
+      else if (*here == '"') {
+    term.push_back(*here);
+    if (!in_phrase) {
+      in_phrase = true;
+    } else {
+      in_phrase = false;
+    }
+      }      
+      else if (in_phrase) {
+        // Found word boundary, but in a phrase, so does not complete term
+    term.push_back(*here);
+      }
+      else {
+        // Found a word boundary
+    if (!orterms.empty()) {
+      orterms += " OR ";
+    }
+    orterms += "value LIKE '%" + term + "%'";
+    term.clear();
+      }
+      ++here;
+    }
+
+    if (!term.empty()) {
+    if (!orterms.empty()) {
+      orterms += " OR ";
+    }
+        orterms += "value LIKE '%" + term + "%'";
+    }
+
+    element_in += orterms;
+  }
+  else {
+    // search on value is "as is" querystring
+    element_in += "value " + sqlcomb + " '" + querystring+"'";
+  }
+
+  
+  querystring = element_in + "))";
     
 }
@@ -901 +1193 @@
     finalquery.push_back(' '); 
   }
-
+ 
   querystring  = finalquery;
+}
+
+
+void format_field_info_sql(text_t &querystring, const text_t &tagseq, 
+               const text_t &sqlcomb, 
+               int argt, int argb) 
+{
+  add_field_info_sql(querystring, tagseq, sqlcomb);
 }