Changeset 91 for trunk/gsdl/src/library/mgsearch.cpp

Timestamp:

1998-12-30T16:50:37+13:00 (25 years ago)

Author:

rjmcnab

Message:

Changed the directory structure (collect.cfg and site.cfg now reside
in the collection/etc directory). Changed all input to the library
software to be converted from utf-8 to unicode (info database, mg,
and display). Got lib.init to read in collect.cfg and build.cfg and
used the information to read in the macrofiles. Made it check for each
macro file in both the collection directory and then the main directory.

File:

• trunk/gsdl/src/library/mgsearch.cpp (modified) (10 diffs)

trunk/gsdl/src/library/mgsearch.cpp

@@ -27 +27 @@
 #include "mgsearch.h"
 #include "locateinfo.h"
+#include "gsdlunicode.h"
+#include "unitool.h"
+
 
 /////////////
@@ -82 +85 @@
           float Weight, void *info) {
 
+  // convert term from utf8 to unicode
+  text_t term;
+  utf8inconvertclass inconvert;
+  convertclass::status_t status;
+  inconvert.reset ();
+  inconvert.setinput (Word, ULen);
+  inconvert.convert (term, status);
+
   queryresultsclass *queryresults = (queryresultsclass *)info;
-  queryresults->termvariants.push_back(Word);
+  queryresults->termvariants.push_back(term);
 
   return 0;
@@ -93 +104 @@
   text_t *output = (text_t *)info;
   if (output == NULL) return 0;
-
-  output->setcarr(Word, ULen);
+  output->clear();
+
+  utf8inconvertclass inconvert;
+  convertclass::status_t status;
+  inconvert.reset ();
+  inconvert.setinput (Word, ULen);
+  inconvert.convert (*output, status);
   
   // replace all control-Cs with spaces
   text_t::iterator here = output->begin();
   text_t::iterator end = output->end();
-  while (here != end)
-    {
-      if (*here == '\x3') *here = ' ';
-      here++;
-    }
+  while (here != end) {
+    if (*here == '\x3') *here = ' ';
+    here++;
+  }
   
   return 0;
@@ -129 +144 @@
 
 
-void mgsearchclass::setindexhome (const text_t &theindexhome)
-{
-  indexhome = theindexhome;
+void mgsearchclass::setcollectdir (const text_t &thecollectdir)
+{
+  collectdir = thecollectdir;
 }
 
 
 bool mgsearchclass::search(const queryparamclass &queryparams, 
-              queryresultsclass &queryresults)
+               queryresultsclass &queryresults)
 {
   bool databaseloaded = true;
@@ -146 +161 @@
   // first check the cache
   if (cache->find(queryparams, queryresults))
-    {
-      return true;
-    }
+    return true;
 
   // make sure there is a query to be processed
@@ -154 +167 @@
   text_t::const_iterator queryend = queryparams.querystring.end();
   while (queryhere != queryend) {
-    if (((*queryhere >= 65) && (*queryhere <= 90)) ||
-      ((*queryhere >= 97) && (*queryhere <= 122)) ||
-      ((*queryhere >= 192) && (*queryhere <= 214)) ||
-      ((*queryhere >= 216) && (*queryhere <= 246)) ||
-      ((*queryhere >= 248) && (*queryhere <= 255)) ||
-      ((*queryhere >= '0') && (*queryhere <= '9'))) break;
+    if (is_unicode_letdig (*queryhere)) break;
     queryhere++;
   }
@@ -169 +177 @@
 
 
-
   // get the names of the index and text suffixes
-  text_t ttidxsuffix, tttxtsuffix;
-  getindexsuffix (queryparams.search_index, 
-          queryparams.collection, ttidxsuffix);
-  gettextsuffix (queryparams.collection, tttxtsuffix);
-  char *idxsuffix = ttidxsuffix.getcstr(); assert (idxsuffix != NULL);
-  char *txtsuffix = tttxtsuffix.getcstr(); assert (txtsuffix != NULL);
+  char *idxsuffix = (getindexsuffix (queryparams.collection, 
+                     queryparams.search_index)).getcstr();
+  assert (idxsuffix != NULL);
+  char *txtsuffix = (gettextsuffix (queryparams.collection)).getcstr();
+  assert (txtsuffix != NULL);
 
 #ifdef __WIN32__
-  char *cindexhome = (indexhome+"\\").getcstr(); assert (cindexhome != NULL);
+  char *ccollectdir = (collectdir+"\\").getcstr(); assert (ccollectdir != NULL);
 #else
-  char *cindexhome = indexhome.getcstr(); assert (cindexhome != NULL);
+  char *ccollectdir = collectdir.getcstr(); assert (ccollectdir != NULL);
 #endif
 
-  if (load_database(cindexhome, idxsuffix, txtsuffix))
+  if (load_database(ccollectdir, idxsuffix, txtsuffix))
     {
       setsearchmode (queryparams);
@@ -195 +201 @@
   delete idxsuffix;
   delete txtsuffix;
-  delete cindexhome;
+  delete ccollectdir;
 
   return databaseloaded;
@@ -348 +354 @@
 // defaultindex is needed to load one
 bool mgsearchclass::docTargetDocument(const text_t &defaultindex, 
-                 const text_t &collection,
-                 int docnum,
-                 text_t &output)
+                      const text_t &collection,
+                      int docnum,
+                      text_t &output)
 {
   bool databaseloaded = true;
@@ -423 +429 @@
     {
       // get the names of the index and text suffixes
-      text_t ttidxsuffix, tttxtsuffix;
-      getindexsuffix (defaultindex, collection, ttidxsuffix);
-      gettextsuffix (collection, tttxtsuffix);
-      char *idxsuffix = ttidxsuffix.getcstr(); assert (idxsuffix != NULL);
-      char *txtsuffix = tttxtsuffix.getcstr(); assert (txtsuffix != NULL);
+      char *idxsuffix = (getindexsuffix (collection, 
+                     defaultindex)).getcstr();
+      assert (idxsuffix != NULL);
+      char *txtsuffix = (gettextsuffix (collection)).getcstr();
+      assert (txtsuffix != NULL);
 
 #ifdef __WIN32__
-      char *cindexhome = (indexhome+"\\").getcstr(); assert (cindexhome != NULL);
+      char *ccollectdir = (collectdir+"\\").getcstr(); assert (ccollectdir != NULL);
 #else
-      char *cindexhome = indexhome.getcstr(); assert (cindexhome != NULL);
+      char *ccollectdir = collectdir.getcstr(); assert (ccollectdir != NULL);
 #endif
       
-
-      if (load_database(cindexhome, idxsuffix, txtsuffix))
-    {
-      databaseloaded = true;
-    }
+      if (load_database(ccollectdir, idxsuffix, txtsuffix))
+    databaseloaded = true;
       else 
-    {
-      databaseloaded = false;
-    }
-      
+    databaseloaded = false;
+      
       // free up the c strings
       delete idxsuffix;
       delete txtsuffix;
-      delete cindexhome;
+      delete ccollectdir;
     }