recoll / Code / Diff of /src/rcldb/rcldb.cpp

Diff of /src/rcldb/rcldb.cpp [0dde29] .. [521013]

Switch to unified view

-a/src/rcldb/rcldb.cpp
+b/src/rcldb/rcldb.cpp
 #ifndef lint
-static char rcsid[] = "@(#$Id: rcldb.cpp,v 1.5 2004-12-17 15:50:48 dockes Exp $ (C) 2004 J.F.Dockes";
+static char rcsid[] = "@(#$Id: rcldb.cpp,v 1.6 2005-01-24 13:17:58 dockes Exp $ (C) 2004 J.F.Dockes";
 #endif
 #include <sys/stat.h>
 #include <iostream>
 ...
 #include "rcldb.h"
 #include "textsplit.h"
 #include "transcode.h"
 #include "unacpp.h"
+#include "conftree.h"
 #include "xapian.h"
-// Data for a xapian database
+// Data for a xapian database. There could actually be 2 different ones for
+// indexing or query as there is not much in common.
 class Native {
  public:
     bool isopen;
     bool iswritable;
-    class Xapian::Database db;
+    // Indexing
-    class Xapian::WritableDatabase wdb;
+    Xapian::WritableDatabase wdb;
     vector<bool> updated;
+    // Querying
+    Xapian::Database db;
+    Xapian::Query query;
     Native() : isopen(false), iswritable(false) {}
 };
 Rcl::Db::Db()
 ...
     pdata = new Native;
+}
 Rcl::Db::~Db()
+{
+    cerr << "Rcl::Db::~Db" << endl;
     if (pdata == 0)
     return;
     Native *ndb = (Native *)pdata;
     cerr << "Db::~Db: isopen " << ndb->isopen << " iswritable " <<
     ndb->iswritable << endl;
     try {
     // There is nothing to do for an ro db.
     if (ndb->isopen == false || ndb->iswritable == false) {
+      cerr << "Deleting native database" << endl;
         delete ndb;
         return;
+    }
     ndb->wdb.flush();
     delete ndb;
 ...
         ndb->iswritable = true;
         break;
     case DbRO:
     default:
         ndb->iswritable = false;
-      cerr << "Not ready to open RO yet" << endl;
+      ndb->db = Xapian::Auto::open(dir, Xapian::DB_OPEN);
-      exit(1);
+      break;
+    }
     ndb->isopen = true;
     return true;
     } catch (const Xapian::Error &e) {
     cout << "Exception: " << e.get_msg() << endl;
 ...
     Xapian::termpos curpos;  // Last position sent to callback
     wsData(Xapian::Document &d) : doc(d), basepos(1), curpos(0)
     {}
 };
+// Callback for the document to word splitting class during indexation
-bool splitCb(void *cdata, const std::string &term, int pos)
+static bool splitCb(void *cdata, const std::string &term, int pos)
+{
     wsData *data = (wsData*)cdata;
     // cerr << "splitCb: term " << term << endl;
     //string printable;
 ...
+{
     string inter;
     out.erase();
     if (!unac_cpp(in, inter))
     return false;
-    out.resize(inter.length());
+    out.reserve(inter.length());
     for (unsigned int i = 0; i < inter.length(); i++) {
     if (inter[i] >= 'A' && inter[i] <= 'Z')
         out += inter[i] + 'a' - 'A';
     else
         out += inter[i];
 ...
     if (1 /*dupes == DUPE_replace*/) {
     // If this document has already been indexed, update the existing
     // entry.
     try {
-      Xapian::docid did = ndb->wdb.replace_document(pathterm,
+#if 0
-                            newdocument);
+      Xapian::docid did =
+#endif
+      ndb->wdb.replace_document(pathterm, newdocument);
 #if 0
         if (did < updated.size()) {
         updated[did] = true;
         //cout << "updated." << endl;
         } else {
 ...
     return true;
+    }
     return true;
+}
+#include <vector>
+class wsQData {
+ public:
+    vector<string> terms;
+};
+// Callback for the document to word splitting class during indexation
+static bool splitQCb(void *cdata, const std::string &term, int )
+{
+    wsQData *data = (wsQData*)cdata;
+    cerr << "splitQCb: term '" << term << "'" << endl;
+    cerr << "splitQCb: term length: " << term.length() <<  endl;
+    //string printable;
+    //transcode(term, printable, "UTF-8", "ISO8859-1");
+    //cerr << "Adding " << printable << endl;
+    data->terms.push_back(term);
+    return true;
+}
+bool Rcl::Db::setQuery(const std::string &querystring)
+{
+    wsQData splitData;
+    TextSplit splitter(splitQCb, &splitData);
+    string noacc;
+    if (!dumb_string(querystring, noacc)) {
+  return false;
+    }
+    //    noacc = querystring;
+    splitter.text_to_words(noacc);
+    Native *ndb = (Native *)pdata;
+    //        splitData.terms.resize(0);
+    //        splitData.terms.push_back(string("le"));
+    ndb->query = Xapian::Query(Xapian::Query::OP_OR, splitData.terms.begin(),
+                 splitData.terms.end());
+    return true;
+}
+bool Rcl::Db::getDoc(int i, Doc &doc)
+{
+    // cerr << "Rcl::Db::getDoc: " << i << endl;
+    Native *ndb = (Native *)pdata;
+    Xapian::Enquire enquire(ndb->db);
+    enquire.set_query(ndb->query);
+    Xapian::MSet matches = enquire.get_mset(i, 1);
+    // cerr << "Query `" << ndb->query.get_description() << "'" <<
+    // "Estimated results: " << matches.get_matches_lower_bound() << endl;
+    if (matches.empty())
+  return false;
+    Xapian::Document xdoc = matches.begin().get_document();
+    // Parse xapian document's data and populate doc fields
+    string data = xdoc.get_data();
+    ConfSimple parms(&data);
+    parms.get(string("mtype"), doc.mimetype);
+    parms.get(string("mtime"), doc.mtime);
+    parms.get(string("url"), doc.url);
+    return true;
+}