Open Source Projects Europe

 *   Free Software Foundation, Inc.,

 *   Free Software Foundation, Inc.,

 *   59 Temple Place - Suite 330, Boston, MA  02111-1307, USA.

 *   59 Temple Place - Suite 330, Boston, MA  02111-1307, USA.

 */

 */

#ifndef _DB_H_INCLUDED_

#ifndef _DB_H_INCLUDED_

#define _DB_H_INCLUDED_

#define _DB_H_INCLUDED_

#include <string>

#include <string>

#include <list>

#include <list>

#include <vector>

#include <vector>

#ifndef NO_NAMESPACES

#ifndef NO_NAMESPACES

namespace Rcl {

namespace Rcl {

#endif

#endif

class SearchData;

class SearchData;

class TermIter;

class TermIter;

class TermMatchEntry {

class TermMatchEntry {

public:

public:

    TermMatchEntry() : wcf(0) {}

    TermMatchEntry() : wcf(0) {}

    TermMatchEntry(const string&t, int f) : term(t), wcf(f) {}

    TermMatchEntry(const string&t, int f) : term(t), wcf(f) {}

/**

/**

 * Wrapper class for the native database.

 * Wrapper class for the native database.

 */

 */

class Db {

class Db {

 public:

 public:

    /* General stuff (valid for query or update) ****************************/

    /* General stuff (valid for query or update) ****************************/

    Db();

    Db();

    ~Db();

    ~Db();

    enum OpenMode {DbRO, DbUpd, DbTrunc};

    enum OpenMode {DbRO, DbUpd, DbTrunc};

    bool open(const string &dbdir, const string &stoplistfn, 

    bool open(const string &dbdir, const string &stoplistfn, 

    bool close();

    bool close();

    bool isopen();

    bool isopen();

    /** Retrieve main database directory */

    /** Retrieve main database directory */

    string getDbDir();

    string getDbDir();

    /* Query-related methods ************************************/

    /* Query-related methods ************************************/

    /** Return total docs in db */

    /** Return total docs in db */

    int  docCnt(); 

    int  docCnt(); 

    /** Return a list of index terms that match the input string

    /** Return a list of index terms that match the input string

     * Expansion is performed either with either wildcard or regexp processing

     * Expansion is performed either with either wildcard or regexp processing

     * Stem expansion is performed if lang is not empty */

     * Stem expansion is performed if lang is not empty */

    enum MatchType {ET_WILD, ET_REGEXP, ET_STEM};

    enum MatchType {ET_WILD, ET_REGEXP, ET_STEM};

    bool termMatch(MatchType typ, const string &lang, const string &s, 

    bool termMatch(MatchType typ, const string &lang, const string &s, 

           list<TermMatchEntry>& result, int max = -1);

           list<TermMatchEntry>& result, int max = -1);

    /* Build synthetic abstract out of query terms and term position data */

    /* Build synthetic abstract out of query terms and term position data */

    bool makeDocAbstract(Doc &doc, string& abstract);

    /** Get document for given filename and ipath */

    /** Get document for given filename and ipath */

    bool getDoc(const string &fn, const string &ipath, Doc &doc, int *percent);

    bool getDoc(const string &fn, const string &ipath, Doc &doc, int *percent);

    /** Get a list of existing stemming databases */

    /** Get a list of existing stemming databases */

    std::list<std::string> getStemLangs();

    std::list<std::string> getStemLangs();

    /** Set parameters for synthetic abstract generation */

    /** Set parameters for synthetic abstract generation */

             const string& base);

             const string& base);

    /** Filename wildcard expansion */

    /** Filename wildcard expansion */

    bool filenameWildExp(const string& exp, list<string>& names);

    bool filenameWildExp(const string& exp, list<string>& names);

private:

private:

    // Internal form of close, can be called during destruction

    // Internal form of close, can be called during destruction

    bool i_close(bool final);

    bool i_close(bool final);

    string m_reason; // Error explanation

    string m_reason; // Error explanation

    // This is how long an abstract we keep or build from beginning of

    // This is how long an abstract we keep or build from beginning of

    // text when indexing. It only has an influence on the size of the

    // text when indexing. It only has an influence on the size of the

    // db as we are free to shorten it again when displaying

    // db as we are free to shorten it again when displaying

    int          m_idxAbsTruncLen;

    int          m_idxAbsTruncLen;

    // This is the size of the abstract that we synthetize out of query

    // This is the size of the abstract that we synthetize out of query

    // term contexts at *query time*

    // term contexts at *query time*

    int          m_synthAbsLen;

    int          m_synthAbsLen;

    // This is how many words (context size) we keep around query terms

    // This is how many words (context size) we keep around query terms

    // when building the abstract

    // when building the abstract

    int          m_synthAbsWordCtxLen;

    int          m_synthAbsWordCtxLen;

    // Flush threshold. Megabytes of text indexed before we flush.

    // Flush threshold. Megabytes of text indexed before we flush.

    int          m_flushMb;

    int          m_flushMb;

    // Text bytes indexed since beginning

    // Text bytes indexed since beginning

    long long    m_curtxtsz;

    long long    m_curtxtsz;

    // Text bytes at last flush

    // Text bytes at last flush

    long long    m_flushtxtsz;

    long long    m_flushtxtsz;

    // Text bytes at last fsoccup check

    // Text bytes at last fsoccup check

    long long    m_occtxtsz;

    long long    m_occtxtsz;

    // Maximum file system occupation percentage

    // Maximum file system occupation percentage

    int          m_maxFsOccupPc;

    int          m_maxFsOccupPc;

    // Database directory

    // Database directory

    string       m_basedir;

    string       m_basedir;