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

Diff of /src/rcldb/rcldb.h [ee7d0f] .. [3d59c6]

Switch to unified view


...
     * Side-effect: set the existence flag for the file document 
     * and all subdocs if any (for later use by 'purge()') 
     */
    bool needUpdate(const string &udi, const string& sig);

    /** Add or update document identified by unique identifier.
     * @param config Config object to use. Can be the same as the member config
     *   or a clone, to avoid sharing when called in multithread context.
     * @param udi the Unique Document Identifier is opaque to us. 
     *   Maximum size 150 bytes.
     * @param parent_udi the UDI for the container document. In case of complex
     *  embedding, this is not always the immediate parent but the UDI for
     *  the container file (which may be a farther ancestor). It is
     *  used for purging subdocuments when a file ceases to exist and
     *  to set the existence flags of all subdocuments of a container
     *  that is found to be up to date. In other words, the
     *  parent_udi is the UDI for the ancestor of the document which
     *  is subject to needUpdate() and physical existence tests (some
     *  kind of file equivalent). Empty for top-level docs. Should
     *  probably be renamed container_udi.
     * @param doc container for document data. Should have been filled as 
     *   much as possible depending on the document type. 
     */
    bool addOrUpdate(RclConfig *config, const string &udi, 
             const string &parent_udi, Doc &doc);
#ifdef IDX_THREADS
    void waitUpdIdle();
#endif
...
    /** Delete document(s) for given UDI, including subdocs */
    bool purgeFile(const string &udi, bool *existed = 0);

    /** Remove documents that no longer exist in the file system. This
     * depends on the update map, which is built during
     * indexing (needUpdate() / addOrUpdate()). 
     *
     * This should only be called after a full walk of
     * the file system, else the update map will not be complete, and
     * many documents will be deleted that shouldn't, which is why this
     * has to be called externally, rcldb can't know if the indexing
     * pass was complete or partial.
     */
    bool purge();

    /** Create stem expansion database for given languages. */
    bool createStemDbs(const std::vector<std::string> &langs);
    /** Delete stem expansion database for given language. */
    bool deleteStemDb(const string &lang);

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

    /** Return total docs in db */
    int  docCnt(); 
    /** Return count of docs which have an occurrence of term */
    int termDocCnt(const string& term);
    /** Add extra Xapian database for querying. 
     * @param dir must point to something which can be passed as parameter 
     *      to a Xapian::Database constructor (directory or stub).
     */
    bool addQueryDb(const string &dir);
    /** Remove extra database. if dir == "", remove all. */
    bool rmQueryDb(const string &dir);
    /** Look where the doc result comes from. 
     * @param doc must come from a db query so that "opaque" xdocid is set.
     * @return: 0 main index, (size_t)-1 don't know, 
     *   other: order of database in add_database() sequence.
     */
    size_t whatDbIdx(const Doc& doc);
    /** Tell if directory seems to hold xapian db */
...
     *        in the TermMatchResult header
     */
    enum MatchType {ET_WILD, ET_REGEXP, ET_STEM};
    bool termMatch(MatchType typ, const string &lang, const string &term, 
           TermMatchResult& result, int max = -1, 
           const string& field = cstr_null);

    /** Return min and max years for doc mod times in db */
    bool maxYearSpan(int *minyear, int *maxyear);

    /** Wildcard expansion specific to file names. Internal/sdata use only */
    bool filenameWildExp(const string& exp, vector<string>& names, int max);
...
    {
    return m_synthAbsLen;
    }
    /** Get document for given udi
     *
     * Used by the 'history' feature, and to retrieve ancestor documents.
     */
    bool getDoc(const string &udi, Doc &doc);

    /* The following are mainly for the aspell module */
    /** Whole term list walking. */

	a/src/rcldb/rcldb.h		b/src/rcldb/rcldb.h
	...		...
245	* Side-effect: set the existence flag for the file document	245	* Side-effect: set the existence flag for the file document
246	* and all subdocs if any (for later use by 'purge()')	246	* and all subdocs if any (for later use by 'purge()')
247	*/	247	*/
248	bool needUpdate(const string &udi, const string& sig);	248	bool needUpdate(const string &udi, const string& sig);
249		249
250	/** Add or update document. The Doc class should have been filled as much as	250	/** Add or update document identified by unique identifier.
		251	* @param config Config object to use. Can be the same as the member config
		252	* or a clone, to avoid sharing when called in multithread context.
		253	* @param udi the Unique Document Identifier is opaque to us.
		254	* Maximum size 150 bytes.
		255	* @param parent_udi the UDI for the container document. In case of complex
		256	* embedding, this is not always the immediate parent but the UDI for
		257	* the container file (which may be a farther ancestor). It is
		258	* used for purging subdocuments when a file ceases to exist and
		259	* to set the existence flags of all subdocuments of a container
		260	* that is found to be up to date. In other words, the
		261	* parent_udi is the UDI for the ancestor of the document which
		262	* is subject to needUpdate() and physical existence tests (some
		263	* kind of file equivalent). Empty for top-level docs. Should
		264	* probably be renamed container_udi.
		265	* @param doc container for document data. Should have been filled as
251	* possible depending on the document type. parent_udi is only	266	* much as possible depending on the document type.
252	* use for subdocs, else set it to empty */	267	*/
253	bool addOrUpdate(RclConfig *config, const string &udi,	268	bool addOrUpdate(RclConfig *config, const string &udi,
254	const string &parent_udi, Doc &doc);	269	const string &parent_udi, Doc &doc);
255	#ifdef IDX_THREADS	270	#ifdef IDX_THREADS
256	void waitUpdIdle();	271	void waitUpdIdle();
257	#endif	272	#endif
	...		...
259	/** Delete document(s) for given UDI, including subdocs */	274	/** Delete document(s) for given UDI, including subdocs */
260	bool purgeFile(const string &udi, bool *existed = 0);	275	bool purgeFile(const string &udi, bool *existed = 0);
261		276
262	/** Remove documents that no longer exist in the file system. This	277	/** Remove documents that no longer exist in the file system. This
263	* depends on the update map, which is built during	278	* depends on the update map, which is built during
264	* indexing (needUpdate()).	279	* indexing (needUpdate() / addOrUpdate()).
265	*	280	*
266	* This should only be called after a full walk of	281	* This should only be called after a full walk of
267	* the file system, else the update map will not be complete, and	282	* the file system, else the update map will not be complete, and
268	* many documents will be deleted that shouldn't, which is why this	283	* many documents will be deleted that shouldn't, which is why this
269	* has to be called externally, rcldb can't know if the indexing	284	* has to be called externally, rcldb can't know if the indexing
270	* pass was complete or partial.	285	* pass was complete or partial.
271	*/	286	*/
272	bool purge();	287	bool purge();
273		288
274	/** Create stem expansion database for given language. */	289	/** Create stem expansion database for given languages. */
275	bool createStemDbs(const std::vector<std::string> &langs);	290	bool createStemDbs(const std::vector<std::string> &langs);
276	/** Delete stem expansion database for given language. */	291	/** Delete stem expansion database for given language. */
277	bool deleteStemDb(const string &lang);	292	bool deleteStemDb(const string &lang);
278		293
279	/* Query-related methods ************************************/	294	/* Query-related methods ************************************/
280		295
281	/** Return total docs in db */	296	/** Return total docs in db */
282	int docCnt();	297	int docCnt();
283	/** Return count of docs which have an occurrence of term */	298	/** Return count of docs which have an occurrence of term */
284	int termDocCnt(const string& term);	299	int termDocCnt(const string& term);
285	/** Add extra database for querying */	300	/** Add extra Xapian database for querying.
		301	* @param dir must point to something which can be passed as parameter
		302	* to a Xapian::Database constructor (directory or stub).
		303	*/
286	bool addQueryDb(const string &dir);	304	bool addQueryDb(const string &dir);
287	/** Remove extra database. if dir == "", remove all. */	305	/** Remove extra database. if dir == "", remove all. */
288	bool rmQueryDb(const string &dir);	306	bool rmQueryDb(const string &dir);
289	/** Look where the doc result comes from.	307	/** Look where the doc result comes from.
		308	* @param doc must come from a db query so that "opaque" xdocid is set.
290	* @return: 0 main index, (size_t)-1 don't know,	309	* @return: 0 main index, (size_t)-1 don't know,
291	* other: order of database in add_database() sequence.	310	* other: order of database in add_database() sequence.
292	*/	311	*/
293	size_t whatDbIdx(const Doc& doc);	312	size_t whatDbIdx(const Doc& doc);
294	/** Tell if directory seems to hold xapian db */	313	/** Tell if directory seems to hold xapian db */
	...		...
310	* in the TermMatchResult header	329	* in the TermMatchResult header
311	*/	330	*/
312	enum MatchType {ET_WILD, ET_REGEXP, ET_STEM};	331	enum MatchType {ET_WILD, ET_REGEXP, ET_STEM};
313	bool termMatch(MatchType typ, const string &lang, const string &term,	332	bool termMatch(MatchType typ, const string &lang, const string &term,
314	TermMatchResult& result, int max = -1,	333	TermMatchResult& result, int max = -1,
315	const string& field = cstr_null	334	const string& field = cstr_null);
316	);
317	/** Return min and max years for doc mod times in db */	335	/** Return min and max years for doc mod times in db */
318	bool maxYearSpan(int minyear, int maxyear);	336	bool maxYearSpan(int minyear, int maxyear);
319		337
320	/** Wildcard expansion specific to file names. Internal/sdata use only */	338	/** Wildcard expansion specific to file names. Internal/sdata use only */
321	bool filenameWildExp(const string& exp, vector<string>& names, int max);	339	bool filenameWildExp(const string& exp, vector<string>& names, int max);
	...		...
330	{	348	{
331	return m_synthAbsLen;	349	return m_synthAbsLen;
332	}	350	}
333	/** Get document for given udi	351	/** Get document for given udi
334	*	352	*
335	* Used by the 'history' feature (and nothing else?)	353	* Used by the 'history' feature, and to retrieve ancestor documents.
336	*/	354	*/
337	bool getDoc(const string &udi, Doc &doc);	355	bool getDoc(const string &udi, Doc &doc);
338		356
339	/* The following are mainly for the aspell module */	357	/* The following are mainly for the aspell module */
340	/** Whole term list walking. */	358	/** Whole term list walking. */