zebraapi.c

harvest是一个下载html网页得机器人

        zebra_set_state (zh, 'o', seqno);
    }
    else
    {
        logf (LOG_LOG, "nothing to commit");
    }
    bfs_destroy (bfs);

    zebra_unlock (zh->lock_shadow);
    zebra_unlock (zh->lock_normal);
    return 0;
}

int zebra_clean (ZebraHandle zh)
{
    yaz_log(LOG_API,"zebra_clean");
    return zebra_commit_ex(zh, 1);
}

int zebra_commit (ZebraHandle zh)
{
    yaz_log(LOG_API,"zebra_commit");
    return zebra_commit_ex(zh, 0);
}

int zebra_init (ZebraHandle zh)
{
    const char *rval;
    BFiles bfs = 0;
    ASSERTZH;
    yaz_log(LOG_API,"zebra_init");
    zh->errCode=0;

    if (!zh->res)
    {
        zh->errCode = 109;
        return -1;
    }
    rval = res_get (zh->res, "shadow");

    bfs = bfs_create (res_get (zh->service->global_res, "register"),
                      zh->path_reg);
    if (rval && *rval)
        bf_cache (bfs, rval);
    
    bf_reset (bfs);
    bfs_destroy (bfs);
    zebra_set_state (zh, 'o', 0);
    return 0;
}

int zebra_compact (ZebraHandle zh)
{
    BFiles bfs;
    ASSERTZH;
    yaz_log(LOG_API,"zebra_compact");
    zh->errCode=0;
    if (!zh->res)
    {
        zh->errCode = 109;
        return -1;
    }
    bfs = bfs_create (res_get (zh->res, "register"), zh->path_reg);
    inv_compact (bfs);
    bfs_destroy (bfs);
    return 0;
}

int zebra_record_insert (ZebraHandle zh, const char *buf, int len, int *sysno)
{
    int sysn=0;
    ASSERTZH;
    yaz_log(LOG_API,"zebra_record_insert");
    if (sysno)
	*sysno=0;
    zh->errCode=0;
    if (zebra_begin_trans (zh, 1))
        return -1;
    extract_rec_in_mem (zh, "grs.sgml",
                        buf, len,
                        "Default",  /* database */
                        0 /* delete_flag */,
                        0 /* test_mode */,
                        &sysn /* sysno */,
                        1 /* store_keys */,
                        1 /* store_data */,
                        0 /* match criteria */);
    if (zebra_end_trans (zh))
        return -1;
    if (sysno)
	*sysno=sysn;
    return 0;
}

int zebra_set_group (ZebraHandle zh, struct recordGroup *rg)
{
    ASSERTZH;
    yaz_log(LOG_API,"zebra_set_group");
    zh->errCode=0;
    memcpy (&zh->rGroup, rg, sizeof(*rg));
    return 0;
}

int zebra_result (ZebraHandle zh, int *code, char **addinfo)
{
    ASSERTZH;
    yaz_log(LOG_API,"zebra_result");
    *code = zh->errCode;
    *addinfo = zh->errString;
    return 0;
}

int zebra_shadow_enable (ZebraHandle zh, int value)
{
    ASSERTZH;
    yaz_log(LOG_API,"zebra_shadow_enable");
    zh->errCode=0;
    zh->shadow_enable = value;
    return 0;
}

int zebra_record_encoding (ZebraHandle zh, const char *encoding)
{
    ASSERTZH;
    yaz_log(LOG_API,"zebra_record_encoding");
    zh->errCode=0;
    xfree (zh->record_encoding);

    /*
     * Fixme!
     * Something about charset aliases. Oleg???
     */

    if (zh->iconv_to_utf8 != 0)
        yaz_iconv_close(zh->iconv_to_utf8);
    if (zh->iconv_from_utf8 != 0)
        yaz_iconv_close(zh->iconv_from_utf8);
    
    zh->record_encoding = xstrdup (encoding);
    
    logf(LOG_DEBUG, "Reset record encoding: %s", encoding);
    
    zh->iconv_to_utf8 =
        yaz_iconv_open ("UTF-8", encoding);
    if (zh->iconv_to_utf8 == 0)
        yaz_log (LOG_WARN, "iconv: %s to UTF-8 unsupported", encoding);
    zh->iconv_from_utf8 =
        yaz_iconv_open (encoding, "UTF-8");
    if (zh->iconv_to_utf8 == 0)
        yaz_log (LOG_WARN, "iconv: UTF-8 to %s unsupported", encoding);

    return 0;
}

int zebra_set_resource(ZebraHandle zh, const char *name, const char *value)
{
    ASSERTZH;
    yaz_log(LOG_API,"zebra_set_resource %s:%s",name,value);
    zh->errCode=0;
    res_put(zh->res, name, value);
    return 0;
}

const char *zebra_get_resource(ZebraHandle zh,
                               const char *name, const char *defaultvalue)
{
    const char *v;
    ASSERTZH;
    v= res_get_def( zh->res, name, (char *)defaultvalue);
    zh->errCode=0;
    yaz_log(LOG_API,"zebra_get_resource %s:%s",name,v);
    return v;
}

/* moved from zebra_api_ext.c by pop */
/* FIXME: Should this really be public??? -Heikki */

int zebra_trans_no (ZebraHandle zh)
{
    ASSERTZH;
    yaz_log(LOG_API,"zebra_trans_no");
    return zh->trans_no;
}

int zebra_get_shadow_enable (ZebraHandle zh)
{
    yaz_log(LOG_API,"zebra_get_shadow_enable");
    return (zh->shadow_enable);
}

int zebra_set_shadow_enable (ZebraHandle zh, int value)
{
    yaz_log(LOG_API,"zebra_set_shadow_enable %d",value);
    zh->shadow_enable = value;
    return 0;
}

int init_recordGroup (struct recordGroup *rg)
{
    assert(rg);
    yaz_log(LOG_API,"init_recordGroup");
    rg->groupName = NULL;
    rg->databaseName = NULL;
    rg->path = NULL;
    rg->recordId = NULL;
    rg->recordType = NULL;
    rg->flagStoreData = -1;
    rg->flagStoreKeys = -1; 
    rg->flagRw = 1;
    rg->databaseNamePath = 0;
    rg->explainDatabase = 0; 
    rg->fileVerboseLimit = 100000; 
    rg->followLinks = -1;
    return 0;
} 


/* This is from extract.c... it seems useful, when extract_rec_in mem is 
   called... and in general... Should be moved to somewhere else */
void res_get_recordGroup (ZebraHandle zh,
			  struct recordGroup *rGroup,
			  const char *ext)
{
    char gprefix[128];
    char ext_res[128]; 
    
    yaz_log(LOG_API,"res_get_recordGroup e=%s",ext);
    if (!rGroup->groupName || !*rGroup->groupName)
	*gprefix = '\0';
    else 
	sprintf (gprefix, "%s.", rGroup->groupName);
    
    /* determine file type - depending on extension */
    if (!rGroup->recordType) {
	sprintf (ext_res, "%srecordType.%s", gprefix, ext);
	if (!(rGroup->recordType = res_get (zh->res, ext_res))) {
	    sprintf (ext_res, "%srecordType", gprefix);
	    rGroup->recordType = res_get (zh->res, ext_res);
	}
    }
    /* determine match criteria */
    if (!rGroup->recordId) { 
	sprintf (ext_res, "%srecordId.%s", gprefix, ext);
	if (!(rGroup->recordId = res_get (zh->res, ext_res))) {
	    sprintf (ext_res, "%srecordId", gprefix);
	    rGroup->recordId = res_get (zh->res, ext_res);
	}
    } 
    
    /* determine database name */
    if (!rGroup->databaseName) {
	sprintf (ext_res, "%sdatabase.%s", gprefix, ext);
	if (!(rGroup->databaseName = res_get (zh->res, ext_res))) { 
	    sprintf (ext_res, "%sdatabase", gprefix);
	    rGroup->databaseName = res_get (zh->res, ext_res);
	}
    }
    if (!rGroup->databaseName)
	rGroup->databaseName = "Default";
    
    /* determine if explain database */
    sprintf (ext_res, "%sexplainDatabase", gprefix);
    rGroup->explainDatabase =
	atoi (res_get_def (zh->res, ext_res, "0"));
    
    /* storeData */
    if (rGroup->flagStoreData == -1)
    {
	const char *sval;
	sprintf (ext_res, "%sstoreData.%s", gprefix, ext);
	if (!(sval = res_get (zh->res, ext_res)))
	{
	    sprintf (ext_res, "%sstoreData", gprefix);
	    sval = res_get (zh->res, ext_res);
	}
	if (sval)
	    rGroup->flagStoreData = atoi (sval);
    }
    if (rGroup->flagStoreData == -1) 
	rGroup->flagStoreData = 0;
    
    /* storeKeys */
    if (rGroup->flagStoreKeys == -1)
    {
	const char *sval;
	
	sprintf (ext_res, "%sstoreKeys.%s", gprefix, ext);
	sval = res_get (zh->res, ext_res);
	if (!sval)
	{
	    sprintf (ext_res, "%sstoreKeys", gprefix);
	    sval = res_get (zh->res, ext_res);
	}
	if (!sval)
	    sval = res_get (zh->res, "storeKeys");
	if (sval)
	    rGroup->flagStoreKeys = atoi (sval);
    }
    if (rGroup->flagStoreKeys == -1)
	rGroup->flagStoreKeys = 0;
} 


/* almost the same as zebra_records_retrieve ... but how did it work? 
   I mean for multiple records ??? CHECK ??? */
void api_records_retrieve (ZebraHandle zh, ODR stream,
			   const char *setname, Z_RecordComposition *comp,
			   oid_value input_format, int num_recs,
			   ZebraRetrievalRecord *recs)
{
    ZebraPosSet poset;
    int i, *pos_array;
    yaz_log(LOG_API,"api_records_retrieve s=%s n=%d",setname,num_recs);

    if (!zh->res)
    {
        zh->errCode = 30;
        zh->errString = odr_strdup (stream, setname);
        return;
    }
    
    zh->errCode = 0; 

    if (zebra_begin_read (zh))
	return;

    pos_array = (int *) xmalloc (num_recs * sizeof(*pos_array));
    for (i = 0; i<num_recs; i++)
	pos_array[i] = recs[i].position;
    poset = zebraPosSetCreate (zh, setname, num_recs, pos_array);
    if (!poset)
    {
        logf (LOG_DEBUG, "zebraPosSetCreate error");
        zh->errCode = 30;
        zh->errString = nmem_strdup (stream->mem, setname);
    }
    else
    {
	for (i = 0; i<num_recs; i++)
	{
	    if (poset[i].term)
	    {
		recs[i].errCode = 0;
		recs[i].format = VAL_SUTRS;
		recs[i].len = strlen(poset[i].term);
		recs[i].buf = poset[i].term;
		recs[i].base = poset[i].db;
		recs[i].sysno = 0;
	    
	    }
	    else if (poset[i].sysno)
	    {
	      /* changed here ??? CHECK ??? */
	      char *b;
		recs[i].errCode =
		    zebra_record_fetch (zh, poset[i].sysno, poset[i].score,
					stream, input_format, comp,
					&recs[i].format, 
					&b,
					&recs[i].len,
					&recs[i].base);
		recs[i].buf = (char *) odr_malloc(stream,recs[i].len);
		memcpy(recs[i].buf, b, recs[i].len);
		recs[i].errString = 0; /* Hmmm !!! we should get this */ 
		recs[i].sysno = poset[i].sysno;
		recs[i].score = poset[i].score;
	    }
	    else
	    {
	        char num_str[20];

		sprintf (num_str, "%d", pos_array[i]);	
		zh->errCode = 13;
                zh->errString = odr_strdup (stream, num_str);
                break;
	    }

	}
	zebraPosSetDestroy (zh, poset, num_recs);
    }
    zebra_end_read (zh);
    xfree (pos_array);
}


/* ---------------------------------------------------------------------------
  Record insert(=update), delete 

  If sysno is provided, then it's used to identify the reocord.
  If not, and match_criteria is provided, then sysno is guessed
  If not, and a record is provided, then sysno is got from there
NOTE: Now returns 0 at success and updates sysno, which is an int*
  20-jun-2003 Heikki
*/

int zebra_insert_record (ZebraHandle zh, 
			 struct recordGroup *rGroup,
			 const char *recordType,
			 int *sysno, const char *match, const char *fname,
			 const char *buf, int buf_size,
			 int force_update) /* This one is ignored */
 
{
    int res;
    yaz_log(LOG_API,"zebra_insert_record sysno=%d", *sysno);

    if (buf_size < 1) buf_size = strlen(buf);

    zebra_begin_trans(zh, 1);
    res=bufferExtractRecord (zh, buf, buf_size, rGroup, 
			     0, /* delete_flag  */
			     0, /* test_mode */
			     recordType,
			     sysno,   
			     match, fname,
			     force_update, 
			     0); /* allow_update */
    zebra_end_trans(zh); 
    if (res < 0) return (res);
    return res; 
}

int zebra_update_record (ZebraHandle zh, 
			 struct recordGroup *rGroup,
			 const char *recordType,
			 int* sysno, const char *match, const char *fname,
			 const char *buf, int buf_size,
			 int force_update)
{
    int res;

    yaz_log(LOG_API,"zebra_update_record sysno=%d", *sysno);

    if (buf_size < 1) buf_size = strlen(buf);

    zebra_begin_trans(zh, 1);
    res=bufferExtractRecord (zh, buf, buf_size, rGroup, 
			     0, /* delete_flag */
			     0, /* test_mode */
			     recordType,
			     sysno,   
			     match, fname,
			     force_update, 
			     1); /* allow_update */
    zebra_end_trans(zh); 
    return res; 
}

int zebra_delete_record (ZebraHandle zh, 
			 struct recordGroup *rGroup, 
			 const char *recordType,
			 int *sysno, const char *match, const char *fname,
			 const char *buf, int buf_size,
			 int force_update) 
{
    int res;
    yaz_log(LOG_API,"zebra_delete_record sysno=%d", *sysno);

    if (buf_size < 1) buf_size = strlen(buf);

    zebra_begin_trans(zh, 1);
    res=bufferExtractRecord (zh, buf, buf_size, rGroup, 
			     1, /* delete_flag */
			     0, /* test_mode */
			     recordType,
			     sysno,
			     match,fname,
			     force_update,
			     1); /* allow_update */
    zebra_end_trans(zh);
    return res;   
}

/* ---------------------------------------------------------------------------
  Searching 
*/

int zebra_search_PQF (ZebraHandle zh, const char *pqf_query,
		      const char *setname, int *numhits)
{
    int hits = 0;
    int res=-1;
    Z_RPNQuery *query;
    ODR odr = odr_createmem(ODR_ENCODE);

    yaz_log(LOG_API,"zebra_search_PQF s=%s q=%s",setname, pqf_query);
    
    query = p_query_rpn (odr, PROTO_Z3950, pqf_query);
    
    if (!query)
        yaz_log (LOG_WARN, "bad query %s\n", pqf_query);
    else
        res=zebra_search_RPN (zh, odr, query, setname, &hits);
    
    odr_destroy(odr);

    yaz_log(LOG_API,"Hits: %d",hits);

    if (numhits)
	*numhits=hits;

    return res;
}

/* ---------------------------------------------------------------------------
  Sort - a simplified interface, with optional read locks.
  FIXME - This is a horrible name, will conflict with half the applications
*/
int zebra_sort_by_specstr (ZebraHandle zh, 
			   ODR stream,
			   const char *sort_spec,
			   const char *output_setname,
			   const char **input_setnames) 
{
    int num_input_setnames = 0;
    int sort_status = 0;
    Z_SortKeySpecList *sort_sequence = yaz_sort_spec (stream, sort_spec);
    yaz_log(LOG_API,"sort (FIXME) ");
    if (!sort_sequence)
    {
        logf(LOG_WARN,"invalid sort specs '%s'", sort_spec);
        zh->errCode = 207;
	return -1;
    }
    
    /* we can do this, since the perl typemap code for char** will 
       put a NULL at the end of list */
    while (input_setnames[num_input_setnames]) num_input_setnames++;

    if (zebra_begin_read (zh))
        return -1;
    
    resultSetSort (zh, stream->mem, num_input_setnames, input_setnames,
                   output_setname, sort_sequence, &sort_status);
    
    zebra_end_read(zh);
    return sort_status;
}