⭐ 欢迎来到虫虫下载站! | 📦 资源下载 📁 资源专辑 ℹ️ 关于我们
⭐ 虫虫下载站

📄 zsets.c

📁 harvest是一个下载html网页得机器人
💻 C
📖 第 1 页 / 共 2 页
字号:
			    if (psysno == sort_info->entries[i]->sysno)				break;			if (i >= 0)			    continue;		    }		    position++;		    assert (num_i < num);		    if (position == positions[num_i])		    {			sr[num_i].sysno = psysno;			yaz_log (LOG_DEBUG, "got pos=%d (unsorted)", position);			sr[num_i].score = -1;			num_i++;		    }		}	    }	    rset_close (rset, rfd);	}    }    return sr;}void zebraPosSetDestroy (ZebraHandle zh, ZebraPosSet records, int num){    xfree (records);}struct sortKeyInfo {    int relation;    int attrUse;    int numerical;};void resultSetInsertSort (ZebraHandle zh, ZebraSet sset,			  struct sortKeyInfo *criteria, int num_criteria,			  int sysno){    struct zset_sort_entry this_entry;    struct zset_sort_entry *new_entry = NULL;    struct zset_sort_info *sort_info = sset->sort_info;    int i, j;    sortIdx_sysno (zh->reg->sortIdx, sysno);    for (i = 0; i<num_criteria; i++)    {	sortIdx_type (zh->reg->sortIdx, criteria[i].attrUse);	sortIdx_read (zh->reg->sortIdx, this_entry.buf[i]);    }    i = sort_info->num_entries;    while (--i >= 0)    {	int rel = 0;	for (j = 0; j<num_criteria; j++)	{            if (criteria[j].numerical)            {                double diff = atof(this_entry.buf[j]) -                              atof(sort_info->entries[i]->buf[j]);                rel = 0;                if (diff > 0.0)                    rel = 1;                else if (diff < 0.0)                    rel = -1;            }            else            {	        rel = memcmp (this_entry.buf[j], sort_info->entries[i]->buf[j],			  SORT_IDX_ENTRYSIZE);            }	    if (rel)		break;	}		if (!rel)	    break;	if (criteria[j].relation == 'A')        {	    if (rel > 0)	        break;	}        else if (criteria[j].relation == 'D')	{	    if (rel < 0)		break;	}    }    ++i;    j = sort_info->max_entries;    if (i == j)	return;    if (sort_info->num_entries == j)	--j;    else	j = (sort_info->num_entries)++;    new_entry = sort_info->entries[j];    while (j != i)    {	sort_info->entries[j] = sort_info->entries[j-1];	--j;    }    sort_info->entries[i] = new_entry;    assert (new_entry);    for (i = 0; i<num_criteria; i++)	memcpy (new_entry->buf[i], this_entry.buf[i], SORT_IDX_ENTRYSIZE);    new_entry->sysno = sysno;    new_entry->score = -1;}void resultSetInsertRank (ZebraHandle zh, struct zset_sort_info *sort_info,			  int sysno, int score, int relation){    struct zset_sort_entry *new_entry = NULL;    int i, j;    i = sort_info->num_entries;    while (--i >= 0)    {	int rel = 0;	rel = score - sort_info->entries[i]->score;	if (relation == 'D')	{	    if (rel >= 0)		break;	}	else if (relation == 'A')	{	    if (rel <= 0)		break;	}    }    ++i;    j = sort_info->max_entries;    if (i == j)	return;    if (sort_info->num_entries == j)	--j;    else	j = (sort_info->num_entries)++;        new_entry = sort_info->entries[j];    while (j != i)    {	sort_info->entries[j] = sort_info->entries[j-1];	--j;    }    sort_info->entries[i] = new_entry;    assert (new_entry);    new_entry->sysno = sysno;    new_entry->score = score;}void resultSetSort (ZebraHandle zh, NMEM nmem,		    int num_input_setnames, const char **input_setnames,		    const char *output_setname,		    Z_SortKeySpecList *sort_sequence, int *sort_status){    ZebraSet sset;    RSET rset;    if (num_input_setnames == 0)    {	zh->errCode = 208;	return ;    }    if (num_input_setnames > 1)    {	zh->errCode = 230;	return;    }    yaz_log (LOG_DEBUG, "result set sort input=%s output=%s",	  *input_setnames, output_setname);    sset = resultSetGet (zh, input_setnames[0]);    if (!sset)    {	zh->errCode = 30;	zh->errString = nmem_strdup (nmem, input_setnames[0]);	return;    }    if (!(rset = sset->rset))    {	zh->errCode = 30;	zh->errString = nmem_strdup (nmem, input_setnames[0]);        return;    }    if (strcmp (output_setname, input_setnames[0]))    {	rset = rset_dup (rset);	sset = resultSetAdd (zh, output_setname, 1);        sset->rset = rset;    }    resultSetSortSingle (zh, nmem, sset, rset, sort_sequence, sort_status);}void resultSetSortSingle (ZebraHandle zh, NMEM nmem,			  ZebraSet sset, RSET rset,			  Z_SortKeySpecList *sort_sequence, int *sort_status){    int i, psysno = 0;    struct it_key key;    struct sortKeyInfo sort_criteria[3];    int num_criteria;    int term_index;    RSFD rfd;    yaz_log (LOG_LOG, "resultSetSortSingle start");    sset->sort_info->num_entries = 0;    sset->hits = 0;    num_criteria = sort_sequence->num_specs;    if (num_criteria > 3)	num_criteria = 3;    for (i = 0; i < num_criteria; i++)    {	Z_SortKeySpec *sks = sort_sequence->specs[i];	Z_SortKey *sk;	if (*sks->sortRelation == Z_SortRelation_ascending)	    sort_criteria[i].relation = 'A';	else if (*sks->sortRelation == Z_SortRelation_descending)	    sort_criteria[i].relation = 'D';	else	{	    zh->errCode = 214;	    return;	}	if (sks->sortElement->which == Z_SortElement_databaseSpecific)	{	    zh->errCode = 210;	    return;	}	else if (sks->sortElement->which != Z_SortElement_generic)	{	    zh->errCode = 237;	    return;	}		sk = sks->sortElement->u.generic;	switch (sk->which)	{	case Z_SortKey_sortField:	    yaz_log (LOG_DEBUG, "Sort: key %d is of type sortField", i+1);	    zh->errCode = 207;	    return;	case Z_SortKey_elementSpec:	    yaz_log (LOG_DEBUG, "Sort: key %d is of type elementSpec", i+1);	    zh->errCode = 207;	    return;	case Z_SortKey_sortAttributes:	    yaz_log (LOG_DEBUG, "Sort: key %d is of type sortAttributes", i+1);	    sort_criteria[i].attrUse =		zebra_maps_sort (zh->reg->zebra_maps,				 sk->u.sortAttributes,                                 &sort_criteria[i].numerical);	    yaz_log (LOG_DEBUG, "use value = %d", sort_criteria[i].attrUse);	    if (sort_criteria[i].attrUse == -1)	    {		zh->errCode = 116;		return;	    }	    if (sortIdx_type (zh->reg->sortIdx, sort_criteria[i].attrUse))	    {		zh->errCode = 207;		return;	    }	    break;	}    }    rfd = rset_open (rset, RSETF_READ);    while (rset_read (rset, rfd, &key, &term_index))    {        if (key.sysno != psysno)        {	    (sset->hits)++;            psysno = key.sysno;	    resultSetInsertSort (zh, sset,				 sort_criteria, num_criteria, psysno);        }    }    rset_close (rset, rfd);    for (i = 0; i < rset->no_rset_terms; i++)	yaz_log (LOG_LOG, "term=\"%s\" nn=%d type=%s count=%d",                 rset->rset_terms[i]->name,                 rset->rset_terms[i]->nn,                 rset->rset_terms[i]->flags,                 rset->rset_terms[i]->count);    *sort_status = Z_SortStatus_success;    yaz_log (LOG_LOG, "resultSetSortSingle end");}RSET resultSetRef (ZebraHandle zh, const char *resultSetId){    ZebraSet s;    if ((s = resultSetGet (zh, resultSetId)))	return s->rset;    return NULL;}void resultSetRank (ZebraHandle zh, ZebraSet zebraSet, RSET rset){    int kno = 0;    struct it_key key;    RSFD rfd;    int term_index, i;    ZebraRankClass rank_class;    struct rank_control *rc;    struct zset_sort_info *sort_info;    sort_info = zebraSet->sort_info;    sort_info->num_entries = 0;    zebraSet->hits = 0;    rfd = rset_open (rset, RSETF_READ);    yaz_log (LOG_LOG, "resultSetRank");    rank_class = zebraRankLookup (zh, res_get_def(zh->res, "rank", "rank-1"));    rc = rank_class->control;    if (rset_read (rset, rfd, &key, &term_index))    {	int psysno = key.sysno;	int score;	void *handle =	    (*rc->begin) (zh->reg, rank_class->class_handle, rset);	(zebraSet->hits)++;	do	{	    kno++;	    if (key.sysno != psysno)	    {		score = (*rc->calc) (handle, psysno);		resultSetInsertRank (zh, sort_info, psysno, score, 'A');		(zebraSet->hits)++;		psysno = key.sysno;	    }	    (*rc->add) (handle, key.seqno, term_index);	}	while (rset_read (rset, rfd, &key, &term_index));	score = (*rc->calc) (handle, psysno);	resultSetInsertRank (zh, sort_info, psysno, score, 'A');	(*rc->end) (zh->reg, handle);    }    rset_close (rset, rfd);    for (i = 0; i < rset->no_rset_terms; i++)	yaz_log (LOG_LOG, "term=\"%s\" nn=%d type=%s count=%d",                 rset->rset_terms[i]->name,                 rset->rset_terms[i]->nn,                 rset->rset_terms[i]->flags,                 rset->rset_terms[i]->count);        yaz_log (LOG_LOG, "%d keys, %d distinct sysnos", kno, zebraSet->hits);}ZebraRankClass zebraRankLookup (ZebraHandle zh, const char *name){    ZebraRankClass p = zh->reg->rank_classes;    while (p && strcmp (p->control->name, name))	p = p->next;    if (p && !p->init_flag)    {	if (p->control->create)	    p->class_handle = (*p->control->create)(zh);	p->init_flag = 1;    }    return p;}void zebraRankInstall (struct zebra_register *reg, struct rank_control *ctrl){    ZebraRankClass p = (ZebraRankClass) xmalloc (sizeof(*p));    p->control = (struct rank_control *) xmalloc (sizeof(*p->control));    memcpy (p->control, ctrl, sizeof(*p->control));    p->control->name = xstrdup (ctrl->name);    p->init_flag = 0;    p->next = reg->rank_classes;    reg->rank_classes = p;}void zebraRankDestroy (struct zebra_register *reg){    ZebraRankClass p = reg->rank_classes;    while (p)    {	ZebraRankClass p_next = p->next;	if (p->init_flag && p->control->destroy)	    (*p->control->destroy)(reg, p->class_handle);	xfree (p->control->name);	xfree (p->control);	xfree (p);	p = p_next;    }    reg->rank_classes = NULL;}

⌨️ 快捷键说明

复制代码 Ctrl + C
搜索代码 Ctrl + F
全屏模式 F11
切换主题 Ctrl + Shift + D
显示快捷键 ?
增大字号 Ctrl + =
减小字号 Ctrl + -