/**
* 关键词分词
*
* @Author tianyunzi
* @DateTime 2015-10-16T10:38:46+0800
*
* @param [type] $keyword
* [description]
* @return array [description]
*/
public function splitKeywordHelper($keyword)
{
if (empty($keyword)) {
return array();
}
$url = $this->di['config']->base->split . "/wd/" . urlencode($keyword);
$num = 3;
while ($num) {
$res = Tools::curl($url, '', '', true, 5);
if ($res[0] == 200) {
$res[1] = str_replace("+", "", $res[1]);
$keywords = explode(" ", $res[1]);
$keywords = array_filter($keywords);
return $keywords;
}
$num--;
}
return array($keyword);
}
/**
* 功能描述 通过sphinxQL方式查询
* @author 吕小虎
* @datetime ${DATE} ${TIME}
* @version
* @param
* @return
*/
public function getDataFromSpinx($data)
{
$this->data = $data;
//分词
$this->data['split'] = \Xz\Func\Common\Tools::curlGetContentMs($this->di['config']->base->split . '/wd/' . urlencode($this->data['wd']), 50);
if (empty($this->data['split'])) {
$this->data['split'] = $data['wd'];
}
$sphinxConfig = $this->di["config"]["combusinessearchsphinx"];
$conn = new Connection();
$indexTable = $sphinxConfig->table;
$conn->setParams(array('host' => $sphinxConfig->host, 'port' => $sphinxConfig->port));
$fieldArr = array('id');
$query = SphinxQL::create($conn)->select($fieldArr)->from($indexTable);
//企业名称和企业法人搜索 企业注册号搜索
if (!empty($this->data['wd'])) {
//处理搜索词
$keyArr = explode(' ', $this->data['split']);
$keyArr = array_filter($keyArr);
$keyStr = '';
if (is_array($keyArr) && !empty($keyArr)) {
foreach ($keyArr as $value) {
$keyStr .= '"' . $value . '" ';
}
}
if (is_numeric($this->data['wd']) && mb_strlen($this->data['wd'], 'UTF-8') == 15) {
//注册号全匹配搜索
$query->where('regno', '=', $this->data['wd']);
} else {
//企业名称和法人搜索
$query->match(array('comname', 'legal'), $keyStr, true);
}
}
//一级分类筛选
if (!empty($this->data['cate1id']) && intval($this->data['cate1id']) > 0) {
$query->where('cate1', '=', intval($this->data['cate1id']));
}
//二级分类筛选
if (!empty($this->data['cate2id']) && intval($this->data['cate2id']) > 0) {
$query->where('cate2', '=', intval($this->data['cate2id']));
}
//地区筛选
if (!empty($this->data['areaid']) && intval($this->data['areaid']) > 0) {
if ($this->data['areaid'] % 10000 == 0) {
$start = intval($this->data['areaid'] / 10000) * 10000;
$end = $start + 9999;
$query->where('areaid', 'BETWEEN', array($start, $end));
} elseif ($this->data['areaid'] % 100 == 0) {
$start = intval($this->data['areaid'] / 100) * 100;
$end = $start + 99;
$query->where('areaid', 'BETWEEN', array($start, $end));
} else {
$query->where('areaid', '=', intval($this->data['areaid']));
}
}
//成立时间筛选
if (isset($this->data['foundstart']) && $this->data['foundstart'] > 0) {
$query->where('startdate', '>=', intval($this->data['foundstart']));
if (isset($this->data['foundend']) && $this->data['foundend'] > 0 && $this->data['foundend'] > $this->data['foundstart']) {
$query->where('startdate', '<=', intval($this->data['foundend']));
}
}
//查询条数
if (isset($data['max'])) {
$query->option('max_matches', $data['max']);
}
if (isset($this->data['offset']) && isset($this->data['limit'])) {
$query->limit($this->data['offset'], $this->data['limit']);
}
//分类
$facet = Facet::create($conn)->facet('cate1');
$query->facet($facet);
$facet = Facet::create($conn)->facet('cate2');
$query->facet($facet);
//地区
$facet = Facet::create($conn)->facet('areaid');
$query->facet($facet);
$result = array('data' => array(), 'cate1' => array(), 'cate2' => array(), 'areaid' => array());
$batchResult = $query->executeBatch();
if (is_array($batchResult) && count($batchResult) > 0) {
$result['data'] = $batchResult[0];
$result['cate1'] = isset($batchResult[1]) ? $batchResult[1] : array();
$result['cate2'] = isset($batchResult[2]) ? $batchResult[2] : array();
$result['areaid'] = isset($batchResult[3]) ? $batchResult[3] : array();
}
if (!empty($result)) {
if (!empty($result['cate1'])) {
$result['cate1'] = array_combine(array_column($result['cate1'], 'cate1'), array_column($result['cate1'], 'count(*)'));
}
if (!empty($result['cate2'])) {
$result['cate2'] = array_combine(array_column($result['cate2'], 'cate2'), array_column($result['cate2'], 'count(*)'));
}
if (!empty($result['areaid'])) {
$result['areaid'] = array_combine(array_column($result['areaid'], 'areaid'), array_column($result['areaid'], 'count(*)'));
}
}
$cidArr = array();
if (!empty($result['data'])) {
foreach ($result['data'] as $key => $value) {
$cidArr[] = $value['id'];
}
}
//根据id获取缓存
if (!empty($cidArr)) {
$combusObj = new CacheCombusiness();
$fieldArr = array('cid', 'comname', 'regno', 'province', 'address', 'legal', 'gccid', 'RegistCapi', 'businessstart', 'startdate');
$result['data'] = $combusObj->getMore($cidArr, $fieldArr);
} else {
$result['data'] = array();
}
//重新实例化以获取查询到的总数
$sphinxql = new SphinxQL($conn);
$total = $sphinxql->query('show meta');
$total = $this->formatSphQLArray($total->execute());
$result['total_found'] = $total['total_found'];
$result['time'] = $total['time'];
$result['split'] = $this->data['split'];
return $result;
}
/**
* 功能描述
* @author 吕小虎
* @datetime ${DATE} ${TIME}
* @version
* @param
* @return
*/
public function getDataFilterFromSpinx($data)
{
return array('cate1' => array(), 'cate2' => array(), 'areaid' => array());
$this->data = $data;
//分词
$this->data['split'] = \Xz\Func\Common\Tools::curlGetContentMs($this->di['config']->base->split . '/wd/' . urlencode($this->data['wd']), 50);
if (empty($this->data['split'])) {
$this->data['split'] = $data['wd'];
}
$sphinxConfig = $this->di["config"]["combusinessearchsphinx"];
$conn = new Connection();
$indexTable = $sphinxConfig->table;
$conn->setParams(array('host' => $sphinxConfig->host, 'port' => $sphinxConfig->port));
$fieldArr = isset($data['field']) && !empty($data['field']) ? $data['field'] : array('id');
$query = SphinxQL::create($conn)->select($fieldArr)->from($indexTable);
//企业名称和法人搜索
$query->match(array('comname'), $this->data['split']);
//一级分类筛选
if (!empty($this->data['cate1id']) && intval($this->data['cate1id']) > 0) {
// $query->match(array('cate1'), $this->data['cate1id']);
$query->where('cate1', '=', intval($this->data['cate1id']));
}
//二级分类筛选
if (!empty($this->data['cate2id']) && intval($this->data['cate2id']) > 0) {
// $query->match(array('cate2'), $this->data['cate2id']);
$query->where('cate2', '=', intval($this->data['cate2id']));
}
//地区筛选
if (!empty($this->data['areaid']) && intval($this->data['areaid']) > 0) {
if ($this->data['areaid'] % 10000 == 0) {
$start = intval($this->data['areaid'] / 10000) * 10000;
$end = $start + 9999;
$query->where('areaid', 'BETWEEN', array($start, $end));
} elseif ($this->data['areaid'] % 100 == 0) {
$start = intval($this->data['areaid'] / 100) * 100;
$end = $start + 99;
$query->where('areaid', 'BETWEEN', array($start, $end));
} else {
$query->where('areaid', '=', intval($this->data['areaid']));
}
}
//分类
$facet = Facet::create($conn)->facet('cate1');
$query->facet($facet);
$facet = Facet::create($conn)->facet('cate2');
$query->facet($facet);
//地区
$facet = Facet::create($conn)->facet('areaid');
$query->facet($facet);
$result = array();
// echo $query->compileSelect()->getCompiled();die();
$batchResult = $query->executeBatch();
if (is_array($batchResult) && count($batchResult) > 0) {
// $result['data'] = $batchResult[0];
$result['cate1'] = isset($batchResult[1]) ? $batchResult[1] : array();
$result['cate2'] = isset($batchResult[2]) ? $batchResult[2] : array();
$result['areaid'] = isset($batchResult[3]) ? $batchResult[3] : array();
}
if (!empty($result)) {
if (!empty($result['cate1'])) {
$result['cate1'] = array_combine(array_column($result['cate1'], 'cate1'), array_column($result['cate1'], 'count(*)'));
}
if (!empty($result['cate2'])) {
$result['cate2'] = array_combine(array_column($result['cate2'], 'cate2'), array_column($result['cate2'], 'count(*)'));
}
if (!empty($result['areaid'])) {
$result['areaid'] = array_combine(array_column($result['areaid'], 'areaid'), array_column($result['areaid'], 'count(*)'));
}
}
return $result;
}
/**
* 从sphinx获取数据
* @author 刘建辉
* @datetime 2015-08-19T18:05:19+0800
* @return [type] [description]
*/
public function getDataFromSpinx($data)
{
$this->data = $data;
if (isset($this->data['redword']) && !empty($this->data['redword'])) {
$this->data['split'] = $this->data['redword'];
} else {
//分词
$this->data['split'] = $this->data['wd'];
$splitWd = \Xz\Func\Common\Tools::curlGetContentMs($this->di['config']->base->split . '/wd/' . urlencode($this->data['wd']), 50);
if ($splitWd) {
$this->data['split'] = $splitWd;
}
}
$sphinxConfig = $this->di["config"]["prosearchsphinx"];
$conn = new Connection();
//$indexTable = "product_distri";
$indexTable = "product_m_distri";
//TODO 索引范围
if (!isset($this->data['cateid'])) {
if (isset($this->data['cate3'])) {
$this->data['cateid'] = $this->data['cate3'];
}
}
//逻辑判断必须加>0
if (isset($this->data['cateid']) && $this->data['cateid'] > 0) {
$cateinfo = \Xz\Lib\Cate::getCateInfo(array($this->data['cateid']), array('nav'));
if (!empty($cateinfo)) {
$this->data['cate1'] = $cateinfo[$this->data['cateid']]['nav'][0]['cateid'];
$indexTable = "product_distri" . '_' . $this->data['cate1'];
}
//$indexTable = "product_distri";
}
$conn->setParams(array('host' => $sphinxConfig->host, 'port' => $sphinxConfig->port));
$gcdweight = "weight()+IF(id>900000000, tradenum*100, 0)+inquirynum*20+star*2+basescore*5+creditscore+IF(is_op=1, all_uv*10+all_pv, 0)+IF(id>900000000, weight()*0.1, 0) as gcpdweight";
//$gcdweight = "weight() + gcdweight as gcpdweight";
$query = SphinxQL::create($conn)->select('id', 'cid', $gcdweight)->from($indexTable);
$query->match('*', $this->data['split']);
$query->option('max_matches', 200);
if (!empty($this->data['cateid']) && intval($this->data['cateid']) > 0) {
$query->where('cate3', '=', intval($this->data['cateid']));
}
if (!empty($this->data['brand']) && intval($this->data['brand']) > 0) {
$query->where('brand', '=', intval($this->data['brand']));
}
if (!empty($this->data['province']) && intval($this->data['province']) > 0) {
$query->where('province', '=', intval($this->data['province']));
}
if (!empty($this->data['city']) && intval($this->data['city']) > 0) {
$query->where('city', '=', intval($this->data['city']));
}
if (!empty($this->data['iscertify']) && intval($this->data['iscertify']) > 0) {
$query->where('is_gccertify', '=', intval($this->data['iscertify']));
}
if (!empty($this->data['isprice']) && intval($this->data['isprice']) > 0) {
$query->where('price', '>', 0);
}
if (!empty($this->data['feature'])) {
$featureArr = explode('_', $this->data['feature']);
foreach ($featureArr as $value) {
$query->where('feature', '=', intval($value));
}
}
if (!empty($this->data['sort'])) {
switch ($this->data['sort']) {
case 1:
$query->orderBy('tradenum', 'DESC');
//销量
break;
case 2:
$query->orderBy('visitnum', 'DESC');
//访问量/热度
break;
case 3:
$query->orderBy("price", "DESC");
break;
case 4:
$query->orderBy("price", "ASC");
break;
case 6:
$query->orderBy("integral", "ASC");
break;
default:
$query->orderBy("gcpdweight", "DESC");
break;
}
} else {
$query->orderBy("gcpdweight", "DESC");
}
//$facet = Facet::create($conn)->facet('cate3');
//$query->facet($facet);
$facet = Facet::create($conn)->facet('feature')->limit(20)->orderby("count(*)", "desc");
$query->facet($facet);
//品牌
$brand = Facet::create($conn)->facet('brand')->limit(20)->orderby("count(*)", "desc");
$query->facet($brand);
if (!empty($this->data['province'])) {
//市
$facet = Facet::create($conn)->facet('city')->limit(20)->orderby("count(*)", "desc");
$query->facet($facet);
} else {
//省
$facet = Facet::create($conn)->facet('province')->limit(20)->orderby("count(*)", "desc");
$query->facet($facet);
}
//查询条数
//$query->limit($this->data['offset'], $this->data['limit']);
$query->limit(0, 200);
//匹配设定
$query->option('field_weights', array('proname' => 300));
//$query->option('ranker', 'sph04');
$batchResult = $query->executeBatch();
$result = array();
if (is_array($batchResult) && count($batchResult) > 0) {
$result['data'] = $batchResult[0];
//$result['cate3'] = isset($batchResult[1]) ? $batchResult[1] : array();
$result['property'] = isset($batchResult[1]) ? $batchResult[1] : array();
$result['brand'] = isset($batchResult[2]) ? $batchResult[2] : array();
if (!empty($this->data['province'])) {
$result['city'] = isset($batchResult[3]) ? $batchResult[3] : array();
} else {
$result['province'] = isset($batchResult[3]) ? $batchResult[3] : array();
}
}
//if (!empty($result['data'])) {
foreach ($result['data'] as $key => &$value) {
//拿到产品ID，企业ID
$value = array($value['id'], $value['cid']);
}
unset($value);
//防霸屏
$result['data'] = $this->sortData($result['data']);
/*if (!empty($result['cate3'])) {
$result['cate3'] = json_encode(array_combine(array_column($result['cate3'], 'cate3'), array_column($result['cate3'], 'count(*)')));
} else {
$result['cate3'] = '[]';
}*/
if (!empty($result['province'])) {
$result['province'] = json_encode(array_combine(array_column($result['province'], 'province'), array_column($result['province'], 'count(*)')));
} else {
$result['province'] = '[]';
}
if (!empty($result['city'])) {
$result['city'] = json_encode(array_combine(array_column($result['city'], 'city'), array_column($result['city'], 'count(*)')));
} else {
$result['city'] = '[]';
}
if (!empty($result['brand'])) {
$result['brand'] = json_encode(array_combine(array_column($result['brand'], 'brand'), array_column($result['brand'], 'count(*)')));
} else {
$result['brand'] = '[]';
}
if (!empty($result['property'])) {
$result['property'] = json_encode(array_combine(array_column($result['property'], 'feature'), array_column($result['property'], 'count(*)')));
} else {
$result['property'] = '[]';
}
//}
//重新实例化以获取查询到的总数
$sphinxql = new SphinxQL($conn);
$total = $sphinxql->query('show meta');
$total = $this->formatSphQLArray($total->execute());
$result['total_found'] = $total['total_found'];
$result['time'] = $total['time'];
$result['split'] = $this->data['split'];
return $result;
}
/**
* 获取分词&&核心词
* @author 刘建辉
* @datetime 2015-10-31T10:25:30+0800
* @param [type] $string [description]
* @param [type] $type [description]
* @return [type] [description]
*/
public static function splitWord($string, $type)
{
$url = 'http://tj.yw.gongchang.com/api/uid/?mark=jieba_output&type=cut_all&secret=7232275&content=';
$result = \Xz\Func\Common\Tools::curlGetContentMs($url . urlencode($string), 500);
if ($result) {
$result = json_decode($result, true);
}
if ($type == 'split') {
return $result['data']['cut_all'];
}
if ($type == 'core') {
if (!empty($result['data']['TF-IDF_keyword_extract'])) {
return $result['data']['TF-IDF_keyword_extract'][0];
}
}
return $string;
}