PHP智能識(shí)別收貨地址信息實(shí)例

?php
class DistinguishAddress {
/**
 * 類(lèi)的入口方法
 * 傳入地址信息自動(dòng)識(shí)別，并返回最高匹配結(jié)果
 * 如果地址新增，則需要?jiǎng)h除緩存文件重新緩存
 * @param $address
 **/
function getAddressResult($address){
 // 優(yōu)先第一種方法
 $result = $this->getAddressArrar($address);
 // 如果結(jié)果不理想，再模糊去匹配
 if($result['level'] != 3){
  $result_sub = $this->addressVague($address);
  // 只有全匹配對(duì)才替換，否則不做任何改變
  if($result_sub['level'] == 3){
   $result = $result_sub;
  }
 }
 // 聯(lián)系方式-優(yōu)先匹配電話
 if(preg_match('/1\d{10}/', $address, $mobiles)){ // 手機(jī)
  $result['mobile'] = $mobiles[0];
 } else if(preg_match('/(\d{3,4}-)?\d{7,8}/', $address, $mobiles)){ // 固定電話
  $result['mobile'] = $mobiles[0];
 }
 // 識(shí)別姓名-必須空格分享的--概率
 preg_match_all('/[\x{4e00}-\x{9fa5}]{2,}/iu', $address,$names);
 if($names){
  $name_where = '';
  foreach ($names[0] as $name){
   // 必須是大于1個(gè)字符且小于5個(gè)字符的
   if(1  mb_strlen($name,'utf-8')  mb_strlen($name, 'utf-8')  5){
    $sub_name = mb_substr($name, 0, 1, 'utf-8');
    $name_where .= "name like '{$sub_name}%' or ";
   }
  }
  if(!empty($name_where)){
   $name_where = substr($name_where, 0, -3);
   $names_sql = "select name from surname where {$name_where} order by sort desc";
   $list = Db::getInstance('DbTrade')->getAll($names_sql);
   // 統(tǒng)計(jì)有多少種可能性-姓名
   $result['name_num'] = count($list);
   if($list) {
    $name_first = $list[0]['name'];
    foreach ($names[0] as $name){
     $len = mb_strlen($name_first, 'utf-8');
     if (mb_substr($name, 0, $len, 'utf-8') == $name_first){
      $result['name'] = $name;
     }
    }
   }
  }
 }
 // 去掉詳細(xì)里面的姓名和電話
 $result['info'] = str_replace($result['mobile'], '', $result['info']);
 $result['info'] = str_replace($result['name'], '', $result['info']);
 $result['info'] = $result['province']['region_name'] . $result['city']['region_name'] . $result['district']['region_name'] . $result['info'];
 return $this->getCityLevelList($result);
}
/**
 * 獲取對(duì)應(yīng)城市等級(jí)列表
 **/
function getCityLevelList($result){
 // 獲取所有地址遞歸列表
 $regions = $this->getRegionTreeList();
 // 獲取省份列表- 只有存在值才返回對(duì)應(yīng)列表
 $province_id = $result['province']['region_id'];
 if ($province_id) {
  foreach ($regions as $region){
   unset($region['childs']);
   $result['province_list'][] = $region;
  }
 }
 // 獲取城市列表- 只有存在值才返回對(duì)應(yīng)列表
 $city_id = $result['city']['region_id'];
 if ($city_id) {
  foreach ($regions[$province_id]['childs'] as $region){
   unset($region['childs']);
   $result['city_list'][] = $region;
  }
 }
 // 獲取地區(qū)列表- 只有存在值才返回對(duì)應(yīng)列表
 $district_id = $result['district']['region_id'];
 if ($district_id) {
  foreach ($regions[$province_id]['childs'][$city_id]['childs'] as $region){
   unset($region['childs']);
   $result['district_list'][] = $region;
  }
 }
 return $result;
}
/**
 * 獲取所有地址遞歸列表
 **/
function getRegionTreeList(){
 // IO
 $file_name = 'regions.json';
 if(is_file($file_name)){
  $regions = file_get_contents($file_name);
  $regions = json_decode($regions, true);
 } else {
  $region_sql = "select region_id,region_name,parent_id from region";
  $regions = Db::getInstance('DbTrade')->getAll($region_sql);
  $regions = $this->arrayKey($regions);
  file_put_contents($file_name, json_encode($regions));
 }
 return $regions;
}
/**
 * 第一種方法
 * 根據(jù)地址列表遞歸查找準(zhǔn)確地址
 * @param $address
 * @return array
 **/
function getAddressArrar($address){
 // 獲取所有地址遞歸列表
 $regions = $this->getRegionTreeList();
 // 初始化數(shù)據(jù)
 $province = $city = $district = array();
 // 先查找省份-第一級(jí)地區(qū)
 $province = $this->checkAddress($address, $regions);
 if($province){
  // 查找城市-第二級(jí)地區(qū)
  $city = $this->checkAddress($address, $province['list']);
  if($city){
   // 查找地區(qū)-第三級(jí)地區(qū)
   // 西藏自治區(qū)那曲市色尼區(qū)遼寧南路西藏公路 第三個(gè)參數(shù)因?yàn)檫@個(gè)地址沖突取消強(qiáng)制
   $district = $this->checkAddress($address, $city['list']);
  }
 }
 return $this->getAddressInfo($address, $province, $city, $district);
}
 /**
  * 第二種方法
  * 地址模糊查找
  **/
function addressVague($address){
 $res = preg_match_all('/\S{2}[自市區(qū)鎮(zhèn)縣鄉(xiāng)島州]/iu', $address,$arr);
 if(!$res) return false;
 $where = ' where ';
 foreach ($arr[0] as $value){
  if(strpos($value, '小區(qū)') === false  strpos($value, '開(kāi)發(fā)區(qū)') === false){
   $where .= "region_name like '%{$value}' or ";
  }
 }
 $where = substr($where,0,-3);
 $region_sql = "select region_id,region_name,parent_id,region_type from region " . $where;
 $citys = $GLOBALS['db']->getAll($region_sql);
 // 匹配所有地址
 $result = array();
 foreach ($citys as $city){
  // 所有相關(guān)聯(lián)的地區(qū)id
  $city_ids = array();
  if($city['region_type'] == 2) {
   $city_ids = array($city['parent_id'], $city['region_id']);
   // 嘗試能不能匹配第三級(jí)
   $region_sql = "select region_id,region_name,parent_id,region_type,left(region_name,2) as ab_name from region where parent_id='{$city['region_id']}'" ;
   $areas = $GLOBALS['db']->getAll($region_sql);
   foreach ($areas as $row){
    if(mb_strpos($address,$row['ab_name'])){
     $city_ids[] = $row['region_id'];
    }
   }
  } else if($city['region_type'] == 3){
   $region_sql = "select parent_id from region where region_id='{$city['parent_id']}'" ;
   $city['province_id'] = $GLOBALS['db']->getOne($region_sql);
   $city_ids = array($city['parent_id'], $city['region_id'], $city['province_id']);
  }
  // 查找該單詞所有相關(guān)的地區(qū)記錄
  $where = " where region_id in(" . join(',', $city_ids) . ")";
  $region_sql = "select region_id,region_name,parent_id,region_type,left(region_name,2) as ab_name from region " . $where . ' order by region_id asc';
  $city_list = $GLOBALS['db']->getAll($region_sql);
  sort($city_ids);
  $key = array_pop($city_ids);
  $result[$key] = $city_list;
  sort($result);
 }
 if($result){
  list($province, $city, $area) = $result[0];
  return $this->getAddressInfo($address, $province, $city, $area);
 }
 return false;
}
/**
 * 匹配正確的城市地址
 * @param $address
 * @param $city_list
 * @param int $force
 * @param int $str_len
 * @return array
 **/
function checkAddress($address, $city_list, $force=false, $str_len=2){
 $num = 0;
 $list = array();
 $result = array();
 // 遍歷所有可能存在的城市
 foreach ($city_list as $city_key=>$city){
  $city_name = mb_substr($city['region_name'], 0, $str_len,'utf-8');
  // 判斷是否存包含當(dāng)前地址字符
  $city_arr = explode($city_name, $address);
  // 如果存在相關(guān)字眼，保存該地址的所有子地址
  if(count($city_arr) >= 2){
   // 必須名稱(chēng)長(zhǎng)度同時(shí)達(dá)到當(dāng)前比對(duì)長(zhǎng)度
   if(strlen($city['region_name'])  $str_len){
    continue;
   }
   $num ++;
   $list = $list + $city['childs'];

   $result[] = array(
    'region_id' => $city['region_id'],
    'region_name' => $city['region_name'],
    'list' =>$list,
   );
  }
 }
 // 如果有多個(gè)存在，則加大字符匹配長(zhǎng)度
 if($num > 1 || $force){
  $region_name1 = $result[0]['region_name'];
  $region_name2 = $result[1]['region_name'];

  if(strlen($region_name1) == strlen($region_name2)  strlen($region_name1) == $str_len){
   $region_id1 = $result[0]['region_id'];
   $region_id2 = $result[1]['region_id'];
   $index = $region_id1 > $region_id2 ? 1 : 0;
   $result = $result[$index];
   return $result;
  }
  return $this->checkAddress($address, $city_list, $force, $str_len+1);
 } else {
  $result[0]['list'] = $list;
  return $result[0];
 }
}
/**
 * 根據(jù)原地址返回詳細(xì)信息
 * @param $address
 * @param $province
 * @param $city
 * @param $area
 * @return array
 **/
function getAddressInfo($address, $province, $city, $district){
 // 查找最后出現(xiàn)的地址 - 截取詳細(xì)信息
 $find_str = '';
 if($province['region_name']){
  $find_str = $province['region_name'];
  if($city['region_name']){
   $find_str = $city['region_name'];
   if($district['region_name']){
    $find_str = $district['region_name'];
   }
  }
 }
 // 截取詳細(xì)的信息
 $find_str_len = mb_strlen($find_str,'utf-8');
 for($i=0; $i$find_str_len-1; $i++){
  $substr = mb_substr($find_str,0,$find_str_len - $i, 'utf-8');
  $end_index = mb_strpos($address, $substr);
  if ($end_index){
   $address = mb_substr($address, $end_index + mb_strlen($substr) , mb_strlen($address) - $end_index);
  }
 }
 !empty($find_str)  $find_str = '|\S*' . $find_str;
 $area['info'] = preg_replace("/\s*|,|，|:|：{$find_str}/i", '', $address);
 $level = 0;
 if($district['region_name']){
  $level = 3;
 } else if($city['region_name']){
  $level = 2;
 } else if ($province['region_name']) {
  $level = 1;
 }
 return array(
  'province' => array('region_id'=>$province['region_id'], 'region_name'=>$province['region_name']),
  'city'  => array('region_id'=>$city['region_id'], 'region_name'=>$city['region_name']),
  'district'  => array('region_id'=>$district['region_id'], 'region_name'=>$district['region_name']),
  'info'  => $area['info'],
  'level'  => $level,
 );
}
/**
 * 遞歸所有地址成無(wú)限分類(lèi)數(shù)組
 * @param $data
 * @param int $region_id
 * @return array
 **/
function arrayKey($data, $region_id=1){
 $result = array();
 foreach ($data as $row){
  if($region_id == $row['parent_id']){
   $key = $row['region_id'];
   $row['childs'] = $this->arrayKey($data, $row['region_id']);
   $result[$key] = $row;
  }
 }
 return $result;
}
}
?>

DROP TABLE IF EXISTS `surname`;
CREATE TABLE `surname` (
 `id` int(11) NOT NULL AUTO_INCREMENT,
 `name` char(10) CHARACTER SET utf8 COLLATE utf8_general_ci NULL DEFAULT NULL,
 `sort` int(11) NULL DEFAULT NULL,
 PRIMARY KEY (`id`) USING BTREE,
 INDEX `name`(`name`) USING BTREE,
 INDEX `sort`(`sort`) USING BTREE
) ENGINE = InnoDB AUTO_INCREMENT = 481 CHARACTER SET = utf8 COLLATE = utf8_general_ci COMMENT = '姓氏表' ROW_FORMAT = Compact;

CREATE TABLE `region` (
 `region_id` smallint(5) UNSIGNED NOT NULL AUTO_INCREMENT,
 `parent_id` smallint(5) UNSIGNED NOT NULL DEFAULT 0,
 `region_name` varchar(120) CHARACTER SET utf8 COLLATE utf8_general_ci NOT NULL DEFAULT '',
 `region_type` tinyint(1) NOT NULL DEFAULT 2,
 `agency_id` smallint(5) UNSIGNED NOT NULL DEFAULT 0,
 PRIMARY KEY (`region_id`) USING BTREE,
 INDEX `parent_id`(`parent_id`) USING BTREE,
 INDEX `region_type`(`region_type`) USING BTREE,
) ENGINE = InnoDB AUTO_INCREMENT = 3956 AVG_ROW_LENGTH = 44 CHARACTER SET = utf8 COLLATE = utf8_general_ci ROW_FORMAT = Compact;