First add the library to your project:
composer require netgen/query-translator:^1.0
Demo 如下：
app/Service/Query/ESQueryGenerator.php
<?php
/**
* Created by PhpStorm.
* User: admin
* Date: 2020/1/14
* Time: 11:22
*/
namespace App\Service\Query;
use App\Models\ES\Search;
use Illuminate\Support\Arr;
class ESQueryGenerator
{
/**
* 模糊查詢的字符串最大長度
* @var int
*/
const QUERY_STRING_LIMIT = 200;
/**
* 精准查询
* @param $field
* @param $val
* @return array
*/
public static function termQuery($field, $val)
{
return [
'term' => [$field => $val],
];
}
/**
* 嵌套精准查询
* @param $path
* @param $field
* @param $val
* @return array
*/
public static function nestedTermQuery($path, $field, $val)
{
return [
'nested' => [
'path' => $path,
'query' => [
'term' => ["{$path}.{$field}" => $val],
],
],
];
}
public static function queryStringQuery($val, $fields = [])
{
if (strlen($val) > self::QUERY_STRING_LIMIT) {
$val = substr($val, 0, self::QUERY_STRING_LIMIT);
}
//转义
$val = addcslashes($val, ":/?&()'");
//空格替换（解决：查询英文单词时，有空格无法查出结果）
$val = str_replace(' ', '?', $val);
//不存在*追加一下
if (strpos($val, '*') === false) {
$val = "*{$val}*";
}
if (!$fields) {
return [
'query_string' => [
'query' => "{$val}",
],
];
}
return [
'query_string' => [
'query' => "{$val}",
'fields' => (array)$fields,
"default_operator" => "AND",
],
];
}
/**
*
* 嵌套query string
*
* @param $path
* @param $field
* @param $val
* @return array
*/
public static function nestedQueryStringQuery($path, $val, $field = '')
{
if (strlen($val) > self::QUERY_STRING_LIMIT) {
$val = substr($val, 0, self::QUERY_STRING_LIMIT);
}
//转义
$val = addcslashes($val, ":/?&()'");
//空格替换（解决：查询英文单词时，有空格无法查出结果）
$val = str_replace(' ', '?', $val);
//不存在*追加一下
if (strpos($val, '*') === false) {
$val = "*{$val}*";
}
if (!$field) {
return [
'nested' => [
'path' => $path,
'query' => [
'query_string' => [
'query' => "{$val}",
],
],
],
];
}
return [
'nested' => [
'path' => $path,
'query' => [
'query_string' => [
'query' => "{$val}",
'fields' => [$path . "." . $field],
"default_operator" => "AND",
],
],
],
];
}
/**
*
* 递归生成查询语句
*
* @param $query
* @param $query_arr
* @param $field_map array 查询字段与实际字段的映射关系
* @param $OP
* @return mixed
*/
public static function makeQuery(&$query, $query_arr, $field_map, $OP = 'DEFAULT')
{
foreach ($query_arr as $type => $child_query_arr) {
if (is_numeric($type)) {
//数字索引跳过继续
self::makeQuery($query, $child_query_arr, $field_map, $OP);
} elseif (in_array($type, ['OR', 'AND'], true)) {
//翻译操作符
switch ($type) {
case 'OR':
$op = 'should';
break;
case 'AND':
$op = 'must';
break;
default:
$op = 'must';
break;
}
//带上操作符继续
self::makeQuery($query[]['bool'][$op], $child_query_arr, $field_map, $op);
} elseif ($type == "GROUP") {
//获取组内容、操作符继续
self::makeQuery($query, $child_query_arr['value'][0], $field_map, $OP);
} elseif (in_array($type, ['WORD', 'PHRASE'], true)) {//最终语句拼接
//精准字段搜
if ($child_query_arr['field'] !== '~') {
$child_query_arr['field'] = Arr::get($field_map, $child_query_arr['field'],
$child_query_arr['field']);
//是否是嵌套类型
$nested = false;
if (strpos($child_query_arr['field'], '@') !== false) {
$nested = true;
list($path, $field) = explode('@', $child_query_arr['field']);
} else {
$path = null;
$field = $child_query_arr['field'];
}
//term or query_string
switch ($type) {
case 'PHRASE':
if ($nested) {
$query[] = self::nestedTermQuery($path, $field, $child_query_arr['value']);
} else {
$query[] = self::termQuery($field, $child_query_arr['value']);
}
break;
default:
if ($nested) {
$query[] = self::nestedQueryStringQuery($path, $child_query_arr['value'],
$field);
} else {
$query[] = self::queryStringQuery($child_query_arr['value'], $field);
}
break;
}
} else {
$query[] = self::queryStringQuery($child_query_arr['value']);
}
}
}
return $query;
}
/**
*
* 获取查询语句对应的ES请求体
*
* @param $q
* @param $field_map //搜索语法字段映射
* @param $index
* @param $type
* @param $size
* @param int $page
* @param string $order_by
* @param string $order_type
* @return array
*/
public static function getQuery(
$q,
$field_map,
$index,
$type,
$size,
$page = 1,
$order_by = 'updated_at',
$order_type = 'desc'
) {
$from = max(($page - 1) * $size, 0);
if (!$q) {
return [
'index' => $index,
'type' => $type,
'size' => $size,
'from' => $from,
'body' => [
'sort' => [[$order_by => ['order' => $order_type]]],
],
];
}
$syntax_tree = ParserService::syntaxTree($q)['QUERY'];
self::makeQuery($query, $syntax_tree, $field_map);
$request_query = [
'index' => $index,
'type' => $type,
'size' => $size,
'from' => $from,
'body' => [
'query' => $query[0],
'sort' => [[$order_by => ['order' => $order_type]]],
],
];
return $request_query;
}
/**
* 对ES结果进行分页结果格式化
*
*
* @param $src
* @param $page
* @param $page_size
* @return array
*/
public static function paginate($src, $page, $page_size)
{
//组装返回数据
if (Arr::get($src, 'hits.total', 0)) {
$list = [];
$hits = Arr::get($src, 'hits.hits');
if ($hits) {
foreach ($hits as $hit) {
$source = Arr::get($hit, '_source');
$source['index_id'] = Arr::get($hit, '_id');
$source['index_index'] = Arr::get($hit, '_index');
$list[] = $source;
}
}
$total = Arr::get($src, 'hits.total');
$data = [
'current_page' => (int)$page,
'total_page' => (int)ceil(max($total, 1) / $page_size),
'total' => (int)$total,
'list' => $list,
];
} else {
$data = [
'current_page' => (int)1,
'total_page' => (int)1,
'total' => (int)0,
'list' => [],
];
}
return $data;
}
}app/Service/Query/ParserService.php
<?php
/**
* Created by PhpStorm.
* User: admin
* Date: 2020/1/13
* Time: 10:56
*/
namespace App\Service\Query;
use App\Models\ES\Search;
use QueryTranslator\Languages\Galach\Tokenizer;
use QueryTranslator\Languages\Galach\TokenExtractor\Full as FullTokenExtractor;
use QueryTranslator\Languages\Galach\Parser;
use QueryTranslator\Languages\Galach\Generators;
class ParserService
{
const defaultFieldName = 'error';
public static function syntaxTree($queryString, $fieldMap = Search::FIELD_MAP)
{
// 2. Query string is given to Tokenizer which produces an instance of TokenSequence
// Note that Tokenizer needs a TokenExtractor, which is an extension point
// Here we use Full TokenExtractor which provides full Galach syntax
$tokenExtractor = new FullTokenExtractor();
$tokenizer = new Tokenizer($tokenExtractor);
$tokenSequence = $tokenizer->tokenize($queryString);
// 3. TokenSequence instance is given to Parser which produces an instance of SyntaxTree
$parser = new Parser();
$syntaxTree = $parser->parse($tokenSequence);
return SyntaxTreeRenderer::render($syntaxTree);
}
public static function queryString($queryString, $fieldMap = Search::FIELD_MAP)
{
// 2. Query string is given to Tokenizer which produces an instance of TokenSequence
// Note that Tokenizer needs a TokenExtractor, which is an extension point
// Here we use Full TokenExtractor which provides full Galach syntax
$tokenExtractor = new FullTokenExtractor();
$tokenizer = new Tokenizer($tokenExtractor);
$tokenSequence = $tokenizer->tokenize($queryString);
// 3. TokenSequence instance is given to Parser which produces an instance of SyntaxTree
$parser = new Parser();
$syntaxTree = $parser->parse($tokenSequence);
// If needed, here you can access corrections
//foreach ($syntaxTree->corrections as $correction) {
// echo $correction->type;
//}
// 4. Now we can build a generator, in this example an ExtendedDisMax generator to target
// Solr's Extended DisMax Query Parser
// This part is a little bit more involving since we need to build all visitors for different
// Nodes in the syntax tree
$visitors = [];
$visitors[] = new Generators\Lucene\Common\Prohibited();
$visitors[] = new Generators\Lucene\Common\Group(
$fieldMap,
self::defaultFieldName
);
$visitors[] = new Generators\Lucene\Common\Mandatory();
$visitors[] = new Generators\Lucene\Common\LogicalAnd();
$visitors[] = new Generators\Lucene\Common\LogicalNot();
$visitors[] = new Generators\Lucene\Common\LogicalOr();
$visitors[] = new Generators\Lucene\Common\Phrase(
$fieldMap,
self::defaultFieldName
);
$visitors[] = new Generators\Lucene\Common\Query();
$visitors[] = new Generators\Lucene\Common\Tag('tag_ms');
$visitors[] = new Generators\Lucene\Common\User('user_s');
$visitors[] = new Generators\Lucene\QueryString\Word(
$fieldMap,
self::defaultFieldName
);
$aggregate = new Generators\Common\Aggregate($visitors);
$generator = new Generators\QueryString($aggregate);
// Now we can use the generator to generate the target output
$targetString = $generator->generate($syntaxTree);
return $targetString;
}
}app/Service/Query/SyntaxTreeRenderer.php
<?php
/**
* Created by PhpStorm.
* User: admin
* Date: 2020/1/13
* Time: 15:17
*/
namespace App\Service\Query;
use QueryTranslator\Languages\Galach\Values\Node\Group;
use QueryTranslator\Languages\Galach\Values\Node\Query;
use QueryTranslator\Languages\Galach\Values\Node\LogicalAnd;
use QueryTranslator\Languages\Galach\Values\Node\LogicalOr;
use QueryTranslator\Languages\Galach\Values\Node\LogicalNot;
use QueryTranslator\Languages\Galach\Values\Node\Mandatory;
use QueryTranslator\Languages\Galach\Values\Node\Prohibited;
use QueryTranslator\Languages\Galach\Values\Node\Term;
use QueryTranslator\Languages\Galach\Values\Token\Phrase;
use QueryTranslator\Languages\Galach\Values\Token\Tag;
use QueryTranslator\Languages\Galach\Values\Token\User;
use QueryTranslator\Languages\Galach\Values\Token\Word;
use QueryTranslator\Values\Node;
use QueryTranslator\Values\SyntaxTree;
class SyntaxTreeRenderer
{
/**
* Renders structured tree representation of the given $syntaxTree.
*
* @param \QueryTranslator\Values\SyntaxTree $syntaxTree
*
* @return array
*/
public static function render(SyntaxTree $syntaxTree)
{
return self::convert($syntaxTree->rootNode);
}
private static function getNodeName(Node $node)
{
switch (true) {
case $node instanceof Term && $node->token instanceof Phrase:
return 'PHRASE';
case $node instanceof Term && $node->token instanceof Tag:
return 'TAG';
case $node instanceof Term && $node->token instanceof User:
return 'USER';
case $node instanceof Term && $node->token instanceof Word:
return 'WORD';
case $node instanceof LogicalAnd:
return 'AND';
case $node instanceof LogicalOr:
return 'OR';
case $node instanceof LogicalNot:
return 'NOT';
case $node instanceof Mandatory:
return 'MANDATORY';
case $node instanceof Prohibited:
return 'PROHIBITED';
case $node instanceof Group:
return 'GROUP';
case $node instanceof Query:
return 'QUERY';
}
throw new \RuntimeException('Did not recognize given node');
}
private static function convert(Node $node)
{
$subObjects = [];
if ($node instanceof Term) {
$subObjects = self::getTermSubObjects($node);
} else {
foreach ($node->getNodes() as $subNode) {
$subObjects[] = self::convert($subNode);
}
}
if ($node instanceof Group) {
$subObjects = [
'field' => $node->tokenLeft->domain ?: '~',
'value' => $subObjects,
];
}
return [self::getNodeName($node) => $subObjects];
}
private static function getTermSubObjects(Term $term)
{
$token = $term->token;
switch (true) {
case $term->token instanceof Phrase:
/** @var \QueryTranslator\Languages\Galach\Values\Token\Phrase $token */
return [
'field' => $token->domain ?: '~',
'value' => $token->phrase,
];
case $term->token instanceof Tag:
/** @var \QueryTranslator\Languages\Galach\Values\Token\Tag $token */
return [
'field' => $token->marker ?: '~',
'value' => $token->tag,
];
case $term->token instanceof User:
/** @var \QueryTranslator\Languages\Galach\Values\Token\User $token */
return [
'field' => $token->marker ?: '~',
'value' => $token->user,
];
case $term->token instanceof Word:
/** @var \QueryTranslator\Languages\Galach\Values\Token\Word $token */
return [
'field' => $token->domain ?: '~',
'value' => $token->word,
];
}
throw new \RuntimeException('Did not recognize given node');
}
}