Solr search

solr Is based on Lucene Enterprise search server , Put it war Package is deployed in web You can run directly in the container

java operation solr Use solrJ(solr Of java client ),spring data solr frame ( Encapsulates the solrJ) Provides solrTemplate Simplify the development of search services

The content we can search through the search engine is stored in the index library , So you can see the stored content after searching , So these must create domains (field), The stored content does not have to be indexed , Maybe just showing , Such as images , What is stored is a picture url, No one will search this url Also not to url Perform sorting and other operations , So the image field only stores images without indexing

The search process can be simplified , A primitive understanding , Is to query what I want from the database sku data , This process is not restricted by database search , We can search wherever we want , And it has multiple conditions at the same time . What comes out is certainly not a specific , But a relatively specific , Including the brand price range, specifications and models , We can choose more specifically by clicking in .

So I understand , Which domains to set
among id This domain solr It has been defined , It's the only one , We don't define it , But in JavaBean You still need to make notes in id Add up @Field annotation , If the incoming bean There is no id It's a mistake
title ( Search and display )
brand ( Search and display )
Price ( Search for ( For example, search according to the price range ) And display )
picture ( Show )
Brand classification ( Search and display )
The seller ( Search for ( The brand and the seller are not necessarily the same , For example, Huawei is both a seller and a brand , But glory is the seller's brand , If you search Huawei, you will find many Huawei brands ) And display )
spuid( Although not shown to the user , But got , And enter the product details page according to it )

Copying a domain is to copy the values of other domains to this domain , The replication domain is no longer stored , Because other domains already store , Replication domains are multivalued , Because it doesn't make sense to replicate a domain .
For example, I created a project called keywords Replication domain for , Put the title , brand , The seller , The contents of the category field are copied , Then I enter a keyword in the search box , Then it can search results from multiple domains

Dynamic domains are constantly changing domains , Such as specifications , There's a lot of , Such as The Internet :5G Color : Red
In this way, a domain represents all kinds of possibilities , Dynamic adjustment

Use spring-data-solr First, import. spring-data-solr Of jar package Then create solr Of spring The configuration file ( Define two bean) It's equipped with solrTemplate And what it depends on solrServer

In the revision of schema After the configuration file ,solr Server restart will take effect . In addition, the addition, deletion and modification operations must be submitted to take effect

solrTemplate.saveBean(object) Add or modify data
solrTemplate.getById(1, TbItem.class) according to id Inquire about
solrTemplate.deleteById(“1”); Delete... According to the primary key
solrTemplate.saveBeans(list) Save collection , Set to store objects
solrTemplate.queryForPage(query, TbItem.class); Paging query
query Query criteria object Query query=new SimpleQuery(":"); Query all query.setOffset(20);// Start index （ Default 0） query.setRows(20);// Records per page ( Default 10)
Criteria criteria =new Criteria (“item_title”)contains(“2”); Create a query condition object query.addCriteria(criteria); Add the query criteria object to the query criteria
Query query=new SimpleQuery(":"); solrTemplate.delete(query); Delete according to the query object

When we highlight some fields in the search , The front page will be displayed as it is html Content , This is because Angularjs To prevent html Security measures taken for attacks , We can use $sceclothingserviceinOftrustAShtml(data)FangLawCome\; onYesthishtmlInto\; theThat\text{'}s\; okdischargeThat\text{'}s\; okbutyesExaminationConsiderationTothisworkcanyesOneindividualthroughuseOfworkcan,theWithstraightPick\; uptakeitsetThe\; righteousToOneindividualtoofilterdeviceinapp.filte{(}^{\prime }toofilterdeviceOfname{word}^{\prime },{[}^{\prime }$sce’,function($sce){
return function(data){
return $sce.trustAsHtml(data);
}
}])
Use... On the page ng-bind-html=“item.title | The name of the filter ” Call the filter

service The layer can also call service layer , But what is called must be started first , Judge redis Is there any one key It can be used hasKey To judge

Search out the content to be displayed according to the keyword , And highlight , This is a basic search . We want to show the classification of the search content , The searched content has the field of classification , Let's query the category field separately , Then get the value of the group and put it into a classification set , Classification set and highlight set are our resultMap A value of , Show the brand later , The specifications are all added to resultMap Medium will do .
After obtaining the commodity classification, you can find the template according to the classification id, According to the template id You can also find information about brands and specifications , After processing, we get the data we want , Put it in resultMap in .
Because the template id Query for , Too many queries about brand specifications , We store it in redis in , And then from redis Get data in , To reduce the pressure of database access ( Be careful redis Storage problems , Guarantee redis The data in are all up-to-date , Synchronize with database , Therefore, when adding, deleting and modifying, it is necessary to store them in the third category , We can go back to redis The stored code is written in the query all , Query once after adding, deleting and modifying , Every update can be guaranteed redis The data in are all up-to-date )

redis-server.exe redis.windows.conf windows in redis Start command for loading configuration file , Be sure to log in with administrator privileges cmd In the implementation of

It's going on solr When querying , Each query criteria object query All need to have a criteria Conditions of the object , But there can only be one . Each query has only one master query object , All of them are filter query objects ( It also has its own criteria object ), Must be added to the main query object

When searching the page for queries , The display of commodity classification is only based on keyword query , Do not consider the subsequent conditions , By default, I display the specifications and brands of the first category , If the classification is passed , According to the classification of biography , The code in the highlighted part is our theme display code , You need to add filter conditions here

solr Remember to submit after the call solrtemplate.commit(); The calling of services should have a clear division of labor , Unified by controller Call the service layer , Each service layer is responsible for its own business , Don't mix with each other